User guide for PRELES, a simple model for the assessment of gross
Contents
1. User guide for PRELES a simple model for the assessment of gross primary production and water balance of forests Mikko Peltoniemi Tuomo Kalliokoski Antti Jussi Lindroos Egbert Beuker Annikki Makela DELIVERABLE LIFE09 ENV FI 00057 1 Climate change induced drought effects on forest growth and vulnerability Climforisk METLA www metla fi Working Papers of the Finnish Forest Research Institute publishes preliminary research results and conference proceedings The papers published in the series are not peer reviewed The papers are published in pdf format on the Internet http www metla fi julkaisut workingpapers ISSN 1795 150X Post Box 18 FI 01301 Vantaa Finland tel 358 10 2111 fax 358 10 211 2101 e mail julkaisutoimitus metla fi Finnish Forest Research Institute Post Box 18 Fl 01301 Vantaa Finland tel 358 10 2111 fax 358 10 211 2101 e mail info metla fi http www metla fi Authors Mikko Peltoniemi Tuomo Kalliokoski Antti Jussi Lindroos Egbert Beuker Annikki M kel Title User guide for PRELES a simple model for the assessment of gross primary production and water balance of forests Year Pages ISBN ISSN 2012 0 ISBN 978 95 1 40 2395 8 PDF 1795 150X Regional Unit Research programme Projects Vantaa 8534 Climforisk EU Life EU EN V FI 00571 Accepted by Risto Siev nen 5 10 2012 Abstract Simple models of ecosystem processes ar2. vciccccsccscieccceiwcneaccsenerecetsiacerecanenansiucstacntennenctestenrecenoecciniees 9 3 1 Program strctUte i wises sec scd scat ehes dada vt ga dass voeeabca sce sasendcesasdnve EEEN aE AAEE ET EANA ATTE 9 3 2 Modell US ssicsc sscedeicsdecietitesns Goleta atborvaeleideiaieiaoens leedieiibnig TEE ditions cadena 10 Model useand tm puts sses econ tsestves Manet ou degusitudandetdascusctelvends 10 Input Of model parameter a sisecsssieecgaccesnncciadevasseceaubedecesanssacacedeseotand canes EEA EE EEn ia r esat 12 4 Example simulations with the model sssssunnnnnsnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn 13 4 1 Model predictions and climate change eesseseeeseeeeseeessesressesrrssrsrieserrrssresesreesesresseeres 13 4 2 Model tests at ICP level TI sites cesscssscvigesscdsseesesieaieagattaeed iennng scab oouendeiteireesttoant Gennsnenadinge 16 5 CONnCIUSIONS a E 17 1 Introduction Estimates of ecosystem carbon sinks and water balances are a starting base for many ecosystem impact studies These two primary variables are influenced by the prevailing weather and as such they are vulnerable to changes in climate Increasing temperature and CO2 of air means higher gross primary production GPP but increasing summer temperatures will also increase the evaporative demand Increasing evaporation may not be fully compensated by increasing summer precipitation partially because variability of summer rains is expected to increase IPCC 20
3. Hydrol Pub 6352 62 20 IPCC 2007 Climate Change 2007 The Physical Science Basis Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change Cambridge University Press Cambridge United Kingdom and New York NY USA Jarvis P G 1976 The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the field Philosophical Transactions of the Royal Society London B 273 1976 593 610 Jylh K Ruosteenoja K R is nen J Ven l inen A Tuomenvirta H Ruokolainen L et al 2009 The changing climate in Finland estimates for adaptation studies ACCLIM project report 2009 Finnish Meteorological Institute Reports 4 Kellom ki S 1995 Computations on the influence of changing climate on the soil moisture and productivity in scots pine stands in southern and northern Finland Climatic Change 29 1 35 51 Kuusisto E 1984 Snow accumulation and snowmelt in Finland Publications of Water Research Institute 551 149 Landsberg J J amp Waring R H 1997 A generalised model of forest productivity using simplified concepts of radiation use efficiency carbon balance and partitioning Forest Ecology and Management 95 3 209 228 Lu J Sun G McNulty S G amp Amatya D M 2005 A comparison of six potential evapotranspiration methods for regional use in the southeastern United States Journal of the American Water R
