Home

ARTS User Guide

1. 1 2 5 Source code documentation 2 40604 440824 04004 aswaa 1 260 Build instructions s e s esra Sle eke ee Sed eee a a wa 1 2 7 Command line parameters oo oo oc ce ecra esa ARTS as a scripting language o o 1 3 1 Workspace variables lt os core mo tooien y a 1 3 2 Workspace methods ss ss seo ea oe bee eee h ae LES AZGNdaS e i ac ae le a e ds ld Include controlfles o o asac a e a rs a a ee ad Testiconttollles o srs aa ee hw ee ee we ee oe we ok OE a Verbosity levels s v se m aya eS Se ee Rae ee ge A ee Importing and exporting data 21 Data fOrmatss 2 0 6 dle ad A Ae te Oe ee A EA A S DA IMIG AMINES aa ewe ape wt Go Gel ZA NACDE NIES sose aoc ws wh ee e e ee A we a 21 3 Gridded fields ica toi hee ae ee Hed ae a Roh a A Naming convention for grids o Description of the atmosphere Sl 32 3 3 3 4 3 5 3 6 au 3 8 Altitude coordinates a eR EE eR a Atmospheric dimensionality o 0 005 Atmospheric grids and fields o o Geo location of 1D and 2D suo css nos sas sra Hydrostatic equilibrium 2 cuna rada eee eed The reference ellipsoid and the surface 4 Whe cloud DORs tac le amp tk Boe Mh we ea a Wind and magnetic vector fields o oo YA ADO Ur Ur ur Un Rh ooo wo 13 13 13 13 13 13 II CONTENTS 4 Radiative transfer basics 23 4 1 Stokes
2. b2 x3 2 b3 xa 2 b4 16 3 where xo 1 are the coefficients to be retrieved elements of x The interpretation of a retrieval is simplified if the average of f equals xpo and the scalars b1 bo are selected schematically as 1 fe bp dz n gt o0 16 4 According to this expression b is zero for odd n However z is in practise a discrete variable z not necessarily symmetric around 0 and b is taken as the average of z all bn can be non zero The normalisation of z is not only made for interpretation reasons it can be required for pure numerical reasons such as when z represent frequency in Hz In practise the basis functions are vectors denoted below as z Element 7 of z is zi j 2 5 bi 16 5 16 3 Atmospheric variables common expressions The analytically oriented calculation procedure to obtain the Jacobian for atmospheric quantities is here outlined The expressions are based on the chain rule and can be applied for absorption constituents atmospheric temperatures and winds It is important to notice that only local effects are considered and the expressions have limitations as discussed below in Sec 16 3 6 16 3 1 Matrix derivatives Some of the expressions of this section treat derivatives involving vectors and matrices Such derivatives must be handled with some care the standard calculation rules do not 16 3 ATMOSPHERIC VARIABLES COMMON EXPRESSIONS 111
3. 03 668 03 686 03 667 03 676 Options a and b are equivalent you will have one ozone species that represents all iso topologues and that will be associated with a single VMR field in vmr_field With option c you have five different ozone species so you have to supply five different VMR fields If those five fields are identical exactly same numerical values you will get the same total absorption as with options a and b Overall the tag mechanism allows quite complex absorption setups The built in docu mentation for abs_speciesSet gives a detailed explanation of the tag syntax and some exam ples Particularly note that order of the species list matters as absorption line data is assigned to species in their order within the abs_species list and no line record is assigned to more than one species It is furthermore important to note that there is no intelligence in ARTS that checks that the chosen tag combinations make sense so the user should know what s he is doing or follow one of the many examples in the ARTS controlfiles directory 6 5 2 Explicit line by line calculations For absorption species with explicit line by line calculation the calculation involves the steps summarized in Table 6 2 which contains the steps that are common to all the three contexts in which explicit line by line calculations can occur as well as the steps that are specific to each of those cases The list of variables an
4. Example of correct input data for using the second order approach L2 lt xml version 1 0 gt lt arts format ascii version 1 gt lt Array type LineMixingRecord nelem 38 gt lt LineMixingRecord gt lt SpeciesTag gt 02 66 x x lt SpeciesTag gt lt QuantumNumberRecord gt lt Upper gt lt QuantumNumbers nelem 3 gt J 1 1 N 1 1 v1 0 1 lt QuantumNumbers gt lt Upper gt lt Lower gt lt QuantumNumbers nelem 3 gt J 2 1 N 1 1 vl 0 1 lt QuantumNumbers gt lt Lower gt lt QuantumNumberRecord gt lt Vector nelem 10 gt 2 65e 06 1 32e 07 8 35e 12 50 GAS ABSORPTION 8 43e 13 0 000545 0 00017 300 0 8 1 6 1 6 lt Vector gt lt LineMixingRecord gt lt Array gt lt arts gt The format of the above data is thusly SpeciesTag QuantumNumberRecord Vector The above input will cause the 02 66 x x line with upper quantum numbers J 1 N 1 and n 0 and with lower quantum numbers J 2 N 1 and n 0 to match If this is the only input file the described line is the only line experiencing line mixing in your calculations If there were more lines in the file these would also experi ence line mixing In order of appearance in the vector its format is first order zeroth phase correction Pa first order first phase correction Pa7 second order zeroth absorption correction Pa second order first absorption correction Pa7 second order zeroth line
5. Unless you work on continuum model development or verification you should never have to modify these default settings The core method to calculate continua and complete absorption models is abs_xsec_per_speciesAddConts Users normally do not have to call this method explicitly since it is used implicitly by higher level methods such as propmat_clearsky AddOnTheFly and propmat_clearsky_fieldCalc 6 5 4 Collision induced absorption Collisions of molecules centro symmetric molecules e g O2 Na H2 CO2 and CHa possessing no permanent electric dipole create a transient dipole which causes so called collision induced absorption CIA Absorption strength of CIA is characterized by its de pendency on the molecular density of both molecular species involved in the collision Recently the well known HITRAN spectral line catalogue has started to offer also tab ulated binary absorption cross sections for CIA This is described in detail in Richard et al 2012 and also in the documentation that comes with the data themselves Binary absorption cross sections have to be multiplied with the number densities of both involved molecular species to yield absorption coefficients Qi j Kij Ni Nj 6 3 where 7 and j denote the two different absorbing species As a consequence K j has units of m molec in ARTS the original HITRAN units are different Using CIA in ARTS is easy First of all include one or more CIA tags in your abs
6. lg dn dg 15 12 where d is the along path delay due to decreased propagation velocity and dy is delay due to deviation from geometric propagation refraction The total extra path delay in units of seconds is At 15 13 C and is provided by iyRadioLink as an auxiliary variable 15 2 6 Bending angle The bending angle is a measure on the deviation from geometric propagation Using the nomenclature defined in Figure 15 2 6 the bending angle a can be calculated as e g Schreiner et al 1999 Eq 6 a paps LEO 8 T This equation is derived assuming what in ARTS is denoted as a 1D atmosphere ARTS operates with zenith angles and the equation above becomes a Y Ur 0 15 14 where 4 and 7 is the zenith angle of propagation path at the transmitter and the receiver respectively This equation is applied for all atmospheric dimensionalities hence any bend ing in the horizontal plane that can occur for 3D is neglected A remark here is that the ARTS zenith angles for the path are the ones to observe the propagating radiation and e g Y T PLEO Equation 7 8 in ARTS Theoryshows that that without refraction 0 Y Y note the difference in notation and Equation 15 14 is con sistent with a zero bending angle for propagation in vacuum Consequently ARTS should theoretically return a bending angle of zero if geometrical path calculations are selected but due to limited calculation
7. o 99 15 2 Radio link calculations 2 ocorra a a oro s s 100 15 21 Practical usage ic ia barda RE ee he YP oi ad 100 15 2 2 Determination of the radio link path 100 15 2 3 Definition of attenuation and loss terms 101 15 2 4 Free space loss and sensor characteristics 102 15 23 Extra patiidelay eiii a a he ome ware 103 152 6 Bending angle wiss tage ad a o ek we tw 104 152270 DetOCusin ge osc oe a ko we a a ok ed ee T 105 Detocusing method ll vicios ee ee 105 Defocusing method 2 xs ms aaa 105 15 2 8 Atmospheric xtinctiom 4 soe e e i a a ee we ew a aa 106 16 Clear sky Jacobians 107 LGI Intro GU COM s e a s aes a ES ee a ee we A R 108 16 1 1 Perturbations o sota a A a a et a R T A 108 16 1 2 Analytical expressions lt a 108 16 1 3 Workspace variables and methods 108 16 2 Basis functions sss eges 4 a aa da Dee a Ba a a 109 16 2 1 Basis functions for piece wise linear quantities 109 16 2 2 Polynomial basis functions o a aaae 110 16 3 Atmospheric variables common expressions ooo aa 110 16 3 1 Matrix derivatiyes soor ens ee Me ep g a i a e ANT A a 110 CONTENTS y 16 3 2 Separation of terms 2b soco ssa cocia aa de ee 111 16 3 3 OS Cts eneral CASE aio a y a SBS 112 16 3 4 Including the surface oaoa eee 113 16 3 5 Os Ozx locally unpolarised absorption 113
8. sorption vector is given by Ag as in this case the emission vector can be calculated as je Aab 4 10 26 RADIATIVE TRANSFER BASICS where b can be seen as the emission source vector defined as b B 0 0 0 4 11 Hence as only the first element of b is non zero the absorption vector is in this case equal to the first column of Aa The absorption vector can not be extracted from K as this propagation matrix covers also scattering and scat_data must also contain such data In summary the total absorption vector in ARTS is obtained as Bia l 4 12 where a is the first column of A and a is the absorption vector due to particles 4 2 4 Main cases The two equations below are discussed thoroughly in 6 of ARTS Theory This including that for some conditions also the n2 law of radiance must be considered to obtain completely exact results see also Section 9 5 The equations below treat a single frequency and a single direction at a time and can be said to describe monochromatic pencil beam radiative transfer For simplicity the frequency and direction are left out from many of the equations in this user guide Clear sky radiative transfer If we for the moment assume that scattering can be totally neglected then Equation 4 6 can be simplified to ds v r fi dl Cases where Equation 4 13 is valid are in ARTS denoted as clear sky radiative transfer im plying LTE if nothing else is stated The discussio
9. 1 the stored position is identical to the sensor position and that position can be used to determinate the radiative background see below constant Numeric The propagation path constant It is defined as the product rn sin q see Chapter 7 4 1 of ARTS Theory at the position of the sensor This is a true con stant for the path just for 1D atmospheres but can be useful also in other cases For example it equals the impact parameter normally used to define limb radio occulta tions This field is initiated 1 to indicate that the constant is not yet set background String The radiative background for the propagation path The possible options for this field are space surface cloud box interior and cloud box level where the source of radiation should be clear the content of the strings start_pos Vector The practical start position of the propagation path This vector equals in general the last row of pos The exception is radio link calculations where the transmitter is placed above the model atmosphere where this field gives the position of the transmitter start_los Vector Line of sight at start point of propagation path Set and used in the same way as start pos start_Istep Numeric The distance between start_pos and the last position in pos This value is zero except for a transmitter placed above the top of the atmosphere Hence this length corresponds to propgation if free space n 1 end_p
10. E OT 27 i a relationship that follows from Eqs 16 30 and 16 31 Hydrostatic equilibrium and limitations A changed temperature has non local effects originating from refraction and hydrostatic equilibrium The expressions above ignore totally refraction effects As mentioned if hse is set to off the term 0AI Ox is set to zero That is the path length through the layer is not affected by a temperature change With hse set to on the complete expressions above are used If this treatment of hydrostatic equilibrium is sufficient or not depends on the observa tion geometry It should be insufficient for limb sounding where changes even at altitudes below the tangent point can have an influence as the geometrical altitudes of all higher layers is changed through hydrostatic equilibrium However this effect vanishes for ground based observations at zenith and satellite measurements at nadir giving a full account of hydro static equilibrium even with the analytical expressions In practise it should be possible to apply the expressions outside zenith and nadir as long as the observations are of up or down ward type The same applies to measurements from inside the atmosphere e g aircraft ones if the reference pressure for hydrostatic equilibrium p hse is matched to the pressure of the observation point 118 CLEAR SKY JACOBIANS 16 7 Sensor pointing The term sensor pointing refers to deviations betw
11. In standard usage of the method there is no signal reflected by the surface and the radiative transfer calculations are started at the surface See the built in documentation of iyTransmissionStandard for a full list of possible aux iliary quantities These data include quantities that make it possible to determine the trans mission for different parts of the propagation path For example the state of iy at each point of the propagation path can be extracted and also the transmission matrix Eq 9 8 for each path step is accessible 15 2 Radio link calculations 15 2 1 Practical usage This type of calculations is handled by iyRadioLink A sensor position must be specified as usual The sensor takes the place as the receiver of the radio link The observation angles of the receiver are not user input if any given they are just ignored they are de termined by the position of the transmitter This position is given as transmitter_pos or rte_pos2 dependent on if yCalc or iyCalc is used Specification of receiver and transmitter properties beside their positions is discussed in Section 15 2 3 The quantities returned by iyRadioLink are also described in Section 15 2 3 It is allowed to use iyRadioLink together with pure geometrical propagation paths but this option does not give realistic results as the this set up misses the impact of defocus ing Sec 15 2 7 With refraction the propagation path can not be determined in
12. The 2D latitude can then 3 3 ATMOSPHERIC GRIDS AND FIELDS 17 Atmospheric grids Ground ee Geoid DA AF Atmospheric field ate ar Figure 3 2 Schematic of a 2D atmosphere The radii for the surface and the pressure levels vary here linear between the latitude grid points The atmospheric fields vary linearly along the pressure levels and the latitude grid points that is along the dotted lines Inside the grid cells the fields have a bi linear variation No cloud box is included in this figure Geoid in the legend should be Ellipsoid be taken as the angular distance along the satellite track A 2D latitude of e g 100 will then correspond to a 3D latitude of 80 The atmosphere is normally treated to be undefined outside the considered plane but some scattering calculations may treat the surrounding atmosphere in an simplified manner A 2D atmosphere is shown in Figure 3 2 3D In this the most general case the atmospheric fields vary in all three spatial co ordinates as in a true atmosphere Figures 3 3 A spherical coordinate system is used where the dimensions are radius r latitude a and longitude 8 and a po sition is given as r a 3 With other words the standard way to specify a geo graphical position is followed The valid range for latitudes is 90 90 where 90 corresponds to the North pole etc Longitudes are counted from the Greenwich meridian with p
13. This chapter discusses different practical aspects of absorption in ARTS Sections 6 2 and 6 3 introduce the main physical quantities and the main agendas respectively Section 6 4 explains how absorption is handled inside radiative transfer calculations Section 6 5 discusses how absorption is actually calculated and how the calculation is set up Finally Section 6 6 describes how absorption is stored in a lookup table and how it is extracted again Here in the User Guide we focus on practical aspects of absorption in ARTS But ab sorption calculations also have a deep theoretical background particularly the line by line calculations and the continuum models Some of this background is discussed in ARTS Theory Chapter 2 6 2 Key physical quantities The scalar gas absorption coefficient a has units of 1 m You can think of it as being defined by the Lambert Beer law h Ibe 6 1 History 2013 06 21 New intro and general revision Added CIA part Stefan Buehler 2012 08 28 Updated to propmat_clearsky by Richard Larsson 2011 07 05 Added intro and sections on abs in RT and abs calculation Also revised lookup table section First attempt of a complete absorption chapter for ARTS2 Stefan Buehler 2003 03 28 Documentation for WSM abs_fieldCalc extended by Stefan Buehler af ter comment from Sreerekha T R 2003 03 10 Lookup tables added by Stefan Buehler 2002 06 04 Restarted for ARTS 1 1 by Stefan Buehler 36 GAS
14. always apply The basic relationships used here are da SR 16 6 O Ab aoe when A independent of a 16 7 da a 0b c b c o n where a b and c are vectors and A is a matrix Matrix exponential derivatives can not be expressed analytically in a general manner but some cases can be handled If two matrices A and B are commutative i e AB BA then the following is valid Dattorro 201 1 A B Z Agere 16 9 Ot where t is a scalar variable The cases where A and B are commutative includes the sit uation where one of them is a scalar matrix can be written as al i e diagonal with all non zero elements equal 16 3 2 Separation of terms The overall task is to calculate Eq 16 1 with subscript p left out oy Ox where x is the element of x for which we want to obtain the weighting function This is a column of the complete weighting function matrix A first step is to identify how sensor characteristics can be incorporated To make the nomenclature simpler we assume here that the simulations cover only a single measurement block and we have cf Eq 5 2 y Hip 16 10 We can then apply the chain rule for a first time to obtain Oy Oy Oin We Ox Oy Ox Ox Hence sensor characteristics can be handled by calculating the weighting function column matching all monochromatic pencil beam calculation of the measurement block k and perform a multiplication with H a parallel procedure compared
15. and water vapor respectively Such high orders are only appropriate because the function to be interpolated the is very smooth 6 6 3 Workspace variables and methods The gas absorption lookup table is implemented by the class GasAbsLookup which re sides in the files gas_abs_lookup cc and gas_abs_lookup h The lookup table itself is stored in the workspace variable abs_lookup It can be gen erated with the method abs lookupCalc ARTS also includes some methods that automati cally set input parameters for abs_lookupCalc such as grid ranges and reference profiles of pressure temperature and trace gas concentrations These methods are abs_lookupSetup abs_lookupSetupBatch and abs_lookupSetup Wide The first two will take into account the actual atmospheric state or set of atmospheric states for the calculation The third alter native simply sets up a table that should cover most reasonable atmospheric conditions Buehler et al 2011 as well as the built in documentation contains more information on these setup methods Alternatively the table can be loaded from a file with ReadXML After loading the method abs_lookupAdapt has to be called It will make sure that the lookup table agrees exactly with your calculation For example it has to check that the frequencies that you want to use are included in the set of frequencies for which the table has been calculated There is no interpolation in frequency This is on purpose because t
16. for all pencil beam directions of the block do Section 5 3 call iy_main_agenda giving i Algorithm 2 copy i to correct part of i end for put the product Hpi in correct part of y end for Algorithm 2 The main operations for methods to be part of iy_main_agenda determine the propagation path by ppath_agenda Section 9 2 determine the radiation at the start of the propagation path Section 9 3 perform radiative transfer along the propagation path Section 9 4 unit conversion of i following iy_unit Section 9 5 a call of yCalc and this section outlines this method and the associated main variables The calculation approach fits with the formalism presented in Sections 1 1 1 2 of ARTS Theory where the separation between atmospheric radiative transfer and inclusion of sensor effects shall be noted especially and a similar nomenclature is used here y Complete measurement vector In addition to atmospheric radiative transfer the vec tor can include effects by sensor characteristics and data reduction operations The corresponding workspace variable is y i Monochromatic pencil beam data for a measurement block The definition of a mea surement block is found in Section 5 3 This vector is only affected by atmospheric radiative transfer As workspace variable denoted as iyb but can be considered as a pure internal variable and should not be of concern for the user i Monochromatic data for one line of sight i
17. see e g Eriksson 2000 A retrieval characterisation following Rodgers 1990 2000 raises the same demand A column of the Jacobian Kx is defined as oy 16 1 On where y is the vector of measurement data and x is one forward model scalar variable See further Section 1 3 of ARTS Theory The nomenclature of that section is also used here The quantity in Eq 16 1 is in the atmospheric sounding community frequently denoted as a weighting function and accordingly Kx is called the weighting function matrix In the documentation of ARTS both terms Jacobian and weighting functions are used These names refer normally to Kx the partial derivatives with respect to the variables to be retrieved forming the state vector x However in the context of retrieval characterisation the same matrix for the remaining model parameters is of equally high interest denoted as Kp In the same manner the terms inversion and retrieval are used interchangeably The main task of the user is to select which quantities that shall be retrieved and to define the associated retrieval grids These aspects must be considered for successful in versions but are out of scope for this document Beside for the most simple retrievals it is further important to understand how the Jacobian is calculated A practical point is the calculation speed primarily determined if perturbations or analytical expressions are used Sec 16 1 The derivation of the different Ja
18. the sensor for a specified line of sight Propagation paths are introduced in Section 9 2 and this section provides further details For a general usage of ARTS 1t should suffice to read Section 10 1 The remaining sub sections deal with more low level aspects of the calculations and are of interest only 1f you want to understand the finer details of ARTS The actual equations applied are found in Chapter 7 of ARTS Theory 10 1 Practical usage The overall calculation approach for finding the propagation path is specified by ppath_agenda The standard choice for this agenda is ppathStepByStep applying ppath_step_agenda repeatedly in order to trace the path backwards starting at the sensor This set up is assumed throughout this chapter A slighltly different selection of workspace methods is required for radio link calculations see further Section 15 2 The exact ray tracing algorithm to be applied for the calculation of propagation path is selected through ppath step agenda see further Section 9 2 The fastest calculations are obtained if refraction is neglected denoted as geometrical calcutions The workspace method to apply if this assumption can be made is ppath_stepGeometric The main consideration for using ppath_stepGeometric is to select a value for ppath Imax This variable controls to some extent the calculation accuarcy as described in Section 9 9 This variable sets the maximum distance between points of the propagation path Set
19. treated to act as blackbody the surface is then not reflecting any radiation Reflections from the surface can also be neglected if the zenith optical thickness of the atmosphere between the surface and cloud box is sufficiently high 3 8 WIND AND MAGNETIC VECTOR FIELDS 21 3 8 Wind and magnetic vector fields The atmospheric fields discussed above are scalar quantities while some atmospheric vari ables can be seen as vector fields However in ARTS input vector fields are broken down into the zonal meridional and vertical components and are given as three scalar fields This division into scalar values is used to allow that one or several of the components easily can be set to zero which is done be setting the corresponding workspace variable to be empty Following the standard naming scheme for winds the components are denoted as u The zonal component A positive value signifies an Eastward direction v The meridional component A positive value signifies a Northward direction w The vertical component A positive value signifies an upward direction The workspace variables to describe the wind vector field are wind_u_field wind_v_field and wind_w field To clarify the definition of the vector components above the winds components are defined as follows Uy A positive wind is defined as air moving from west to east i e towards higher longi tudes Uy A positive wind is defined as air moving from south to north i e towards high
20. workspace method doit_za_grid_optCalc Please refer to the online documentation of this method The filename of the optimized zenith angle grid can be given as a generic input If a filename is given the equidistant grid is taken for the calculation of the scattering integrals and the optimized grid is taken for the radiative transfer part Otherwise if no filename is specified za_grid_opt_file the equidistant grid is taken for the calculation of the scattering integrals and for the radiative transfer calculations This option makes sense for down looking cases to speed up the calculation The main agenda for a DOIT calculation is doit mono_agenda The agenda is executed by the workspace method DoitCalc DoitCalc 18 2 1 The DOIT main agenda Although there are alternatives the most elegant usage of DOIT involves specifying it within the agenda doit mono agenda which calculates the radiation field inside the cloud box The agenda requires the incoming clearsky field on the cloudbox boundary as an input and gives as output the scattered field on the cloudbox boundary if the sensor is placed out side the cloudbox or the full scattered field in the cloudbox if the sensor is placed inside the cloudbox 18 2 DOIT FRAME 127 AgendaSet doit_mono_agenda Prepare scattering data for DOIT calculation Optimized method DoitScatteringDataPrepare Alternative method needs less memory scat_data_monoCalc Perform iterations 1 sc
21. 16 36 Limitations ss faa aaa ee eae d Hee eee ee oe 113 16 4 Absorption SPECIES o s E a aE ead ed a Re a eee 114 16 4 1 Common practicalities o 0 002 202 0054 114 16 4 2 Perturbation calculations ooo o 114 16 4 3 Analytical expressions o onerosa ra da ad 114 Unpolarised absorption 2 ee cu ee eee 114 General case o o esie s ws Bk OR Ee Eee eee ee eee 115 16 3 We ses os EA PREPS eS EE PS ee SS es 116 16 6 Atmospheric temperatures gt o oac eee BA eae Be Oya de we ha 116 16 6 1 Common practicalities ss s ee eee pie ee Eee ok ae 116 16 6 2 Perturbation calculations o o 116 16 6 3 Analytical expressions o e e 116 Unpolarised absorption o o e e 116 General Case oca ea a e he ee ed a a a ha ok a 117 Hydrostatic equilibrium and limitations 117 16 7 Sensor pontig 2 21 424 6b et ede dba edhe bee eh eee 118 16 8 Sensorirequencies cu a sae a eae a BAe Bae we a we we 118 16 9 Polynomial baseline fit o o 119 16 10Simusoidal baseline fit mo rd a a Re 119 17 Batch calculations 121 17 1 Batch calculations of measurement vector Y o o 121 17 2 Control file examples oia a dd RS 122 IV Radiative transfer dedicated scattering methods 123 18 Scattering calculations The DOIT module 125 18 1 The 1D control Hleexample sse seana bs a iaie ee ee a i oa 125 18 2 D
22. 2 ooo o ee 62 horizontally alisned 2 4 0 04 cda 0442484 608 63 GENET 2 saca ae a ack a ba eo Eee ee eS 63 8 2 4 Generating single scattering properties 64 8 3 Particle SIZS dISTEIDULIONS suis a Be ee hae aa ew a 64 8 3 1 Mono disperse particle distribution 64 8 3 2 Gamma size distribution 65 8 3 3 McFarquhar and Heymsfield parametrization 66 8 4 dmplementation sas sas 04 daw a ae aw gd a 66 8 4 1 Work space methods and variables 67 HI Radiative transfer clear sky general functionality 69 9 Clear sky radiative transfer 71 9 1 Overall calculation procedure o o 71 9 2 Propasation paths sa iba da a a Dee Ba aR 73 9 3 The radiative background o ca oo spcco eo towo e e o 73 9 4 Basic radiative transfer variables and expressions 75 9 4 1 Unpolarised absorption o o 75 9 4 2 Polarised absorption coa a da ate 76 9 4 3 Blackbody and cosmic background radiation 77 9 5 Output unit and the n law 2 444465 46044 b4 0s Haw ea Oe we 77 9 6 Single pencil beam calculations o o 78 9 7 DISPRrSION ce ani a a dd Be ae a 78 OS Auxiliary datas ea ds a ac ia ra a Se do o 79 9 9 Caleulation ACCUTACY io 206 4 sas sana a a 80 10 Propagation paths 81 10 1 PracticalUsale u cocira sosca aoe car d a ee ee 81 10 2 Caleulati
23. 8 9 8 AUXILIARY DATA 79 9 8 Auxiliary data The core output of the radiative calculations is y iy if iyCalc is used jacobians discussed in Sec 16 but different auxiliary data can be extracted First of all yCalc outputs automat ically y_f y_pol y_pos and y_los These data give information about the frequency polar isation sensor position and sensor bore sight respectively corresponding to each value in y The content of the variables are governed by the sensor settings and the order calculated radiances are stored discussed in Sec 5 3 A more general mechanism for extracting auxiliary data is controlled by the iy_aux_vars workspace variable This mechanism is most useful together with iyCalc and for the mo ment we assume that this method is used limitations for yCalc are discussed below The quantities that can be extracted differ see the built in documentation for the options for each workspace method of concern e g arts d iyEmissionStandard The options for this particular method iyEmissionStandard can be divided into different groups more variables will can be added Atmosphere along the path The pressure temperature and volume mixing rations along the propagation path Attenuation along the path Total and species specific absorption coefficients along the propagation path Radiative properties along the path The radiance at each propagation path point Overall radiative properties The total
24. Clear sky In ARTS the term clear sky refers to the case when the influence of parti cles can be ignored and only absorbing species are of relevance Hence a clear sky calculation can involve e g cloud water droplets but on the condition that the wavelength is such that scattering can be neglected The propagation effects of absorbing species and of scattering particles are kept separated The total propagation matrix is K A K 4 8 where a and p refers to absorbing species and particles respectively and the symbol A is used for the first term to remind about that the only extinction process covered by this matrix is absorption but including magneto optical effects As workspace variable Ag is denoted as propmat_clearsky obtained by the propmat_clearsky_agenda agenda K is obtained internally from scat_data 4 2 3 Emission and absorption vectors One of the general assumptions in ARTS is that local thermodynamic equilibrium LTE can be assumed For the moment there exists no method in ARTS to handle deviations from LTE but this can be changed in the future On the condition of LTE the emission vector Je in Equation 4 6 can be written as je Ba 4 9 where B is the Planck function a scalar value describing blackbody radiation The related workspace variables are blackbody radiation and blackbody radiation agenda The quantity a is denoted as the absorption vector For clear sky calculations the ab
25. GNU General Public License along with the program if not write to the Free Software Foundation Inc 59 Temple Plac Suite 330 Boston MA 02111 1307 USA Contributing authors Author email Main contribution s Stefan Buehler Editor Sections 1 6 and 17 sbuehler at uni hamburg de Claudia Emde Sections 8 and 18 claudia emde at dlr de Patrick Eriksson Editor Sections 3 4 5 7 9 10 11 12 16 patrick eriksson at chalmers se 13 14 15 and 20 Oliver Lemke Latex fixes olemke at core dump info The present address is given for active contributors while for others the address to the institute where the work was performed is given Department of Computer Science Electrical and Space Engineering Division of Space Technology Lule University of Technology Box 812 SE 98128 Kiruna Sweden Department of Earth and Space Sciences Chalmers University of Technology SE 41296 Gothenburg Sweden Institute of Environmental Physics University of Bremen P O Box 33044 D 28334 Bremen Germany Contents I Overview 1 Introduction 2 3 1 1 1 2 113 1 4 15 1 6 Whatis ARTS 22 028 44 8648 44 a as a ew a Documentation soc sas asriarda a eR ae ae He aa 1 21 Guide documents os ace ek Sep das ats ae Ae a 22 O te ose he a eA ie eh ke Byte BE ttm eh E A Be de 1 2 3 Built indocumentation nenesie rar Es 1 2 4 Test and include controlfiles
26. PATHS Chapter 11 Reference ellipsoid and surface properties 11 1 The reference ellipsoid The ellipsoid is an imaginary surface used as a reference when specifying the surface al titude and the altitude of pressure levels As the name indicates this reference surface is defined as an ellipsoid The reference ellipsoid should normally be set to some global geode tic datum such as WGS 84 Inside ARTS the ellipsoid is represented as a vector denoted as refellipsoid This vector must have length two where the two elements are equatorial radius re and eccentricity e respectively 11 1 1 Ellipsoid models A geodetic datum is based on a reference ellipsoid The ellipsoid is rotationally symmetric around the north south axis That is the ellipsoid radius has no longitude variation it is only a function of latitude The ellipsoid is described by an equatorial radius re and a polar radius rp These radii are indicated in Figure 11 1 The definition of the ellipsoid used in ARTS is based on re and the eccentricity e e l rip 11 1 Workspace methods to set the reference ellipsoid for a particular planet include refellip soidEarth refellipsoidMars and refellipsoidMars The radius of the ellipsoid for a given geocentric latitude a is 2y2 rola ME 2 11 2 2 m2 2 0022 r2 sin a r2 cos a ysin a 1 e2 cos a The radius given by Equation 11 2 can be directly applied for 3D atmospheres For 2D ca
27. Set by the user Unit Hz Group Vector Figure 1 2 Built in documentation for variable f grid obtained by command arts d f_grid or on page http www sat ltu se arts docserver variables f_grid ARTS has built in documentation for all workspace methods and variables which can be accessed as described in Section 1 2 3 In this user guide just clicking on the name of a variable or method will take you directly to the built in documentation for that object Below we will discuss workspace variables and methods in some more detail and give more examples 1 3 1 Workspace variables Workspace variables such as the variable s in Figure 1 1 are the variables that are manip ulated by the workspace methods during the execution of an ARTS controlfile Workspace variables belong to different groups Index String Vector Matrix etc The built in docu mentation lists all groups at the time of writing there are approximately 60 of them As the example in Figure 1 1 shows workspace variables can be freely created by the user with methods like StringCreate VectorCreate and so on Each group has its own create method However in most cases it is not necessary to create new variables in this way since a lot of variables are predefined in ARTS The built in documentation describes all prede fined variables As an example Figure 1 2 shows the description for the variable f_grid which stores the frequency grid and is use
28. This is contrast to DOM methods where the whole radiation field is calculated CPU and memory cost scales more slowly than other methods with grid size so that large or detailed 3D scenarios are not a problem This stems from the suitability of Monte Carlo Integration MCI for evaluating integrals over highly dimensioned spaces As well as CPU cost increasing dramatically in 3D DOM applications with the number of grid points in each dimension the memory requirements becomes pro hibitive at moderate grid sizes due to the requirement that the radiance in every direc tion must be stored at each grid point Optically thick media are no problem A feature of reversed Monte Carlo algorithms is that only parts of the atmosphere that actually contribute to the observed radiance are considered in the computation So where the medium is optically thick due to absorption or scattering only the parts of the atmosphere closest to the sensor are visited by the algorithm This contrasts with DOM methods where as mentioned above the whole radiation field is computed Also a requirement of DOM methods is that the optical thickness between adjacent grid points must be j 1 which increases the grid size and hence cost of the method History 140924 Started by Patrick Eriksson 132 SCATTERING CALCULATIONS THE MONTE CARLO SCATTERING MODULE For the theory behind the ARTS MC module see Section 9 of ARTS Theory Chapter 20 Cloud radar simu
