Home

User's Guide - Numerical Terradynamic Simulation Group

Contents

1. Figure 3 2 Choosing the time range a java applet you must have a java enabled browser for this approach This is fairly straightforward Next you need to choose the time range of your search To do this you may enter the date in normal or standard date range Y Y Y Y MM DD Julian date range Y Y Y Y DDD or you can do an annually repeating time range Fig 3 1 and Fig 3 2 After indicating the time range you must make sure that you have allowed the search engine to return the proper number of results This is done by entering the number of granules to return in the field entitled Return a Maximum of lt blank gt data granules The maximum number of granules that can be returned at the time of this writing are 1000 Fig 3 2 Not much more is required to get MOD17A2 data just click on the start button and you re in business 3 2 Search In Progress page This page shows you the results of the search as it is happening If there are any errors this page will indicate them with Error in the status field The status field will also tell you Version 2 0 12 2 2003 Page 46 of 57 MODI7 User s Guide MODIS Land Team tek gt DAD Quee Great res Y oh W d Address http f edcrevwew cr uses gov iims diriana og tend orm fue sr 5d 104 395382008 2D 1 1444078 mode e SEARO aubmodeat OMMTRACE Data Data Sets Data Cranules Status Data Center Center Beturmed Returned Comments O Eces 1 ns Recerving Resta Success
2. LAI O mo 0 O Sy e FPAR noGc FPAR filling 300 360 PSN noGc e PSN filling e GPP_noQc GPP filling 8 day Figure 2 1 Comparison of temporal profiles of 2001 Collection 4 MOD15A2 with original values FPAR_noQc LAI_noQc and temporally linearly filled FPAR and LAI FPAR filling LAI filling and of temporal profiles of MOD17A2 with original MOD15A2 inputs GPP_noQc PSN_noQc and MOD17A2 with filled MOD15A2 GPP filling PSN filling The pixel is located in the Amazon rainforest lat 1 0 lon 60 with the MODIS land cover Evergreen Broadleaf Forest EBF eliminate DAO cell boundary lines in a MOD17 image and thus we utilize a modified cosine function of the form D cos n 2 d d i 1 2 3 4 2 1 where D is the non linear distance between the 1 km MODIS pixel and any one of four surrounding DAO cells d is the great circle distance between the 1 km pixel and the same DAO cell and d being used This ensures that D 21 when d 0 and D 0 when d d Based on the non is the great circle distance between the two farthest DAO cells of the four max linear distance D the weighted value W can be expressed as W D Y D 2 2 and therefore for a given pixel the corresponding smoothed value V i e interpolated Tmin Version 2
3. Relationship between woody biomass and PAR conversion efficiency for estimating net primary production from NDVI International Journal of Remote Sensing 15 1725 1730 Maier C S Zarnoch et al 1998 Effects of temperature and tissue nitrogen on dormant season stem and branch maintenance respiration in a young loblolly pine Pinus taeda plantation Tree Physiology 18 1 11 20 Monteith J 1972 Solar radiation and productivity in tropical ecosystems Journal of Applied Ecology 9 747 766 Monteith J 1977 Climate and efficiency of crop production in Britain Philosophical Transactions of the Royal Society of London Ser B 277 294 Nemani R R C D Keeling et al 2003 Climate Driven Increases in Global Terrestrial Net Primary Production from 1982 to 1999 Science 300 1560 1563 Olson R J K R Johnson et al 2001 Global and Regional Ecosystem Modeling Databases of Model Drivers and Validation Measurements ORNL TM 2001 196 1 84 Poorter L 2001 Light dependent changes in biomass allocation and their importance for growth of rain forest tree species Functional Ecology 15 1123 123 Prince S and S Goward 1995 Global primary production a remote sensing approach Journal of Biogeography 22 4 5 815 835 Reeves MR Zhao M Heinsch FA and Running SW in prep Characterizing moisture driven rangeland biomass fluctuations using MODIS primary productivity estimates Reich P B Kloepp