4. thanked for the review of the manuscript References Bergh J Freeman M Sigurdsson B Kellom ki S Laitinen K Niinist S et al 2003 Modelling the short term effects of climate change on the productivity of selected tree species in Nordic countries Forest Ecology and Management 183 1 3 327 340 Briimmer C Black T A Jassal R S Grant N J Spittlehouse D L Chen B et al 2012 How climate and vegetation type influence evapotranspiration and water use efficiency in Canadian forest peatland and grassland ecosystems Agricultural and Forest Meteorology 15314 30 Bugmann H amp Cramer W 1998 Improving the behaviour of forest gap models along drought gradients Forest Ecology and Management 103247 263 Duursma R amp M kel A 2007 Summary models for light interception and light use efficiency of non homogeneous canopies Tree physiology 27 6 859 870 Fischlin A Ayres M Karnosky D F Kellomaki S Louman B Ong C et al 2009 Future environmental impacts and vulnerabilities Julkaisussa Sepp l R Buck A amp Katila P toim Adaptation of forests and people to climate change A global assessment report IUFRO World Series 22 Helsinki s 53 100 Granier A 1987 Evaluation of transpiration in a Douglas fir stand by means of sap flow measurements Tree physiology 3 4 309 320 Hamon W R 1963 Computation of direct runoff amounts from storm rainfall Int Assoc Sci
5. 07 Fischlin ym 2009 Jylh ym 2009 This means that water cannot be neglected from impact assessments as has been frequently assumed e g Bergh ym 2003 Water effects on GPP may turn out to be important for any prognosis of future forests A natural framework for this is linking GPP to transpiration as the carbon and water fluxes are inherently bound to each other due to stomata of leaves For purposes of large scale regional analysis a simplified modelling approach is preferable because it can be quick and it can be applied at high spatial resolution Although more complex models would provide a more accurate description of processes governing e g water balance their use is often uncertain due to missing input data at this scale Many researches have earlier applied simplified models of potential evapotranspiration PET defined at daily to monthly time steps and scaled these down to actual evapotranspiration on the basis of the availability of water in the soil over the same period which itself is influenced by the evapotranspiration estimate Running ja Coughlan 1988 Kellom ki 1995 Bugmann ja Cramer 1998 Sun ym 2008 Commonly used PET models include Thornthwaite Thorthwaite 1948 and Hamon Hamon 1963 models based on temperature and the Turc Turc 1961 and Priestley and Taylor Priestley ja Taylor 1972 models based on global radiation However these methods have been shown to produce mutually different results that do not necess
6. 80 1 1 1 2 100 0 172224 7 5989583333 999 0 1 380 01 0 7936249222 180 0 0 0 0 328482 4 2079166667 999 2 380 01 0 7936249222 180 0 0 0 0 29592 1 0316666667 999 1 1806 380 01 0 7936249222 180 0 0 0 Box 2 Colums of the input data file PPFD Photosynthetic photon flux density mol m2 day Tair Mean temperature C VPD Mean vapour pressure deficit of the day kPa Precip Precipitation above canopy mm CO2 CO2 of air ppm fAPAR Fraction of absorbed photososynthetic radiation between 0 1 Estimated from LAI of the stand SW Soil water mm First row value used for model initialization 10 Canopywater Canopy water mm First value used for model initialization SOG Snow on ground mm of water First value used for model initialization S Temperature acclimation state C First value used for model initialization Missing data in the input files is represented with 999 If data is missing the program will use previous days values as input In option i only the first row values of input data are used to initialise SW Canopywater SOG and S whereas in option ii simulation is repeatedly initialised for each row of the file because the model is run for independent sites and not for consequtive days The program generates an output file that is automatically named as lt parameter file gt _predictions The file is is in semicolon separated format and it lists the output column names shown in B
7. arily correlate well with actual evapotranspiration Lu ym 2005 The temperature based methods especially do not easily transfer from one biome or geographic area to another Bugmann ja Cramer 1998 Shaw ja Riha 2011 In this report we describe a model which has been developed and parameterized for the prediction of GPP and soil water balance in boreal conditions and provide guidance for the use of the model program that has been developed The complexity of the model is at intermediate level between the highly mechanistic and the simple index type models based on PET and water availability The set of simple empirical models predicts daily ecosystem gross primary production GPP P evapotranspiration ET E and soil water content 0 Following the principles of leaf level models ET model depends on our canopy level GPP model through canopy conductance Our 6 model in turn influences both the GPP and ET models as soil water is an essential factor in both photosynthesis and ET Finally the model depends on the E model as E is one of important processes determining 0 2 Model 2 1 Ecosystem processes represented by the model Working Papers of the Finnish Forest Research Institute http www metla fi julkaisut workingpapers 2012 mwp247 htm The model is a simple semi process based model representing the inter linkages between photosynthesis of the canopy P and water balance of the ecosystem Figure 1 The ecosystem model