29. and the polarization state of the radiation However in order to describe scattering of radiation by a particle or a particle ensemble it makes sense to define another coordinate system taking into consideration the symmetries of the particle or the scattering medium as one gets much simpler expressions for the single scattering properties For macroscopically isotropic and mirror symmetric scattering media it is con venient to use the scattering frame in which the incidence direction is parallel to the z axis and the x axis coincides with the scattering plane that is the plane through the unit vectors e and A The scattering frame is illustrated in Figure 8 1 For symmetry reasons the single scattering properties defined with respect to the scattering frame can only depend on the scattering angle O inc sin sca O arccos n 28 a 8 1 between the incident and the scattering direction Z A inc At sca X Figure 8 1 Illustration of the scattering frame The z axis coincides with the incident di rection i The scattering angle O is the angle between anda 8 2 2 Scattering data structure The single scattering data is stored in a specific structure format the the SingleScatteringData class The format allows space reduction due to symme try for certain special cases e g random orientation or horizontal alignment The class consists of the following fields compare also Table 8 2 2 Definition r
30. angles and the second column is azimuth angles For 1D and 2D there is only one column in the matrix while for 3D a row i of the matrix is Yi wi The number of rows for sensor_los must be the same as for sensor_pos The correspondance to rte_pos is rte_los 5 2 3 Sensor characteristics and data reduction The term sensor characteristics is used here as a comprehensive term for the response of all sensor parts that affect how the field of monochromatic pencil beam intensities are trans lated to the recorded spectrum For example the antenna pattern the side band filtering and response of the spectrometer channels are normally the most important characteristics of a microwave heterodyne radiometer Any processing of the spectral data that takes place before the retrieval is denoted as data reduction The most common processing is to rep resent the original spectra with a smaller set of values that is a reduction of the data size The most common data reduction techniques is binning and Hotelling transformation by an eigenvector expansion In ARTS the influence of sensor characteristics and data reduction is incorporated by transfer matrices The application of these transfer matrices assumes that each step is a linear operation which should be the case for the response of the parts of a well designed instrument Non linear data reduction could be handled by special workspace methods The sensor and data reduction are described as a seri
31. aspect of this definition is that it is possible to define true rectangular responses This is achieved by setting the end points of the grid where the response drops to zero The sensor parts are normally associated with some loss of power and sensor contains also some amplification device In general it is not needed to consider these aspects as such effects are cancelled out by the calibration process The sensor should then be modelled as having no losses which is ensured by setting sensor norm to 1 The different responses are then normalised in an appropriate manner With sensor_norm set to 0 all normalisation issues must be handled when defining the response files 12 2 Some comments Some text removed from other chapters to be intgrated into this chapter For each sensor position a number of monochromatic pencil beam spectra are calcu lated The monochromatic frequencies are given by f grid The pencil beam directions are obtained by summing the sensor line of sight angles sensor_los for the position and the values of mblock_dlos_grid For example pencil beam zenith angle z is calculated as Yi Yo Ayi 12 1 where o is the sensor line of sight for the position of concern and Ay is value of the first column of mblock_dlos_grid With other words mblock_dlos_grid gives the grid relative to the sensor line of sight for the calculation of the intensity field that will be weighted with the antenna response As the s
32. be described as being spherically symmetric The term 1D is used here for simplicity and historical reasons not because it is a true 1D case a strictly 1D atmosphere would just extend along a line A spherical symmetry means that atmospheric fields and the surface extend in all three dimensions but they have no latitude and longitude variation This means that for example atmospheric fields vary only as a function of altitude and the surface constitutes the surface of a sphere The radial coordinate is accordingly sufficient when dealing with atmospheric quantities The latitude and longitude of the sensor are normally of no concern but when required the sensor is considered to be placed at latitude and longitude zero a 6 0 0 The sensor is assumed to by directed towards the North pole corre sponding to an azimuth angle of 0 A 1D atmosphere is shown in Figure 3 1 2D In contrast to the 1D and 3D cases a 2D atmosphere is only strictly defined inside a plane More in detail this case be seen as observations restricted to the plane where the longitude equals 0 or 180 A polar system consisting of a radial and an angular coordinate is applied The angular coordinate is denoted as latitude and matches the 3D latitude in the range 90 909 but for 2D there is no lower or upper limit for the latitude coordinate The 2D case is most likely used for satellite measurements where the atmosphere is observed inside the orbit plane
33. center correction Hz Pa second order first line center correction Hz Pa7 standard temperature for corrections K first order phase temperature correction exponential term second order absorption temperature correction exponential term and second order line center temperature correction exponential term The format of the vector is fixed according to necessary information found in Makarov et al 2011 the naming scheme of the variables above can also be understood from the same work Note that first order line mixing as per the MPM PWR complete oxygen models can be achieved with the correct input by letting the second order terms be nil Example of correct input data for using the LBLRTM approach LL and NR lt xml version 1 0 gt lt arts format ascii version 1 gt lt Array type LineMixingRecord nelem 38 gt lt LineMixingRecord gt lt SpeciesTag gt 02 66 x x x lt SpeciesTag gt lt QuantumNumberRecord gt lt Upper gt lt QuantumNumbers nelem 3 gt m a J 1 1 N 1 1 v1 0 1 lt QuantumNumbers gt lt Upper gt a 1 u lt Lower gt lt QuantumNumbers nelem 3 gt J 2 1 N 1 1 v1 0 1 lt QuantumNumbers gt lt Lower gt lt QuantumNumberRecord gt lt Vector nelem 12 gt 200 250 296 340 0 00443303429558 4e 4 0 003982500863558e 4 0 003628075993092e 4 0 003341074759437e 4 6 5 CALCULATING GAS ABSORPTION 51 0 0 0 lt Vector gt
34. data However some operations require that the positions is known and this is handled by lat_true and lon_true See the built in documentation for further information on how to specify these variables 3 5 Hydrostatic equilibrium There is no general demand that the model atmosphere fulfils hydrostatic equilibrium That is t_field and z_field can be specified independently of each other On the other hand ARTS provides means for ensuring that a model atmosphere matches hydrostatic equilibrium by the method z_fieldFromHSE The method considers that gravitation varies with latitude and altitude and lat_true and lon_true must be set for 1D and 2D Hydrostatic equilibrium gives only constrain for the distance between the pressure lev els not for the absolute geometrical altitudes For this reason a reference point must be introduced This is done by setting the pressure of this point by p_hse common for all lati tude and longitudes The geometrical altitudes matching p_hse are taken from the original values in z field 3 6 The reference ellipsoid and the surface Geometrical altitudes are specified as the vertical distance to the reference ellipsoid refel lipsoid discussed further in Section 11 1 The lower boundary of the atmosphere is de noted as the surface The surface is specified by its geometrical altitude on the latitude and longitude grids The workspace variable holding these data is called z_surface It is not allowed
35. density field in xml format class GF ie1d3 If the cloud is composed of several different scattering elements ParticleTypeAdd can be used repeatedly for all scattering elements for instance to the randomly oriented spherical particles from above one could add horizontally aligned cylindrical particles ParticleTypeAdd scat_data pnd_field_raw atmosphere_dim f_grid ssd_cylinder_30um_horizontally_aligned xml ond_cylinder_30um_horizontally_aligned xml Alternatively it is possible to use the method ParticleTypeAddAll which is convenient to generate a size distribution using several size bins In this case one needs to define one scat tering element for each size bin If the num ber of size bins is large the control file becomes very lengthy when using Particle TypeAdd repeatedly ParticleTypeAddAll requires as input an array of string including the filenames of the single scattering data files for all individ ual scattering elements and the variable pnd_field_raw which includes the particle number density fields for all scattering elements at once Using this function one has to make sure that the order of the filenames containing the single scattering data corresponds to the order of the particle number density fields in pnd_field_raw After reading the data the workspace variable pnd_field is calculated using pnd_fieldCalcFrompnd_field_raw 68 DESCRIPTION OF CLOUDS Calculate the particle number density field
36. different ways but it has always four elements The Stokes vector s is here written as I Q U 4 1 S V where the first component 1 is the full intensity of the radiation the second component Q is the difference between vertical and horizontal polarisation the third component U is the difference for 45 polarisation and the last component V is the difference between left and right circular polarisation That is I Iy In Iyase Laso Ihe rhc 4 2 Q l h 4 3 U T4450 l 450 4 4 V Iihe rhc 4 5 where I Ip Iya50 and I 45 are the intensity of the component linearly polarised at the vertical horizontal 45 and 45 direction respectively and Ihc and ihe are the intensity for the right and left hand circular components Further details on polarisation and the definition of the Stokes vector are found in ARTS Theory Section 5 ARTS is a fully polarised forward model but can be run with a smaller number of Stokes components The selection is made with the workspace variable stokes_dim For example gaseous absorption and emission are in general unpolarised and if not particle History 130218 First version by Patrick Eriksson 24 RADIATIVE TRANSFER BASICS and surface scattering have to be considered it is sufficient to only include the first Stokes components in the simulations ie stokes_dim set to 1 In this case to include higher order Stokes components resul
37. dimension 1D to History 110505 Complete revision by Stefan Buehler Also integrated text from ARTS2 article first submission 2002 10 Written mainly by Stefan Buehler some parts by Patrick Eriksson 4 INTRODUCTION have no longitude variation 2D or vary in all three spatial dimensions 3D The surface is by default assumed to be spherical For 2D and 3D a complete reference ellipsoid is used and the surface can have arbitrary shape Polarisation is fully described by using the Stokes formalism Scattering can be considered in several manners Extinction from scatterers can be in cluded in transmission type calculations For radiance calculations of thermal emis sion in contrast to solar radiation there are two modules at hand to take the scattering into account DOIT Chapter 18 and MC Chapter 19 The scattering particles are for efficiency reasons confined to a region of the atmosphere denoted as the cloudbox Observation geometry is free That is the forward model can be used to simulate ground based down looking limb sounding and balloon aircraft measurements Sensor characteristics can be incorporated in a flexible and efficient manner Jacobians the partial derivatives of simulated measurement with respect to forward model variables can be provided for a number of variables where analytical expressions are used as far as possible Details are found in later parts of the user guide Use the table of c
38. dimensionality i 52 4 506 540 45 gue pwede een 23 4 2 The radiative transfer equation o e e 24 4 2 1 Propagation effects es ss cs escra q dr a a 24 4 2 2 Absorbing species and scattering particles 24 4 2 3 Emission and absorption vectors o a 25 4 24 Maincases oos es saae a dE Ea ed A 26 Clear sky radiative transfer we goa catau mari a ian 26 Radiative transfer with scattering oaoa 26 5 Complete calculations 27 O sarira y es ede tee aue Seed Shee eb Popa a E 27 5 2 Compulsory sensor and data reduction variables 29 S2 Sensor POSON sos a cal a dd eee we 29 3 22 Linc orse iE iin ce aa Rw eG YS a 30 5 2 3 Sensor characteristics and data reduction 31 5 3 Measurement sequences and blocks o 31 II Atmospheric properties 33 6 Gas absorption 35 OWL IMTOdUCHON suse ad Sma ea eal ld WAG ee Gg ee wwe a 35 6 2 Key physical quantities o esasi dat ta kagai 35 Od ASENdAS as or tee ne gee ada 36 6 4 Gas absorption in radiative transfer simulations 37 6 5 Calculating gas absorpulony s os oos oe acns a a a E a 37 6 5 1 ADSOTPUON SPECIES sx s sre edd ed Pl a A 39 6 5 2 Explicit line by line calculations 41 6 5 3 Continua and complete absorption models 41 6 5 4 Collision induced absorption o o 43 655 Zeeman caleulavions 5 aims o
39. do not reach relativistic values the Doppler shift can be calculated as VVo COS Av Swa 13 6 c where vo is the rest frequency and c is the speed of light The core part of these calculations is implemented in the general internal function dotprod with_los The negative sign is caused by the fact the line of sight is the observation direction not the direction of the EM waves Chapter 14 Faraday rotation The polarisation state of an electromagnetic wave propagating through a plasma with a static magnetic field will be changed normally denoted as Faraday rotation The effect is present for both passive and active signals but Faraday rotation is proportional to v and can in general be neglected for emission measurements due to the relatively high frequencies applied For Earth the effect can in general be neglected above 5 GHz Hence Faraday rotation is a special consideration for radio microwave radiative transfer A brief theoretical description is found below in Section 14 2 14 1 Practical usage The first step is to ensure that the magnetic field and field of free electrons are non zero See Section 3 8 for how to introduce a magnetic field Free electrons are treated as an absorbing species and hence are part of vmr_field Sec 4 2 2 A further constrain is that the Stokes dimensionality stokes_dim is set to 3 or 4 This can be understood by Equation 14 4 Faraday rotation affects Stokes elements 2 and 3
40. in fact the only option implemented so far and the rest of this section outlines how this workspace method works The upwelling radiation from the surface can be written as Figure 11 4 Ti 1 Y RI 11 7 l where I indicates the Stokes vector for one frequency J is the total upward travelling in tensity from the surface along the propagation path J is the emission from the surface J 2 is the downward travelling intensity reaching the surface along direction l and R is the reflection coefficient matrix from direction to the present propagation path The emission from the surface Te is stored in surface_emission the directions l for which downward trav elling intensities are given by surface los and the reflection coefficients R are stored in surface_rmatrix Some special cases are discussed below Section 6 8 of ARTS Theory provides the theoretical background 11 3 1 Blackbody surface If the surface can be assumed to act as a blackbody the workspace method surfaceBlack body can be used This method sets surface emission to B 0 0 ol and surface_los and surface_rmatrix to be empty 11 3 2 Specular reflections Several methods to incorporate a flat surface exist including surfaceFlatRefractiveIndex and surfaceFlatScalarReflectivity The methods differ in how the dielectric properties of the surface are given and if these are constant or not with frequency In the case of specular reflections surface_los has the len
41. inside yCalc See the built in doc umentation of the workspace method you have selected for iy_main_agenda for comments on practical aspects and available output units The most extensive support for conversion to other units is provided by iyEmissionStandard while other methods have no support at all ie they ignore iy_unit It is also possible to change the unit as a post processing step by yApplyUnit or iyApplyUnit but some restrictions apply and there are no automatic checks if the input data have correct unit Further considerations and expressions for the unit conversion are discussed in the ARTS 2 journal paper Eriksson et al 2011 Sec 5 7 The n law of radiance is introduced in Section 6 5 of ARTS Theory As shown in that section the main impact of the law is handled by consistently using the vacuum speed in the definition of the Planck radiation law as done inside ARTS Eq 9 9 This suffices if the sensor is placed in space where the refractive index is 1 or if you use brightness temperatures Remaining cases are also handled exactly if iyEmissionStandard is used For those remaining cases radiance data shall be scaled with the refractive index squared at the observation position For Earth the maximum value of this factor is about 0 1 and can anyhow normally be neglected In summary there is normally no need for you as an user to consider the n law The exception is if you extract radiance data for a point inside an atm
42. line calculation the complete models are intended to be used alone To select a continuum or complete absorption model simply use the corresponding tag with abs_speciesSet Cur rently available models are listed in Table 6 3 The names should be fairly self explanatory and can be used to find background infor mation on the various models in ARTS Theory The condensate absorption models are a bit special and perhaps need some extra explanation They are absorption parameterizations by Liebe and allow the inclusion of condensate in the rare cases where scattering is not important Their general applicability is therefore fairly limited The behavior of the continua and complete absorption models can be modified by passing them some additional parameters stored in the variables abs_cont_names abs_cont_models and abs_cont_parameters Basically abs_cont_names identifies the model abs_cont_models contains switches that select different behavior for example taking only the lines or only the continuum part of a complete model and abs_cont_parameters can contain numerical parameters Yes the nomenclature for these additional continuum parameters particularly abs_cont_models is confusing However most users will never have to deal with these variables explicitly They are set to default values in the include file continua arts Users should therefore always include this file at the start of their controlfiles with INCLUDE continua arts
43. linked by specifying the geometrical altitudes z field 3 1 Altitude coordinates Pressure The main altitude coordinate is pressure This is most clearly manifested by the fact that the vertical atmospheric grid consists of equal pressure levels The vertical grid is accordingly denoted as the pressure grid and the corresponding workspace variable is p grid The choice of having pressure as main altitude coordinate results in that atmospheric quantities are retrieved as a function of pressure Pressure altitude A basic assumption in ARTS is that atmospheric quantities tempera ture geometric altitude species VMR etc vary linearly with the logarithm of the pressure This corresponds roughly to assuming a linear variation with altitude Radius Geometrical altitudes are needed to determine the propagation path through the at mosphere etc The main geometrical altitude coordinate is the distance to the centre of the coordinate system used the radius This is a natural consequence of using a spherical or polar coordinate system The radius is used inside ARTS for all geomet rical calculations Geometrical altitude The term geometrical altitude signifies here the difference in radius between a point and the reference ellipsoid Sec 11 1 along the vector to the centre of the coordinate system Equation 11 6 This is consistent with the usage of geocen tric latitudes see below Hence the altitude is not measured along the local zenith di
44. measurements is emission in the atmosphere or by the Earth s surface Thermal IR radiation is governed by the same basic physical principles and therefore this wavelength region is also well handled in ARTS now But ARTS contains so far no dedicated methods for scattering of solar radiation and there is therefore a restriction to simulations of long wave radiation microwave to thermal IR However ARTS can be used for basic studies of lonwave radiation fluxes as for example in Buehler et al 2006 or John et al 2006 More lately some support for handling radio link calculations have been added One main application of ARTS should be to perform retrievals for remote sensing data A special feature of ARTS in this context is its high flexibility when defining observation geometry including scanning features and sensor characteristics Jacobians weighting functions are also provided There exist two versions of ARTS This user guide deals with the later of the two ver sions Eriksson et al 2011 here denoted as just ARTS ARTS 1 the first version of ARTS Buehler et al 2005 can only handle 1D atmospheres with unpolarised radiation and sit uations where scattering can be neglected These restrictions have been removed in the current version A short summary of ARTS s main features is The atmosphere can be 1D 2D or 3D That is atmospheric variables temperature gas concentrations etc can be assumed to only vary in the vertical
45. of the first kind is H20 18 which identifies a particular isotopologue of water vapor An example of the second kind is H20 ForeignContCKDMT100 which identifies a particular continuum model An example of the third is 02 Z which identifies that special Zeeman routines should be used Tags can be combined if they refer to the same 40 GAS ABSORPTION Initialization Spectroscopic parameters Explicit line by line calculation Continuum parameters Continua and complete absorption models HITRAN HITRAN CIA data collision induced absorption Figure 6 4 An inside view of abs_xsec_agenda 6 5 CALCULATING GAS ABSORPTION 41 molecule different isotopologues are allowed Even continuum tags can be combined with explicit line by line tags if they refer to the same molecule It should be noted that isotopologue ratios are taken into account implicitly when line strengths are calculated so even if you make calculations for individual isotopologues the VMR numbers in the variable vmr_field should not be adjusted for the isotopologue ratio the isotopologue ratio can be changed instead see Section 6 5 9 As an example to make a line by line calculation for all ozone isotopologues you could represent them in different ways by abs_speciesSet a abs_speciesSet species 03 b abs_speciesSet species 03 666 03 668 03 686 03 667 O3 676 c abs_speciesSet species 03 666
46. of variables required for a DOIT calculation using DoitInit DoitInit As the next step we have to calculate the incoming field on the boundary of the cloudbox This is done using the workspace method DoitGetIncoming DoitGet Incoming The method doit_i_fieldSetClearsky interpolates the incoming radiation field on all points inside the cloudbox to obtain the initial field doit_i_field for the DOIT calculation As a test one can alternatively start with a constant radiation field using the method doit_i_fieldSetConst doit_i_fieldSetClearsky The grid discretization plays a very significant role in discrete ordinate methods In spherical geometry the zenith angular grid is of particular importance cf ARTS Theory Section 8 6 1 The angular discretization is defined in the workspace method DoitAngu larGridsSet DoitAngularGridsSet doit_za_grid_size scat_aa_grid scat_za_grid 19 10 doit_za_grid_opt xml For down looking geometries it is sufficient to define the generic inputs N_za_grid Number of grid points in zenith angle grid recommended value 19 N_aa grid Number of grid points in azimuth angle grid recommended value 37 From these numbers equally spaced grids are created and stored in the work space variables scat_za_grid and scat_aa_grid For limb simulations it is important to use an optimized zenith angle grid with a very fine resolution about 90 for the RT calculations Such a grid can be generated using the
47. on the atmospheric state variables e Pressure e Temperature e Concentrations of absorbing matter i e gases absorbing particles free electrons e Magnetic field 54 GAS ABSORPTION The basic idea of the lookup table is to pre calculate absorption for discrete combi nations of these variables and then use interpolation to extract absorption for the actual atmospheric state Due to the nature of the Zeeman and Faraday effects also particle ab sorption particularly due to their directional dependence those are not implemented in the lookup table Thus we can ignore the magnetic field The lookup table concept and implementation is described only very briefly here in the user guide Much more details and validation results can be found in Buehler et al 2011 6 6 2 Lookup table concept The fundamental law of Beer states that extinction is proportional to the intensity of radi ation and to the amount of absorbing substance dl D rn I 0 I otoa 6 4 i i dl where the meaning of the symbols is defined in Table 6 1 As one can see from the above equation a large part of the pressure dependence of Qi comes from n If one assumes constant volume mixing ratio of species 2 then n is proportional to the total pressure according to the ideal gas law Therefore the lookup table should store rather than a We then have to worry only about the dependence of on the atmospheric state variables Pressur
48. other planets is available from the art s xml data package It is 52 GAS ABSORPTION appealing to apply this data in non scattering calculations e g at low frequencies where the scattering contribution is negligible in a consistent manner The ARTS method for that is propmat_clearskyAddParticles and its application is described in the following To consider grey body particle absorption the user has to include prop mat_clearsky AddParticles in the propmat_clearsky_agenda Furthermore for each scattering element see Section 8 1 for how a scattering element is defined 1 a particles tag needs to be added to abs_species 2 the corresponding concentration field has to be added to vmr field and 3 its single scattering data have to be added to scat_data This can be done each by each using ReadXML and Append methods but a dedicated method Particle Type2abs speciesAdd is available performing these three steps for one scattering element at once ParticleType2abs_speciesAdd adds the raw number density field to vmr_field_raw i e the raw concentration fields can be converted to internal atmospheric grids together with the gas concentration fields using e g AtmFieldsCalc Single scattering data of all individual scattering elements is added to one and the same scattering species specifically to the last one of these in the scat_data array Note that ParticleType2abs_speciesAddq is essentially do ing the same as ParticleTypeAdd but fo
49. path to next For the first Stokes element the following expres sion is applied compare ARTS Theory Equation 6 55 Lim l e Bi 1 7 9 2 76 CLEAR SKY RADIATIVE TRANSFER with Bi B T B T 1 2 9 3 Al a Qi 1 2 9 4 Ti where I T and a are the radiance temperature and absorption coefficient respectively at point 7 of the propagation path and Al is the distance along the path between point i and i 1 That is B is an average of the Planck function at the path step end points and the absorption is assumed to vary linearly between the two points The start value of I is governed by the radiative background Section 9 3 A consequence of unpolarised absorption is that also the emission is unpolarised and the emission term vanishes for higher Stokes elements Accordingly the expression for the second Stokes component is Qi V Qi vje 9 5 The third and forth Stokes component are handled likewise The expressions above are implemented in the workspace method iyEmissionStandard intended to be part of iy_main_agenda An alternative way to perform the calculations for the first Stokes element would be I 5 timBi l e 9 6 where J is the final intensity and t is the transmission between the sensor and point i This calculation approach is not used as it fits poorer with the calculation of weighting functions J must be known Section 16 However the calculation of weighting fu
50. path to transmitter and path point derived If the trans mitter is placed inside the model atmosphere the closest point is found by an interpolation between the points defining the path If the transmitter is above the top of the atmosphere TOA the closest point is derived by geometrically extend the path from its end point at TOA For each iteration of the search the zenith angle is updated The first choice is to deter mine the new zenith angle by a least squares fit using data from the three last test paths On the same time the zenith angles giving the smallest positive and negative angular miss are kept in memory giving the end points for new range of new zenith angles to consider If the least squares fit gives an angle outside this range bi section between the end points of the search range is used for updating the angle A special consideration to ground inter sections must be given see the code for details If it is determined that the ground makes a link impossible a dummy propagation path of length 1 having surface as radiative background is returned The radiative background is otherwise set to transmitter The above is repeated until the miss term is smaller than the given accuracy criterion defined by the argument za_accuracy For a receiver in a low Earth orbit a tangent altitude accuracy of 1 m corresponds roughly to 2 1075 for za_accuracy If the selected criterion is not met the calculations are re
51. pnd_fieldCalcFrompnd_field_raw The definition of the single scattering data along with the corresponding particle number density fields is common in both scattering modules the DOIT module described in Chapter 18 and the Monte Carlo module in Chapter 19 Part II Radiative transfer clear sky general functionality Chapter 9 Clear sky radiative transfer This section discusses variables and the approach used to handle the actual radiative transfer calculations This includes how effects caused by the sensor and surface are incorporated Measurements of thermal emission in absence of particle scattering are used as example and the basic theory for such simulations is also covered The first ARTS version was developed for emission measurements and such observations remain the standard case in ARTS A basic assumption for this chapter is thus that there is no particle scattering This is denoted as clear sky calculations Scattering is restricted to the cloud box Sec 3 7 In short the more demanding calculations are restricted to a smaller domain of the model at mosphere and the radiative transfer in that domain is mainly treated by dedicated workspace methods For pure transmission measurements where scattering into the line of sight is neglected see Chapter 15 This chapter discusses only the direct radiative transfer partial derivatives i e the Jacobian or weighting functions are discussed in Section 16 Absorption
52. precision a small deviation from zero can be found 15 2 RADIO LINK CALCULATIONS 105 15 2 7 Defocusing A special effect for radio links is defocusing This effect is caused by the fact that refraction varies over the wavefront In short defocusing occurs if neighbouring ray paths diverge more quickly than for free space propagation This is the typical situation for limb sounding due to the vertical variation of the refractive index causing an additional divergence of the transmitted power The opposite focusing is also possible In fact already for an ideal spherically symmetric atmosphere a focusing effect takes place in the horizontal dimension caused by the curvature of the planet In this case the middle point of the beam experiences the highest refractive index and has then the lowest propagation speed This in addition to symmetry around the middle point counteracts partly the free space divergence of the power Defocusing can be estimated by two different algorithms simple denoted as method 1 and 2 The user selects the method to apply by the argument defocus method Defocusing method 1 This method is intended to be general using a very simple calculation approach The optical path length between the transmitter and the receiver is determined here denoted as lo This is followed by that two propagation paths are calculated with slight shifts in the zenith angle The size of the angular shift is an user input defocus_shi
53. scattering angle dummy grid 3 horizontally_aligned Only half of the grid is required Range 0 0 lt aa lt 180 0 The angular grids have to satisfy the following conditions They have to be equidistant FIXME is this actually still true The value of the data must be the same for the first and the last grid point FIXME of the azimuthal grid doesn t make sense for zenith angle grid This condition is required for the integration routine If we only have to store a part of the grid for example za_grid only from 0 to 90 these two values 0 90 must be grid points Tensor7 pha_mat data Phase matrix data Z Unit m The dimensions of the data array are frequency temperature za_sca aa_sca za_inc aa_inc matrix_element The order of matrix elements depends on the chosen case For most cases we do not need all matrix elements see description of cases below 62 DESCRIPTION OF CLOUDS Table 8 1 Structure of single scattering data files Symbol Type Dimensions Description enum ptype specification String short description of the scattering element v Vector v frequency grid T Vector T temperature grid Y Vector w zenith angle grid Ww Vector w azimuth angle grid Z Tensor7 v T y w Y w i phase matrix K Tensor5 v T w i extinction matrix a Tensor5 v T Y w i absorption vector e Tensor5 ext_mat_data Extinction matrix data K Unit m The dimensi
54. several variables This is unproblematic for quantities that are of discrete nature including scalar variables However for atmospheric fields and other continuous model quantities the discrete representation inside the forward model requires consideration To avoid in consistencies between model input and output it is important that the mapping from the discrete variables to the continuous view of the quantity is well defined and applied con sistently through the forward model This mapping is given by the basis functions Similar arguments and nomenclature are found in Read et al 2006 The basis functions are discussed explicitly in few places in this user guide but it shall be noted that all interpolations imply an underlying set of basis functions On the other hand an understanding of both the derivation and the obtained Jacobians require direct consideration of the basis functions ARTS operates with two types of basis functions 16 2 1 Basis functions for piece wise linear quantities To treat an one dimensional quantity to be piece wise linear or to say that a linear interpo lation shall be applied are identical definitions The basis functions matching this definition have triangular shape sometimes denoted as tenth functions Such functions are exem plified in Fig 16 1 see also Buehler et al 2005 110 CLEAR SKY JACOBIANS In most cases the quantity is considered to be undefined outside the end points of
55. started with a smaller ray tracing step length ppath lraytrace and with last value of rte los as first guess The decrease of the ray tracing step length follows the argument pplrt_factor If the ray tracing length is smaller than pp1rt_lowest the calculations are halted and a dummy propagation path of length 1 flagged as undefined is returned The method returns ppath_lraytrace set to the value used for the final calculation A suitable start value for ppath lraytrace depends on the size of refractive index For satellite to satellite links close to the Earth s surface a suitable value can be as low as 100 m 15 2 3 Definition of attenuation and loss terms The radiance law is not applicable for the signal from a point source The power received Pr by an antenna with an effective aperture size Ae located a distance l from a transmitter with power p and gain g can be expressed as e g Ippolito 2008 Prot Ari where ta is the transmission due to losses caused by the atmosphere This total atmospheric loss is the product between losses due to defocusing t Sec 15 2 7 and gaseous and particle extinction te Sec 15 2 8 pr l ta Ae 15 2 ta tate 15 3 102 TRANSMISSION CALCULATIONS The factor 1 t TP represents the pure geometrical dilution of the power following the inverse square law This effect is encountered also for propagation in free space and accordingly it is denoted as the fr