4. MOD17 algorithm also requires input of climate data to derive GPP and respiration These data are obtained from the DAO Data Assimilation Office Atlas et al 2000 modeled daily meteorological obersvations as a 1 x 1 25 scale At the end of each year MOD17A3 annual NPP is calculated from the 8 day MOD17A2 i e NPP is the summation of 8 day PsnNet minus growth respiration There are two main problems with the Collection 4 MOD17 data set The first is that in some cases 8 day MVC MODI5A2 is still contaminated by clouds or other noise As a result in regions with higher frequencies of cloud cover such as tropical rain forests values of FPAR and Version 2 0 12 2 2003 Page 37 of 57 MODI7 User s Guide MODIS Land Team LAI will be greatly reduced Fig 2 1 To distinguish between good quality and contaminated data MOD15A2 contains Quality Control QC fields which allow users to determine which pixels are suitable for further analysis The use of contaminated FPAR and LAI inputs will produce incorrect 8 day GPP and PsnNet and consequently unreliable annual NPP The second problem arises from the use of DAO meteorological data in the algorithm Currently the DAO data version used by MODI7 is GEOS402 which has a spatial resolution of 1 x 1 25 All 1 km MODIS pixels located within the same large DAO cell will use the same meteorological data without spatial variation In other words each 1 km pixel retains the characte
5. NPP of the entire terrestrial earth surface at 1 km spatial resolution 150 million cells each having GPP and NPP computed individually Running et al 2000 Thornton et al 2002 The MOD17A2 A3 User s Guide provides a description of the Gross and Net Primary Productivity algorithms MOD17A2 A3 designed for the MODIS sensor aboard the Aqua and Terra platforms The resulting 8 day products are archived at a NASA DAAC Distributed Active Archive Center The document is intended to provide both a broad overview and sufficient detail to enable the successful use of the data in research and applications CHAPTER I THE MODIS ALGORITHM 1 The Algorithm Background and Overview 1 1 Estimating vegetative productivity from absorbed radiation A conservative relationship between absorbed photosynthetically active radiation APAR and net primary productivity NPP was first proposed by Monteith Monteith 1972 Monteith 1977 This original logic known as radiation use efficiency suggested that the NPP of well watered and fertilized annual crop plants was linearly related to the amount of absorbed photosynthetically active solar radiation APAR APAR depends upon 1 the geographic and seasonal variability of daylength and potential incident radiation as modified by cloudcover and aerosols and 2 the amount and geometry of displayed leaf material Monteith s logic therefore combines the meteorological constraint of available sunlight at a sit
6. it can be used to estimate total annual livewood maintenance respiration This approach relies on empirical studies relating the annual growth of leaves to the annual growth of other plant tissues The compilation study by Cannell 1982 is an excellent resource providing the basis for many of the relationships developed for this portion of the MOD17 Algorithm and tested with the BIOME BGC ecosystem process model Leaf longevity is required to estimate annual leaf growth for evergreen forests but it is assumed to be less than one year for deciduous forests which replace all foliage annually This logic further assumes that there is no litterfall in deciduous forests until after maximum annual leaf mass has been attained The parameters relating annual leaf growth respiration costs to annual fine root live wood and dead wood growth respiration were calculated directly from similar parameters developed for the BIOME BGC model White et al 2000 Thornton et al 2002 To create the annual NPP term the MOD17 algorithm maintains a series of daily pixel wise terms to appropriately account for plant and soil respiration To determine livewood maintenance respiration the mass of livewood Livewood Mass kg C is calculated as Livewood Mass ann leaf mass max livewood leaf ratio 1 9 where ann leaf mass max is the annual maximum leaf mass for a given pixel kg C obtained from the daily Leaf Mass calculation Equation 1 4 The livewood leaf ratio i
7. 0 12 2 2003 Page 39 of 57 MODI7 User s Guide MODIS Land Team Collection 4 MOD17A2 Collection 4 MOD17A3 Collection 4 5 MOD17A2 Collection 4 5 MOD17A3 Figure 2 2 Comparison of Collection 4 and Collection 4 5 MOD17A2 GPP composite period 241 and MOD17A3 NPP for 2001 Tavg VPD SWrad is 4 V W v 2 3 i l Theoretically this DAO spatial interpolation can improve the accuracy of meteorological data for each 1 km pixel because it is unrealistic for meteorological data to abruptly change from one side of DAO boundary to the other as seen in Collection 4 Fig 2 2 shows how this method works for MOD17A2 A3 The degree to which this interpolated DAO will improve the accuracy of meteorological inputs however is largely dependent on the accuracy of DAO data and the properties of local environmental conditions such as elevation or weather patterns To explore the above question we use observed daily weather data from World Meteorological Organization WMO daily surface observation network gt 5000 stations Fig 3 1 to compare changes in Root Mean Squared Error RMSE and Correlation COR between the original and enhanced DAO data As a result of the smoothing process on average RMSE is reduced and COR increased for 72 996 and 8496 of the WMO stations respectively when comparing original and enhanced DAO data to WMO observations for 2001 and 2002 Fig 3 2 Clearly the nonlinear spatial interpol