8. ay k and its value for the previous day Xx 1 Tis a constant related to the speed of response of the current acclimation status to changes in Ty Xo C is a threshold for X defining the low limit above which S starts to increase 7 fs Smar C is the value where the acclimation modifier reaches its optimum temperature state related effects are modelled using a modifier for temperature acclimation fs fow p min fp fw Effect of stomata to GPP is assumed to be the fo e amp 2 D Effect of vapour pressure deficit D kPa on GPP e a which accounts for the effect of atmospheric CO C ppm Cao is the reference CO2 380 ppm High D decreases canopy conductance which decreases photosynthesis WP min ee text for explanations fw in 1 pp S for explanati co2 Cy Cyo C ean response of cano to CO See text for 1 Ca Ca0 C M p f py GPP to CO S fi CaotCm explanations The evapotranspiration model Evapotranspiration is composed of transpiration of vegetation and free evaporation from non photosynthetic surfaces The transpiration part is represented in the model by an equation assuming that the canopy is well connected to the atmosphere such that transpiration can be predicted with canopy conductance multiplied with vapour pressure deficit ie transpiration gD Jarvis 1976 Whitehead 1998 Br mmer ym 2012 Canopy conductance g is predicted with an equation loosely
9. e from 0 to 1 See Table 1 for explanations of f In the previous version of the model M kel ym 2008 water vapour pressure deficit of atmosphere reduced P through an exponential relationship fp e x is negative and a separate modifier was introduced to account for soil water Here we merged the fp and fw modifiers following the principle of one constraint of stomatal conductance Landsberg ja Waring 1997 fow p min fp fw p where fp is estimated from relative extractable water W defined as _ _9 Owp O Owp where ypis the soil water content at wilting point and Op is the water content at field capacity For the soil water modifier we adopted the widely used threshold model proposed by Granier Granier 1987 where fwp min 1 W pp i e fwp increases linearly with increasing W between Oyp and pp after which it is set to value 1 Using previous day s estimate for soil water is justifiable because changes in soil water are small during a day when soil water is constraining GPP CO influences GPP in two ways in the model 1 Modifier foo2 1 Ca Ca0 Ca CaotCm Which represents the mean increase of GPP with increasing COs C The base level C 9 380 ppm and Cm is a parameter 2 Increasing CO influences also the stomatal conductance which becomes less reactive to VPD This is the reason for introducing a multiplier for x in the fp modifier which takes the form Caol Ca where cx is a unit
10. e useful tools for various kind of ecosystem impact studies We built a simple model of ecosystem gross primary production evapotranspiration and soil water content which requires minimal input data and which is efficient to run In this report we briefly describe the model equations document the model program and provide user guide for the current version of the model We also use the model to run a few example simulations that describe how the model responds to the environment and test the model predictions of soil water in reference conditions with ICP level II data on soil water The model is intended to be used in large scale prediction of GPP ET and drought in the Climforisk EU Life project Keywords carbon climate change drought ecosystem model photosynthesis water balance Available at http www metla fi julkaisut workingpapers 20 12 mwp247 htm Replaces Is replaced by Contact information Mikko Peltoniemi mikko peltoniemi metla fi Other information Deliverable of the Action 3 of the Climforisk project Modelling software and documentation 30 6 2012 Contents Contents D MALPOGU CHOON TT 5 2 MOGGL A T A T E EEE T 5 2 1 Ecosystem processes represented by the Model cece eeeeeeeeeeeeseeeeeeeeeeeeaeeeaeeeaeeeaeeenaees 5 2 2 Model ParaMetertZatioms scacennsicizeseansshedossghtevenubsseiscsiueesrasehseans sdeddvonugesitlsscysseaesvecsersesndesbane 9 3 Model implementation ssc