56. that for 1D cases a motion along a constant altitude has no influence on the simulated spectra as the same atmospheric fields are seen for a given viewing direction It is favourable if possible to handle all spectra as a single block instead of using a block for each sensor position This is the case as the antenna patterns for the different line of sights are normally overlapping and a pencil beam spectrum can be used in connection with 32 COMPLETE CALCULATIONS several measurement spectra to estimate the intensity field If a measurement sequence is divided into several blocks even if a single block would be sufficient pencil beam spectra for basically identical propagation paths can be calculated several times which of course will increase the computational time To summarise for cases when the sensor is not in motion or with a 1D atmosphere and a sensor not moving vertically the aim should be to use a single block for the measurement sequence If not a single block is used the standard option should be that the blocks cover one spectrum each There could exist reasons to select an intermediate solution to let the extent of the blocks be several spectra but not the full measurement sequence This could be the case when the atmospheric dimensionality is 2D or 3D and the sensor is moving but the movement during some subsequent spectra can be neglected The pencil beam spectra for each line of sight are appended vertically to form a c
57. the grid Hence the basis function for a grid end point is then just half a tenth The exception to this rule is retrieval grids of piece wise linear variables To avoid that retrieval grids must cover the complete atmosphere end point values are assumed to be valid to the end of the atmosphere or data range of concern That is the basis functions for end points of retrieval grids follow the tenth shape inside the grid range and have a constant value of 1 outside In terms of interpolation this matches to allow extrapolation the applying a nearest interpolation for positions outside the covered range the end values are valid all the way to infinity The basis functions are defined likewise for higher dimensions but the tenth functions are then 2D or 3D tenths 16 2 2 Polynomial basis functions Some retrieval quantities are expressed using a polynomial basis Sensor zenith angle point ing off set is one such quantity The off set is then treated to have a polynomial variation as a function of time If the offset is assumed to be constant in time a zero order polynomial shall be selected If there is also a linear drift with time use a first order polynomial etc For these basis functions the explanatory variable time in the example above is nor malised to cover the range 1 1 here denoted as z and the continuous representation f of the variable of concern can be written as f z x0 x1 2 b1 z2 2
58. the practical differ ence between this equation and Eq 16 27 should be small Or reversely Eq 16 27 should be approximately correct as C and C are close to identical matrices 116 CLEAR SKY JACOBIANS 16 5 Winds Calculation of wind weighting functions are triggered by jacobianAddWind Each wind component see Sec 13 1 is treated as an individual retrieval variable That is if you want to retrieve all three wind components jacobianAddWind must be called three times Only the analytically inclined calculation approach is at hand for winds Theoretically the Doppler shift induced by winds affects the emission source term Bj in the scalar case but this impact is extremely small and the related terms are ignored This gives a case basically identical to the one above for absorption species The only difference is that also the term a x is obtained in a pure numerical manner by recalculating the absorption coefficient with the wind perturbed slightly 16 6 Atmospheric temperatures 16 6 1 Common practicalities To obtain WFs for absorption species use jacobianAddTemperature The calculations can either be done in analytical or perturbation manner Retrieval grids must be specified A special consideration for temperature is hydrostatic equilibrium If effects originating in hydrostatic equilibrium shall be included in the WFs or not is selected by an argument denoted as hse A full account of hydrostatic equi
59. the propagation path at each point The number of rows of the matrix is np For 1D and 2D the matrix has a single column holding the zenith angle For 3D there is an additional column giving the azimuth angle The zenith and azimuth angles are defined in Section 5 2 2 If the radiative background is the cloud box the last position in pos and line of sight give the relevant information needed when extracting the radiative background from the cloud box intensity field r Vector The radius for each path position The length of this vector is accordingly np This is a help variable for plotting and similar purposes Istep Vector The length along the propagation path between the positions in pos The first value is the length between the first and second point etc nreal Vector The real part of the refractive index at each path position This index corre sponds to the phase velocity ngroup Vector The group index of refraction This index corresponds to the group veloc ity gp_p ArrayOfGridPos Index position with respect to the pressure grid The structure for grid positions is described in ARTS Developer Guide Section 5 4 gp_lat ArrayOfGridPos As gp_p but with respect to the latitude grid gp_lon ArrayOfGridPos As gp_p but with respect to the longitude grid 10 6 Further reading The implementation calculation approaches and the numerical expressions used are dis cussed in Chapter 7 of ARTS Theory 86 PROPAGATION
60. tion Figure 5 2 The valid range for 1D and 3D cases is 0 180 In the case of 2D zenith angles down to 180 are also allowed where the distinction is that positive angles mean a viewing direction towards higher latitudes and negative angles mean a viewing direction towards lower latitudes It should be mentioned that the zenith and nadir directions are here defined to be along the line passing the centre of the coordinate system and the point of concern Section 11 1 1 A nadir observation 7 180 is thus a measurement towards the centre of the coordinate system The azimuth angle w is given with respect to the meridian plane That is the plane going through the north and south poles of the coordinate system 90 and the sensor The valid range is 180 180 where angles are counted clockwise 0 means that the viewing or propagation direction is north wise and 90 means that the direction of concern goes eastward This definition does not work for position on the poles To cover these special cases the definition is extended to say that for positions on the poles the azimuth angle equals the longitude along the viewing direction For example if standing on 5 3 MEASUREMENT SEQUENCES AND BLOCKS 31 any of the poles and the viewing direction is towards Greenwich the azimuth angle is 0 The sensor line of sights are stored in sensor_los This workspace variable is a matrix where the first column holds zenith
61. to how i is compiled to obtain y The calculation procedure expands to allow that the complete weighting function matrix for quantities covered by the analytical calculation procedure is calculated as oy Y _ HK 16 12 Ox where K Oi 0x One column of this matrix is k Hki 16 11 See e g en wikipedia org wiki Matrix_calculus Available at https ccrma stanford edu dattorro mybook html 112 CLEAR SKY JACOBIANS The vector iy consists of a number of Stokes vectors appended i sT s7 s2 2 and the calculation of k can schematically be written as n Oi Os k 2 16 13 i 2 Os Ox The terms 0i 0s are formally matrices However these matrices are not calculated explic itly as they only contain information on where s is stored inside i That is these matrices are of bookkeeping character consisting only of zeros and ones In practice results match ing Os Ox are simply inserted in correct place of k mimicking how is s put into i Accordingly the core task is to calculate Os Ox where for simplicity the subscript j is dropped below This calculation is expanded as Os Os 02 Be 2a Ba Ba 16 14 where x is the value of the quantity of concern at each point of the propagation path That is indexes the path points The actual radiative transfer enters by the terms Os 0x and they are discussed separately below The term Ox Ox appears due to the fact that
62. until the starting point of the propagation path is found The path is determined by starting at the end point and moving backwards to the starting point The calculations are initiated by filling ppath_step with the practical end point of the path This is either the position of the sensor true or hypothetical or some point at the top 10 3 SPACING OF ADDITIONAL PATH POINTS 83 Grid cells Sensor position Propagation path Tangent point Figure 10 2 As Figure 10 1 but with a length criterion for the distance between the points defining the path Note Tangent points are no longer included automatically of the atmosphere determined by geometrical calculations starting at the sensor The agenda performs only calculations to next crossing of a grid all other tasks are performed by ppath_calc with one exception If there is an intersection with the sur face the calculations stop at this point This is flagged by setting the background field of ppath_step Beside this ppath_calc checks if the starting point of the calculations is inside the cloud box or below the surface level and check if the last point of the path has been reached 10 3 Spacing of additional path points The strategy when considering ppath_Imax differs somewhat between the workspace meth ods For pure geometrical calculations the points are spaced evenly inside the grid box That is the points are separated with the same distance lt ppa
63. with the available methods described in the Sections below ARTS does not con sider refraction by solid and liquid particle as their refractive index is not known to ARTS when scattering is considered ARTS gets their single scattering properties as input see Chapter 8 However the contribution of solid and liquid constituents to the refractive in dex is commonly neglected in radiative transfer models and is expected to have only small effects Refractivity N describes the deviation of the refractive index of a medium nfrom the vacuum refractive index Nyacuum 1 N n 1 Contributions of the different components to refractivity are additive Therefore all ARTS methods that provide refractive index calculate refractivity and sum it up with the input refractive index Within ARTS refractive index is required for calculations of refracted propagation paths and related parameters e g deriving viewing angle for a given tangent altitude FIXME for more Whenever refractive index is required e g at each point along a propagation path it is evaluated according to the mechanism specified by refr_index_air_agenda refr_index_air_agenda provides both the monochromatic refractive index refr index air in the following denoted as n as well as the group refractive index refr_index_air_group de noted as ny refr_index_air differs from refr index_air_group in case of dispersion which e g leads to diverging propagation paths at different
64. x and x are placed at different positions and the representation of the atmospheric fields must be considered here In practise the term is calculated as the value of the basis function for x at the location of x further discussed in Buehler et al 2005 This is a slight approximation with respect to the goal of fully incorporating the piece wise linear representation in the weighting functions Buehler et al 2005 A low value of Al decreases the degree of approximation It can be noted that Ox Ox is normally non zero for more than one element of x The exception is if the positions of x and x are identical Reversely the weighting function for element x can have contributions from several propagation path points x as well as from several radiance spectra Accordingly the practical calculations are done by first determine all Os Ox of the given propagation path These data are then used to determine Os Ox for the retrieval quantity of concern That is each Os Ox is combined with 0x 0x for all elements of x Most of these combinations yield a zero result The terms 0x Ox are determined with help of ARTS internal interpolation grid position routines 16 3 3 0s 0x general case The term 0s 0x is here outlined for the general case of vector radiative transfer In this case the Stokes elements can not be treated separately and matrix vector notation is re quired The final Stokes vector obtained through Eq 9 7 can be expr
65. 03 Haugstad B S Turbulence in planetary occultations iii effects on atmospheric profiles derived from intensity measurements Icarus 3 422 433 1978 Ippolito L J Satellite communications systems engineering chap The RF link John Wi ley and Sons Ltd UK 2008 John V O S A Buehler A von Engeln P Eriksson T Kuhn E Brocard and G Koenig Langlo Understanding the variability of clear sky outgoing long wave radiation based on ship based temperature and water vapor measurements Quarterly Journal of the Royal Meteorological Society 132 2675 2691 2006 Kraus J D Radio astronomy McGraw Hill Book Company 1966 Kuntz M G Hochschild and R Krupa Retrieval of ozone mixing ratio profiles from ground based millimeter wave measurements disturbed by standing waves Journal of Geophysical Research 102 21965 21975 1997 Kursinski E R G A Hajj S S Leroy and B Herman The gps radio occultation tech nique Terrestrial Atmosperic Oceanic Studies 11 53 114 2000 Larsson R S A Buehler P Eriksson and J Mendrok A treatment of the Zeeman effect using Stokes formalism and its implementation in the Atmospheric Radiative Transfer Simulator ARTS Journal of Quantitative Spectroscopy and Radiative Transfer submit ted 2013 Makarov D S M Y Tretyakov and P W Rosenkranz 60 GHz oxygen band Precise experimental profiles and extended absorption modeling in a wide temperature range Journal
66. 1 where h is the Planck constant c the speed of light and ky the Boltzmann constant This ex pression gives the total power per unit frequency per unit area per solid angle The Planck function can also be defined as a function of wavelength The expression in Equation 9 9 deviates from the exact definition see Eq 6 16 in ARTS Theory as it includes c instead of the local propagation speed v The reason for this is the n law of radiance discussed in the section below As long as cosmic background radiation is the only type of non telluric radiation that has to be considered the standard method for inclusion in iy_space_agenda is MatrixCBR together with some calls of Ignore Please note that blackbody_radiation_agenda and ly space agenda as well as iy_surface_agenda must be defined in a consistent manner that they use the same unit for B B T 9 9 9 5 Output unit and the n law First of all it should be noticed that ARTS does not enforce any fixed unit for calculated spectra y it depends on the calculation set up For example if emission is considered or if just transmissions are calculated The primary unit for emission data radiances is W Hz m sr The emission inten sity corresponds directly with the definition of the Planck function Eq 9 9 Conversion to other units is selected by the iy_unit workspace variable The standard manner is to apply the unit conversion as part of the calculations performed
67. 16 2 The perturbation shall given following the unit selected The same value is applied for all WFs which can cause practical problems 16 4 3 Analytical expressions If not made clear above the only term that differs between the Jacobian quantities is Os 0x Unpolarised absorption The term 0B 0x of Eq 16 20 is zero and the chain rule expression reduces to Or r OT e The absorption coefficient at point 7 a is here written as Bi Ii 16 23 Qi tii 0 16 24 where g is the absorption cross section for the species at point 7 and in a unit matching the unit of x and a covers the summed absorption of all other atmospheric constituents Considering Eq 9 4 the derivative of 7 with respect to x is then 07 OT Oai _ Al Ox o r 2 16 25 16 4 ABSORPTION SPECIES 115 General case So far the term 0T 0x is only estimated by performing a small perturbation of the ab sorption associated with the species to be retrieved However there exists a potential to use an analytical expression as discussed below but this is left for future consideration The relationship between x and the propagation matrix of the path step can be written as ee wi Ci naci where C can be seen as a matrix absorption cross section corresponding to in Eq 16 24 and K is the propagation matrix summed for all other atmospheric constituents On the condition that Eq 16 9 is valid we have tha
68. 89 6400 T r m 6390 7 6380 Figure 11 2 The ellipsoid ra dius rq and curvature radius re for the WGS 84 reference ellipsoid The curvature radii are valid for the north south di Radius km rection 6340 ue 4 i cause ellipsiod radius curvature radius 6330 y y y y 7 i 7 0 10 20 30 40 50 60 70 80 90 Latitide deg 150 Figure 11 3 The change of the WGS 84 ellipsoid radius for 1 latitude differences 200 250 Radius difference m deg 300 350 10 20 30 40 50 60 70 80 90 Latitide deg 400 l 0 90 REFERENCE ELLIPSOID AND SURFACE PROPERTIES 11 2 Surface altitude The surface altitude zy is given as the geometrical altitude above the reference ellipsoid The radius for the surface is accordingly Ts T0 2 s 11 6 As also mentioned in Section 3 6 a gap between the surface and the lowermost pressure level is not allowed The ARTS variable for the surface altitude z_ surface is a matrix For 1D the sur face constitutes a sphere by definition as the ellipsoid while for 2D and 3D any shape is allowed and a rough model of the surface topography can be made 11 3 Surface radiative properties If there is an interception of the propagation path by the surface emission and scattering by the surface must be considered This is the task of iy_surface_agenda The standard method for this agenda is iySurfaceRtpropAgenda
69. ABSORPTION Table 6 1 Examples of symbols used in this chapter the corresponding notation in the ARTS source code and a short description of the quantity Here Unit In ARTS Description Q ma Scalar gas absorption coefficient I Ww iy Intensity l m Path length element Ni molec m Number density of species 1 Ki m molec abs_xsec_per_species Absorption cross section of absorbing species 2 Kij m molec abs_cia_data Binary absorption cross section of absorbing species pair 2 7 K m7 4x4 propmat_clearsky Clear sky propagation matrix where J is intensity and is the distance through a homogeneous medium with absorption coefficient a Absorption is additive so the total absorption is the sum of the partial absorptions of all absorbers For an individual absorber we can define another important quantity the absorption cross section Kj as Qi 6 2 ki Ni where subscript denotes the absorber and n is the partial number density of that absorber Absorption cross sections depend less strongly on pressure than absorption coefficients and are therefore more suitable for storing in a lookup table Some processes create polarized absorption which is described by a 4x4 matrix in ARTS called propagation matrix propmat clearsky already introduced in Section 4 2 This variable is used by the ARTS RT functions to describe absorption In the absence of polarizing effects it is simply equal to 14 where 14 is t
70. ARTS This later effect originates also on the fact that right and left hand circular polarisation have different refractive index This can result in that the two polarisations obtain different propagation paths For frequencies close to the plasma frequency the birefringance can be a strong effect but for higher frequencies it should be secondary to Faraday rotation Expressed roughly a difference in optical path for the two circular polarisation of a quarter of a wave length changes the polarisation state strongly by Faraday rotation while additional effects coming from a difference in propagation path birefringance should be negligible According to Rybicki and Lightman 1979 using Gaussian cgs units the angle of rotation 0 F of a plane polarised wave can be described as 3 e de 2rc2m y d nals Beasley ods 0 which converted to SI units becomes Kraus 1966 e 23648 d d YE 303 e s Bgeo i x e s Bgeo gt 14 1 P aromas J Mels Byeo s ds EE f ne s Bgeo s ds 14 1 where e is the charge of an electron ey is the permittivity of vaccum m is the electron mass ne s is the density of electrons at point s Bgeo s is the geomagnetic field at point s and denotes the dot scalar product Accordingly the Faraday rotation is proportional to the part of the magnetic field along the propagation path the field normal to the path gives no effect The change in rotation angle along the propagation pat
71. ARTS User Guide edited by Patrick Eriksson and Stefan Buehler April 28 2015 ARTS Version 2 3 193 The content and usage of ARTS are not only described by this document An overview of ARTS documentation and help features are given in Section 1 2 For continuous reports on changes of the source code and this user guide subscribe to the ARTS developers mailing list at http www sat ltu se arts support We welcome gladly comments and reports on errors in the software or the document Send then an e mail to patrick eriksson at chalmers se or sbuehler at uni hamburg de If you use data generated by ARTS in a scientific publication then please mention this and cite the most appropriate of the ARTS publications The relevant publications are summarised at http www sat ltu se arts docs Copyright C 2000 2015 Stefan Buehler lt sbuehler at uni hamburg de gt Patrick Eriksson lt patrick eriksson at chalmers se gt The ARTS program is free software you can redistribute it and or modify it under the terms of the GNU General Public License as published by the Free Software Foundation either version 2 or at your option any later version This program is distributed in the hope that it will be useful but WITHOUT ANY WARRANTY without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE See the GNU General Public License for more details You should have received a copy of the
72. ARTS built in values mentioned above using isotopologue_ratiosInitFromBuiltin Alternatively they can be read from file using ReadXML For easy manipulation the user might initial ize isotopologue_ratios from built in data write the isotopologue_ratios structure to file using WriteXML modify the data accordingly and read in the manipulated file Files with isotopologue ratios for a couple of planetary atmospheres are provided with the arts xml data package It shall be noted that only isotopologue ratios of the species used in the absorption calculation need to be given Reading in from file resets the full list of isotopologue ratios i e the values for all absorption species known to ARTS with species not given in the input data set to NaN Partition functions Partition functions are currently still hard coded into ARTS That is no settings related to those have to be or can be done on the user level For more information on theoretical background as well as the source and implementation in ARTS see Section 2 1 4 of ARTS Theory 6 6 The gas absorption lookup table 6 6 1 Introduction Calculating gas absorption matrix spectra in a line by line way is quite an expensive thing to do Sometimes contributions from thousands or ten thousands of lines have to be summed up To make matters worse this has to be done over and over again for each point in the atmosphere Actually the absorption matrix depends not directly on position but
73. ClA 02 Al N2 CIA N2 0 N2 CIA N2 1 N2 CIA H2 0 N2 CIA CH4 0 10 H2 CIA H2 0 de H2 CIA He 0 e A H2 CIA CH4 0 E 02 CIA 02 0 CO2 CIA CO2 0 CO2 CIA CO2 0 CH4 CIA CH4 0 N2 CIAfunCKDMT100 02 CIAfunCKDMT100 107 CO2 CKDMT100 e ee a x d 10 de ci Me 57 ee A A 10 4 a li Fifi LY ee M 4 l PA WW i gt So Y ION A SN i l Y NI la is 1000 1500 Wavenumber em 4 N2 CIArotCKDMT1 00 CH4 CIA CH4 0 CH4 CIA Ar 0 Abs Cross section m molec Abs Cross section m molec 2000 2500 0 500 1 000 1 500 2000 2500 Wavenumber lem Figure 6 5 Left All HITRAN CIA continua that are implemented in ARTS each gas here has a partial pressure of 1000 hPa Right Only the ones that are relevant for Earth for Earth surface conditions 6 5 CALCULATING GAS ABSORPTION 47 lt arts format ascii version 1 gt lt SpeciesAuxData version 1 nelem 1 nparam 3 gt 02 66 2 002064 1 1 lt SpeciesAuxData gt lt arts gt In this example indicates the beginning of a new data record 02 66 specifies what iso topologue is associated with the data the number 2 002064 is the relative Land factor and the molecule got S 1 The last value indicates what Hund case is used in the calcu lations For oxygen Hund case b is used indicated b the number 1 Pres
74. Create 6 StringSet 6 surfaceBlackbody 90 surfaceFlatRefractivelndex 90 surfaceFlatScalarReflectivity 90 surfaceLambertianSimple 91 TangentPointExtract 84 wind_u_fieldIncludePlanetRotation 95 148 INDEX WriteXML 7 yApplyUnit 77 ybatchCalc 121 yCalc 27 29 71 78 80 yCloudRadar 135 z field FromHSE 19 workspace variable 6 workspace variables 7 abs_cia_data 36 43 abs_coef 42 abs_cont_models 43 abs_cont_names 43 abs_cont_parameters 43 abs_lines 42 abs_lines_per_species 42 abs_lineshape 42 47 abs_lookup 42 55 abs_lookup_is_adapted 55 abs_species 24 39 41 43 51 abs_vec 128 abs_xsec_per_ species 36 antenna_dim 31 atmosphere_dim 16 blackbody_radiation 25 cloudbox limits 20 cloudbox_on 20 doit_i_field 126 doit_scat field 127 ext_mat 128 ext_mat_spt 128 f_grid 7 73 isotopologue quantum 45 isotopologue_ratios 53 iy 27 36 72 78 134 ty_aux 79 iy_aux_vars 79 80 iy_unit 72 77 134 iyb 72 jacobian 108 jacobian_indices 108 jacobian_quantities 108 lat_grid 17 lat_true 19 lon_grid 18 lon_true 19 mag_u_field 21 47 mag_v _field 21 47 mag_w_field 21 47 mblock_dlos_grid 31 output_file_ format 8 p grid 15 17 p hse 19 pha_mat 127 pnd field 20 25 59 67 pnd_field_raw 67 ppath 79 82 ppath_Imax 80 81 83 ppath lraytrace 81 83 101 ppath_step 82 propmat_clearsky 25 36 37 42 51 propmat_clearsky_field 42 range_bins 135 r
75. Faraday rotation is treated by propmat clearsky agenda For the moment there exists a single workspace method for the task and that is propmat_clearskyAddFaraday That is this method must be included in propmat_clearsky_agenda if you want to include Faraday rotation in your calculations Information on the actual rotation can be obtained as auxiliary data by iyTransmis sionStandard and iyRadioLink The total rotation along the path is selected as Faraday rotation The unit is deg The rotation per length unit Eq 14 2 is selected as Faraday speed The unit is deg m 14 2 Theory A wave propagating through the ionosphere will force free electrons to move in curved paths If the incident wave is circularly polarised the motion of the electrons will be circu lar The refractive index will then not be a single constant but depending on polarisation i e anisotropic More precisely left and right hand polarised waves will propagate with History 121217 Written PE partly based on text written originally by Bengt Rydberg 98 FARADAY ROTATION different speeds Moreover as a plane polarised wave can be thought of as a linear superpo sition of a left and a right hand polarised wave with equal amplitudes but different phase the plane of polarisation will then rotate as the wave is propagating through the media This is denoted as Faraday rotation Birefringance is an associated mechanism but it is not yet treated by
76. For a given set of retrieved x the simplest way to determine the estimated baseline is to perform a multiplication between the relevant parts of the Jacobian and the state vector K Xp 16 34 where p indicates the n 1 columns elements corresponding to z 16 10 Sinusoidal baseline fit If the baseline has components of sinusoidal character there is also a second option pro vided the method jacobianAddSinefit The baseline is in this method modelled as Kuntz et al 1997 y y y a sin 7e b cos eS 16 35 i l li where a and b are the coefficients to be retrieved v is frequency v is a reference frequency and l is period length The reference frequency vi is in practice taken as the first frequency of the spectrometer The period lengths l are user input For each given period length the corresponding sine and cosine terms are included in the Jacobian As for the polynomial fit there exist options to set the baseline to be common between different polarisations viewing directions or measurement blocks Equation 16 34 is also applicable for sinusoidal baseline fits An alternative way to express the expression above is n 2 _ y y Aisin ES d 16 36 i 1 li where Aj fa b 16 37 120 CLEAR SKY JACOBIANS and dr 16 38 b Equation 16 35 is used to define the weighting functions as this gives a linear retrieval problem in contrast to if Equation 16 36 would be used that would require a
77. FromHitranPre2004 matter abs_linesReadFromJpl abs_linesReadFromMytran2 different methods are for different catalogue formats For the ARTS internal format the standard method ReadXML works also but does not allow to select a frequency range as the others do 3 Split line data for Variable abs_lines_per_species different absorption Methods abs_lines_per_speciesCreateFromLines Alternatively read species lines from different catalogues for different species directly with abs_lines_per_speciesReadFromCatalogues 4 Optimize line data Variable abs_lines_per_species optional Methods Add mirror lines for VVW line shape with abs_lines_per speciesAddMirrorLines see ARTS Theory Chapter 2 Remove lines that are outside the line shape cutoff with abs_lines_per_speciesCompact The first four steps are preparation and typically have to be done only once per ARTS run The fifth step is the actual absorption calculation which can occur in different contexts Sa Sb Sc Calculate absorption on the fly Calculate absorption lookup table Calculate absorption only no RT Agenda propmat_clearsky_agenda Variable propmat_clearsky which is nitialized in propmat_clearskyInit Methods propmat _clearskyAddOnTheFly the core method for on the fly absoprtion calculation inlcuding line by line and continuum absorption propmat_clearskyAddZeeman see Sec 6 5 5 propmat_clearskyAddFaraday see Sec 6 5 7 propmat_clear
78. OIT frames a aa ke ke ee a eS de ib 126 18 2 1 The DOIT Mn agenda s soes eor aed adela ee ee ee eG 126 18 2 2 Agendas used in doit i field monolterate 127 Calculation of the scattering integral 127 Radiative transfer with fixed scattering integral term 128 Convergence testi ei ee cd be ee eee a 129 18 2 3 Propagation of the DOIT result towards the sensor 129 18 3 3D DOIT calculations acta aaa anh Bed ba eR eae a G aa 130 19 Scattering calculations The Monte Carlo scattering module 131 20 Cloud radar simulations 133 201 THEY gt se creas dee Dis dow a de e do HR aa ee Gee a 133 202 UDS ee ek ws he ee Ce ee EE IT 134 20 3 Practical Usage oo a Sow e ie ee ae ee ee a a 134 VI CONTENTS V Bibliography and Appendices 137 VI Index 143 Part I Overview Chapter 1 Introduction This section introduces and describes the basic ideas underlying the ARTS program It also presents some terminology You should read it if you want to understand how the program works and how it can be used efficiently 1 1 What is ARTS The Atmospheric Radiative Transfer Simulator ARTS is a software for performing simu lations of atmospheric radiative transfer ARTS is a relatively general and flexible program where new calculation features can be easily added Originally the development of ARTS was initiated to deal with passive mm and sub mm measurements The radiation source for such
79. Tech rep ESTEC Contract No AO 1 4320 03 NL FF 2004 Emde C S A Buehler C Davis P Eriksson T R Sreerekha and C Teichmann A polarized discrete ordinate scattering model for simulations of limb and nadir longwave measurements in 1D 3D spherical atmospheres Journal of Geophysical Research 109 2004 Eriksson P Analysis and comparison of two linear regularization methods for passive atmospheric observations Journal of Geophysical Research 105 D14 18157 18167 2000 140 BIBLIOGRAPHY Eriksson P F Merino D Murtagh P Baron P Ricaud and J de La No Studies for the Odin sub millimetre radiometer 1 Radiative transfer and instrument simulation Cana dian Journal of Physics 80 321 340 2002 Eriksson P M Ekstr m S Biihler and C Melzheimer Efficient forward modelling by matrix representation of sensor responses International Journal of Remote Sensing 27 1793 1808 2006 Eriksson P S A Buehler C P Davis C Emde and O Lemke ARTS the atmospheric ra diative transfer simulator version 2 Journal of Quantitative Spectroscopy and Radiative Transfer 112 1551 1558 2011 Evans K F S J Walter A J Heymsfield and M N Deeter Modeling of submillimeter passive remote sensing of cirrus clouds J Appl Met 37 184 205 1998 Goldstein D Polarized light chap The Stokes parameters and Mueller matrices for optical activity and Faraday rotation Marcel Dekker Inc USA 20
80. The workspace method opt_prop_sptFromMonoData requires that the raw data is already interpolated on the fre quency of the monochromatic calculation This requirement is fulfilled when DoitScat teringDataPrepare or scat_data_monoCalc is executed before doit_i_field_monolterate see Section 18 2 1 The work space method ext_matAddPart and abs_vecAddPart are used to extract the absorption vector abs_vec and extinction matrix ext_mat from the workspace variable ext_mat_spt The gas absorption is added internally AgendaSet spt_calc_agenda opt_prop_sptFromMonoData AgendaSet opt_prop_part_agenda ext_matInit 18 2 DOIT FRAME 129 abs_vecInit ext_matAddPart abs_vecAddPart Convergence test After the radiative transfer calculations with a fixed scattering integral field are com plete the newly obtained radiation field is compared to the old radiation field by a con vergence test The functions and parameters for the convergence test are defined in the agenda doit_conv_test_agenda There are several options The workspace methods doit conv flagAbsBT and doit conv flagAbs compare the absolute differences of the ra diation field element wise as described in ARTS Theory Equation 8 19 The convergence limits are specified by the generic input epsilon which specifies the convergence limit A limit must be given for each Stokes component In doit_conv_flagAbsBT the limits must be specified in Rayleigh Jeans brightness temperatures wh
81. XML ascii v s where in the latter case variable s must be already defined The built in documentation further states that argument filename has a default value in this case the empty string Because of this the argument can actually be omitted so WriteXML ascii v will also work Alternatively workspace methods can be called with named arguments All omitted 1 3 ARTS AS A SCRIPTING LANGUAGE 9 Workspace variable propmat_clearsky_agenda This agenda calculates absorption coefficients for all gas species as a function of the given atmospheric state for one point in the atmosphere The result is returned in propmat_clearskyx the atmospheric state has to be specified by rtp_pressurex rtp_temperaturex rtp_mag and rtp_vmrx H he methods inside this agenda may require a lot of additional nput variables such as f_grid species etc pe Group Agenda Output propmat_clearsky Input f_grid rtp_doppler rtp_mag rtp_pressure rtp_temperature rtp_vmr AgendaSet AgendaSet propmat_clearsky_agenda propmat_clearsky_agenda propmat_clearskyInit Ignore rtp_mag propmat_clearskyAddOnTheF ly propmat_clearskyInit propmat_clearskyAddZeeman propmat_clearskyAddFromLookup Figure 1 4 Top Built in documentation for variable propmat_clearsky_agenda obtained by command arts d propmat clearsky agenda or on page http www sat 1ltu se arts docserver a
82. a Pa that gives w a cos 26 cos 26 11 10 Thus this value is a combination of the surface reflectivity and an solid angle weight The emission surface emission becomes raB b 11 11 0 0 0 92 REFERENCE ELLIPSOID AND SURFACE PROPERTIES Chapter 12 Sensor characteristics A sensor model is needed because a practical instrument gives consistently spectra deviating from the hypothetical monochromatic pencil beam spectra provided by the atmospheric part of the forward model For a radio heterodyne instrument the most influential sensor parts are the antenna the mixer the sideband filter and the spectrometer The response as a function of frequency zenith angle of such instrument parts is here denoted as the sensor characteristics The treatment of sensor variables is introduced in Section 5 2 As described in that section the position and line of sight sensor_pos and sensor_los are treated separately while remaining sensor characteristics are summarised by a response matrix denoted as sensor_response This matrix is applied following Equation 5 2 The purpose of this section is to describe how data on sensor characteristics are inputted to obtain a sensor_response that matches your particular instrument The implementation follows closely Eriksson et al 2006 That article provides also a more careful description of the approach of applying a response matrix and the equations used to conver
83. a strict analytical manner and some search algorithm is required as discussed in Section 15 2 2 15 2 2 Determination of the radio link path ARTS contains so far just a quite simple algorithm for this purpose activated by ap plying ppathFromRtePos2 in ppath_agenda The input rte_los is taken as the first guess for the observation angles of the receiver One way to set this initial value use rte_losGeometricFromRtePosToRtePos2 that sets rte_los to the angles for the geometrical path For satellite to satellite links with tangent points in the lower troposphere the geo metrical zenith angle can give an intersection with the surface and it is recommended to use a lower start value for the zenith angle For 1D cases this is simply achieved 15 2 RADIO LINK CALCULATIONS 101 rte _losGeometricFromRtePosToRtePos2 VectorAddScalar rte_los rte_los 2 5 Another option is to use the angle from the last link but this requires that the links of an occultation is treated in a specified order which is not guaranteed when running with multi threading New propagation path are calculated from the receiver backwards towards the transmit ter until a path is found that passes the transmitter sufficiently close The search and stop criterion is based on the smallest distance between the transmitter and the path However this distance is evaluated as an angular miss The miss term is taken as the difference in zenith angle between the geometrical