11. ely available in any operating system environment We intend to use this model for predicting GPP and water balances of forest ecosystems at the scale of Finland for both retrospective analyses of past climate and for analyses of the effects of climate change Soil water predictions of the model are intended to be used as the basis of an ecosystem wetness index Predictions of this index will then be compared to data on 19 pest pathogen damages of which many have been earlier related to drought periods As such the model offers us a tool for ecosystem impact assessments Model development continues which means that new features may be included and computational logic may be improved APPENDIX Input data and modelled and measured soil water at the ICP level II sites Air T is mean daily air temperature RH is relative humidity Cumulative rainfall is in units of mm Global radiation measured at ICP plots has been converted to PPFD and compared to PPFD predicted at closest weather grid points of FMI SW is scaled soil water content in Ticks on x axis show missing observations in daily ICP level II data Black dots are ICP level II plot measurements and red dots are model simulated results Acknowledgements We thank all people that have been participitating in the development or testing the current model or its predessors Minna Pulkkinen Tapio Linkosalo Pasi Kolari Remko Duursma Teemu H ltt Mika Aurela Risto Siev nen is
12. esources Association 41 3 621 633 M kel A Hari P Berninger F H nninen H amp Nikinmaa E 2004 Acclimation of photosynthetic capacity in Scots pine to the annual cycle of temperature Tree physiology 24 4 369 376 M kel A Pulkkinen M Kolari P Lagergren F Berbigier P Lindroth A et al 2008 Developing an empirical model of stand GPP with the LUE approach analysis of eddy covariance data at five contrasting conifer sites in Europe Global Change Biology 14 1 92 108 Medlyn B E Duursma R A Eamus D Ellsworth D S Prentice I C Barton C V M et al 2011 Reconciling the optimal and empirical approaches to modelling stomatal conductance Global Change Biology Peltoniemi M Pulkkinen M Kolari P Duursma R A Montagnani L Wharton S et al 2012 Does canopy mean nitrogen concentration explain variation in canopy light use efficiency across 14 contrasting forest sites Tree physiology 32 2 200 218 Priestley C H B amp Taylor R J 1972 On the assessment of surface heat flux and evaporation using large scale parameters Monthly Weather Review 100 2 81 92 Running S W amp Coughlan J C 1988 A general model of forest ecosystem processes for regional applications I Hydrologic balance canopy gas exchange and primary production processes Ecological Modelling 42 2 125 154 Shaw S B amp Riha S J 2011 Assessing temperature based PET equations u
13. following an empirical leaf level stomatal conductance function that uses GPP of the whole canopy as input instead of leaf photosynthesis Medlyn ym 2011 Radiation drives evaporation on non green surfaces Because PPFD is strongly correlated with global radiation we use PPFD 6 here as we did in the GPP model The proportion of radiation incident on non green surfaces can be approximated with 1 f pprp This formulation of evapotranspiration requires minimal input data but allows for a link between P and E and a fairly straightforward and flexible fit to data P E a pa fw pP fcozr x 1 fappro ofw E Where fcoz r 1 1 95 Ca Ca0 Ca Cao 2000 removes the mean effect of elevated on GPP generated by fco2 in the model due to increased concentration of CO2 which does not directly influence transpiration The quantities a and y are fitted parameters which partially determine the fraction of the two water fluxes The parameters v and are needed because VPD and soil water do not equally influence GPP and transpiration The modifier fw x is estimated similarly to that for the GPP submodel but with its own threshold parameter pg If there is any water in the canopy fwe 1 The soil water model Soil water is predicted with a simple bucket model with three parameters field capacity Orc wilting point Owp and the daily drainage fraction tp for water above field capacity No drainage occurs below field capacity The model also
14. he GPP term that is more variable by nature 17 2100 GPP Figure 4 GPP and ET at the hypothetical low canopy cover site 4 2 Model tests at ICP level Il sites The model was also tested against soil water measurements at ICP level II forest sites Appendix A In these tests no attempt to bring plot information to the model was made but the model was rather run in a mode corresponding to Hyytiala eddy covariance site and it was run with the gridded weather data from the Finnish Meteorological Institute FMI The weather grid has the resolution of 10 km x 10 km and the closest grid point was selected when the simulations were conducted for ICP level II plots An attempt to use the weather data measured at ICP level II plots was also made but there turned out to be several gaps in the data required for model runs which means that the data processing would have required a significant effort Furthermore for a broad scale test of the model predictions we were interested whether the model could mimic measured variation in soil water data of these plots if weather data that is broadly available is used Comparisons were also possible for some of the years when soil water measurements exist At the ICP Level II plots soil water content was measured using Theta Probe sensors Each plot has two sensors several meters apart at a depth of 20cm below the surface Data was collected at an hourly interval Here the mean
15. includes water storages for snow and free water in the canopy which are also simple bucket models Each day the soil water content of these storages is updated with the following rules 1 Interception fills the canopy water storage while excess water in the storage drains down to the soil the maximum amount of canopy water is a parameter of the model CWmax li Snow water storage 8 y accumulates when mean daily temperature T lt 0 C and mT Tr gt 0 0 T lt 0 Kuusisto 1984 where m is a parameter in the model Snowmelt is transferred to soil water iii Evapotranspiration decreases water storages in the sequence 1 canopy water 2 snow and 3 soil Evapotranspiration is influenced by soil water iv Drainage from the soil occurs above field capacity only Drained water on day k is estimated as F 0 Opc T melts if temperature of day k Ty gt 0 C M f A more detailed description of the model appears elsewhere Peltoniemi et al 201X manuscript 2 2 Model parameterization The model has been parameterized using Hyyti l eddy covariance derived data on GPP and evapotranspiration and measurements of other variable used to run the model Soil water data from Hyyti l has been used in the parameterization as well The model has also been calibrated with data from Sodankyl eddy covariance site and the Hyyti l parameterization has been tested at the Sodankyl site Peltoniemi et al 2012 man
16. is called PRELES PREdict with LESs or PREdict Light use efficiency Evapotranspiration and Soil water and it runs using standard weather data The required inputs are daily mean temperature T vapor pressure deficit D precipitation R and photosynthetic photon flux density PPFD which can be derived with sufficient accuracy from frequently measured global short wave radiation G The structural information the model requires is the fraction of absorbed PPFD which can be estimated from LAI L4 possibly modified by information about stand structure Duursma ja M kel 2007 Canopy GPP Evapotranspiration Soil water Eored f Pred Oprea f Ered TA Figure 1 Linkages between GPP evapotranspiration and soil water represented by the model In the model the canopy GPP is represented with an empirical equation developed in an earlier study M kel ym 2008 The GPP model has been slightly modified in the current model version The GPP model The GPP model predicts photosynthetic production P P gC m day during the day k Pr BO Ti fix where is the potential LUE gC mol PPFD i e the maximum LUE reached in optimal growing conditions and at low light This parameter can also be related to canopy nitrogen concentration Peltoniemi ym 2012 gis photosynthetic photon flux density PPFD mol m day during day k fi are modifiers that account for the suboptimal conditions i All modifiers rang
17. less parameter A summary of f modifiers is presented in Table 1 Further information can be found elsewhere M kel ym 2008 which is introduced as one of the modifiers fapppp Not all absorbed PPFD can be used in photosynthesis The light modifer f1 describes the saturation of photosynthetic production with high PPFD with f ye Table 1 Environmental modifiers influencing GPP For parameters see text and Box 4 Modifier Function Explanation SapPrp The structure of the forest stand is characterized by the fraction of absorbed PPFD Fraction of absorbed PPFD is estimated usually from LAL or it is directly measured St 1 Photosynthetic efficiency of canopy decreases under yor high irradiance Multiplied with PPFD mol m this gives the canopy GPP and rectangular shaped response function to PPFD Peltoniemi ym 2012 fs fsk ming 1 where This modifier captures the S vor ae Xo 0 where seasonal cycle as well as the X ee ECT X1 where X T variation in daily temperature but so that the a responses of ambient Smax C is temperature when canopy GPP is not eeapernemiecarstielaved M kel ym 2004 M kel ym 2008 constrained by temperature S C is state of acclimation estimated using a first order dynamic delay model for X CC which is the a priori estimate for the state of acclimation It is influenced by the ambient temperature T C on d