84. abs_species Correspondingly a field of electron densities is required in vmr field For usage examples check controlfiles artscomponents faraday that is part of the ARTS distribution 6 5 8 Absorbing particles As pointed out before this chapter deals with absorption by non scattering matter In first place this refers to gases while particles aerosols clouds precipitation are considered to also scatter radiation and are handled differently see Chapters 8 18 and 19 However when particles are small compared to the wavelength of the radiation they act as broadband grey body absorbers and can be treated similarly to continuum absorption by gases This is reflected in ARTS providing a few continuum models for condensed matter see Tab 6 3 which essentially are particles too It is tedious though to implement those kind of particle continua for a wide range of different base materials as become of interest when being interested in other than the Earth s atmosphere In the ARTS scattering modules particles are represented by single scattering property data scat_data and particle concentrations particle number densitity fields pnd_field The single scattering data originate from scattering theory programs e g Mie theory T matrix model Discrete Dipole Approximation and their preparation typically requires significant efforts Comprehensive data for hydrometeors in the Earth atmosphere but also clouds dust and the like for
85. ace methods that in some way or other compute abs_xsec See the built in documentation of the individual methods to learn more As for prop mat_clearsky_agenda file agendas arts predefines some typical alternatives also for abs_xsec_agenda 6 5 1 Absorption species Absorption is additive so the total absorption is simply the sum of all partial absorptions And the partial absorption for gases that have spectral lines can be calculated as a sum over the absorption of each spectral line plus some more or less empirical continuum terms An absorption species in ARTS is an abstract entity that has a partial absorption ma trix associated with it and that usually can be associated with a volume mixing ratio of a corresponding gas the VMRs are stored in variable vmr_field Total absorption is the sum of the partial absorptions of all absorption species Absorption species are defined in the ARTS controlfile by special tags which are stored in the variable abs_species and set by the method abs_speciesSet The absorption species tags specify the different considered absorbers which can be gaseous species but also free electrons and grey body particles For gaseous species they also describe the model that should be used to calculate the absorption for each of the species There are three types of tags those for explicit line by line calculations those for continua and complete absorption models and a special Zeeman effect tag An example
86. amma size distributions for ice particles 1 was assumed 66 DESCRIPTION OF CLOUDS The particle number density for size distributions is obtained by integration of the dis tribution function over all sizes n IMC Reff fo n r dr 8 8 f ar exp br dr aan 8 9 After setting a 1 inserting Equation 8 5 and some simple algebra we obtain 2 IMC P IMC Reff 8 10 n gt eff PRE Comparing Equation 8 2 and 8 10 we see that the particle number density for mono disperse particles with a particle size of R is smaller than the particle number density for gamma distributed particles with Reff R The reason is that in the gamma distribution most particles are smaller than RR y f 8 3 3 Ice particle size parameterization by McFarquhar and Heymsfield A more realistic parameterization of tropical cirrus ice crystal size distributions was derived by McFarquhar and Heymsfield 1997 who derived the size distribution as a function of temperature and IMC The parameterization was made based on observations during the Central Equatorial Pacific Experiment CEPEX Smaller ice crystals with an equal volume sphere radius of less than 50 wm are parametrized as a sum of first order gamma functions n r 2 12 IMC lt 50 5or mpl 5 where lt 50 is a parameter of the distribution and IM C lt s5o is the mass of all crystals smaller than 50 um in the observed size distribution Large ice crystals are represente
87. art somewhere in the middle of your batch input data The next section gives some examples of what could be put inside the ybatch_calc_agenda History 090330 Description of ybatch_start added by Stefan Buehler 070726 Copy edited by Stefan Buehler 070618 Updated by Oliver Lemke 040916 Created by Patrick Eriksson 122 BATCH CALCULATIONS 17 2 Control file examples The WSV Extract can be used to extract an element from an Array a row from a Matrix a Matrix from a Tensor3 and so on The selection is always done on the first dimension So for a Tensor3 as input it copies the page with the given index from the input Tensor3 variable to the output Matrix The idea here is to store the batch cases in tensors or arrays with one dimension extra compared to corresponding workspace variables For example a set of t field which is of type Tensor3 is stored as Tensor4 If the dimension is the same for all batch cases tensors are appropriate If the dimen sions vary for example you have a different number of vertical levels for each case then arrays are appropriate In the following 1D example five atmospheric scenarios have been put into the three first loaded files and a random vector of zenith angles is found in the last file The batch calculations are then performed as ReadXML tensor4_1 batch_t_field xml ReadXML tensor4_2 batch_z field xml ReadXML tensor5_1 batch_vmr_field xml ReadXML tensor3_1 ba
88. at are not readily available in e g HITRAN are necessary These include g the relative Land factor and S the molecular total spin These variables must therefore be read to the WSV isotopologue_quantum This variable is of type SpeciesAuxData and a file providing that kind of data would e g look like 46 GAS ABSORPTION Table 6 4 Absorption species tags frequency ranges and temperature ranges for HITRAN CIA data as implemented in ARTS These data contain some modifications from the origi nal HITRAN data which are described in the text CIA tag Spectral range cm Temp range K No of datasets N2 CIA N2 0 0 02 554 00 40 00 400 00 10 N2 CIA N2 1 1850 00 3000 09 228 20 362 50 10 N2 CIA H2 0 0 02 1886 00 40 00 400 00 10 N2 CIA CH4 0 0 02 1379 00 40 00 400 00 10 H2 CIA H2 0 20 00 10000 00 200 00 3000 00 113 H2 CIA He 0 20 00 20000 00 200 00 9900 00 334 H2 CIA CH4 0 0 02 1946 00 40 00 400 00 10 H2 CIA H 0 100 00 10000 00 1000 00 2500 00 4 He CIA H 0 50 00 11000 00 1500 00 10000 00 10 02 CIA 02 0 1150 00 1950 00 193 40 353 40 15 CO2 CIA CO2 0 1 00 250 00 200 00 800 00 7 CH4 CIA CH4 0 0 02 990 00 40 00 400 00 10 CH4 CIA Ar 0 1 00 697 00 70 00 296 00 5 Temperature 310K panel pressure of each gasi 1000 hPa Earth midlatidue summer T 294 K p 1000 hPa N2 CIA N2 0 Ne CIA N2 1 N2 CIA CH4 O2
89. ation and a double sided disturbance is used this option should in general be preferred 16 8 Sensor frequencies This class of Jacobians treats deviations between nominal and actually recorded frequencies Such differences can originate in several ways but the exact origin can normally be ignored and the effect can be modelled as the backend spectrometer filter bank channels are shifted from their nominal position The workspace methods to define such Jacobians are jacobianAddFreqShift and jacobianAddFreqStretch These Jacobians can so far only be determined by applying an interpolation of existing monochromatic data then shifted df from the nominal values The methods treat either the shift or stretch effects This follows standard nomen clature A shift is an off set that is of the same size for all backend channels That is if only a shift is assumed the nominal distances between the channels are assumed to be valid The stretch term considers the distance between the channels For a backend with all channels equally spaced a stretch signifies that the spacing deviates from the nominal value but all channels still equally spaced More generally a stretch means that the de viation from the nominal channel position increases linearly from the middle point of the backend In terms of the basis functions Section 16 2 2 shift and stretch correspond to polynomial order 0 and 1 Both frequency shift and str
90. attering integral 2 RT calculations with fixed scattering integral field 3 convergence test doit_i_field_monolterate HE Se SE He The first method DoitScatteringDataPrepare prepares the single scattering data for use in a DOIT scattering calculation Namely it interpolates the data on the requested frequency and performs the transformation from the scattering frame into the laboratory frame Alter natively the method scat_data_monoCalc can be used In this case only the frequency inter polation is done and the transformations are done later The advantage is that this method needs less memory For 1D calculation it is recommended to use DoitScatteringDataPrepare because it is much more efficient The iteration is performed in the method doit_i_field_monolterate which includes e the calculation of the scattering integral field doit_scat_field requires agendas pha_mat_spt_agenda and doit_scat_field_agenda e the radiative transfer calculations in the cloudbox with fixed scattering integral requires agendas spt_calc_agenda opt_prop_part_agenda ppath_step_agenda and doit_rte_agenda and e the convergence test requires agenda doit_conv_test_agenda For details on the agendas involved see Section 18 2 2 18 2 2 Agendas used in doit_i_field_monolterate There are several methods which can be used in doit_i_field_monolterate for instance for the calculation of the scattering integral The methods are selected in the con
91. bed by two angles the zenith angle 4 and the azimuth angle w The zenith angle exists for all atmospheric dimension alities while the azimuth angle is defined only for 3D The term line of sight is not only used in connection with the sensor it is also used to describe the local propagation direction along the path taken by the observed radiation Section 9 2 The zenith and azimuth angles are defined in an identical way in both of these contexts sensor pointing direction local propagation direction This is expected as the position of the sensor is the end point of the propagation path The sensor line of sight is the direction the antenna is pointed to receive the radiation The line of sight for propagation paths is defined likewise it is the direction in which a hypothetical sensor must be placed to receive the radiation along the propagation path at the point of interest This means that the line of sight and the photons are going in opposite directions As a true sensor has a finite spatial resolution described by the antenna pattern theoretically there is an infinite number of line of sights associated with the sen sor but in the forward model spectra are only calculated for a discrete set of directions If a sensor line of sight is mentioned without any comments it refers to the direction in which the centre of the antenna pattern is directed The zenith angle 1 is simply the angle between the line of sight and the zenith direc
92. bes the absorption look up table approach used inside ARTS 1 2 3 Built in documentation ARTS contains built in documentation for all functions and variables that are directly visible to the user in ARTS terminology called workspace functions and workspace variables they are explained in more depth further below The easiest way to access this documentation is on the web page http www sat ltu se arts docserver Alternatively start ARTS with arts s or arts docserver and then point your browser to http localhost 9000 This user guide also contains links to the built in documentation If you are reading the pdf file on a computer then names of ARTS objects such as f_grid will be links to the corresponding entries in the built in documentation 1 2 4 Test and include controlfiles ARTS calculations are governed by controlfiles see below The ARTS distribution already comes with a large number of controlfiles which fall into two categories includes and tests They are described in more detail below but already mentioned here as an important source of information for new users In particular ARTS already comes with controlfiles to simulate some well known instruments such as for example MHS or HIRS FIXME Control file structure changed update this section 1 2 5 Source code documentation All internal ARTS functions are documented with DOXYGEN at the source code level This documentation is intended mostly for those
93. bsorbers when applying propmat_clearskyAddParticles Trying to use both in parallel results in a runtime error For a setup example check TestAbsParticle arts in controlfiles artscomponents absorption See the built in documen tation of the individual methods for further information 6 5 9 Further input data and parameters for calculating gas absorption Spectral line data Important input to the line by line calculations is the spectral line data usually provided by spectroscopic catalogues ARTS has its own format for the spectral line data but is also capable of handling data from other catalogues like HITRAN both pre and post 2004 formats and JPL see Table 6 2 step 2 Section 2 1 2 of ARTS Theory contains more 6 6 THE GAS ABSORPTION LOOKUP TABLE 53 information on the internal format of the spectral line data It also contains theoretical background for the calculation itself Isotopologue ratios Isotopologue ratios mostly from HITRAN and valid for Earth atmosphere are stored in ARTS source code These data are necessary for working with HITRAN data as HITRAN line strengths are weighted with isotopologue abundance However it is convenient for the user to be able to change isotopologue ratio values e g when modeling absorption in other planets atmospheres The WSV isotopologue_ratios holds the isotopologue ratios applied in the absorption calculation They have to be set by the user It is possible to apply the
94. by atmospheric gases does normally not depend on polarisation but excep tions exist where Zeeman splitting is one example Both polarised and unpolarised absorp tion is handled Even if the gaseous absorption in itself is unpolarised the expressions to apply must allow that polarisation signals from the surface and the cloud box are correctly propagated to the sensor For an introduction to a complete radiative transfer calculations see Chapter 5 For example the content of this chapter corresponds roughly to the flowchart displayed in Fig ure 5 1 outlining a standard radiative transfer emission calculation In fact this chapter can be seen as a direct continuation of Chapter 5 9 1 Overall calculation procedure The structure handling complete radiative transfer calculations is fixed where the main workspace method is denoted as yCalc Fig 5 1 That is most ARTS control files include History 130220 Revised after parts moved to a new chapter Patrick Eriksson 120831 Added flowchart and sections on polarised absorption iyCalc auxiliary data and dispersion Patrick Eriksson 110611 Extended and general revision Patrick Eriksson 050613 First complete version by Patrick Eriksson 72 CLEAR SKY RADIATIVE TRANSFER Algorithm 1 Outline of the overall clear sky radiative transfer calculations yCalc allocate memory for the matrix y Equation 5 2 allocate memory for the matrix i Equation 5 1 for all sensor positions do
95. cause in this case optical properties can be calculated much faster than for arbitrary size distributions The T matrix code for randomly oriented particles includes several types of analytical size distributions e g the gamma distribution or the log normal distribution This section presents the size distribution parameterizations which were used for the ARTS simulations included in this thesis 8 3 1 Mono disperse particle distribution The most simple assumption is that all particles in an ensemble have the same size In order to study scattering effects like polarization or the influence of particle shape it makes 8 3 PARTICLE SIZE DISTRIBUTIONS 65 sense to use this most simple assumption because one can exclude effects resulting from the particle size distribution itself Along with the single scattering properties we need the particle number density field which specifies the number of particles per unit volume at each grid point for ARTS scatter ing simulations For a given MC and mono disperse particles the particle number density n is simply _ IMC IMC 3 IMC 2 Vp Ar pri me n IMC r where m is the mass of a particle r is its equal volume sphere radius p is its density and V is its volume 8 3 2 Gamma size distribution A commonly used distribution for radiative transfer modeling in cirrus clouds is the gamma distribution n r ar exp br 8 3 The dimensionless parameter a describes the wid
96. clear sky optical depth along the path and flag giving the radiative background Along the path means that data are provided for each point of the propagation path The path is described by ppath that is also returned by iy_main_agenda The ppath variable con tains the information needed to geo position for example along the path temperatures Example on setting of iy_aux_vars again valid for iyEmissionStandard ArrayOfStringSet iy_aux_vars Temperature VMR species 0 Absorption summed Absorption species 0 Absorption species 2 iy Optical depth The data are outputted in a single variable iy aux This variable is an array of Tensor4 All dimensions are used when storing e g the propagation matrix along the path for all frequencies of f_grid For other types of quantities one or several dimensions are set to have length 1 See the built in documentation for further details such as the order of the data dimensions Storage of quantities of along the path type assumes that there exists a common prop agation path This is necessarily not the case for calculations by yCalc This is the case as a calculation considering an antenna response includes radiative transfer along several prop agation paths The points of these paths do not end up on common altitude grid neither 80 CLEAR SKY RADIATIVE TRANSFER are at a fixed distance from the sensor In fact the number of points of t
97. cobians involves some approximations due to theoretical and practical considerations Such approximations can be accepted if of low or moderate size but will result in a slower convergence the inversion will require more iterations Due to these later aspects and to meet the needs of more experienced users this section is relatively detailed and contains a high number of equations This section is restricted to Jacobians for clear sky conditions i e to be applied outside the cloudbox So far none of the scattering methods provide Jacobians Sections 16 1 16 2 contain information of general character while the available quantities are discussed in the remaining sections Section 16 4 and forward History 110826 First complete version by Patrick Eriksson 108 CLEAR SKY JACOBIANS 16 1 Introduction There are two main approaches for calculating Jacobians by analytical expressions and by perturbations We start with the conceptually simplest one but also the more inefficient approach 16 1 1 Perturbations The most straightforward method to determine the Jacobian is by perturbing the model parameter of concern For example the Jacobian corresponding to state variable p can always be calculated as F x Aze b F x b Az where x b is the linearization state e is a vector of zeros except for the p th compo nent that is unity and Az is a small disturbance but sufficiently large to avoid numerical instabilit
98. d better by a log normal function 3 IMCs50 nr 3 2in D 2 3 13 2 py 2exp 3u gt 50 9 2 03509 r0 gt 5070 1 log pss0 exp ro 8 12 2 T gt 50 where IM Cso is the mass of all ice crystals greater than 50 um in the observed size distribution ro 1 um is a parameter used to ensure that the equation does not depend on the choice of unit for r 0x59 is the geometric standard deviation of the distribution and u gt s5o is the location of the mode of the log normal distribution The fitted parameters of the distribution can be looked up in the article by McFarquhar and Heymsfield 1997 The particle number density field is obtained by numerical integration over a discrete set of size bins This parameterization of particle size has been implemented in the PyARTS package which was introduced in Section 8 2 2 Using PyARTS one can calculate the size distributions the corresponding single scattering properties and the particle number density fields for given IMC and temperature exp 2a lt sor 8 11 8 4 Implementation The workspace methods related to the description of clouds in ARTS are implemented in the file m_cloudbox cc Work space methods related to the optical properties of the clouds are implemented in the file m_optproperties cc The coordinate system transforma tions described above reside in the file optproperties cc 8 4 IMPLEMENTATION 67 8 4 1 Work space methods and variables The following contr
99. d as input by many workspace methods for ex ample those that calculate absorption coefficients The built in documentation also lists all methods that take f_grid as input and all methods that produce f_grid as output 1 3 2 Workspace methods As shown in Figure 1 1 names of workspace methods in an ARTS controlfile are followed by their output and input arguments workspace variables in parentheses Print s prints the content of variable s From the methods point of view arguments can be output input or both and addi tionally they can be either specific referring to a predefined variable or generic not referring to a predefined variable To illustrate this Figure 1 3 shows the built in documen tation for method WriteXML the most common method to write ARTS variables to a file 8 INTRODUCTION Workspace method WriteXML Writes a workspace variable to an XML file This method can write variables of any group If the filename is omitted the variable is written to lt basename gt lt variable_name gt xml Synopsis WriteXML output_file_format v filename Authors Oliver Lemke Variables IN output_file_format String Output file format GIN v Any Variable to be saved GIN filename String Default Name of the XML file Figure 1 3 Built in documentation for method WriteXML obtained by command arts d WriteXML or on page http www sat ltu se arts docserv
100. d methods in the table is not complete The idea is to give an overview over the important ones and show how they work together Missing are particularly the input variables that describe the atmospheric conditions and continuum description variables which normally do not have to be set by the user anyway See the built in documentation of the various variables and methods for more informa tion It is on purpose not repeated here for better maintainability If you are viewing this pdf file on a computer just click on a variable or method name to get to the corresponding built in documentation Further input data and parameters required not only for line by line calculations is described in Section 6 5 9 6 5 3 Continua and complete absorption models ARTS includes many absorption continua and complete absorption models which are de scribed in ARTS Theory Chapter 2 The common property of all of these is that they do not use the standard ARTS line by line calculation mechanism They may include spectral 42 GAS ABSORPTION Step Variables and Methods 1 Define line shape Variable abs_lineshape function s to use Methods abs_lineshapeDefine same shape for all species abs_lineshape_per_tgDefine different shapes for different species 2 Read spectral line data Variable abs_lines the order of the first Methods abs_linesReadFromArts abs_linesReadFromSplitArtscat two steps does not abs_linesReadFromHitran abs_linesRead
101. distribution 64 polar coordinate system 16 predefined variables 7 pressure 15 pressure altitude 15 propagation path 73 ptypes 62 radiative background 73 radius 15 ray tracing 73 refractive index 57 report file 11 reporting level 11 retrievals 3 scalar radiative transfer 24 scattering 4 scattering frame 60 scripting language 6 sensor characteristics 31 sensor position 29 sensor transfer matrix 31 sensor the 29 Single scattering properties 60 specific 7 spherical coordinate system 17 surface 4 surface altitude 90 tangent point 83 vector radiative transfer 24 verbosity 11 workspace agendas abs_xsec_agenda 36 blackbody radiation agenda 25 77 doit_conv_test_agenda 127 129 doit mono_agenda 126 127 doit_rte_agenda 127 128 doit scat field agenda 127 iy_cloudbox_agenda 75 99 129 iy_main_agenda 29 72 78 79 99 iy_space_agenda 73 77 99 iy_sub_agenda 78 iy_surface_agenda 75 77 90 99 iy_transmitter_agenda 99 102 134 jacobian_agenda 108 opt_prop_part_agenda 127 pha_mat_spt_agenda 127 ppath_agenda 78 81 100 ppath_step_agenda 81 127 propmat_clearsky_agenda 9 25 36 37 42 56 97 refr_index_air_agenda 57 73 78 81 spt_calc_agenda 127 128 ybatch_calc_agenda 121 workspace methods 6 7 abs_coefCalcFromXsec 42 abs_lines_per_speciesAddMirrorLines 42 abs_lines_per_speciesCompact 42 abs_lines_per_speciesCreateFromLines 42 abs_lin
102. dow We dow RR a 45 6 3 6 Internal hine MiXiNE p ee See ee Bi es a ee we a 48 65 Faraday FOON os soe iia eye eae FE ae ee ee es 51 6 5 8 Absorbing particles candado e ok ee ee aa 51 6 5 9 Further input data and parameters for calculating gas absorption 52 Spectral line dt e ene eesse a a be Sed 52 Tsotopolopule ratios oca coda ee rd a 53 Partition functions oo cs oes ewe ws erered 53 6 6 The gas absorption lookup table o a 53 OOl IMtrOdUCUON ea ee eaa a A AR 53 6 6 2 Lookup table concept o o e 54 Pressure dependence caemos a E eee aa 54 Temperature dependence das 54 Trace gas concentration dependence 54 IMtErpOLatO ii Gk ER a a 55 6 6 3 Workspace variables and methods 55 6 6 4 Format of the lookup table o 56 CONTENTS Ill 6 7 Stand alone gas absorption calculation 56 7 Refractive index 57 Wal ASES ico ee tee he ed ee oA ee teed amp ee ee 58 12 Frec elections sok Ra a Bae ee ee ae Se a ee ER 58 8 Description of clouds 59 Sil INMOAUCUON sua ae gow HM SR a eed Sea oe dh ee Pe de 59 8 2 Single scattering properties o o aG a 60 8 2 1 Coordinate Systems kmo eo eo id ee ea a ete 60 8 2 2 Scattering data structe ene aye alee ee eee ee a 60 38 23 Definitlomol ptypes 200 000 2 0000 ED es 62 macroscopically isotropic 2 2
103. e a single pencil beam calculation The corresponding workspace variable is iy iy consists of one or several i appended H The complete sensor response matrix for a measurement block Can include data reduction The corresponding workspace variable is sensor_response The yCalc method is outlined in Algorithm 1 For further details of each calculation step see the indicated equation or section In summary yCalc appends data from different pencil beam calculations and applies the sensor response matrix H The actual radiative transfer calculations are not part of yCalc Atmospheric radiative transfer is solved for each pencil beam direction line of sight separately It is the task of iy_main_agenda Algorithm 2 to perform a single clear sky radiative transfer calculation This agenda in its turn makes us of other agendas such as ppath_agenda All methods developed for iy_main_agenda adapt automatically to the value of stokes_dim That is yCalc is a common method independent of the details of the radiative trans fer For example yCalc is used both if emission measurements or pure transmission data are simulated that choice is made inside iy_main_agenda The three following sections describes the main calculation steps of iy_main_agenda in the order they are executed 9 2 PROPAGATION PATHS 73 9 2 Propagation paths A pencil beam path through the atmosphere to reach a position along a specific line of sight is denoted as
104. e for randomly oriented particles it makes sense to store the single scattering properties in the so called scattering frame in order to reduce memory requirements The following section describes in detail the SingleScatteringData class The number density field pnd_field contains the number densities of all scattering ele ments at all grid points within the cloudbox Section 8 3 describes how to realize different kinds of size distributions in the ARTS frame by defining appropriate particle number den sity fields pnd_field can be read in from externally prepared data files or be derived from mass density of flux fields provided to the model Section 8 4 below describes the first The latter can be done using pnd_fieldCalcFromscat_speciesFields FIXME document here in AUG History 050913 Created and written by Claudia Emde 60 DESCRIPTION OF CLOUDS 8 2 Single scattering properties 8 2 1 Coordinate systems The laboratory frame and the scattering frame For radiative transfer calculations we need a coordinate system to describe the direction of propagation For this purpose we use the laboratory frame which is shown in ARTS The ory Figure 6 1 The z axis corresponds to the local zenith direction and the x axis points towards the north pole The propagation direction is described by the local zenith angle 9 and the local azimuth angle This coordinate system is the most appropriate frame to describe the propagation direction
105. e three variables It is allowed for all three wind components to set the variable to be empty which is shorthand for saying that the wind is zero throughout the atmosphere Sec 3 8 Otherwise the size of the variable is required to match the atmospheric grids No further input is required a Doppler shift is added as soon as any of the winds is non zero exceptions discussed in Sec 13 4 For clarity even though a setting of wind_u_field always is demanded this wind component has no effect on Doppler shifts for 1D and 2D calculations as the wind moves at an angle of 90 from the observation plane Sec 3 8 13 2 Planet rotation ARTS applies an Earth Centred Earth Fixed ECEF coordinate system This implies that a ground based receiver follows the planet s rotation On the other hand if e g the sensor is placed on another planet or is in a transit orbit the rotation of the observed planet will cause Doppler effects This is treated in ARTS by a short cut the rotational movement can be translated to an imaginary wind by the method wind_u_fieldIncludePlanetRotation History 130223 Restructured to include effects beside winds PE 121218 First version by Patrick Eriksson 96 DOPPLER EFFECTS AND WINDS This pseudo wind v is calculated as 2ncos a r z Uy tp where a is the latitude r is the local planet radius z is the altitude and tp is the planet s rotational period This term is added to the true zo
106. e ARTS is on the screen stdout The meaning of the values is exactly as for the report file The first digit is special It determines how much you will see of the output of agendas other than the main program agenda Normally you do not want to see this output since many agendas are called over and over again in a normal program run The agenda verbosity applies in addition to the screen or file verbosity For example if you set the reporting level to 123 you will get e From the main agenda Level 1 2 outputs to the screen and level 1 3 outputs to the report file e From all other agendas Only level 1 outputs to both screen and report file As another example if you set the reporting level to 120 the report file will be empty The default setting for ARTS if you do not use the command line flag is 010 i e only the important messages to the screen nothing to the report file and no sub agenda output 12 INTRODUCTION Chapter 2 Importing and exporting data Sorry so far just a few words about supported data formats FIXME Extend this chapter 2 1 Data formats 2 1 1 XML files XML is the default file format for exchanging data with ARTS Two flavors are supported Plain text and binary In the plain format all data is stored in the XML file For binary the structure of the data is stored in the XML file and the data itself in binary format in a separate file 2 1 2 NetCDF files NetCDF
107. e can be defined through abs_lineshapeDefine abs_lineshape Faddeeva_Algorithm_916 VVH 750e9 48 GAS ABSORPTION 6 5 6 Internal line mixing Line mixing is implemented as the LM species tag Added to a species ARTS knows that it should look for lines with line mixing data attached to them when it calculates the spectral cross sections Note that each line is initialized without any line mixing data and with a tag state that says they are not to be line mixed If this is not changed before the absorption calculations are performed the line will regardless of species tag still act like if it were not line mixed Also note that it is possible to calculate line mixing in conjuncture with the Zeeman effect but that it is not possible to calculate the line mixing of individual Zeeman lines which as far as we know when writing this is unimportant The possible tags on species X that activates the line mixing module are thus X LM x and X Z LM The supported ways to ensure that the line record contains information on how to calculate line mixing thus not ignoring it are e Read an ARTS catalogue file with line mixing parameters e Read LBLRTM line catalogue containing line mixing parameters using abs_linesReadFromLBLRTM gt e Read a line database that does not have line mixing but use the line_mixing_dataMatch method to inject line mixing data into the line record There is no preferred method in ARTS each wo
108. e dependence The pressure dependence is the most important dependence of It comes from the fact that the width of the line shape functions is governed by pressure broadening We have to store the x on some pressure grid and interpolate if we need them for intermediate values Temperature dependence This is the next effect to take into account Both the line widths and the line intensities depend on temperature Of course only certain combinations of pressure and tempera ture occur in the Earth s atmosphere Hence storing the k in a two dimensional table as a function of pressure and temperature would waste a lot of memory and computa tion time Instead they are stored for a reference temperature and set of temperature perturbations for each pressure level E g if the set of perturbations is 10 0 10 then the k would be stored for three different temperatures for each pressure level Tr p 10K Tr p Tr p 10K where Tp p is the reference temperature for each pressure level Trace gas concentration dependence This is a second order effect The width of the line depends not only on total pressure but also on the partial pressure of one or more trace gases In theory this is always the case because the broadening is different for each combination of collision partners However in practice trace gas concentrations in the Earth s atmosphere are normally so low that this can be safely neglected An important exc
109. e fields varies for gases it is VMR while particle mass concentrations are given in kg m and electron density in m Particles This second class treats all matter causing significant scattering and likely also adding to the absorption The amount of scattering matter is given as particle number density fields m denoted as pnd field The pnd field is provided per scattering element see Chapter 8 for definition The corresponding optical properties of the particles are given by scat_data containing one set of single scattering properties for each scattering element Particles can be grouped into scattering species each char acterized by a mass density kg m or flux kg m s field that is converted into particle number density fields before solving the radiative transfer equation Atmospheric quantities are not hard coded to belong to any of these matter classes as the practical division depends on the conditions of the simulations The general rule is that for the shorter the wavelengths a higher number of atmospheric constituents must be treated as particles For thermal infrared and microwave calculations molecules and electrons and likely also aerosols can be treated as absorbing species It can also be possible to place some hydrometeors in this class For example non precipitating liquid clouds can be treated as purely absorbing in the microwave region Related to the division between absorbing species and particles is
110. e loss discussed above 15 2 5 Extra path delay In the absence of an atmosphere a signal sent from a transmitter in the direction towards a receiver would follow a straight line i e the shortest geometric distance between the two instruments Moreover the signal would propagate with the speed of light The geometric distance lg between the transmitter and receiver is f 1ds 15 8 geometric In the presence of an atmosphere two changes follow the speed of the signal is retarded and the ray path gets bent The optical path length is defined as l i n s ds 15 9 ray The geometrical length of the bent ray path is l j lds 15 10 ray To recall what you probably heard during our physics lessons some quotes from the Wikipedia entry on optical path length Fermat s principle states that the path light takes between two points is the path that has the minimum optical path length An electromag netic wave that travels a path of given optical path length arrives with the same phase shift as if it had travelled a path of that physical length in a vacuum The total delay compared if there would be vacuum between transmitter and receiver expressed in units of length is d lo ly gt 0 15 11 104 TRANSMISSION CALCULATIONS Figure 15 1 Illustration of the bending angle of a satellite to satellite radio occultation copied from Schreiner et al 1999 Fig 2 which can be expressed also as d lo lr lr