18. llowing cases were simulated 1 Hyyti l soil and stand i e 0 0 45 fappro 0 75 2 Dry soil and Hyytiala stand i e 0 0 225 fappro 0 75 3 Hyytiala soil and low canopy cover i e 8 c 0 45 fappro 0 25 Simulation case 1 The model predicted less summer evapotranspiration in 2100 than in 2006 Figure 2 This was mainly caused by the reduced sensitivity of transpiration to increased VPD under higher ambient CO Winter evapotranspiration was little affected Soil water content increased in the 13 changed climate during the winter because of changes in the timings of the snowmelt days Soil water content during the summer remained slightly higher in the changed climate than with 2006 weather This had consequent effects on the GPP which was slightly higher in changed climate during the summer soil water minima GPP 220 5 200 AA I VO 1804 w NA 140 4 VN i 120 4 NV 100 Y W PN SW 80 60 40 4 20 5 0 100 200 300 400 Day Figure 2 Simulation of gross primary production GPP gC m day top panel evapotranspiration ET mm top panel soil water conteent SW mm bottom panel with 2006 weather black and predicted weather in 2100 purple 14 Simulation case 2 and 3 GPP at the dry site case 2 was variable due to smaller soil water holding capacity of the soil The model predicted a stonger drought effect on GPP d
19. lots 10 and 11 see Appendix Both plots are located close to the Hyytiala eddy covariance site The soil on plot 10 is composed of sorted sand typical Scots pine site but on plot 11 of unsorted till with relatively high amount of silt typical Norway spruce site The soil properties at plot 10 favour lower soil water values than those at plot 11 It is possible that if the soil parameters differ considerably from those at the Hyytiala site then the model would not follow soil water measurements as well as in the case where the soil is fairly similar to that in Hyytiala Given the uncertainties in the input data mainly rainfall to the model one could expect that it is possible to capture long dry periods that are prevalent on large regional scales but that it is more difficult to capture occasional droughts generated by stochastic nature of the rainfall events in the summer 5 Conclusions Modelling fundamental ecosystem processes is needed for various kind of ecosystem impact studies Spatial variability and non linearity of ecosystems processes requires that the processes are simulated at high spatial resolution Modelling them however can be challenging and data intensive In this report we described and provided user guidance for a simple model which can be used for such purposes The model represents the core processes and linkages between ecosystem photosynthesis and water balance The model has been implemented in C with tools fre
20. ls h o Include header files globally o Global declaration of structures that contain submodel parameters 3 2 Model use Model use and inputs The PRELES program is run on command line in both Windows and Linux systems It can be executed in several modes depending on the purpose of use It can be run i for an arbitrarily long period for one site by using a weather input data file or 11 run for arbitrary number of sites for one day useMeasurement 10 iii for just one day for one site by providing weather inputs and the initial states of storages as arguments on command line or iv for arbitrary number of sites for arbitrary nu ber of days N i PRELES preles par preles input il PRELES preles par preles input iii PRELES preles par 30 11 1 0 380 0 79 180 0 0 20 iv PRELES preles par fapar csv initvars csv weather csv N The run mode of the model is determined by the number of arguments given to the model and the parameter file preles par of the model see section Model parameters In options i and ii the program requires an input file preles input which lists model inputs in semicolon separated format Box 1describes the contents of the input file for options i and ii In the input file mean or cumulative daily values appear on separate rows The colums of the input file of Box lare presented in Box 2 Box 1 First four rows of a weather input fAPAR file 0 50571 11 0127083333 1 0 380 01 0 7936249222 1
21. nder a changing climate in temperate deciduous forests Hydrological Processes 25 9 1466 1478 21 Sun G Noormets A Chen J amp McNulty S G 2008 Evapotranspiration estimates from eddy covariance towers and hydrologic modeling in managed forests in Northern Wisconsin USA Agricultural and Forest Meteorology 148 2 257 267 Thorthwaite C W 1948 An Approach Toward a Rational Classification of Climate Geograph Rev 38 1 55 94 Turc L 1961 Evaluation des besoins en eau d irrigation Evapotranspiration potentielle Annales Agronomiques 12 1 13 49 Whitehead D 1998 Regulation of stomatal conductance and transpiration in forest canopies Tree physiology 18 8 9 633 644 22