111. e numerical manner by perturbing the temperature Eq 9 4 gives 16 28 OT _ A141 a o gt 16 29 16 6 ATMOSPHERIC TEMPERATURES 117 The path length Al for a given pressure is linearly proportional to the temperature and if T is the average temperature along the path step OAl Al gt 16 aT T 16 30 Following the other variables we set T T T 1 2 and Al Ali 16 31 In summary assuming hse set to on Ol __ Al Toa Qi Qi 1 ae OB e B TI 1 e 16 32 Ox 2 Ox 2T i 7 e Ox The derivative of the Planck function 0B Oxz can be expressed analytically Eriksson et al 2002 but as B is provided by an agenda related terms must anyhow be determined in a pure numerical manner For example if the Rayleigh Jeans approximation is for some reason applied B T the term is just 1 The expression for higher Stokes elements or if emission is totally ignored is obtained by setting B and 9B 0x to zero General case The term OT Ox is calculated in a pure numerical manner This is done in two steps In the first step the transmission matrix is recalculated for a slightly perturbed temperature If hydrostatic equilibrium is considered the transmission is also recalculated with a dis turbed Al To avoid making two identical disturbances of the transmission matrix these calculations are performed as av aT T Ox
112. eatment of the transmitter should be to include p and g in p In any case iyRadioLink expects iy_transmitter_agenda to return p for each frequency in f grid 15 2 RADIO LINK CALCULATIONS 103 So far ARTS has no dedicated support to handle characteristics of receivers in radio link configurations The first step here should be to consider Ae but the simplest way to achieve this is to include Ae in p and it is here where the definition of free space loss comes into play as with the standard definition p should instead include gy Some functionality developed for passive sensors could also be of use for radio links but this has not yet been tested practically even less validated and these options are not commented further Setting I p to be consistent in the nomenclature used elsewhere and taking the derivative of I s gives ats 1 s 15 6 This equals the Beer Lambert law with an attenuation coefficient of 2 s With other words the free space loss can be also treated as an ordinary extinction term and when free space attenuation k f is requested as an auxiliary variable it is calculated as kp 15 7 This extinction coefficient can be seen as non linear as it varies with the distance from the transmitter If the transmitter is placed inside the atmosphere k becomes infinity for the path point corresponding to 1 0 The kf coefficient is not dependent on the definition differences of total free spac
113. ecause it predefines agendas with settings suitable for many applications FIXME Revise this section after general arts has been changed 1 5 Test controlfiles The directory controlfiles in the ARTS distribution contains some test and example controlfiles You should study them to learn more about how the program works You can run these controlfiles like this arts TestAbs arts This assumes that you are in the directory where the control file resides and that the art s executable is in your path Alternatively you can run a standard set of the test controlfiles by going to the build directory and say make check This standard set of test is run by us on every automatic build that means every time a new ARTS version is submitted to the subversion repository If your ARTS installation is healthy make check should run through without any errors See file README in the ARTS distribution for detailed instructions on how to build ARTS FIXME The tests are now structured differently Update the text 1 6 Verbosity levels The command line parameter 1 6 VERBOSITY LEVELS 11 artis E or arts reporting can be used to set how much output ARTS produces You can supply a three digit integer here Each digit can have a value between 0 and 3 The last digit determines how verbose ARTS is in its report file If it is O the report file will be empty if it is 3 it will be longest The middle digit determines how verbos
114. ed in Section 9 2 The path steps are normally from one crossing of the atmospheric grids to next To introduce propagation paths steps was necessary to handle the iterative solution for scattering inside the cloud box as made clear from Figure 8 2 of ARTS Theory A full propagation path is stored in the workspace variable ppath that is of the type Ppath see Section 10 5 The paths are determined by calculating a number of path steps A path step is the path from a point to the next crossing of either the pressure latitude or longitude grid Figure 10 1 There is one exception to this definition of a path step and that is when there is an intersection with the surface which ends the propagation path at that point The starting point for the calculation of a path step is normally a grid crossing point but can also be an arbitrary point inside the atmosphere such as the sensor position The path steps are stored in the workspace variable ppath_step that is of the same type as ppath Propagation paths are calculated with the internal function ppath_calc The commu nication between this method and ppath_step_agenda is handled by ppath_step That variable is used both as input and output to ppath_step_agenda The agenda gets back ppath_step as returned to ppath_calc and the last path point hold by the structure is accordingly the starting point for the new calculations If a total propagation path shall be determined the agenda is called repeatedly
115. ee space loss Please note that other definitions of this quantity exists see Section 15 2 4 With these definitions Equation 15 2 can be written as pr l tatet pp 15 5 where p pigtAc a term discussed further in Section 15 2 4 The method iyRadioLink reports p as iy i e p is treated as the main quantity All loss terms tq te and tf can be obtained as auxiliary data iy_aux The auxiliary variables include also extra path delay bending angle and Faraday rotation Some of these terms can be obtained as attenuation rotation delay per length along the propagation path Note the distinction between loss and attenuation The first term refers here to path integrated transmission factors while the second term refers to local attenuation co efficients That is losses are dimensionless values between O and 1 while the unit of attenuation is 1 m See the following sections for details The quantity returned in iy is also what ends up in y outputted by yCalc For increased flexibility the method iyReplaceFromAux allows the user to replace the content of iy with some of the auxiliary variables For example for radio link calculations bending angle can frequently be seen as the primary observation and this case can be handled by iyReplace FromAux The constrain for allowing this swap of variables is that the auxiliary variable to replace p must have the same data dimensions In practice this means that the auxiliary va
116. een nominal and actual viewing direc tion of the sensor So far only deviations in zenith angle can be considered The workspace method to initiate such Jacobians is jacobianAddPointingZa The pointing deviation is treated as a time varying variable then having a polynomial variation hence the basis functions described in Section 16 2 2 are applied The time is taken from sensor_time If the pointing error is assumed to be constant with time the polynomial order to select is 0 and so on As a special case the polynomial order 1 signifies here that the pointing off set is so highly varying that an off set must be assigned to each spectrum sometime called pointing jitter The Jacobian can be calculated in two manners recalc If this option is selected radiative transfer calculations are performed for a shift of sensor_los perturbation size selected by dza Only a one sided perturbation is applied interp The Jacobian is derived from existing data by an interpolation of existing data This achieved by interpolating pencil beam data to a shifted zenith angle grid This will involve some extrapolation of the data and this aspect should be considered when selecting the zenith angles in mblock_dlos_grid The average of a positive and negative shift is determined The shift to apply dza should be smaller than the minimum spacing of the zenith angles inmblock_dlos_grid for accurate results As interpolation is a relative fast oper
117. efellipsoid 19 87 refr_index_air 57 refr index_air_group 57 rte_alonglos_v 96 rte_los 31 78 99 rte_pos 29 78 99 rte _pos2 100 rtp_mag 21 scat_aa grid 126 scat_data 25 26 52 59 scat_za_grid 126 sensor_los 31 99 sensor_norm 94 sensor_pos 29 99 sensor_response 31 72 93 sensor time 118 stokes_dim 23 72 97 surface_emission 90 surface_los 90 surface _rmatrix 90 t_field 18 transmitter_pos 100 vmr_field 18 24 39 41 97 wind_u_field 21 95 wind_v_field 21 95 wind_w_field 21 95 y 27 72 77 yf 79 y los 79 y pol 79 y pos 79 ybatch index 121 ybatch_n 121 ybatch_start 121 INDEX 149 z field 18 19 z_surface 19 90 zenith angle 30
118. ensor line of sight and block grid values are just added there is an ambiguity of the line of sight It is possible to apply a constant off set to the line of sights if the block grids are corrected accordingly For example if the simulations deal with limb sounding and a 1D atmosphere where normally a single block should be used despite a number of spectra are recorded it could be practical to set the line of sight to the viewing direction of the uppermost or lowermost spectrum and the zenith angles in mblock_dlos_grid will not be centred around zero which is the case when the true line of sight is used Chapter 13 Doppler effects and winds The default assumption in ARTS can be expressed as the atmosphere is assumed to be static and the observation platform is static during each measurement while any relative move ment between the atmosphere and the sensor will cause Doppler effects ARTS handles three sources to Doppler shifts winds rotation of the planet and sensor velocity 13 1 Winds Atmospheric transport is not considered by ARTS but winds can still be of importance due to the Doppler effect they can cause This effect is most significant at high altitudes where the line shape is narrow and a frequency shift of absorption and emission is most easily discerned The workspace variables to specify winds are wind_u field wind_v field and wind_w_field below denoted as vy Uy and vy respectively The user need to set all thes
119. ently Hund case a indicated by number 0 also works for some molecules It is up to the user to assure the correct quantum numbers for each case are set properly The supported molecules in ARTS so far for Zeeman calculations are e O case b NO case a b e OH case a CIO case a HO case b e NO case b but only O is tested beyond initial modeling results Beside the additional line parameters it is also necessary to input the magnetic field into the model This can be done following GriddedField3Create B ReadXML B B_1comp xml gz GriddedFieldLatLonRegrid B lat_grid lon_grid B GriddedFieldPRegrid B p_grid B FieldFromGriddedField mag_w_field p_grid lat_grid lon_grid B where B_1comp xml gz contains the atmospheric profile or field for one components of the magnetic field Since the magnetic field is usually not dependent on the pressure it is also possible to use the GriddedFieldZToPRegrid x functionality to input altitude gridded magnetism Input needs to be provided for each non zero magnetic field component separately into the respective WSVs mag_w_field mag_v_field and mag_u_field For more information on magnetic field format in ARTS see Section 3 8 Lastly it is necessary to take phase effects into account when calculating the Zeeman effect It is therefore important that abs_lineshape is chosen such that it allows for this information to be provided One suitable line shap
120. eption is water vapor in the lower troposphere which According to C Melsheimer Beer s law is The taller the glass the darker the brew the less the amount of light that comes through He might have been quoting someone else there but I do not know whom 6 6 THE GAS ABSORPTION LOOKUP TABLE 55 can reach quite high volume mixing ratios Therefore the effect of water vapor mixing ratio on water vapor absorption self broadening as well as on oxygen absorption for example according to the parameterization by Rosenkranz 1993 may not be negligible This is handled by storing water vapor perturbations In contrast to the temperature case the water vapor perturbations are multiplicative not additive Hence if the set of per turbations is 0 1 10 then the x would be stored for three different H2O VMRs for each pressure temperature grid point 0 VMRr p T 10x VMRr p T where VMRr p T is the reference water vapor VMR for each pressure temperature grid point Interpolation The interpolation scheme is quite important for the accuracy of the lookup table In par ticular higher order interpolation gives considerably better accuracy for the same table grid spacing The interpolation orders in the ARTS implementation of the lookup table can be chosen by the user The settings that are recommended and set as defaults in file general arts are quite high interpolation orders of 5 7 and 5 for pressure temper ature
121. er it is determined by the transmitter position A characterisation of a radio link normally involves several attenuation terms not encountered for passive measurements such as free space loss and defocusing Faraday rotation Sec 14 is an additional physical mechanism that is of special concern for active microwave devices it can normally be neglected at the higher frequencies used for passive observations 15 1 Pure transmission calculations This section discusses the iyTransmissionStandard workspace method that is relevant if you want to calculate the transmission through the atmosphere for a given position and line of sight The set up is largely the same as for simulations involving emission such as that the observation geometry is defined by sensor_pos and sensor los or rte_pos and rte_los if iyCalc is used The first main difference is that iy_transmitter_agenda replaces ly space_agenda and iy_surface_agenda The second main difference is that handling of cloud scattering is built in and the need to define iy_cloudbox_agenda vanishes These dif ferences appear inside iy_main_agenda Further blackbody radiation agenda can be left undefined As for emission measurements Sec 9 1 Algorithm 2 the first main operation is to determine the propagation path through the atmosphere but this is here done without con sidering the cloud box it is simple deactivated during this step The possible radiative backgrounds are accordingl
122. er methods WriteXML The list at the bottom of the documentation shows that output_file_format is a specific input argument and that v and filename are generic input arguments What this means is that output_file_ format already automatically exists as a variable whereas v and filename do not The built in documentation provides descriptions also of these generic arguments and lists the allowed values The predefined variables combined with specific method arguments are meant to help in combining methods into meaningful calculations Predefined variables are typically rel evant for more than one method For example variable output_file_format can be used to change the format of all produced files at the same time However the use of a specific vari able in the controlfile is not mandatory so WriteXML output_file_format v cest xml WriteXML ascii v test xml and WriteXML my_format v test xml are all allowed But in the last example the variable my_format must have been defined before Besides the variable names the built in documentation also lists the allowed vari able groups or types In the example the groups for workspace variable v are Any which means that v can belong to any of the known groups The group for filename is String which means that a string is expected here Method arguments can be a literal as in WriteXML ascii v test xml or a variable as in Write
123. er latitudes Uy A positive wind is defined as air moving upwards i e towards higher altitudes As described above one two or all of these variables can be set to be empty if the corre sponding wind component is zero Winds affect the radiative transfer by inducing Doppler shifts see further Chapter 13 To consider Faraday rotation Sec 14 and Zeeman splitting Sec 6 5 5 also the mag netic field must be specified The three component fields are mag_u_field mag_v_field and mag_w field All three components must be specified but can be set to zero for a part of or the complete atmosphere However some component can be irrelevant for the calcu lations For example the u component has no influence on Faraday rotation for 1D and 2D cases The internal representation of the magnetic field at a specific point is handled by rtp_mag For this variable the three components are stored together and thus the local magnetic field is represented as a vector 22 DESCRIPTION OF THE ATMOSPHERE Chapter 4 Radiative transfer basics This chapter introduces some basic radiative transfer nomenclature and equations The ra diative transfer equation is here presented in general terms while special cases and solutions are discussed in later parts of the document 4 1 Stokes dimensionality The full polarisation state of radiation can be described by the Stokes vector and is the formalism applied in ARTS The vector can be defined in
124. ere Further more Richard et al 2012 point out that for molecules with more than two atoms further mechanisms affecting CIA exist which are not covered by the simple models used They hence state that these data should be used very carefully Conclusion No changes but use with care Also note that no data exists at frequencies below 30 GHz 1 cm though some sig nificant absorption is still present at the limiting frequency For those low frequencies the CO2 SelfContPWR93 continuum see Table 6 3 can be used as an alternative we estimated that to be valid at least up to about 100 GHz but deviating significantly above 500 GHz 02 N2 O2 CO2 These UV Vis only datasets were removed 6 5 5 Zeeman calculations The Zeeman effect is calculated in the method propmat_clearskyAddZeeman If this method is included in the propmat_clearsky agenda then species with the tag Z will be calcu lated as Zeeman species If the method is not included then these species will simply be ignored Note that the order within the tag string is important the Zeeman tag must directly follow the molecular species tag That is 02 Z 66 will be counted as Zeeman splitting on the O16016 molecule whereas 02 66 Z will not work the same way or even at all The physics and internal workings of the Zeeman calculations follow the scheme pre sented in Larsson et al submitted 2013 In order to calculate Zeeman splitting additional line parameters th
125. ereas in doit conv flagAbs they must be defined in the basic radiance unit W m HBz sr Another option is to perform a least square convergence test using the workspace method doit conv flagLsq Test calcula tions have shown that this test is not safe therefore the least square convergence test should only be used for test purposes AgendaSet doit_conv_test_agenda doit_conv_flagAbsBT doit_conv_flag doit_iteration_counter doit_i field doit_i field old f grid f_index 0 1 0 01 0 01 0 01 Alternative Give limits in radiances doit_conv_flagAbs doit_conv_flag doit_iteration_counter doit_i field doit_i field old 0 le 15 0 1le 18 0 le 18 0 le 18 If you want to look at several iteration fields for example to investigate the convergence behavior you Can use the following workspace method Se de de de de de e e e Se 2 4 18 2 3 Propagation of the DOIT result towards the sensor In order to propagate the result of the scattering calculation towards the sensor the fields needs to be interpolated on the direction of the sensor s line of sight This is done in the workspace method iyInterpCloudboxField which has to be put into the agenda iy_cloudbox_agenda AgendaSet iy_cloudbox_agenda iyInterpCloudboxField DoitWriteIterationFields doit_iteration_ counter doit_i field 130 SCATTERING CALCULATIONS THE DOIT MODULE 18 3 3D DOIT calculations The DOIT method is imp
126. erred from the start of the tag name See ARTS Theory Chapter 2 for more information on the various models 6 5 CALCULATING GAS ABSORPTION 45 This will usually be the file hitran_cia2012_adapted xml1 gz which is included in the arts xml data but the original HITRAN data files can also be read Finally use WSM abs_xsec_per_speciesAddCIA in abs_xsec_agenda to add the CIA absorption For usage examples look in directory controlfiles artscomponents cia thatis part of the ARTS distribution Figure 6 5 shows all CIA continua that are currently available in ARTS left and sep arately the ones that are relevant for Earth s atmosphere right The valid frequency and temperature ranges for these data as available in ARTS are listed in Table 6 4 Outside the covered frequency ranges the binary absorption cross sections are set to zero while exceeding the valid temperature range will produce NaN values and eventually trigger a runtime error To make the HITRAN data work in ARTS some modifications were necessary specifi cally N2 N2 The two high frequency datasets were merged into one O gt O Three apparently separate datasets that really belong together were merged UV Vis dataset were removed CO2 CO2 Caveat This is only the self continuum of CO2 The CO2 air continuum has strong features above 250 cm that are present in CKD_MT also available in ARTS as one of the continuum and full absorption models but are missing h
127. es for example radio link calculations require special consideration Anyhow the main message here is that by using the agenda concept a very high degree of flexibility can be achieved and new features can be added fairly simply On the other hand the concept require that the user actually apply methods that make sense for the agenda The code of ARTS performs some consistency checks of the agenda output but this can only catch some types of mistakes Complete radiative transfer calculations are normally performed by yCalc This method incorporates sensor responses and has the variable y as main output The letter y here refers to the measurement vector y found in the formalism of Rodgers 1990 2000 see also Sec 1 3 of ARTS Theory The vector can hold anything from a single value to a high number of spectra appended The spectra in the last case can correspond to a limb sounding sequence hence measured for different zenith angles or even be measured by different sensors In any case the data in y contain likely significant impact of different parts of the sensor used such as the angular weighting by the antenna pattern On the other hand atmospheric radiative transfer is performed for monochromatic fre quencies along pencil beam directions The outcome of one such calculation is the Stokes vector for each frequency considered and as workspace variable this quantity is denoted as iy Please note the distinction to y that can contain
128. es of units each having its own transfer matrix There is only one compulsory transfer matrix and it is sensor_response There are several workspace variables associated with this transfer matrix where an tenna_dim and mblock_dlos grid are the compulsory ones The variable antenna_dim gives the dimensionality of the antenna pattern where the options are 1 and 2 standing for 1D and 2D respectively A 1D antenna dimensionality means that the azimuth extension of the antenna pattern is neglected there is only a zenith angle variation of the response A 2D antenna pattern is converted to a 1D pattern by integrating the azimuth response for each zenith angle See further Chapter 12 where inclusion of sensor characteristics is described in 5 3 Measurement sequences and blocks The series of observations modelled by the simulations is denoted as the measurement se quence That is a measurement sequence covers all spectra recorded at all considered sensor positions A measurement sequence consists of one or several measurement blocks A measurement block can be treated as a measurement cycle that is repeated an integer number of times to form the measurement sequence A measurement block covers one or several recorded spectra depending on the mea surement conditions and the atmospheric dimensionality A block can consist of several spectra when there is no effective motion of the sensor with respect to the atmospheric fields It should be noted
129. es_per_speciesReadFromCatalogues 42 abs_lineshape_per_tgDefine 42 abs lineshapeDefine 42 abs _linesReadFromArts 42 abs linesReadFromHitran 42 abs _linesReadFromHitranPre2004 42 abs linesReadFromJpl 42 abs_linesReadFromMytran2 42 abs linesReadFromSplitArtscat 42 abs _lookupAdapt 42 55 abs _lookupCalc 39 42 55 abs _lookupSetup 55 abs_lookupSetupBatch 55 abs_lookupSetupWide 55 abs_speciesSet 39 41 43 abs_vecAddPart 128 abs_xsec_per_speciesAddCIA 42 45 abs_xsec_per_speciesAddConts 42 43 abs_xsec_per_speciesAddLines 42 abs_xsec_per_speciesInit 42 AtmosphereSet1D 16 AtmosphereSet2D 16 AtmosphereSet3D 16 INDEX 147 blackbody radiationPlanck 77 cloudboxOff 20 cloudboxSetManually 67 cloudboxSetManually Altitude 67 doit_conv_flagAbs 129 doit_conv_flagAbsBT 129 doit_conv_flagLsq 129 doit_i_field_monolterate 127 128 doit_i_fieldSetClearsky 126 doit_i_fieldSetConst 126 doit_i_fieldUpdate1D 128 doit_i_fieldUpdateSeq1D 128 doit_i_fieldUpdateSeq1DPP 128 doit_i_fieldUpdateSeq3D 130 doit_scat_fieldCalc 128 doit_scat_fieldCalcLimb 128 doit_za_grid_optCalc 126 DoitAngularGridsSet 126 128 DoitCalc 126 DoitGetIncoming 126 DoitInit 126 DoitScatteringDataPrepare 127 128 ext_matAddPart 128 Extract 122 isotopologue_ratiosInitFromBuiltin 53 iyApplyUnit 77 78 iyCalc 29 78 iyCloudRadar 134 iyEmissionStandard 29 76 79 1yInterpCloudboxField 129 iyLoopF
130. esides in optproperties h 8 2 SINGLE SCATTERING PROPERTIES 61 String ptype An attribute which contains information about the data type which is a classification of the scattering particle specifically regarding its symmetry prop erties randomly oriented horizontally aligned general case This attribute is needed in the radiative transfer function to be able to extract the physical phase ma trix the physical extinction matrix and the physical absorption vector from the data Possible values of ptype are general macroscopically_isotropic horizontally_aligned A more detailed description of the different cases is given below String description Here the scattering element can be characterized explicitly For example information on the size and shape of the particle or the respective dis tributions of a particle ensemble might be given This can be a longer text describing how the scattering properties were generated It should be formatted for direct print out to screen or file Vector _grid Frequency grid Unit Hz Vector T_grid Temperature grid Unit K Vector za_grid 1 general Zenith angle grid Range 0 0 lt za lt 180 0 2 horizontally_aligned Scattering angle grid Range 0 0 lt za lt 180 0 Vector aa_grid Azimuth angle grid 1 general Range 180 0 lt aa lt 180 0 2 macroscopically_isotropic Not needed since optical properties depend only on
131. essed as s s T Tis 1 T bi 16 15 where s is the Stokes vector for all emission generated between the sensor and point 1 T is the transmission Mueller matrix for the same part of the propagation path and T is defined in Eq 9 8 The quantities s T and s are not function of x This gives Os 1 OT ab Ox 16 16 The terms OT Ox and Ob Ox are both calculated in a pure numerical manner for dif ferent reasons discussed below 16 3 ATMOSPHERIC VARIABLES COMMON EXPRESSIONS 113 16 3 4 Including the surface For scattered down welling radiation the effective transmission matrix is T T gt R T 16 17 where T is the transmission between the surface and the sensor R is defined in Eq 11 7 and T is the transmission between the point 7 1 and the surface 16 3 5 0s 0x locally unpolarised absorption Eq 16 16 can be simplified for some conditions A first case is scalar radiative transfer i e only the first element of the Stokes vector is considered and all terms of Eq 16 16 are scalar quantities For all other cases in principle vector radiative transfer must be performed as T can always have off diagonal elements Even if atmospheric absorption is totally unpo larised T can be non diagonal due to the surface Eq 16 17 However if the absorption locally lacks polarisation the calculations can be handled by analytical expressions in a higher degree To focus o
132. et al New section of the HITRAN database Collision induced absorption CIA Journal of Quantitative Spectroscopy and Radiative Transfer 113 1276 1285 2012 Rodgers C Inverse methods for atmospheric sounding Theory and practise 1st ed World Scientific Publishing 2000 Rodgers C D Characterization and error analysis of profiles retrieved from remote sound ing measurements Journal of Geophysical Research 95 5387 5593 1990 Rosenkranz P W Absorption of microwaves by atmospheric gases in Atmospheric remote sensing by microwave radiometry edited by M A Janssen pp 37 90 John Wiley amp Sons Inc 1993 ftp mesa mit edu phil lbl_rt Rybicki G B and A P Lightman Radiative processes in astrophysics chap Plasma effects John Wiley and Sons Inc USA 1979 Schreiner W S S Sokolovskiy V C Rocken and D C Hunt Analysisand validationof gps met radio occultation data in the ionosphere Radio Science 34 949 966 1999 Wright P S Quegan N Wheadon and C Hall Faraday rotation effects on l band space borne sar data IEEE Transactions on Geoscience and Remote Sensing 41 2735 2744 2003 142 BIBLIOGRAPHY Part VI Index Index 1D 16 2D 16 3D 17 agenda 9 antenna pattern dimensionality 31 ARTS 3 ARTS files agendas arts 10 37 39 continua arts 43 controlfiles 10 41 controlfiles artscomponents absorption 52 controlfiles artscomponents cia 45 controlf
133. etch can be assumed to be time varying where exactly the same polynomial approach as for pointing is applied This including the case of setting the order to 1 16 9 POLYNOMIAL BASELINE FIT 119 16 9 Polynomial baseline fit A baseline is microwave jargon for a disturbance of the spectrum that is not covered by the common sensor characteristics The most common case is that the local oscillator signal leaks into the measurement by reflections occurring inside the sensor causing a pattern in the spectrum of standing wave type Such effects are difficult to model in a physical manner and a more general fitting procedure must be applied A common option is then to model the baseline as a polynomial of a specified order That is assuming a measurement giving a single spectrum the measured spectrum y is modelled as y y tizi 16 33 i 0 where y is the baseline free spectrum and x and z are introduced in Section 16 2 2 The Jacobian for such baseline models are obtained by jacobianAddPolyfit For single spectra measurements the only consideration is the polynomial order to use For mea surements where several spectra are appended to form the measurement vector the default option is that baseline can vary between all spectra In some cases it could be assumed that the baseline is common between data for different polarisations viewing directions or measurement blocks and flags can be set to mimic such assumptions
134. for 2D and 3D In short the paths can only enter and leave the model atmosphere at the top of the atmosphere as the atmospheric fields are treated to be undefined outside the covered latitude and longitude ranges Controlled by ppath_step_agenda propagation paths can be calculated purely geomet rically or considering refraction When considering refraction the refractive index is de termined at each point along the path according to refr_index_air_agenda Details about different methods applicable within refr_index_air_agenda are given in Chapter 7 If nothing else is stated it assumed that all frequency components share a single prop agation path Another way to express this assumption is that dispersion is neglected See Section 9 7 for how to consider dispersion In the non dispersive case the propagation path is valid for average of the first and last element in f_grid as this is the frequency given to refr_index_air_agenda Propagation paths can be calculated separately by the method ppathCalc but for stan dard calculations the propagation paths are calculated internally by yCalc Methods and variables to control the path calculations are discussed in Section 10 1 9 3 The radiative background The radiative intensity at the starting point of the path and in the direction of the line of sight at that point is denoted as the radiative background Four possible radiative back grounds exist Space When the propagation path starts a
135. formulas are given in Section 4 1 1 of ARTS Theory refr_index_airThayer calculates the microwave air refractivity in the Earth s atmo sphere taking into account refractivity of dry air and water vapour All other gases are assumed to have a negligible contribution More details are given in Section 4 1 2 of ARTS Theory refr_index_airIR derives the infrared air refractivity in the Earth s atmosphere consid ering only refractivity of dry air More details are given in Section 4 1 3 of ARTS Theory 7 2 Free electrons Free electrons as exist in the ionosphere affect propagating radio waves in several ways Free electrons will have an impact of the propagation speed of radio waves hence a signal can be delayed and refracted This section consideres only the refraction effect neglecting influences of any magnetic field For effects on polarisation state of the waves in presence of a static magnetic field 1 e Faraday rotation see Section 14 refr_index_airFreeElectrons derives this contribution of free electrons to the refractive index The method is only valid when the radiative transfer frequency is large enough at least twice the plasma frequency Information on theoretical background and details on the applied formulas are provided in Section 4 2 of ARTS Theory Reducing the refrindex_airThayer approach to inverse temperature proportionality as applied by refr_index_airMW general causes significant deviati
136. frequencies FIXME Describe where each variable is used Expand History 120918 Started Patrick Eriksson 58 REFRACTIVE INDEX 7 1 Gases For calculating the contribution to the refractive index from atmospheric gases the fol lowing workspace methods are currently available in ARTS refr_index_airMW general refr_index_airThayer and refr_index_airIR All of them are non dispersive i e monochro matic and group refractive index are identical They are supposed to be applied as alterna tives not in addition to each other refr_index_airMW general provides refractivity due to different gas mixtures as occuring in planetary atmospheres and is valid in the microwave spectral region It uses the method ology introduced by Newell and Baird 1965 for calculating refractivity of the gas mixture at actual pressure and temperature conditions based on the refractivity of the individual gases at reference conditions Reference refractivities from Newell and Baird 1965 are available for N O2 CO2 H2 and He Additionally reference refractivity for H20 has been derived from H20 contribution as described by refr_index_airThayer see below for a reference temperature of Ty 273 15 K Any mixture of these gases can be taken into ac count The missing contribution from further gases is roughly accounted for by normalising the calculated refractivity from the six reference gases to a volume mixing ratio of 1 More details on the applied
137. ft The spatial position where each of the two shifted paths have an optical path length of lo is determined The distance between these two points are compared to what is expected for free space propagation If for example the distance is double as high compared to the free space case this is taken as a defocusing transmission of 0 5 The factor can be higher than 1 which corresponds to conditions of focusing Only the zenith angle dimension is considered any defocusing focusing in the horizontal plane is neglected which in general should be a good approximation The method can be applied for all possible links e g links inside the atmosphere for ground to satellite links and satellite to satellite links In the last case if the path with the highest zenith angle intersects with the ground the defocusing is calculated using the nominal path and the low zenith angle path Defocusing method 2 This method is restricted to satellite to satellite links The defocusing effect in this case of a transmitted signal with intensity Jp can be described as Haugstad 1978 Kursinski et al 2000 I IoMa Mg 15 15 where Ma and Mg is the vertical and horizontal defocusing factors respectively In general the defocusing effect is fairly complex in a modelling perspective although for special cases the defocusing factors can be calculated analytically Such analytic expressions can be used for test cases of more general solutio
138. gendas propmat_clearsky_agenda Bottom left Controlfile agenda definition for line by line absorption calculation Bottom right Controlfile agenda definition to extract absorption from a pre calculated lookup table arguments will be set to their default values WriteXML in v is equivalent to calling WriteXML output_file_format v Note that named arguments can not be mixed with positional arguments One additional rule has to be mentioned here If all arguments to a method are specific and the user wants to use all the predefined variables then the entire argument list including parentheses may be omitted 1 3 3 Agendas Agendas are a special group of workspace variables which allow to modify how a calcula tion is performed A variable of group agenda holds a list of workspace method calls It can be executed which means that the method calls it contains are executed one after another Figure 1 4 gives an example for the agenda propmat clearsky agenda Several radia tive transfer methods use this agenda as input variable When they need local absorption coefficients for a point in the atmosphere they execute the agenda with the local pressure temperature and trace gas volume mixing ratio values as inputs The agenda then provides absorption coefficients as output The bottom of Figure 1 4 shows two different ways how this agenda could be defined in 10 INTRODUCTION the controlfile In the first case a line by line absorp
139. gth 1 The specular direction is calculated by the internal function surface_specular_los Equations 6 72 6 74 in ARTS Theory give the values of surface rmatrix and surface emission Any tilt of the surface is neglected when determining the specular direction If there would be any need to consider surface tilt almost complete code for this task existed in surface_specular_los but was removed in version 1 1 876 The code can be obtained by e g checking out version 1 1 875 11 3 SURFACE RADIATIVE PROPERTIES 91 E d a A S i I d e D I 1 propagation path reflection 7 patterns Figure 11 4 Schematic of Equation 11 7 11 3 3 Lambertian surface A basic treatment of Lambertian surfaces is provided by the method surfaceLambertianSim ple This method assumes that the down welling radiation has no azimuthal dependency which fits the assumptions for 1D atmospheres The number of angles to apply in sur facelos is selected by the user For a Lambertian surface the reflected radiation is unpolarised thus independent of the polarisation of the down welling radiation That is each surface _rmatrix has the structure w 00 0 000 0 Bln do 11 8 000 0 When determining the weight w above the method assumes that the down welling ra diance J is constant inside each zenith angle range 04 0b Hence w equals cf Equa tion 6 75 of ARTS Theory Os ho w I cos 0 f 6 0 01 Q sin 0 do d 11 9
140. h r is then given by dtr e3 n ds 8n2cegm2v2 s Bgeo s 14 2 An magneto optical effect Sec 4 2 1 of this type is mapped to a propagation matrix as 0 0 0 0 0 0 20 Kee ooo a al 14 3 0 0 00 If no other effects are present the effect on the Stokes vector for some part of the propgation path can be expressed as a Mueller rotation matrix Goldstein 2003 Meissner and Wentz 2006 Ip 1 0 0 0 I Qr _ 0 cos 20p sin 20p 0 Q 14 4 Up 0 sin 29p cos 2Yp 0 U Vr 0 0 0 1 V The two versions of this equation are also found in Wikipedia under Faraday effect but expressed using A and an error for the numerical factor 2012 12 17 The value 2 365 10 derived from the fundamental constants is confirmed by Wright et al 2003 Chapter 15 Transmission calculations The term transmission calculations refers here to situations when the emission from the atmosphere and surface can be neglected These calculations can be divided into two main types The first one is when just the transmission of the atmosphere is diagnosed Sec 15 1 The observation geometry is then given exactly as for emission simulations by a position and a line of sight The second main type is radio link budgets Sec 15 2 For this case the propagation path is determined solely by the position of the transmitter and the receiver That is the user does not need to set a line of sight of the sensor receiv
141. hat are aligned with respect to polar angle specifically in horizontal direction but oriented randomly regarding the az imuthal angle For these particles the angular dimension can be reduced by one if we rotate the coordinate system appropriately For this case we use the T matrix code for single particles in fixed orientation and average phase matrix and extinction matrix manually like in the general case The phase matrix and also extinction matrix and absorption vector become indepen dent of the incident azimuth angle in this frame Furthermore regarding the symmetry of this case it can be shown that for the scattered directions we need only half of the angu lar grids as the two halves must contain the same data pha_mat_data therefore has the following size N_ N_T N_za_sca N_aa_sca N_za_inc 2 1 1 16 We store za_sca for all grid points from 0 to 180 aa_sca from 0 to 180 and za_inc from 0 to 90 This means that the zenith angle grid has to include 90 as grid point The order of the matrix elements is the same as in the general case For this case it can be shown that the extinction matrix has only three elements Kjj K12 K21 and K34 K43 Because of azimuthal symmetry it can not depend on the azimuth angle Hence the size of ext_mat_data is N_ N_T N_za 2 1 1 3 The absorption coefficient vector has only two elements al and a2 This means that the size of abs_vec_data is N_ N_T N_za 2 1 1 2 genera
142. he 4x4 identity matrix Both gases and particles in the atmosphere absorb and the total absorption is the sum of these two contributions This chapter only deals with absorption of non scattering matter 1 e gas absorption in the first place but can also include absorption by grey body particles and polarization changes by electrons 6 3 Agendas There are two key agendas related to absorption in ARTS propmat_clearsky_agenda and abs xsec agenda They fulfill different purposes and complement each other Agenda prop mat_clearsky_agenda is called by RT methods when they need absorption more precisely the clear sky propagation matrix as defined above Hence this agenda is the interface between the RT part and the absorption part Is is explained further in the next section The other agenda abs_xsec_agenda is called when absorption is actually calculated either inside propmat_clearsky_agenda 1f absorption is handled on the fly or before when the absorption lookup table is generated This agenda is just for the scalar gas absorption part and describes in detail the line by line absorption calculation and various continua It 6 4 GAS ABSORPTION IN RADIATIVE TRANSFER SIMULATIONS 37 f grid rt_mag propmat_clearsky_agenda Y All RT Methods Figure 6 1 An outside view of propmat_clearsky_agenda is explained further in Section 6 5 You will likely need both of the above agendas for your ARTS RT simulation 6 4 Gas abs
143. he gas absorption spectrum is the quantity that changes most rapidly as a function of frequency Frequency interpolation here could be quite dangerous The abs_lookupAdapt method also checks that all used species apart from Zeeman Faraday and particle species are present in the table reduces the table to the used species and sorts the table species data in exactly the same way that they occur in your calculation It sets the variable abs_lookup_is_adapted to flag that the table is now ok When the table has been successfully adapted one can extract absorption matrices with the method propmat clearskyAddFromLookup This will extract absorption matrices i e the cross sections stored in the table are not only interpolated to the desired atmospheric conditions but are also multiplied with the partial number density of the present absorbers The propmat_clearsky AddFromLookup method is meant to be used inside the agenda 56 GAS ABSORPTION propmat clearsky agenda which is applied in several places where absorption matrices are needed both inside the scattering box and outside 6 6 4 Format of the lookup table Usually the user does not need to bother with it as ARTS provides methods to create read and write and extract data from the lookup table However sometimes one desires to analyze e g the absorption cross section data calculated and stored in the lookup table Therefore we give a short description of the format of the absorpti
144. he geometrical distance between the instrument and the scattering par ticles are assumed to be treated as a separate calibration and the forward model treats only the actual backscattering and atmospheric extinction For the conditions given above the backscattered radiation s can be written as Sh T Z Q 0 Tast 20 1 The terms in this equation are s Stokes vector describing the transmitted pulse T A matrix describing the transmission from the receiver to the scattering point the away direction History 121108 Written byPatrick Eriksson 134 CLOUD RADAR SIMULATIONS Z 9 0 The scattering or phase matrix value for the backward direction T As Tg but for the reversed direction the home direction Note that for vector radia tive transfer in general Ta 4 Th The attenuated backscatter coefficient recorded by the receiver is P p s 20 2 where p is the normalised vector describing the polarisation response of the receiver and signifies the dot product The corresponding unattenuated backscattering coefficient is B p Z Q 0 s 20 3 20 2 Units The unit of 6 and 8 is 1 m sr7 For radar applications this is not the most common choice but is here preferred as it directly matches Z It is also the standard definition in the lidar community The more common definition of radar reflectivity is simply 476 However even more common is to report radar data in unit
145. he paths will likely differ For this reason yCalc will issue an error if you in iy_aux_vars include a quantity of along the path character The same applies to dispersion calculations here the propagation path differs already between the frequencies and also iyLoopFrequencies gives also an error if along the path auxiliary data are selected To simplify the practical usage of this mechanism to extract auxiliary data iyEmis sionStandard also accepts some other variables related to particle properties but does not trigger any calculations The corresponding elements of iy_aux can at a later stage be filled with iy_auxFillParticle Variables This feature can be useful for checking if any particles are found along the propagation path to determine if scattering calculations are required 9 9 Calculation accuracy The accuracy of the calculations depends on many factors For many factors such as spec troscopic parameters there is nothing else to do than using best available data On the other hand for other factors there is a trade off between accuracy and speed More accu rate calculations require normally also more computer memory All different grids and the propagation path step length fall into this category of accuracy factors It could be worth discussing the selection of atmospheric grids and the path step length as there can be some confusion about how that affects the accuracy The main purpose of the atmospheric grids
146. he time The calculations can either be done in in analytical or perturbation manner For gases weighting functions WFs can be provided for several units of the gas abundance vmr Volume mixing ratio a value between 0 and 1 The WF divided by 10 corresponds to that 1 ppm of the gas is added to the atmospheric volume of concern nd Number density The WF corresponds here to that one molecule is added rel Relative fractional change In a perfectly linear case the WF corresponds here to that the gas amount is doubled logrel This option returns the same WFs as rel but is included to flag that the natural logarithm of the rel unit is retrieved For the rel and logrel options it is important to note that ARTS calculate the WFs with respect to the given state ARTS does not know anything about the actual reference state for which the rel unit is valid where normally the a priori state is selected For iterative inversions a rescaling of the WFs provided by ARTS is likely needed to make the WFs valid with respect to the original reference state For the assumption made inside ARTS the WFs for rel and logrel are identical A second main consideration is to select the retrieval grids For analytical calculations there are no other selections to be made 16 4 2 Perturbation calculations For pure numerical calculations also the size of the perturbation must be specified Ax in Eq
147. hickness 84 PROPAGATION PATHS of the path but the point of the highest pressure would be an even more relevant definition in this context Another complication is that with refraction there can in principle exist more than one tangent point Up to ARTS 2 0 minimum radius tangent points were added as extra points to the propa gation paths a reminiscent from ARTS 1 but this feature has now been removed following the discussion in the paragraph above However many internal functions make use of the concept of tangent point as the minimum radius one The altitude based tangent point for a propagation path can be determined with the method TangentPointExtract 10 5 The propagation path data structure A propagation path is represented by a structure of type Ppath This structure holds also auxiliary variables to facilitate the radiative transfer calculations and to speed up the inter polation The fields of Ppath are as follows dim Index The atmospheric dimensionality This field shall always be equal to the workspace variable atmosphere_dim np Index Number of positions to define the propagation path through the atmosphere Allowed values are gt 1 The number of rows of pos and los and the length of Zz 9p_p gp_lat and gp_lon shall be equal to np The length of 1_step is np 1 If np lt 1 the observed spectrum is identical to the radiative background For cases where the sensor is placed inside the model atmosphere and np
148. hods Different methods for the calculation of single scattering properties are reviewed in Emde 2005 Calculations of the single scattering properties can be done by external tools or by in ternally interfaced methods e The ATMLAB package includes functions to generate single scattering properties for spherical particles Mie Theory It is also convenient to use the Python module PyARTS which has been developed especially for ARTS and which is freely available at http www sat ltu se arts tools This module can be used to generate single scattering properties for horizontally aligned as well as for randomly oriented particles in the ARTS data file format PyARTS has been developed by C Davis who has implemented the Monte Carlo scattering algorithm in ARTS see ARTS Theory Section 9 e The WSM scat_data_singleTmatrix provides an ARTS internal interface to the the T matrix code by Mishchenko et al 2002 8 3 Representation of the particle size distribution The particle size has an important impact on the scattering and absorption properties of particle ensembles e g clouds as shown for instance in Emde et al 2004 Particle en sembles typically contain a whole range of different particle sizes which can be described by a size distribution giving the number of particles per unit volume per unit radius interval as a function of radius It is most convenient to parameterize the size distribution by ana lytical functions be
149. ht corner The dotted lines indicate that some methods and agendas can make further calls of iy_main_agenda Further the term iy is not restricted to the final outcome of atmospheric radiative transfer it is also used to indicate that quantities and operations are of monochromatic pen cil beam character For example the agenda returning the radiation entering the atmosphere from space is named as iy_space_agenda to indicate that the agenda output is directly asso ciated with the calculation of iy If only a single iy is of concern the radiative transfer can be performed by iyCalc having a smaller set of input arguments than yCalc The output is not restricted to y or iy also some auxiliary data can be extracted as de scribed in Section 9 8 In addition yCalc can also provide weighting functions Chapter 16 5 2 Compulsory sensor and data reduction variables The instrument real or hypothetical that detects the simulated radiation is denoted as the sensor The forward model is constructed in such way that a sensor must exist For cases when only monochromatic pencil beam radiation is of interest the positions and directions for which the radiation shall be calculated are given by specifying an imaginary sensor with infinite frequency and angular resolution The workspace variables for the sensor that always must be specified are sensor_response sensor_pos sensor_los and mblock_dlos_grid These variables are presented separately belo
150. ies However it is normally not needed to make a recalculation using the total forward model as the variables are either part of the atmospheric or the sensor state but not both In addition in many cases it is possibly to find short cuts For example the perturbed state can be approximated by an interpolation of existing data such as for a perturbed zenith angle Such short cuts are discussed separately for each retrieval quantity K 16 2 16 1 2 Analytical expressions For most atmospheric variables such as species abundance it is possible to derive an an alytical expression for the Jacobians This is advantageous because they result in faster and more accurate calculations Such expressions are derived below Some of the terms involved are calculated as a perturbation This is partly a consequence of the flexibility of ARTS The high level radiative transfer methods e g iyEmissionStandard do not know how the low level methods are defining all quantities and a fixed analytical expression can not be used see e g Sec 16 6 So in practise the calculations are semi analytical The analytical calculations are introduced in Sec 16 3 16 1 3 Workspace variables and methods As a workspace variable the complete Jacobian is denoted as jacobian Auxiliary infor mation is provided by jacobian_quantities and jacobian_indices The actual calculations are made as part of yCalc The retrieval quantities are defined separate
151. iles artscomponents doit 125 controlfiles artscomponents faraday 51 gas_abs_lookup cc 55 gas_abs_lookup h 55 56 general arts 10 55 hitran_cia2012_adapted xml gz 45 m_cloudbox cc 66 125 m_optproperties cc 66 125 m_scatrte cc 125 optproperties cc 66 optproperties h 60 README 5 10 TestAbsParticle arts 52 ARTS 1 3 ARTS INCLUDE PATH 10 atmosphere 3 atmospheric dimensionality 16 atmospheric field 18 azimuth angle 30 basis function 19 109 birefringance 98 built in documentation 5 cloud box 20 command line parameters 6 controlfile 6 Coordinate systems 60 curvature radius 88 data reduction 31 data types GasAbsLookup 55 56 GField3 67 Ppath 84 SingleScatteringData 59 60 67 default value 8 Discrete Ordinate ITerative DOIT method 125 dispersion 57 73 78 Doppler effect 95 DOXYGEN 5 ellipsoid 87 engine 6 example controlfiles 10 Faraday rotation 97 generic 7 geo location 19 geocentric latitude 88 geodetic latitude 88 geometrical altitude 15 groups 7 8 include 10 internal ARTS functions dotprod_with_los 96 ppath_calc 82 83 surface_specular_los 90 laboratory frame 60 latitude 17 line of sight 30 longitude 18 magnetic field 21 measurement block 31 146 INDEX measurement sequence 31 meridian plane 30 methods 6 model atmosphere 18 Monte Carlo scattering module 131 n2 law of radiance 77 parser 6 particle size
152. information from a number of monochromatic pencil beam calculations as shown in Section 5 3 History 130219 First version by Patrick Eriksson 28 COMPLETE CALCULATIONS blackbody_ ppath_agenda radiation_agenda iy_space_agenda 4 ye Cee ciel propmat_clearsky_ iy_cloudbox_ agenda agenda refr_index_agenda B iy surface_agenda adi surface_rtprop_ agenda For caption see top of the next page 5 2 COMPULSORY SENSOR AND DATA REDUCTION VARIABLES 29 Figure 5 1 A flowchart of a radiative transfer calculation on previous page using yCalc or iyCalc The figure assumes that iy_main agenda holds iyEmissionStandard that represents the most complex case Some important input data are shown in orange and main output are shown in green The two main methods are plotted as grey boxes while agendas are shown as ovals For cases when iy_sub_agenda not is defined the most common situation the agendas communicate directly with iy main agenda For connections between methods and agendas the arrows show the direction of output calculated data Every call of an agenda involves also some input data settings but this aspect has been ignored for clarity reasons Data entering an agenda along a common line indicates mutually independent options For example absorption can be calculated on the fly from basic spectroscopic data or be extracted from a pre calculated look up table see bottom rig
153. input and output is supported for a subset of the data types available in ARTS 2 1 3 Gridded fields Naming convention for grids o Grid names for GriddedField variables Keep the description short and on one line Sort alphabetically o A Complex Complex number grid i e exactly 2 elements real and imaginary Frequency Frequency dimension for spectral dependent fields Latitude Latitude dimension for atmospheric fields z T VMRs winds Ne B pnd Longitude Longitude dimension for atmospheric fields z T VMRs winds Ne B pnd Pressure Pressure dimension for atmospheric fields z T VMRs winds Ne B pnd History 14 IMPORTING AND EXPORTING DATA Data field names mere suggestions and not tested anywhere in ARTS o 5 o 5 Temperature Temperature field VMR Volume mixing ratio field Altitude Altitude height field pnd_field Particle number density field Chapter 3 Description of the atmosphere This section discusses the model atmosphere how it is defined its boundaries and the variables describing the basic properties One aspect that can cause confusion is that several vertical coordinates must be used Sec 3 1 The main vertical coordinate is pressure and atmospheric quantities are defined as a function of pressure Sec 3 3 but the effective vertical coordinate from a geometrical perspective such as the determination of propagation paths is the radius Sec 3 2 Pressures and radii are
154. is frame only the scattering angle which is the angle between incident and scattered direc tion is needed Furthermore the number of matrix elements of both matrices phase matrix and extinction matrix can be reduced see Mishchenko et al 2002 p 90 To calculate the particle optical properties it is convenient to use Mishchenko s T matrix code for randomly oriented particles Mishchenko and Travis 1998 which returns the averaged phase matrix 8 2 SINGLE SCATTERING PROPERTIES 63 and extinction matrix The only drawback is that the single scattering data has to be trans formed from the particle frame representation to the laboratory frame representation These transformations are described in the appendix of Emde 2005 Only six elements of the transformed phase matrix which is commonly called scattering matrix F are different Therefore the size of pha_mat_data is N_ NT Nza sca 1 1 1 6 The order of the matrix elements is as follows F11 F12 F22 F33 F34 F44 The extinction matrix is in this case diagonal and independent of direction and polarization That means that we need to store only one element for each frequency Hence the size of ext_mat_data is N_ NT 1 1 1 The absorption vector is also direction and polarization independent Therefore the size of abs_vec_data for this case is the same as ext_mat_data N_ NT 1 1 1 horizontally aligned The ptype value horizontally_aligned refers to particles t
155. l The ptype value general refers to arbitrarily shaped and oriented particles For those generally no symmetries ecist hence all 16 elements of the phase matrix have to be stored The average phase matrix has to be generated from all individual phase matrices of the particles in the distribution outside ARTS The individual phase matrices are calculated using Mishchenko s T matrix code for single particles in fixed orientation Mishchenko 2000 We have to store all elements for all angles in the grids The size of pha_mat_data is therefore 64 DESCRIPTION OF CLOUDS N_f N_T N_za_sca N_aa_sca N_za_inc N_aa_inc 16 The matrix elements have to be stored in the following order Z11 Z12 Z13 Z14 Z21 Z22 Seven extinction matrix elements are independent cp Mishchenko et al 2002 p 55 The elements being equal for single particles should still be the equal for a distribu tion as we get the total extinction just by adding Here we need only the incoming grids so the size of ext_mat_data is N_ N_T N_za_inc N_aa_inc 7 The absorption vector in general has four components cp Equation 2 186 in Mishchenko et al 2002 The size of abs_vec_data is accordingly N_ N_T N_za_inc N_aa_inc 4 8 2 4 Generating single scattering properties The single scattering properties have to be pre calculated before entering the scattering solver This can be done for example by Mie T Matrix or Discrete dipole approximation met
156. lated for the refractive index of water at the specified temperature Hence the basic calculations are performed in standard manner using iy_main_agenda However the deviating data pattern in iy results in that an alternative to yCalc is needed 20 3 PRACTICAL USAGE 135 and it is yCloudRadar The method applies two instrument effects Firstly the polarisation response of the receiever is incorporated Eq 20 2 Secondly the data are averaged as a function of range following the range_bins These range bins can be specified either as geomtrical altitude or the two way propagation time The range binning is described further in the built in documentation Further the data are rearranged into a vector and returned as y A number of auxiliary data can be obtained by iyCloudRadar this including 8 For a list of possible variables arts d iyCloudRadar A difference of yCloudRadar compared to yCalc is that all auxiliary quantities provided by iyCloudRadar are treated 136 CLOUD RADAR SIMULATIONS Part V Bibliography and Appendices Bibliography Buehler S A P Eriksson T Kuhn A von Engeln and C Verdes ARTS the Atmospheric Radiative Transfer Simulator Journal of Quantitative Spectroscopy and Radiative Trans fer 91 65 93 2005 Buehler S A A von Engeln E Brocard V O John T Kuhn and P Eriksson Recent developments in the line by line modeling of outgoing longwave radiation Journal of Quanti
157. lations ARTS provides some support for modelling measurements of backscattering inside the at mosphere This deviates for the active measurements discussed in the previous chapter as in this case the transmitter and receiver are placed at the same position Here backscatter ing is recorded while in the transmission case the attenuation through the atmosphere is probed This has the consequence that the measurement data not only span the frequency and a polarisation dimensions but also has a time or distance dimension That is the backscattering is basically reported as a function of distance from the sensor for one or several frequencies frequencies and combinations of transmitted and received polarisations Accordingly these data differ in nature from the other types of measurements and the stan dard main function yCalc is not applicable A very important restriction applies only single scattering is treated That is no other scattering than the targeted backscattering is considered This is frequently an acceptable simplification for precipitation and cloud radars observations and such observations should be the main applications of the discussed workspace methods The basic measurement ap proach is the same for lidars but the assumption of single scattering is much less commonly met for such instruments 20 1 Theory The transmitted pulses are treated to be monochromatic pencil beams Effects due to an tenna patterns and t
158. lemented for 1D and 3D spherical atmospheres but it is strongly recommended to use it only for 1D calculations because there are several numerical dif ficulties related to the grid discretizations It is difficult to find appropriate discretizations to get sufficiently accurate results in reasonable computation time Therefore only experi enced ARTS users should use DOIT for 3D calculations only for smaller cloud scenarios Please refer to the online documentation for the workspace method for 3D scattering calcu lations doit 1 fieldUpdateSeg3D All other workspace methods adapt automatically to the atmospheric dimensionality Chapter 19 Scattering calculations The Monte Carlo scattering module This is just a stub for this chapter So far just rescuing some text from an old and now deleted Wiki page The ARTS Monte Carlo scattering module offers an efficient method for single viewing direction radiative transfer calculations in arbitrarily complex 3D cloudy cases When sim ulating the observation of detailed 3D cloudy scenarios reversed Monte Carlo algorithms have several advantages over other methods such as discrete ordinate and Forward Monte Carlo methods These features include All computational effort is dedicated to calculating the Stokes vector at the location of interest and in the direction of interest This is particularly relevant for space borne remote sensing where we are only interested in a narrow field of view
159. librium is possible for perturbation cal culations while the analytical approach only treats the local effect see further below 16 6 2 Perturbation calculations The size of the perturbation must be selected in K Complete radiative transfer calculations are done after perturbing the temperature field Hence all possible effects are included such as changed propagation paths through the impact of temperature on the refractive index Please note that hydrostatic equilibrium comes in during the perturbation If hse is set to on also z_field is recalculated as part of the temperature perturbation Section 3 5 If set to off there is no change of z field That is you must make an active choice regarding hydrostatic equilibrium while others effects are included automatically 16 6 3 Analytical expressions Unpolarised absorption Compared to atmospheric species the expressions become here more complex as temper ature also affects the propagation path length Al and the emission source term Bj Accordingly all terms of Eq 16 20 are relevant and the expansion of 07 Ox generates additional terms OL n OTi OU Or OAL Tap e OB ta adn OMG On oe Terms part of expressions found above are not discussed separately here The term Al Ox originates in the constrain of hydrostatic equilibrium and is set to zero when hse is set to off Otherwise it is set as derived below The term a Ox is calculated in a pur
160. llipsoid is Rodgers 2000 1 11 3 ras cos a ral sin a where rns and Te are the north south and east west curvature radius respectively 3 _3 Tns rara r2 cos w ie sin ww 11 4 mfr ana 2012 923 Tew Tr lr costw r sin w 2 11 5 The azimuth angle w is defined in Section 5 2 2 The latitude and azimuth angle to apply in Equations 11 3 11 5 shall rather be valid for a middle point of the propagation paths such as some tangent point instead of the sensor position 11 1 2 Geocentric and geodetic latitudes The fact that Earth is an ellipsoid instead of a sphere opens up for the two different defi nitions of the latitude The geocentric latitude which is the the one used here is the angle between the equatorial plane and the vector from the coordinate system centre to the posi tion of concern The geodetic latitude is also defined with respect to the equatorial plane but the angle to the normal to the reference ellipsoid is considered here as shown in Figure 11 1 It could be mentioned that a geocentric latitude does not depend on the ellipsoid used while the geodetic latitudes change if another reference ellipsoid is selected For Earth the largest difference between geocentric and geodetic latitude is found at mid latitudes where it reaches 12 arc minutes There are yet no methods in ARTS for conversion of data between geodetic and geocentric latitudes 11 1 THE REFERENCE ELLIPSOID
161. lt LineMixingRecord gt lt Array gt lt arts gt In this format the first four values are a temperature grid in Kelvin For LL the following four values are the first order line mixing coefficients at these temperature grid points in units 1 Pa and the last four are the second order line mixing coefficients at these temper ature grid points in units 1 Pa For NR the following four values are the first order non resonant term at these temperature grid points in units 1 Pa and the last four are the second order non resonant terms at these temperature grid points in units 1 Pa ARTS will use linear interpolation to get proper coefficients between and slightly outside grid points Warning There appears to be some errors in the quantum number encoding in HI TRANO4 and HITRANOS that makes the line mixing module wrong The newer HI TRAN2012 has fixed these issues 6 5 7 Faraday rotation Faraday rotation is a change of polarization state of radiation in interaction with free elec trons in presence of a static magnatic field For further details on theory and usage in ARTS see Section 14 Here we only give a short summary how to setup the calculation of Faraday contribution to the absorption or better propagation matrix propmat_clearsky First to include Faraday rotation effects propmat_clearskyAddFaraday must be included in the propmat_clearsky_agenda Second a species tag free_electrons needs to be contained in
162. ly before calling yCalc This process is started by calling jacobianInit The retrieval quantities are then introduced through workspace methods named as jacobianAddSomething For example for atmospheric tem perature the method is jacobianAddTemperature It does not matter in which order these add methods are called The definition of retrieval quantities is finalised by calling jacobianClose To disable the calculation of the Jacobian skip all above and just use jacobianOff The methods named jacobianCalcSomething shall never be used directly Neither needs the user to consider jacobian_agenda 16 2 BASIS FUNCTIONS 109 33 5 T T T T 3h 4 32 57 ei 4 w w u N T 7 1 V I I W I 1 1 Altitude km 2 1 l y 30 51 Da 4 307 J 29 5 4 29 i i i 0 0 2 0 4 0 6 0 8 1 Basis function Figure 16 1 Examples on 1d basis functions for a vertical grid with a 1 km spacing 30 km 31 km and 32 km The input to the add methods differs In some cases you can select between the analytical and perturbation options For all perturbation calculations you must specify the size of the perturbation For atmospheric gases you can use different units For atmospheric fields and some other quantities you must define the retrieval grid s to use 16 2 Basis functions A forward model must use a discrete representation it describes each quantity with one or
163. m_o2 ReadXML 1m_02 02 xm1 Create a way to temporally store CO2 line mixing data ArrayOfLineMixingRecordCreate lm_co2 ReadXML 1m_co2 co2 xml1 Match the line mixing data with 02 and Zeeman effect line_mixing_dataMatch species_tag 02 Z LM line_mixing_records 1m_02 line_mixing_tag LMx Match the line mixing data with 02 and Zeeman effect line_mixing_dataMatch species_tag CO2 LM line_mixing_records 1m_co2 line_mixing_tag LMx The latter method is much more complicated than reading directly from the catalogue This 1s because the user has to ensure what type of line mixing is going on and then parse that data properly to ARTS In order for ARTS to know what line the user wants to line mix a series of line database matching based on the quantum numbers of the line is necessary This means that it only works for those species and catalogue formats for which we have implemented quantum number reading This list is small but growing and 1f you need help adding another species let us know via the mailing list There are three line_mixing_tag supported at this point LL NR and L2 These refer to LBLRTM and second order line mixing Setting these tags will tell ARTS how to calculate the line mixing of any particular line The data is matched from ArrayOfLineMixingRecord which must contain a SpeciesTag a QuantumNumberRecord and a vector that is the correct length for the method selected
164. mentations of discrete ordinate schemes are only applicable for 1D plane parallel or 3D cartesian atmospheres All of these algorithms can not be used for the simulation of limb radiances A description of the DOIT method is given in ARTS Theory Chapter 8 and has been published in Emde et al 2004 and in Emde 2005 The workspace methods required for DOIT calculations are implemented in the files m_scatrte cc m cloudbox cc and m_optproperties cc Here we introduce the steps to be performed in a DOIT calculation along with the relevant workspace methods by means of a controlfile example 18 1 The 1D control file example This example demonstrates how to set up a 1D DOIT calculation A full run ning controlfile example for a DOIT calculation can be found in the ARTS pack age in the controlfiles artscomponents doit directory The file is called TestDOIT arts For detailed descriptions of the workspace methods and variables please refer also to the online documentation start doc server on your own computer with arts sorusehttp www sat ltu se arts docserver stable History 020601 Created and written by Claudia Emde 050223 Rewritten by Claudia Emde mostly taken from Chapter 4 of Claudia s PhD thesis 050929 Included technical part example contol file 141009 Moved general model parts to theory guide 126 SCATTERING CALCULATIONS THE DOIT MODULE 18 2 DOIT frame The first step for a DOIT calculation is the initialization
165. n is not performed This function is effi cient for simulations in up or down looking geometry where the we do not need the fine zenith angle grid resolution about 90 Radiative transfer with fixed scattering integral term With a fixed scattering integral field the radiative transfer equation can be solved ARTS Theory Equation 8 9 The workspace method to be used for this calculation is de fined in doit_rte_agenda The most efficient and recommended workspace method is doit_i_fieldUpdateSeq1D where the sequential update which is described in ARTS The ory Section 8 5 is applied The workspace method doit_i_fieldUpdate1D does the same calculation without sequential update and is therefore much less efficient because the number of iterations depends in this case on the number of pressure levels in the cloudbox Other options are to use a plane parallel approximation implemented in the workspace method doit_i fieldUpdateSeq1DPP This method is not much more efficient than doit_i_fieldUpdateSeq1D therefore it is usually better to use doit fieldUpdateSeq1D since it is more accurate AgendaSet doit_rte_agenda doit_i_fieldUpdateSeq1D Alternatives doit_i_fieldUpdateSeq1DPP 1 fieldUpdatelD The optical properties of the particles i e extinction matrix and absorption vec tor for all scattering elements are required for solving the radiative transfer equa tion How they are calculated is specified in spt_calc_agenda