22. ox 3 Box 3 Output columns of the model Day The running number of the day i e row number GPP Gross primary production gC m2 day ET Evapotranspiration mm SW Soil water mm SOG Snow on ground in mm of water CW Water in the canopy mm Snowmelt Snowmelt estimate mm of water Throughfall Water raining to soil mm Drainage Drainage mm of water S Acclimation state C fS Acclimation temperature modifier 0 1 fD VPD modifier 0 1 fW Soil water modifier 0 1 In run mode iii all input data is given as model arguments which are in the same order as the input data in the files of the modes 1 i1 Mode iii is executed automatically if weather data is given as program arguments Option iv requires separate input files for fAPAR values initial storage values and weather table and an argument that tells for how many days the model is run which must correspond for the number of days there is weather data in weather table This mode is meant for high performance calculation where cumulative output variables for several days are produced for a great number of sites by using weather data point and observations indicated in the fapar csv file Box 4 shows the logic of the calculation and the contents of input and cumulative output 11 Box 4 Logic of mode iv of running Preles and contents of input files Values in boxes are for example only Weather csv is a lookup table which lists daily weather observation
23. r evaporation oz REW threshold when evaporation decreases v Soil water correction for transpiration m Melting coefficient of snow Fraction of intercepted water I_0 fAPAR 0 75 Maximum storage for canopy surfacial water for free evaporation 0 33 IO 4 824704 CWmax 4 Example simulations with the model 4 1 Model predictions and climate change The primary interest in PRELES is in the prediction of climate change effects on forest GPP and soil water balance To give an impression about model predictions under climate change and in comparison to current climate we made a few model simulations As the starting base for simulations we used the data measured in year 2006 in Hyyti l eddy covariance site The year 2006 was a special year because it includes one of the rare sequence of days in the measurement history of the Hyyti l site when drought clearly influence ecosystem GPP and evapotranspiration We asked whether the Hyytiala pine stand would have suffered from drought had there been more CO in the air and had there been higher temperatures as expected under climate change For this example we assumed that precipiration is the same under climate change as it was in 2006 Changes of temperature for each of the four seasons were obtained from A1B scenario in the year 2100 as represented in Jylh et al 2009 Table 10 Appendix 10 and CO concentration was assumed accordingly to be 760 ppm The fo
24. s for FMI grid points for N days fapar csv 2100 0 7 2530 5 3950 8 1 Loop sites in fapar csv Record closest weather grid point Get initial storages from statevar csv from the same row Get weather data for N days for closest weather grid point from weather csv Write period s predictions to a file preles par_predictions and the new state of the model to statevars csv_after e Unique lines indicate unique 2 sites Order of sites lines is irrelevant for Preles Columns weather grid point 3 id fAPAR statevars csv 2100 180 0 0 0 25 100 10 0 9 1 39 190 0 0 2 Columns SW CW SOG S 4 Sites should be in the same order as in fapar csv weather csv 1 1 20 10 1 10 380 1 2 20 10 1 10 380 1 3 20 20 1 1 380 2 1 20 10 1 10 380 2 2 20 10 1 10 380 2 3 2 20 10 1 10 380 3 1 20 20 1 1 380 3 2 20 10 1 10 380 3 3 20 10 1 10 380 4 1 20 10 1 10 380 4 2 20 20 1 1 380 4 3 20 10 1 10 380 Columns Weather grid point number running number of day in period PPFD Temperature VPD Rainfall CO2 preles par predictions 2100 6 54 0 5 187 2 180 25 3 54 0 4 182 2 180 39 3 54 0 4 184 2 180 Grid points should be in increasing order Days in each grid point should be in increasing order for the period of N days Columns Mean GPP Mean ET Mean SW Minimum SW Input of model parameters The PRELES program used a parameter file named preles par in the above example which ha
25. s the same format in each run mode The parameter file lists the parameters of the model and tells how the model is run The box below shows the meaning of each parameter Box 5 Parameter file left cell explanations and symbols of parameters RUN_PARAMETERS_FOR_MODEL 0 useMeasurement 0 mode i 10 mode ii 30 mode iv 0 LOGFLAG Loglevels 0 2 cause increasing quantity of logging SITE_SPECIFIC_PARAMETERS Depth of the soil mm 413 soildepth Orc Field capacity above drainage occurs 0 450 ThetaFC Owe Wilting point 0 118 ThetaPWP tp Fraction 1 tauDrainage of water above ThetaFC drains in a day 3 tauDrainage GPP_MODEL_PARAMETERS 0 748464 betaGPP 12 74915 tauGPP 3 566967 SOGPP Br Light use efficiency gC mol 1 fS model season temperature parameter So fS parameter Smax fS parameter x fD parameter effect of VPD on GPP 12 18 4513 SmaxGPP 0 136732 kappaGPP 0 033942 gammaGPP 0 448975 soilthresGPP 2000 cmC02 0 4 ckappaCO2 EVAPOTRANSPIRATION_PARAMETERS 0 33271 alpha 0 857291 lambda 0 041781 chi 0 474173 soilthresET 0 278332 nuET SNOW_RAIN_PARAMETERS 1 5 Meltcoef 0 y fL parameter sat effect of PPFD op fW parameter GPP reduces after REW lt soilthresGPP Cm parameter Mean effect of CO2 ppm on GPP C parameter Change in CO2 effect on VPD a Parameter for transpiration multiplier VPD correction for transpiration x Parameter fo