166. n iterative process to determine A and Chapter 17 Batch calculations Quite often one wants to repeat a large number of calculations with only a few variables changed Examples of such cases are to perform 1D calculations for a number of atmo spheric states taken from some atmospheric model generate a set of spectra to create a training database for regression based inversions or perform a numeric inversion error anal ysis For such calculations it is inefficient to perform the calculations by calling ARTS repeatedly For example as data must be imported for each call even if the data are identi cal between the cases Cases such as the ones described above are here denoted commonly as batch calculations 17 1 Batch calculations of measurement vector y Batch calculations of measurement vectors are done by the WSM ybatchCalc which takes three inputs the indices ybatch_start and ybatch_n and the agenda ybatch_calc_agenda The method ybatchCalc will do ybatch_n calculations starting at index ybatch start It will run the agenda ybatch_calc_agenda for each case individually The agenda gets an input ybatch index which it should use to extract the right input data from one or more arrays or matrices of input Execution of ybatch_calc_agenda must result in a new spectrum vector y most likely by a call of yCalc The WSV ybatch_start is set to a default of 0 in general arts so you do not have to set this variable unless you want to st
167. n of such radiative transfer calculations is continued in Section 9 A v r fi b v r fm s v r A Aas Bag 4 13 Radiative transfer with scattering Some additional conditions are required to put scattering into the picture If scattering is of incoherent and elastic nature the extension of Equation 4 13 is ds v r fi dl K v r fi s v r Bag v r nm 4 14 Z v r f fi s v r A di An where Z is the scattering or phase matrix ARTS includes several modules to handle scattering introduced in the last part of this document Chapter 5 Complete calculations This chapter outlines how complete radiative transfer simulations are performed ARTS is not only performing atmospheric radiative transfer also sensor characteristics can be considered As a consequence the topics of this chapter include the distinction between monochromatic pencil beam and full calculations and how the sensor is introduced 5 1 Overview An attempt to illustrate a standard ARTS calculation is found in Figure 5 1 The figure shows that most calculation tasks are handled by an agenda For example ppath_agenda has the task of determining the propagation path for the given observation geometry In principle the agenda could solve the task by loading data from a file but most likely it will use some of the dedicated workspace methods These workspace methods are targeted towards different observation typ
168. n the analytical expressions dealing with local radiative transfer effects let us write Eq 16 15 as s s T s 16 18 In the nomenclature of Eq 9 7 Ss s 1 but would be confusing to here use s 41 Hence Os T OSs Ox Ox The first element of ss J can be written as cf Eq 9 2 I e 1 e Bi 16 19 By again using the chain rule the derivative of J with respect to x can be written as o A 2 MN E a any 16 20 The two remaining derivatives 07 Ox and OB 0x depend on the quantity considered and are discussed further below For higher Stokes components of ss or in general if emission is totally ignored Eq 16 19 is simplified to here exemplified for the second Stokes element Q Q e Oj 16 21 and the chain rule expression is correspondingly shorter 0Q en OT mi js 16 22 Ox e Ox Qi 16 3 6 Limitations A constraint for the analytical expressions above is that the effect of the variable must only be local Main examples on non local effects should occur through hydrostatic equilib rium and refraction Significant impact of a gas through these mechanisms should only be found for water vapour in the lower troposphere but is a general concern for temperature as discussed in Sec 16 6 114 CLEAR SKY JACOBIANS 16 4 Absorption species 16 4 1 Common practicalities To obtain the Jacobian for absorption species use jacobianAddAbsSpecies The method handles one species at t
169. n the results ARTS offers a general solution Dispersion can be handled by setting iy_main_agenda as AgendaSet iy_main_agenda iyLoopFrequencies The radiative transfer method you put in iy_main_agenda for non dispersive calculations are now moved to iy_sub_agenda For example if iyEmissionStandard is the method of your choice AgendaSet iy_sub_agenda iyEmissionStandard The approach is simple iyLoopFrequencies calls iy sub agenda for each single frequency in f_grid and appends the output With some details iyLoopFrequencies performs a loop over the f_grid creates an internal f_grid of length 1 holding the frequency of concern and calls iy_sub_agenda with this length 1 frequency grid This has the result that a propagation path is calculated for each frequency component Some more steps are required to correctly include dispersion A basic demand is that ppath agenda considers refraction Further refr index_air_agenda must provide a dispersive refractive index Most methods aimed for refr_index_air_agenda give a refractive index that does not varies with frequency An example on the opposite is refr_index_airFreeElectrons If a method with dispersive refractive index is used for non dispersive calculations it re ceives the mean of the first and last element in f_grid as already commented above A limitation of iyLoopFrequencies is that it can not be combined with auxiliary data of along the path character Sec 9
170. nal wind speed vu 13 1 13 3 Sensor velocity The feature is not yet fully implemented but a rudimentary inclusion of the sensor s velocity can be made by the rte_alonglos_v workspace variable The variable shall be set to the sensor velocity component along the viewing direction For the moment this velocity is assumed to be the same for all pencil beam calculations 13 4 Limitations For efficiency reasons when extracting particle extinction absorption and scattering the mean Doppler shift along the propagation path is applied but the shift varies with fre quency This allows to at least include larger shifts caused by e g satellite velocity The smaller shifts due to winds are here of less importance as the particle extinction is quite smooth with frequency The above is valid for transmission type calculations Doppler shifts are so far totally neglected inside the DOIT and MC scattering modules 13 5 Equations The main equations for deriving the Doppler shift from the winds are given in this section The total wind v is v 4 v2 02 v2 13 2 The zenith angle of the wind direction is Wy arccos vy Vv 13 3 and the azimuth angle is Wy arctan Vu vy implemented by the atan2 function 13 4 The cosine of the angle between the wind vector and the line of sight is cos y cos Yy cos Y sin Wy sin Y cos wy wi 13 5 where y and w are the angles of the line of sight Finally as the winds
171. nctions is simplified if T is at hand and this quantity is also tracked by iyEmissionStandard 9 4 2 Polarised absorption The overall calculation procedure is the same with polarised absorption the only differ ence is the radiative transfer expression applied The calculations for the different Stokes components can here not be separated and matrix vector notation is required Sip e AlKig 1 e AiKop 9 7 where 1 is the identity matrix The K and b at point and 1 are averaged element wise to give K and b respectively in line with Equation 9 4 The calculation of the transmission matrix T e K 9 8 involves a matrix exponential This calculation step is handled for simpler cases with ana lytical expressions while for more complex cases the Pad approximation ARTS Developer Guide Section 7 3 is applied Only the first element of b is non zero and only the first column of the matrix corresponding to the term 1 e is of interest 9 5 OUTPUT UNIT AND THE N LAW 77 9 4 3 Blackbody and cosmic background radiation The term B is set by blackbody radiation agenda The setting of this agenda basically determines the unit of the final outcome of Eq 9 2 see further Sec 9 5 For radiance calculations the standard workspace method to use inside blackbody radiation agenda is blackbody_radiationPlanck The Planck function is in this method and in ARTS generally defined as 2hv c exp hv kyT
172. ned with a single point to know for which position and line of sight the intensity field of the cloud box shall be interpolated 9 4 BASIC RADIATIVE TRANSFER VARIABLES AND EXPRESSIONS 75 atmosphere the observed monochromatic pencil beam intensity i in Algorithm 1 equals the output of iy_space_agenda The surface The sum of surface emission and radiation reflected by the surface is the radia tive background when the propagation path intersects with the surface It is the task of iy_surface_agenda to return this up welling radiation from the surface see further Chapter 11 Surface of cloud box For cases when the propagation path enters the cloud box the radia tive background is the intensities leaving the cloud box This radiation is obtained by iy_cloudbox_agenda Interior of cloud box If the sensor is situated inside the cloud box there is basically no propagation path The radiative background and also the final spectrum equals the internal intensity field of the cloud box at the position of the sensor in the direction of the sensor line of sight This case is also handled by iy_cloudbox_agenda It should be noted that except for the first case above the determination of the radiative back ground involves further radiative transfer calculations For example in the case of surface reflection the down welling radiation could be determined by a new call of iy main agenda and the radiative background for that calculation is then
173. ngitude limits for the cloud box cannot be placed at the end points of the corresponding grid as it must be possible to calculate the incoming intensity field The cloud box is activated by setting the variable cloudbox_on to 1 The limits of the cloud box are stored in cloudbox_ limits It is recommended to use the method cloudboxOff when no scattering calculations shall be performed This method assigns dummy values to all workspace variables not needed when scattering is neglected When the radiation entering the cloud box is calculated this is done with the cloud box turned off This to avoid to end up in the situation that the radiation entering the cloud box depends on the radiation coming out from the cloud box It is the task of the user to define the cloud box in such way that the link between the outgoing and ingoing radiation fields of the cloud box can be neglected The main point to consider here is radiation reflected by the surface To be formally correct there should never be a gap between the surface and the cloud box This is the case as radiation leaving the cloud box can then be reflected back into the cloud box by the surface If it is considered that the surface is a scattering object it is clear that the surface should in general be part of the cloud box However for many cases it can be accepted to have a gap between the surface and the cloud box with the gain that the cloud box can be made smaller Such a case is when the surface is
174. ns For a signal transmitted in a limb geometry through a spherical symmetric atmosphere i e 1D with an exponential scale height H Haugstad 1978 ate Me i 106 TRANSMISSION CALCULATIONS de Ms 1 2 1 5 where a is the bending angle Eq 15 14 L is the distance from the tangent point to the receiver and R is the planet radius An alternative form of the equations above also also valid for an atmosphere with a non exponential scale height is Kursinski et al 2000 da LL TA Ma 1 15 16 da L A a LL 17 Mg 15 17 where L and L is the distance between the tangent point and the transmitter and receiver respectively and a nr sin w is denoted as the impact parameter and is equal to the prop agation path constant defined in Eq 7 30 of ARTS Theory The bending angle a is given by the derived path between the transmitter and receiver The term da da is estimated using calculations with shifted zenith angles with a shift set by the user defocus_shift Hence this term is calculated in a manner similar to the approach used in method 1 15 2 8 Atmospheric extinction Atmospheric absorption and scattering are handled exactly as for iyTransmissionStandard Accordingly only scattering out of the propagation path is considered Chapter 16 Clear sky Jacobians Inversions of both OEM and Tikhonov type require that the Jacobian can be provided by the forward model
175. of Quantitative Spectroscopy and Radiative Transfer 112 1420 1428 2011 McFarquhar G M and A J Heymsfield Parametritation of tropical cirrus ice crystal size distributions and implications for radiative transfer Results from CEPEX Journal of the Atmospheric Sciences 54 2187 2200 1997 Meissner T and E J Wentz Polarization rotation and the third stokes parameter the effects of spacecraft attitude and faraday rotation IEEE Transactions on Geoscience and Remote Sensing 44 506 515 2006 BIBLIOGRAPHY 141 Mishchenko M I Calculation of the amplitude matrix for a nonspherical particle in a fixed orientation Appl Opt pp 1026 1031 2000 Mishchenko M I and L D Travis Capabilities and limitations of a current fortran imple mentation of the t matrix method for randomly oriented rotationally symmetric scatterers J Quant Spectrosc Radiat Transfer 60 309 324 1998 Mishchenko M I L D Travis and A A Lacis Scattering Absorption and Emission of Light by Small Particles Cambridge University Press 2002 ISBN 0 521 78252 Newell A C and R C Baird Absolute determination of refractive indices of gases at 47 7 Gigahertz J Appl Phys 36 1965 Read W G Z Shippony M J Schwartz N J Livesey and W V Snyder The clear cky unpolarized forward model for the EOS Aura Microwave Limb Sounder MLS EEE Transactions on Geoscience and Remote Sensing 44 1367 1379 2006 Richard C
176. of equivalent reflectivity Ze This quantity is defined as e g Donovan and van Lammeren 2001 Ar 20 4 Kp 20 4 where A is wavelength of the radar and the reference dielectric factor is calculated using the complex refractive index of ice or liquid water n 20 5 20 3 Practical usage As for other measurements the main radiative transfer calculations are performed inside ly main agenda The workspace method to be used inside this agenda is iyCloudRadar This method returns s for each point of the propagation path and for each frequency in f_grid The calculated data are packed into iy Here the difference to other measurement types emerge For example the second row holds s corresponding to the second point of the propagation path not the second frequency If f_grid contains several frequencies the data for the second frequency are placed below in the row dimension the data for the first frequency etc The polarisation of the transmitted pulses sp are taken from iy transmitter agenda The agenda shall here return a Stokes vector for each frequency in f_grid of unit intensity The unit of returned data is selected by iy unit There are two options 1 and Ze For the first option no unit conversion is performed while for the second option Equation 20 4 is applied on all Stokes elements of s In the later case liquid water at a user specified temperature is used as reference That is K is calcu
177. olfile section illustrates how a simple cloud can be included in an ARTS calculation First we have to define the cloudbox region i e the region where scattering objects are found To do this we can use the method cloudboxSetManually Altitude cloudboxSetManuallyAltitude cloudbox_on cloudbox_limits atmosphere_dim z_field lat_grid lon_grid 8000 120000 0 O O O If we want to do a simulation for a cirrus cloud at an altitude from 9 to 11 km the cloudbox limits can be set to 8 and 12km The latitude and longitude limits are set to an arbitrary value for a 1D calculation For 3D calculations they are also needed Alternatively one can use the method cloudboxSetManually where one has to provide pressure instead of altitude limits Now we have to specify the cloud particles inside the scattering region Initialisation ParticleTypelnit Only one scattering element is added in this example ParticleTypeAdd scat_data pnd_field_raw atmosphere_dim f_grid ssd_sphere_50um_macroscopically_isotropic xml pnd_sphere_50um_macroscopically_isotropic xml In the workspace method ParticleTypeAdd the single scattering properties for one scattering element are read The generic input filename scat_data must be set to the filename of a datafile containing the single scattering data class SingleScatteringData in xml format The generic input filename pnd field must contain the filename of the corresponding particle number
178. om mon vector ip Values are put in following the order in f grid Hence the frequencies for this vector are k l 5 1 Un where v is element 7 of f grid and n the length of the same vector The order of the angles inside mblock_dlos_grid is followed when looping the pencil beam directions The workspace variable sensor response is here denoted as Hy It is applied on each iy and the results are appended vertically following the order of the positions in sensor_pos Hbio1 Hbib 2 y 5 2 Hbibn where 1 indicates the first sensor position etc It should be noted that the compulsory sensor variables give no information about the content of the obtained y as it is not clear which parts and features the block transfer matrix covers If H only incorporates the antenna pattern the result is a set of hypothetical spectra corresponding to a point inside the sensor On the other hand if H includes the whole of the sensor and an eigenvector data reduction the result is not even a spectrum in traditional way it is just a column of coefficients with a vague physical meaning Part II Atmospheric properties Chapter 6 Gas absorption 6 1 Introduction When calculating radiative transfer the local absorption at each point in the atmosphere has to be known Furthermore if one also wants to calculate Jacobians then the partial absorption for different atmospheric components different absorption species also has to be known
179. on approach 2 ssor eos a we ea a 82 10 3 Spacing of additional path points oaaae 83 104 Tangent polis buon p i a DO aoa GS a 83 10 5 The propagation path data structure ooa a 84 10 6 Purther reading 2 lt moa dange rear a a R a ak ae 85 IV CONTENTS 11 Reference ellipsoid and surface properties 87 11 1 The reference ellipsoid kc iiacibaeeaeanagevaw ta adeadr 87 11 1 1 Ellipsoid models lt s courses Mee a bales 87 11 1 2 Geocentric and geodetic latitudes 88 11 2 Surface MIOS o i os 2a naka gata Gaes a da eas 90 11 3 Surface radiative properties secere cowe rss 90 11 31 Blackbody surface sa ua gon aj ate eae dom ee eds ee a a 90 113 2 Specular reflections via SE a ee we 90 1133 Lambertian SUECA 91 12 Sensor characteristics 93 Di General pec vec Bd sete ae ee ke hp ee hs wr ck Se a E e a Ee eB 93 12 2 SOME COMMONS s Yow nse seule a earns alee athe oie te WEE aes a a oe 94 13 Doppler effects and winds 95 BA Winds sieer e 5 kb kab be eA CREE BA EAE ae eh 95 13 2 Planet rotation suas ca el Wee a ea he Wet a hale es 95 13 3 Sensor velocity uo 24 20 3 6 2a 655 SS bo HOHE HO ee eae 96 io ss e a eee te ae a wr ee a ee Be eR ee ode wh aed a 96 13 3 06216 02 1 en er cr o da 96 14 Faraday rotation 97 14 1 Practical usage 002 Rw A RAR we aw ea 97 142 THON i a eee ee Ae ee eR ee Re eh ee we ee 97 15 Transmission calculations 99 15 1 Pure transmission calculations lt se ss sos o o
180. on lookup table here More detailed information can be found in the source code where the GasAbsLookup class is implemented specifically in gas_abs_lookup h The absorption lookup table is a compound type variable comprising of in this order variable type of each entry shown in parantheses e species an array of the species tags the lookup table is valid for ArrayOfArrayOf SpeciesTag nonlinear_species an array indicating the species that require non linear treatment ArrayOfIndex e f grid the frequency grid Vector p grid the pressure grid Vector vmrs_ref the reference profiles of volume mixing ratios VMRs for all species asso ciated with the pressure Matrix dimension number of species number of pressure levels e t ref the reference temperature profile associated with the pressure grid Vector e t_pert the temperature perturbations Vector nis_pert the VMR perturbations of the non linear species in terms of fractional units of the reference VMRs Vector e xsec the absorption cross sections Tensor4 dimension number of temperature perturbations number of species and non linear species perturbations number of frequencies number of pressure levels 6 7 Stand alone gas absorption calculation Within the RT calculations gas absorption is calculated or extracted locally i e for a spe cific point in the atmosphere or in other words for a specific set of pressure temperature and trace ga
181. ons are frequency temperature za inc aa_inc matrix element Again the order of matrix elements depends on the chosen case e Tensor5 abs vec data Absorption vector data a Unit m The absorption vector is also precalculated It could be calculated from extinction matrix and phase matrix But this calculation takes long computation time as it requires an angular integration over the phase matrix For the cases with symmetries e g random orientation the data files will not become too large even if we store additionally the absorption vector The dimensions are frequency temperature za inc aa_inc vector element 8 2 3 Definition of ptypes Ptype essentially classifies the scattering elements regarding their orientation As described above this effects the optimal choice of the coordinate system to represent the scattering element in Below we give a more detailed description of each possible ptype classification in the model macroscopically isotropic The ptype value macroscopically isotropic refers to macroscopically isotropic and mirror symmetric scattering media It covers totally randomly oriented particles as well as spher ical particles For this type of scattering media the optical properties are calculated in the so called scattering frame as shown in Figure 8 1 In this coordinate system the z axis corre sponds to the incident direction and the xz plane coincides with the scattering plane Using th
182. ons from H20 refractivity from refr_index_airThayer How ever they are smaller than when refraction by H20 is not accounted for Chapter 8 Description of clouds 8 1 Introduction In the Earth s atmosphere we find liquid water clouds consisting of approximately spherical water droplets and cirrus clouds consisting of ice particles of diverse shapes and sizes We also find different kinds of aerosols In order to take into account this variety the model allows to define several scattering elements A scattering element is either a specific single particle or a particle ensemble e g an ensemble following a certain size or shape distribution The scattering element can represent particles that are completely randomly oriented azimuthally randomly oriented or arbitrarily oriented Each scattering element is characterized by its single scattering properties and a field of particle number densities For each grid point in the cloud box the atmospheric volume that enclosed all scattering particles the single scattering properties of all scattering elements weighted by their respective particle number density at this location are summed up to derive the cloud ensemble optical properties The scat_data structure contains the single scattering properties K a and Z for each of the scattering elements In scat_data the single scattering properties are stored in different coordinate systems depending on the kind of particle For instanc
183. ontents and the index for navigating through the user guide 1 2 Documentation We know that the ARTS documentation is far from perfect It is quite complete in some areas but patchy in others It also contains bugs and more serious errors We are struggling to make it as good as possible but it is ongoing work and we do not have any direct funding for it All help from users to extend or correct the documentation is highly appreciated Having said that the documentation that already is available for ARTS is described in the following subsections 1 2 1 Guide documents ARTS User Guide This document ARTS Developer Guide Guide for ARTS developers ARTS Theory Describes the theoretical basis for some parts of ARTS 1 2 2 Articles Buehler et al 2005 General description of the old ARTS version without scattering Many basic features are still the same so this article is relevant also for the current ARTS version Eriksson et al 2011 Introduction and overview to ARTS 2 Emde et al 2004 Describes the Discrete Ordinate Iterative method DOIT for handling scattering 1 2 DOCUMENTATION 5 Davis et al 2005 Describes the Monte Carlo scattering method Eriksson et al 2006 Describes the calculation approach for the incorporation of sensor characteristics Buehler et al 2010 Describes a method to efficiently handle broadband infrared chan nels that is implemented in ARTS Buehler et al 2011 Descri
184. orption species list abs_species All valid tags are listed in Table 6 4 Secondly read in WSV abs_cia_data which contains the tabulated binary absorption cross sections from a file 44 GAS ABSORPTION Class Tag name Water vapor continua Complete water vapor models Carbon dioxide continua Oxygen continua Complete oxygen models Nitrogen continua Condensate absorption models H20 SelfContStandardType H20 ForeignContStandardType H20 ForeignContMaTippingType H20 ContMPM93 H20 SelfContCKD222 H20 ForeignContCKD222 H20 SelfContCKD242 H20 ForeignContCKD242 H20 SelfContCKD24 H20 ForeignContCKD24 H20 SelfContCKDMT 100 H20 ForeignContCKDMT100 H20 SelfContCKDMT252 H20 ForeignContCKDMT252 H20 ForeignContATMO1 H20 CP98 H20 MPM87 H20 MPM89 H20 MPM93 H20 PWR98 CO2 CKD241 CO2 CKDMT100 CO2 CKDMT252 CO2 SelfContPWR93 CO2 ForeignContPWR93 CO2 SelfContHo66 CO2 ForeignContHo66 02 CIAfunCKDMT100 02 v0vOCKDMT100 02 v1vOCKDMT100 02 visCKDMT252 02 SelfContStandardType 02 SelfContMPM93 02 SelfContPWR93 02 PWR98 02 PWR93 02 PWR88 02 MPM93 02 MPM92 02 MPM89 02 MPM87 O2 MPM85 02 TREOS N2 SelfContMPM93 N2 SelfContPWR93 N2 SelfContStandardType N2 SelfContBorysow N2 CIArotCKDMT100 N2 CIAfunCKDMT100 N2 CIArotCKDMT252 N2 CIAfunCKDMT252 N2 DryContATMO1 liquidcloud MPM93 icecloud MPM93 rain MPM93 Table 6 3 ARTS continua and complete absorption models The molecular species can be inf
185. orption in radiative transfer simulations The interface between the RT part of ARTS and the absorption part of ARTS is the agenda propmat_clearsky_agenda see Figure 6 1 RT functions execute this agenda whenever they need local absorption matrices In a typical ARTS run the agenda will be executed many times over for different points in the atmosphere See the built in documentation for the ex act input and output arguments of the agenda The idea is that input arguments are the local atmospheric conditions temperature pressure trace gas volume mixing ratios magnetic field etc The output of the agenda is a single variable propmat_clearsky a tensor with dimen sions of absorption species frequency Stokes dimension and Stokes dimension Stokes dimension of one thus emulates scalar absorption The physical quantity corresponding to this variable is the clear sky propagation matrix A as defined in Equation 4 8 It describes all non scattering extinction effects that is absorption and related polarization effects An inside view of propmat_clearsky_agenda is given in Figure 6 2 The agenda can contain a number of different workspace methods that in some way or other compute prop mat_clearsky See the built in documentation of the individual methods to learn more File agendas arts one of the standard include controlfiles predefines some typical alterna tives how propmat_clearsky_agenda can be set for different purposes 6 5 Calcula
186. os Vector The end position of the propagation path If the point is placed inside the atmosphere this field is redundant as it is equal to the first row of pos but identifies the sensor position for observations from space 10 6 FURTHER READING 85 end los Vector The line of sight at the end point of the propagation path Provides addi tional information if the sensor is placed above the top of the atmosphere and gives then the observation direction of the sensor end_Istep Numeric The distance between end_pos and the first position in pos This value is non zero just if the sensor is placed above the top of the atmosphere Hence this length corresponds to propgation 1f free space n 1 pos Matrix The position of the propagation path points inside the atmosphere This matrix has np rows and up to 3 columns Each row holds a position where column 1 is the radius column 2 the latitude and column 3 the longitude cf Section 5 2 1 The number of columns for 1D and 2D is 2 while for 3D it is 3 The latitudes are stored for 1D cases as these can be of interest for some applications and are useful 1f the propagation path shall be plotted The latitudes for 1D give the angular distance to the sensor see further Section 3 2 The propagation path is stored in reversed order that is the position with index 0 is the path point closest to the sensor and equals start pos if it is inside the atmosphere los Matrix The line of sight of
187. ositive values towards the east Longitudes can have values from 360 to 360 When the difference between the last and first value of the longitude grid is 360 then the whole globe is considered to be covered The user must ensure that the atmospheric fields for 6 and P 360 are equal If a point of propagation path is found to be outside the range of the longitude grid this will results in an error if not the whole globe is covered When possible the longitude is shifted with 360 in the relevant direction 3 3 Atmospheric grids and fields As mentioned above the vertical grid of the atmosphere consists of a set of layers with equal pressure the pressure grid p_grid This grid must of course always be specified The upper end of the pressure grid gives the practical upper limit of the atmosphere as vacuum is assumed above With other words no absorption and refraction take place above the uppermost pressure level A latitude grid lat_grid must be specified for 2D and 3D For 2D the latitudes shall be treated as the angular distance along the orbit track as described above in Section 3 2 The 18 DESCRIPTION OF THE ATMOSPHERE Figure 3 3 Schematic of a 3D atmosphere Plotting symbols as in Figure 3 2 Radii and fields are here defined to vary linearly along the latitude and longitude grid points This means that the radius of a pressure level has a bi linear variation inside the area limited by two latitude and longitude g
188. osphere and the refractive index deviates significantly from 1 at this point 78 CLEAR SKY RADIATIVE TRANSFER 9 6 Single pencil beam calculations The text above assumes that yCalc is used This method can always be used but yCalc is not mandatory if the simulations only deal with monochromatic data for a single line of sight In this case it could be more handy to use iyCalc which basically is a direct call of 1y_main_agenda A reason for selecting iyCalc is that a larger set of auxiliary quantities can be extracted Sec 9 8 On the input side the main difference when using iyCalc is that the observation position and line of sight are specified by rte pos and rte los instead of sensor pos and sensor_los The calculated radiances are returned as the matrix iy instead of the vector y No automatic unit conversion is made inside iyCalc This is instead handled separately by iyApplyUnit 9 7 Dispersion The clear sky radiative transfer methods handle all frequencies in f_grid in parallel for efficiency reasons One consequence of this feature is that only a single propagation path is calculated that is assumed to be common for all frequencies With other words dispersion is not considered This is in general an acceptable simplification but exceptions exist where one example is radiative transfer through the ionosphere at frequencies approaching the plasma frequency When dispersion is expected to give a significant impact o
189. p_grid lat_grid and lon_grid is to build up the mesh on which the atmospheric fields are defined This means that the spacing of these grids shall be selected having the representation of the atmospheric fields in mind That is the spacing shall be fine enough that the atmospheric field is sufficiently well approximated by the piece wise multi linear representation between the grid crossings The result is that a finer spacing must be used to represent correctly atmospheric fields with a lot of structure while the grids can have fewer points when the atmospheric fields are smooth The accuracy when performing the actual radiative transfer calculations depends on the refinement of the expressions used and the discretisation of the propagation path If Equa tion 9 2 is used the underlying assumption is that the Planck function and the absorption vary linearly along the propagation path step These assumptions are of course less vio lated if the path step length is made small An upper limit of the path step length is set by ppath_Imax In many cases it should suffice to just include path points at the crossings of the atmospheric grids ppath_Imax lt 0 An exception can be limb sounding where the path step length can be very long around the tangent point but a limit of about 25 km should suffice normally See also Section 10 3 Chapter 10 Propagation paths A propagation path is the name given in ARTS to the way the radiation travels to reach
190. r non scattering instead for scattering in cloudbox cases where in non scattering setups the concentration data is stored together with gas concentrations in vmr_field while for scattering setups it is stored separately in pnd_field and that for the single scattering data and concentration fields the exact same data can be applied Beside being able to re use particle data from scattering cases this method is also ad vantageous compared to the particles as continuum models implementations as it allows for directional dependent absorption and for polarization effects that occur e g with non spherical particles In the default case absorption by particles is applied both in the extinction and emission terms of the radiative transfer equation i e both right hand terms in Equation 4 13 How ever using a flag propmat_clearsky AddParticles can apply total particle extinction instead It shall be noted that while this applies the correct extinction it also creates an unphysical emission term Hence this option shall only be applied when the source term is negligible as e g for occultation measurements Be aware that both propmat_clearskyAddParticles and the scattering methods use scat_data to store the particle single scattering data Hence it is straight forward that these methods can not be applied simultaneously In one ARTS run all particles are handled either as scattering entities when using the scattering modules or as grey body a
191. rection History 130219 Revised by Patrick Eriksson 020315 First version by Patrick Eriksson 16 DESCRIPTION OF THE ATMOSPHERE Atmospheric grids Ground Geoid A Cloud box aN g Atmospheric field Figure 3 1 Schematic of a 1D atmosphere The atmosphere is here spherically symmetric This means that the radius of the ellipsoid the surface and all the pressure levels are constant around the globe The fields are specified by a value for each pressure level The extension of the cloud box is either from the surface up to a pressure level or between two pressure levels which is the case shown in the figure The figure indicates also that the surface must be above the lowermost pressure level Geoid in the legend should be Ellipsoid 3 2 Atmospheric dimensionality The structure of the modelled atmosphere can be selected to have different degree of com plexity the atmospheric dimensionality There exist three options 1D 2D and 3D where 1D and 2D can be seen as special cases of 3D The significance of these different atmo spheric dimensionalities and the geometrical coordinate systems used are described below in this section The atmospheric dimensionality is selected by setting the workspace vari able atmosphere_dim to a value between and 3 The atmospheric dimensionality is most easily set by the functions AtmosphereSet1D AtmosphereSet2D and AtmosphereSet3D 1D A 1D atmosphere can
192. requencies 78 80 iyRadioLink 100 102 iyReplaceFromAux 102 1ySurfaceRtpropA genda 90 iyTransmissionStandard 99 jacobianAddAbsSpecies 114 jacobianAddFreqShift 118 jacobianAddFreqStretch 118 jacobianAddPointingZa 118 jacobianAddPolyfit 119 jacobianAddSinefit 119 jacobianAdd Temperature 116 jacobianAddWind 116 jacobianClose 108 jacobianInit 108 jacobianOff 108 MatrixCBR 77 Matrix UnitIntensity 100 102 opt prop sptFromMonoData 128 ParticleType2abs_speciesAdd 52 ParticleTypeAdd 67 ParticleTypeAddAll 67 pha_mat_sptFromDataDOITOpt 127 pha mat sptFromMonoData 127 pnd_fieldCalcFrompnd_field_raw 67 pnd_fieldCalcFromscat_speciesFields 39 ppath_stepGeometric 81 ppath_stepRefractionBasic 81 ppathCalc 73 ppathFromRtePos2 100 ppathStepByStep 81 Print 6 propmat_clearsky_fieldCalc 42 43 56 propmat_clearskyAddFaraday 42 51 97 propmat clearskyAddFromLookup 42 55 propmat clearskyAddOnTheFly 38 39 42 43 propmat clearskyAddParticles 42 52 propmat_clearskyAddZeeman 42 45 propmat_clearskyInit 42 ReadXML 42 refellipsoidEarth 87 refellipsoidForAzimuth 88 refellipsoidMars 87 refellipsoidOrbitPlane 88 refr_index_airFreeElectrons 58 78 refr_index_airIR 58 refr_index_airMW general 58 refr_index_airThayer 58 rte_losGeometricFromRtePosToRtePos2 100 scat_data_monoCalc 127 128 scat_data_singleT matrix 64 sensor responselnit 93 sensorOff 93 String