26. uring early summer in the changed climate than with 2006 weather data Figure 3 top panel This was due to the higher evapotranspiration in the beginning of the season which reduced the soil water content in the changed climate more than with 2006 data Figure 3 bottom panel The most severe drought during the summer was still less pronounced in the changed climate than it was with 2006 data GPP 200 gt 180 4 160 4 140 5 120 4 100 i NA 80 i Y Y 60 ha SW 40 20 4 o4 T T T T 0 100 200 300 400 Day Figure 3 GPP and SW at the hypothetical dry site simulation case 2 15 The level of GPP at the low canopy cover site was generally low due to small amount of light harvested 10 ___ 2100 GPP T l 200 300 400 Day Figure 4 top panel GPP at the low canopy cover site was less influenced by the soil water dynamics than in the other sites This can be attributed to evapotranspiration that was a bit lower and more evenly distributed during the season 16 10 __ 2100 GPP T T l l 100 200 300 400 D Day Figure 4 bottom panel The reason for more uniform distrubution is that the evapotranspiration at low canopy cover site is influenced relatively more by the evaporation term of the evapotranspiration equation than by the transpiration part that is influenced by t
27. uscript Based on these tests the Hyyti l parameterized model predicted Sodankyl fluxes of GPP and evapotranspiration surprisingly well R for GPP was 0 82 and for evapotranspiration it was 0 61 These are fairly close to the values we obtained when the model was parameterized with Sodankyl s own data R 0 88 for GPP and R 0 76 for evapotranspiration 3 Model implementation PRELES has been implemented in C programming language and the model code is available for any use in the Project web pages www metla fi life climforisk Readily compiled executables have been provided for Windows 32 bit platforms and Linux 64 platforms For other platforms we propose that the user installs Qt development platform http qt nokia com and compiles the source code In the compilation the gcc compiler provided by the Qt development environment was used 3 1 Program structure The model code is organised in the following files main c o Read input files o Call the workhorse preles o Write output files preles c The main workhorse for calculation of ecosystem processes o Estimate GPP o Estimate evapotranspiration o Update water balance Spp c o Functions to estimate GPP and f modifiers water c o Functions to estimate evapotranspiration and ecosystem water balance initruns c o Utility functions to initialize variables with irrational or missing values o Functions for handling model parameters prelesgloba
28. value for both sensors was used Due to variation in the soil structure there may be considerable differences in the measured values for both 18 sensors and thus the mean of only two sensors may not provide an accurate value that is representative for the whole plot Comparisons revealed that generally mean daily air temperature measured at the plots best corresponds to what is estimated in the FMI grid whereas there is more difference between estimates of photosynthetically active photon flux density PPFD derived with a model from global radiation measuments There is considerable difference in the FMI rainfall estimates and what has been measured at the plots This is understandable due to the high spatial variability of summer rainfalls which seems to make any predictions of drought fairly uncertain Another reason is that the automatic tipping bucket rain gauges used for rainfall measurements at the Level II plots Model RG306 Lakewood Systems got easy plugged Generally speaking soil water predicted by the model follows what has been measured at the sites although there are large level differences between the predictions and the measurements However one should not look at the level differences as the model was run with Hyytiala soil parameters but rather how the soil water estimates follow each other in time The effect of soil properties on the measured soil water values is probably the cause of the differences shown between the p
Download Pdf Manuals
Related Search