193. riable is a function of frequency possibly being a Stokes vector for each frequency Vari ables fulfilling this constrain are defocusing loss t atmospheric extinction loss te free space loss tf bending angle extra path delay and total Faraday rotation 15 4 15 2 4 Free space loss and sensor characteristics It is repeated that total free space loss is in ARTS defined as given above in Equation 15 4 o tp T It is stressed that this deviates from what appears to be a more common even standard definition of free space loss A 471 where A is the wavelength of the signal This expression encompasses the relationship between the receiver s effective antenna area and its gain Ae gyA 47 This later definition was avoided to keep atmospheric and sensor effects clearly separated in ARTS Hence the definition of free space loss has consequences for how to specify sensor characteristics A possible choice is to treat pure multiplicative factors such as p g and Ae separately and let ARTS handle the pure radiative propagation effects In this case p Eq 15 5 shall be set to unity MatrixUnitIntensity handles this case If e g Faraday rotation or particle scattering is of concern stokes_dim must be gt 1 and also the polarisation of the transmitted signal is of concern and must be included in the specification of p e g p 1 1 0 0 when V or His transmitted The next step towards a more complete tr
194. rid values while the atmospheric fields have a tri linear variation inside the grid cells latitude angle is throughout calculated for the vector going from the centre of the coordinate system to the point of concern Hence the latitudes here correspond to the definition of the geocentric latitude and not geodetic latitudes Sec 11 1 1 This is in accordance to the definition of geometric altitudes Sec 3 1 For 3D a longitude grid lon grid must also be specified Valid ranges for latitude and longitude values are given in Section 3 2 If the longitude and latitude grids are not used for the selected atmospheric dimensionality then the longitude grid for 1D and 2D and the latitude grid for 1D must be set to be empty The atmosphere is treated to be undefined outside the latitude and longitude ranges covered by the grids if not the whole globe is covered This results in that a propagation path is not allowed to cross a latitude or longitude end face of the atmosphere if such exists it can only enter or leave the atmosphere through the top of the atmosphere the uppermost pressure level See further Section 9 2 The volume covered by the grids is denoted as the model atmosphere The basic atmospheric quantities are represented by their values at each crossing of the involved grids indicated by thick dots in Figure 3 2 or for 1D at each pressure level thick dots in Figure 3 1 This representation is denoted as the field of the quanti
195. rk the same when they reach the ab sorption calculations but we urge the user to be sure that line mixing data and the line database are of the same origin as the calculations will be of poor quality if this is not the case Example control file meta flow for code calculating line mixing using abs_linesReadFromLBLRTM Define your atmospheric species abs_speciesSet species 02 Z LM CO2 LM read from a line database abs_linesReadFromLBLRIM filename aer fmin lel fmax 1e20 Note that this meta flow is very similar to using an ARTS catalogue with line mixing These methods are easy because they identity what line mixing method will be used directly from the catalogue It should be noted that LBLRTM also gives the non resonant term of the molecular oxygen spectra as a low frequency line If it is desired to calculate this term the lower frequency range must contain the line as per the definition of our reading routines Example control file meta flow for code calculating line mixing using line_mixing_dataMatch Define your atmospheric species abs_speciesSet species 02 Z LM CO2 LM Read from a line database abs_linesReadFromHITRAN filename HITRAN2012 par fmin lel fmax 1e20 Sort the line records into the format ARTS wants 6 5 CALCULATING GAS ABSORPTION 49 abs_lines_per_speciesCreateFromLines Create a way to temporally store 02 line mixing data ArrayOfLineMixingRecordCreate l
196. s VMR However sometimes it is of interest to explicitly calculate and output absorption e g for testing and validating modules of the absorption calculation for model comparisons for plotting and analyzing absorption coefficients etc Table 6 2 step 5c lists high and low level workspace methods for this purpose In particular the method prop mat_clearsky_fieldCalc provides the absorption matrices i e polarized absorption coeffi cients per species tag group for an entire atmospheric scenario and the complete frequency grid Chapter 7 Refractive index FIXME Write a proper introduction Comment on that this is not the complex re fractive index also covering absorption Refractive index here restricted to the real part of the refractive index which is basically a complex quantity with the imaginary part expressing absorption describes several effects of matter on propagation of electromagnetic waves This particularly includes changes of the propagation speed of electromagnetic waves which leads to a delay of the signal as well as a change of the propagation direction a bending of the propagation path The latter is commonly called refraction Several components in the atmosphere contribute to refraction hence to the refractive index the gas mixture air solid and liquid constituents clouds precipitation aerosols and electrons ARTS includes mechanisms for deriving the contributions from gases and electrons
197. ses the ellipsoid data must be adopted The assumption inside ARTS is that the 2D plane goes through the North and South poles The polar radius to apply should then match History 120224 Extended and revised Patrick Eriksson 050613 First version finished by Patrick Eriksson 88 REFERENCE ELLIPSOID AND SURFACE PROPERTIES geoid ellipsoid zenith Figure 11 1 Definition of the local tangent ellipsoid radii re and rp geo centric latitude a and geodetic latitude a The dotted line is the normal to the local tangent of the ellipsoid The zenith and nadir directions and geometri cal altitudes are here defined to ES follow the solid line r e x the real ellipsoid radius at the highest latitude inside the satellite orbit plane The method refellipsoidOrbitPlane performs this operation Further for 1D cases the reference ellipsoid is by definition a sphere and the radius of this sphere shall be selected in such way that it represents the local shape of a reference ellipsoid This is achieved with refellipsoidForAzimuth that sets re to the local radius of curvature of the ellipsoid and e to zero The curvature radius differs from the local radius except at the equator and an east west direction For example at the equator and a north south direction the curvature radius is smaller then the local radius while at the poles for all directions it is greater see further Figure 11 2 The curvature radius re of an e
198. skyAddParticles see Sec 6 5 8 Alternative propmat_clearskyAddFromLookup instead of propmat_clearsky AddOnTheFly extract absorption from pre calculated lookup table see Sec 6 6 Note that the lookup table cannot contain absorption for Zeeman tagged species Faraday rotation and particles due to their directional dependencies Variable abs_lookup Methods abs lookupCalc Alternative Load lookup table from file with ReadXML it then has to be adapted to the current calculation and checked with abs_lookupAdapt Variable propmat_clearsky_field abs_coef Methods high level propmat_clearsky_fieldCalc Methods low level abs_xsec_per _speciesInit abs_xsec_per_speciesAddLines the core method for the actual line by line calculation used internally by all higher level methods abs_xsec_per_speciesAddConts add continua or complete absorption models see Section 6 5 3 abs_xsec_per_speciesAddCIA add collision induced absorption see Section 6 5 4 abs_coefCalcFromXsec calculate absorption coefficients from absorption cross sections Table 6 2 Steps for line by line absorption calculation and associated ARTS workspace variables and methods 6 5 CALCULATING GAS ABSORPTION 43 lines but then these lines are hardwired into the absorption model itself Consequently the first four steps in Table 6 2 are not needed for these models The pure continua are intended to be used together with an explicit ARTS line by
199. space or the cloud box The inten sity field entering the cloud box is in some cases calculated by calls of iy_main_agenda with cloud box deactivated and the radiative background for these calculations is then space or the surface This results in that space is normally the ultimate radiative background for the calculations The exception is for propagation paths that intersects with the surface and the surface is treated to act as a blackbody For such cases the propagation path effectively starts at the surface 9 4 Basic radiative transfer variables and expressions This section describes how the core radiative transfer equation is solved practically in ARTS As mentioned in this chapter focus is put on emission measurements Local thermo dynamic equilibrium LTE is throughout assumed The equation to solve is Equation 4 13 ds a Aq b s Aas Bag 9 1 where the involved quantities are defined and discussed in Section 4 2 9 4 1 Unpolarised absorption Let s start with the simpler case of non polarised absorption that is the absorption is inde pendent of polarisation state For unpolarised absorption the matrix A is diagonal with all diagonal elements equal and only the first of the elements of aa is non zero The radiative transfer equation above can be solved in many ways and with different level of refinement The standard approach in ARTS is to solve the radiative transfer from one point of the propagation
200. t K 16 26 T Al R Al e Ox Ox 2 CT 16 27 The question is when Eq 16 27 can be applied We have three main cases e The equation is valid as long as the species to be retrieved has unpolarised absorption This as C and C 1 can then both be written as 0 1 and xC i 1Ci 1 2 is commutative with any K e Eq 16 27 is not valid when all three matrices involved K and the two C matrices have off diagonal elements at the same time However this is is unlikely case The main causes to polarised absorption are free electrons Zeeman splitting and par ticles and these mechanisms are mainly found in the ionosphere mesosphere and troposphere respectively e The remaining case is when K is diagonal but the species to be retrieved has po larised absorption The likely situation is that both C and C are non diagonal on the same time and the condition of Eq 16 9 is not fulfilled However as argued below Eq 16 27 should still be approximately valid and this is used as working hypothesis in ARTS An alternative way to express the propagation matrix is i Ti 1 ma AiR with CG Ci With this formulation even the third case above would be handled as we now only have two matrices to consider and C is commutative with a diagonal K We would then have OT E TAL OT Ox 2 As C and C 1 are very similar matrices and then also close to C
201. t different type of characteristics to response values As a user of ARTS the main practical issue is to understand the different file formats used for the different sensor parts For the moment this is only described mainly through the on line documentation 12 1 General In principle a sensor must always be specified However if this shall be a hypothetical sensor just providing the monochromatic pencil beam data coming out of the atmospheric radiative transfer calculations use sensorOff sensor_response is in this case just an identity matrix For other cases the definition of the sensor characteristics is initiated by calling sen sor_responselnit The natural order to call the main functions for the different sensor parts should be to follow the radiation through the instrument That is the antenna should nor mally be the first part to consider If the order can be changed depends on the conditions For example for a double side band receiver the antenna must be considered before the mixer if the antenna response differs between the two bands If the same antenna response History 110826 A simple version by Patrick Eriksson 94 SENSOR CHARACTERISTICS can be assumed for both bands the same result is obtained even 1f the mixer is introduced before the antenna Each response is defined for some grid All responses are assumed to be zero outside the range covered by the grid even if the end values deviate from zero A positive
202. t the top of the atmosphere space is the radia tive background The normal case should be to set the radiation at the top of the atmosphere to be cosmic background radiation An exception is when the sensor is directed towards the sun The radiative background at the top of the atmosphere is determined by iy_space_agenda If a propagation path is totally outside the model 74 CLEAR SKY RADIATIVE TRANSFER Figure 9 1 Examples on allowed propagation paths for a 2D atmosphere The atmosphere is plotted as in Figure 3 2 beside that the points for the atmospheric fields are not emphasised The position of the sensor is indicated by an asterisk x the points defining the paths are plotted as circles o joined by a solid line The part of the path outside the atmosphere not included in the path structure is shown by a dashed line Path points corresponding to a tangent point are marked by an extra plus sign but note that these no longer are explicitly included as path point in contrast to ARTS 2 0 and earlier The shown paths include the minimum set of definition points There exists also the possibility to add points inside the grid cells for example to ensure that the distance between the path points does not exceed a specified limit Figure 9 2 Examples on allowed propagation paths for a 1D atmosphere with an activated cloud box Plotting symbols as in Figure 9 1 When the sensor is placed inside the cloud box the path is defi
203. tative Spectroscopy and Radiative Transfer 98 446 457 2006 Buehler S A V O John A Kottayil M Milz and P Eriksson Efficient radiative transfer simulations for a broadband infrared radiometer Combining a weighted mean of repre sentative frequencies approach with frequency selection by simulated annealing Journal of Quantitative Spectroscopy and Radiative Transfer 111 602 615 2010 Buehler S A P Eriksson and O Lemke Absorption lookup tables in the radiative transfer model arts Journal of Quantitative Spectroscopy and Radiative Transfer 112 1559 1567 2011 Dattorro J Convex Optimization amp Euclidean Distance Geometry Meboo Publishing USA 2011 Davis C C Emde and R Harwood A 3D polarized reversed monte carlo radiative trans fer model for mm and sub mm passive remote sensing in cloudy atmospheres IEEE Transactions on Geoscience and Remote Sensing 43 1096 1101 2005 Donovan D P and A C A P van Lammeren Cloud effective particle size and water content profile retrievals using combined lidar and radar observations 1 Theory and ex amples Journal of Geophysical Research 106 27425 27448 2001 Emde C A polarized discrete ordinate scatterig model for radiative transfer simulations in spherical atmospheres with thermal source Ph D thesis University of Bremen 2005 Emde C and T R Sreerekha Development of a RT model for frequencies between 200 and 1000 GHz WP1 2 Model Review
204. tch_za xml IndexSet ybatch_n 5 AgendaSet ybatch_calc_agenda Print ybatch_index 0 Extract t_field tensor4_1 ybatch_index Extract z_ field tensor4 2 ybatch_index Extract vmr_field tensor5_1 ybatch_index Extract sensor_los tensor3_1 ybatch_index yCalc ybatchCalc WriteXML ascii ybatch If you then want to repeat the calculations for example with another propagation path step length e g 25 km it is sufficient to add the lines AgendaSet ppath_step_agenda ppath_stepGeometric ppath_step atmosphere_dim lat_grid lon_grid z_field refellipsoid z_surface 25e3 ybatchCalc WriteXML ascii ybatch ybatch_run2 xml Part IV Radiative transfer dedicated scattering methods Chapter 18 Scattering calculations The DOIT module The Discrete Ordinate ITerative DOIT method is one of the scattering algorithms in ARTS The DOIT method is unique because a discrete ordinate iterative method is used to solve the scattering problem in a spherical atmosphere Although the DOIT module is implemented for 1D and 3D atmospheres it is strongly recommended to use it only for 1D because the Monte Carlo module Chapter 19 is much more appropriate for 3D calulations More appropriate in the sense that it is much more efficient A literature review about scatter ing models for the microwave region which is presented in Emde and Sreerekha 2004 shows that former imple
205. th of the distribution The other two pa rameters can be linked to the effective radius Resp and the ice mass content IMC as fol lows at3 b 8 4 _ IMC a F 3npb DT a 8 5 where p is the density of the scattering medium and T is the gamma function For cirrus clouds p corresponds to the bulk density of ice which is approximately 917 kg m Generally the effective radius Refs is defined as the average radius weighted by the particles cross sectional area 1 Reff a o A r rn r dr 8 6 Tmin where A is the area of the geometric projection of a particle The minimal and maximal particle sizes in the distribution are given by Tmin and Tmar respectively In the case of spherical particles A rr The average area of the geometric projection per particle A is given by ay a Teg Alma AT Se n rjdr Tmin 8 7 The question is how well a gamma distribution can represent the true particle size distri bution in radiative transfer calculations This question is investigated by Evans et al 1998 The authors come to the conclusion that a gamma distribution represents the distribution of realistic clouds quite well provided that the parameters Reff IMC and a are chosen cor rectly They show that setting a 1 and calculating only Resp gave an agreement within 15 in 90 of the considered measurements obtained during the First ISCCP Regional Experiment FIRE Therefore for all calculations including g
206. th_Imax When refraction is considered the ray tracing moves forward in steps following ppath_Iraytrace When another step of this size would result in a distance gt ppath_Imax the present point is added to lt ppath step A consequence of this is that additional points are likely not evenly spaced The distance between most points will be ppath_lraytrace times an integer value but the distance between the last additional point and the grid border can be any number lt ppath_Imax If ppath_step is set to be negative no additional points are included as for geometrical calculations As points are always included in the propagation paths at the crossings of the atmo spheric grids making these grids finer will give shorter path steps However it is neither good practise or efficient to use the atmospheric grids to control the accuracy of the radia tive transfer calculations An upper limit on the step length ppath_Imax shall be applied for this purpose 10 4 Tangent points The term tangent point refers to the point of a limb sounding path with the lowest altitude For 1D cases this definition is clear but for 2D and 3D calculations there are complications First of all there are two different possible definitions the point having the lowest radius ie distance to the planets centre or the point with the lowest altitude ie vertical distance to the reference ellipsoid The later is the more important with respect to optical t
207. that there is an altitude gap between the surface and the lowermost pressure level That is the surface pressure must be smaller than the pressure of the lower most vertical grid level On the other hand it is not necessary to match the surface and the first pressure level the pressure grid can extend below the surface level 20 DESCRIPTION OF THE ATMOSPHERE Figure 3 4 A latitudinal or longitudinal cross section of a 3D atmosphere Plotting sym bols as in Figure 3 1 Radii and fields inside the cross section match the definitions for 2D The vertical extension of the cloud box is defined identical for 1D and 3D The horizontal extension of the cloud box is between two latitude and longitude grid positions where only one of the dimensions is visible in this figure 3 7 The cloud box In order to save computational time calculations involving scattering are limited to a special atmospheric domain This atmospheric region is denoted as the cloud box The distribution of scattering matter inside the cloud box is specified by pnd field see further Section 4 2 2 The cloud box is defined to be rectangular in the used coordinate system with limits exactly at points of the involved grids This means for example that the vertical limits of the cloud box are two pressure levels For 3D the horizontal extension of the cloud box 1s between two points of the latitude grid and likewise in the longitude direction Fig 3 4 The latitude and lo
208. the propagation path Propagation paths are described by a set of points on the path and the distance along the path between the points These quantities and a number of auxiliary variables are stored together in a structure described in Section 10 5 The path points are primarily placed at the crossings of the path with the atmospheric grids p grid lat_grid and lon grid A path point is also placed at the sensor if it is placed inside the atmosphere Points of surface reflections are also included if such exist More points can also be added to the propagation path for example by setting an upper limit for the distance along the path between the points This is achieved by the variable ppath_Imax see further Sections 9 9 and 10 1 The propagation paths are determined basically by starting at the sensor and following the path backwards by some ray tracing technique If the sensor is placed above the model atmosphere geometrical calculations are used as there is no refraction in space to find the crossing between the path and the top of the atmosphere where the ray tracing then starts Paths are tracked backwards until the top of the atmosphere or to an intersection with the cloud box or the surface The propagation path or paths before a surface reflection is calculated when determining the up welling radiation from the surface Section 11 3 Example on propagation paths are shown in Figures 9 1 and 9 2 Not all propagation paths are allowed
209. this variable to e g 1 if you don t want to apply such a length criterion A straightforward but inefficient treatment of refraction is provided by ppath_stepRefractionBasic This method divides the propagation path into a series of geomtrical ray tracing steps The size of the ray tracing steps is selected by ppath_Iraytrace This variable affects only the ray tracing part the distance between points of the propaga tion path actually returned is controled by ppath_Imax as above At each ray tracing step the refractive index is evaluated according to the specification of refr_index_air_agenda Several methods to determine refractive index are available see Chapter 7 History 120202 Revised and parts moved to ARTS Theory Patrick Eriksson 030310 First complete version written by Patrick Eriksson 82 PROPAGATION PATHS Figure 10 1 Tracking of propagation paths For legend see Figure 10 2 The figure tries to visualize how the calculations of propagation paths are performed from one grid cell to next In this example the calculations start directly at the sensor position x as it placed inside the model atmosphere The circles give the points defining the propagation path Path points are always included at the crossings of the grid cell boundaries Such a point is then used as the starting point for the calculations inside the next grid cell 10 2 Calculation approach The propagation paths are calculated in steps as outlin
210. ting gas absorption This section deals with calculating gas absorption matrices in ARTS This can typically occur in three different contexts as on the fly absorption matrix calculation within the radiative transfer calculation when preparing a gas absorption lookup table see Section 6 6 or when the user is only interested in the absorption itself see Section 6 7 In all these cases the same agenda is used to actually calculate absorption abs_xsec_agenda Outside views of this agenda are shown in Figure 6 2 as input to prop 38 GAS ABSORPTION onl oa a a ea a Initialization line by line or from lookup table OR 7 Additional spectroscopic parameters Agenda Output Single scattering data Figure 6 2 An inside view of propmat_clearsky_agenda At the same time this gives an outside view of abs_xsec_agenda as input to propmat_clearskyAddOnTheFly in the context of on the fly absorption generation 6 5 CALCULATING GAS ABSORPTION 39 Spectroscopic data i I z Y abs_species f_grid abs_xsec_agenda abs_lookupCalc Figure 6 3 An outside view of abs_xsec_agenda in the context of absorption lookup table generation mat_clearskyAddOnTheFly in the context of on the fly absorption generation and Figure 6 3 as input to abs_lookupCalc in the context of absorption lookup table generation An inside view of abs_xsec_agenda is given in Figure 6 4 The agenda can contain a number of different worksp
211. tion calculation is performed when the agenda is executed every time absorption coefficients are needed In the second case the absorption coefficients are extracted from a pre calculated lookup table On invocation an agenda executes its methods one after the other The inputs and outputs defined for the agenda must be satisfied by the invoked workspace methods E g if an agenda has propmat clearsky in its list of output workspace variables at least one workspace method which generates propmat_clearsky must be added to the agenda in the controlfile 1 4 Include controlfiles ARTS controlfiles can include other ARTS controlfiles which is achieved by statements such as INCLUDE general arts This mechanism is used to predefine gen eral default settings settings for typical applications and or settings for the simulation of well known instruments A variety of include controlfiles are collected in directory controlfiles of the ARTS distribution You should normally at least include the file general arts which contains general default settings Because giving the full path for every include file is inconvenient ARTS will look for include files in a special directory This can be set by the command line option I lt includepath gt or by the environment variable ARTS_INCLUDE_PATH If none of these are set ARTS will assume the include path to point to the includes directory in the ARTS distribution The file agendas arts is also useful b
212. trol file by defining several agendas Calculation of the scattering integral To calculate the scattering integral ARTS Theory Equation 8 7 the phase matrix pha_mat is required How the phase matrix is calculated is defined in the agenda pha_mat_spt_agenda Calculation of the phase matrix AgendaSet pha_mat_spt_agenda Optimized option pha_mat_sptFromDataDOITOpt Alternative option pha_mat_sptFromMonoData If in doit_mono_agenda the optimized method DoitScatteringDataPrepare is used we have to use here the corresponding method pha_mat_sptFromDataDOITOpt Otherwise we have to use pha_mat_sptFromMonoData To do the integration itself we have to define doit_scat_field_agenda 128 SCATTERING CALCULATIONS THE DOIT MODULE AgendaSet doit_scat_field_agenda doit_scat_fieldCalcLimb Alternative doit_scat_fieldCalc Here we have two options One is doit_scat_fieldCalcLimb which should be used for limb simulations for which we need a fine zenith angle grid resolution to represent the radi ation field This method has to be used if a zenith angle grid file is given in DoitAngu larGridsSet The scattering integral can be calculated on a coarser grid resolution hence in doit_scat_fieldCalcLimb the radiation field is interpolated on the equidistant angular grids specified in DoitAngularGridsSet by the generic inputs Nza and Naa Alternatively one can use doit_scat_fieldCalc where this interpolatio
213. ts only in slower calculations Simulations where stokes_dim is two or higher are denoted as vector radiative transfer while scalar radiative transfer refers to the case when only the first Stokes component is considered 4 2 The radiative transfer equation The radiative transfer problem can only be expressed in a general manner as a differential equation One version for vector radiative transfer is ds v r fi di where y is frequency r represents the atmospheric position ni is the propagation direction at r l is distance along K is the propagation matrix je represents the emission at the point and js covers the scattering from other directions into the propagation direction K v r fi s v r f je v r j v r n 4 6 4 2 1 Propagation effects Three mechanisms contribute to the elements of the propagation matrix absorption scat tering and magneto optical effects Absorption and scattering can together be denoted as extinction referring to that these two mechanisms result in a decrease of the intensity 1 A common name for K is also the extinction matrix The extinction processes also affect the Q U and V elements of the Stokes vector If the degree of polarisation p VR U V p EEE 4 7 is kept constant or not depend on symmetry properties of the attenuating media For ex ample absorption of atmospheric gases is not changing p as long as the molecules have no preferred orientation which is a valid ass
214. ty The field must at least be specified for the geometric altitude of the pressure levels z field the temperature t_field and considered atmospheric species vmr_field The content and units of vmr_field are discussed in Section 4 2 2 3 4 GEO LOCATION OF 1D AND 2D 19 All the fields are assumed to be piece wise linear functions vertically with pressure altitude as the vertical coordinate and along the latitude and longitude edges of 2D and 3D grid boxes For points inside 2D and 3D grid boxes multidimensional linear interpolation is applied that is bi linear interpolation for 2D etc Note especially that this is also valid for the field of geometrical altitudes z_field Fields are rank 3 tensors For example the temperature field is T T P a 8 That means each field is like a book with one page for each pressure grid point one row for each latitude grid point and one column for each longitude grid point In the 1D case there is just one row and one column on each page The representation of atmospheric fields and other quantities is discussed further in Section 16 2 where the concept of basis functions is introduced In short the basis functions give the mapping from the set of discrete values to the continuous representation of the quantity 3 4 Geo location of 1D and 2D For 1D and 2D atmospheres lat_grid and lon_grid do not contain true geographical positions they are either empty or lat grid contains transformed
215. uctions are executed sequentially Controlfile somefile arts is executed by running arts somefile arts A minimal ARTS controlfile example the well known Hello World program is given in Figure 1 1 In this example the variable s is called a workspace variable We use this name to distinguish it from the variables that appear internally in the ARTS source code In a similar spirit the functions StringCreate StringSet and Print in the example are called workspace methods We use this name to distinguish them from the functions that appear internally in the ARTS source code For brevity we may sometimes drop the workspace qualifier and refer to them just as methods ARTS consists roughly of three parts Firstly the ARTS core contains the controlfile parser and the engine that executes the controlfile This part is quite compact and consti tutes only a small fraction of the total source code Secondly there is a large collection of workspace methods that can be used to carry out various sub tasks at the time of writing approximately 300 Thirdly there is a large number of predefined workspace variables at the time of writing more than 200 These predefined variables make it easier to set up controlfiles since they provide hints on how the different workspace methods fit together 1 3 ARTS AS A SCRIPTING LANGUAGE 7 Workspace variable f_grid The frequency grid for monochromatic pencil beam calculations Usage
216. umption beside when there is a significant Zeeman effect Non polarising absorption corresponds to that the propgation matrix can be written as al where a is the absorption coefficient and 1 is the identity matrix There are also examples on effects that change the polarisation state of the Stokes vector without affecting J These effects are caused by an interaction with the magnetic field and are thus denoted magneto optical Ionospheric Faraday rotation can approximately be seen as a pure magneto optical effect while the Zeeman effect cause both non isotropic absorption and has magneto optical aspects As Equation 4 6 indicates there are two source mechanisms that can act to increase the intensity emission and scattering 4 2 2 Absorbing species and scattering particles The complexity of the radiative transfer is largely dependent on whether scattering must be considered or not For this reason ARTS operates with two classes of atmospheric matter Absorbing species This class covers atmospheric matter for which scattering can be ne glected The set of species to consider is described by the workspace variable abs_species and the associated atmospheric fields are gathered into vmr_field As the name indicates this later variable is mainly containing volume mixing ratio VMR 4 2 THE RADIATIVE TRANSFER EQUATION 25 data but as this unit is not applicable in all cases also other units are accepted That is the unit for th
217. w 5 2 1 Sensor position The observation positions of the sensor are stored in sensor_pos This is a matrix where each row corresponds to a sensor position The number of columns in the matrix equals the atmospheric dimensionality 1 column for 1D etc The columns of the matrix from first to last are geometrical altitude latitude and longitude Accordingly row i of sensor pos for a 3D case is zi a 3 The sensor position can be set to any value but the resulting propagation paths also dependent on sensor los must be valid with respect to the model atmosphere see Section 9 2 An obviously incorrect choice is to place the senor below the surface altitude If the sensor is placed inside the model atmosphere any sensor line of sight is allowed this including the cases that the sensor is placed on the surface looking down and that the sensor is placed inside the cloud box One or several spectra can be calculated for each position as described in Section 5 3 The corresponding workspace variable for single pencil beam calculations is rte_pos that is an input argument to e g iy_main_agenda and iyCalc 30 COMPLETE CALCULATIONS zenith A line of sight Figure 5 2 Definition of zenith an gle 4 and azimuth angle w for a line of sight The figure shows a line of sight with a negative azimuth angle gt south north 5 2 2 Line of sight The viewing direction of the sensor the line of sight is descri
218. who want to do ARTS development work It can be accessed at http www sat ltu se arts misc arts doc doxygen html 1 2 6 Build instructions Instructions on how to configure and compile the ARTS source code can be found in the file README in the top directory of the ARTS distribution 6 INTRODUCTION Arts2 arts 1 14 122 StringCreate s Executing Arts StringSet s Hello World StringCreate Print s StringSet Print Hello World This run took 0 03s 0 03s CPU time Everything seems fine Goodbye Figure 1 1 Left A minimal ARTS controlfile example Right ARTS output when running this controlfile 1 2 7 Command line parameters ARTS offers a number of useful command line parameters In general there is a short form and a long form for each parameter The short form consists of a minus sign and a single letter whereas the long form consists of two minus signs and a descriptive name To get a full list of available command line parameters type arts h or arts help 1 3 ARTS as a scripting language One of the main goals in the ARTS development was to make the program as flexible as pos sible so that 1t can be used for a wide range of applications and new features can be added in a relatively simple manner As a result ARTS behaves like a scripting language An ARTS controlfile contains a sequence of instructions When ARTS is executed the controlfile is parsed and then the instr
219. y the surface or space i e where the propagation path starts The next step is to call iy_transmitter_agenda It should be noted that the same agenda is called independently if the radiative background is space or the surface It is up to the user to History 130206 Written PE partly based on text written originally by Bengt Rydberg 100 TRANSMISSION CALCULATIONS decide if these two cases shall be distinguished in some manner no workspace method for this task exists yet For these calculations the standard choice for iy _transmitter_agenda is MatrixUnitIntensity If this workspace method is used the output of iyTransmission Standard shows you the fraction of unpolarised radiation that is transmitted through the atmosphere and the polarisation state of the transmitted part The radiative transfer expression applied is cf Eq 9 7 Siy1 e Ms 15 1 where the extinction matrix is determined in the same manner as for emission cases Sec 9 3 In situations where the matrix K is diagonal the scalar form shown in Eq 9 5 is used The method determines automatically if any part of the propagation path is inside the cloud box if active If this is the case particle extinction is included in K following the same path step averaging as applied for pure absorbing species For this method scattering is purely a loss mechanism there is no gain by scattering into the line of sight A related concern is the treatment of the surface

ARTS User Guide

Contents

Download Pdf Manuals

Related Search

Related Contents