8. 26 2 21 5 33 8 QI10 mr unitless 2 0 2 0 2 0 2 0 2 0 2 0 Annual froot leaf ratio kgC kgC 1 3 14 1 3 14 1 1 1 8 livewood_leaf_ratio kgC kgC 0 031 0 162 0 152 0 203 0 132 0 107 leaf mr base kgC kgC day 200 0 00604 0 00604 0 00305 0 00778 0 00677 0 00869 froot mr base kgC kgC day 20C 0 00519 0 00519 0 00519 0 00519 0 00519 0 00519 livewood mr base kgC kgC day 20C 0 00322 0 00397 0 00297 0 00371 0 00372 0 00312 leaf gr base kgC kgC day 200 0 3 0 3 0 3 03 03 03 froot leaf gr ratio kgC kgC 1 3 4 1 13 1 1 1 1 13 livewood leaf gr ratio kgC kgC 0 16 0 20 0 15 0 19 0 19 0 15 deadwood leaf gr ratio kgC kgC 1 6 141 1 1 6 1 8 1 0 ann turnover prop Cunitless 0 25 0 50 1 00 1 00 0 50 0 25 Corresponding UMD Land Cover 1 2 3 4 5 3 Classification MOD1201 BIOME CLASSIFICATION PARAMETER Werass Cshrub Oshrub Grass Crop Epsilon_max 0 000768 0 000888 0 000774 0 000680 0 000620 Daily Tmin _ max 0 11 39 2 61 3 80 12 02 12 02 Tmin_min 0 8 00 3 00 3 00 3 00 3 00 VPD_max Pa 3100 3100 3600 3500 4100 VPD min Pa 650 650 650 650 650 SLA projected m2 kg leaf C 33 8 12 0 19 0 40 0 36 0 O10 mr unitless 2 0 2 0 2 0 2 0 2 0 Annual froot_leaf_ratio kgC kgC 1 8 1 0 1 2 2 0 2 0 livewood_leaf_ratio kgC kgC 0 051 0 079 0 040 0 000 0 000 leaf mr base kgC kgC day 2005 0 00269 0 00519 0 00714 0 01280 0 00930 froot mr base kgC kgC day 20C 0 00519 0 00519 0 00519 0 00719 0 00519 lrvewood mr base kgC kgC day 2005 0 00100 0 00436 0 00218 0 00000 0 000
9. Forest 3 Deciduous Needleleaf Forest 4 Deciduous Broadleaf Forest 5 Mixed Forest 6 Closed Shrubland 7 Open Shrubland 8 Woody Savanna 9 Savanna 10 Grassland 12 Cropland 13 Urban or Built Up 16 Barren or Sparsely Vegetated 254 Unclassified 255 Missing Data 4 Practical Considerations for Processing and Use of MODIS Data Two considerations paramount to understanding the MODIS data stream are the unique projection and tiling system and the file format inherent to all MODIS land products 4 1 MODIS tile projection characteristics All MODIS land products are projected on the Integerized Sinusoidal ISIN 10 grid where the globe is tiled for production and distribution purposes with 36 tiles along the east west axis and 18 tiles along the north south axis each about 1200x1200 kilometers Fig 4 1 MODIS is meeting the stated geolocation requirement of 0 1 pixels at 2 standard deviations for the 1 km bands Wolfe et al 2002 A The Collection 4 projection is Sinusoidal SIN while Collections 1 3 use a Integerized Sinusoidal Projection ISIN At a 1 km spatial resolution the difference between the SIN and ISIN projections is negligible The decision to switch from the ISIN to the SIN projection was made to make the data more compatible with current image processing software For many applications it may be convenient to reproject MODIS data from the ISIN or SIN projection to a different projection that is more suited to the area
10. Pious ae 555 hartian priere de aiat a TES im Ficibir T ik aT Sapper ccm L Gam Ed CE EDU ads Jems HHI Tera graxala SZ BODITAI 001 AANIHIN Horse T Lzcaj Gramais JD BODQ d2 RINSINEI bed H I 2002101161216 hdr ACA Tete met FOG TREE NET PTS 8 D47 L4 a IER OSS ZEID a oot OS ForuskoBedsm Hakim Crd Peral FIDERORRGT loma 401 Dara goassis mc abala 004 Ja Comida Local Granis i Kiar Wn id Ail num 13134 har ECLOpaI IE 13238 Geib H I Dura mar POIG RET lROGOCTNTERZIZ ia Foresiedia Wakiew Grenade Fipfull FTILETOURGT diera 000 Lasa remei CE Eaj 830 Ifa hd d lamni Gramale ie hotadas i882333 boned ani 1833182335333 kdj lura a TTE TEES EET HASTE bal Da Gobel sith 1218 GD Tiii frat Formato adia Watzsm Gransia FrpFaii FILEPOERHRT Tews HH Date grams a CC BODI7RI BN 28831388 H Local Scale ID 00170 RIEBINZS babe nd NUJ 2392132281477 kd Puke met PODIS TEREA BET PRCTOESHTEERiR B Da 4 Ghote SEN 2538 Gap fad cok FE ORE HH FzsmatocHedla Felice Osassa Frtpbull FILEPORRAT Tome HE Tura grama El BcDG HH JHHSIREER Local Gramaiis ID RoD2732 181 hiNwH HNT IHETINIIIRAE hil take mri KDE TERNI eC a L4 SOOMEL IER THE SID CNN m PIT x EINEN L1 Figure 4 5 Verifying and submitting the order 5 The DataPool The DataPool was developed as an alternative distribution point for EOS data The DataPool is a large disk cache where EOS data are temporarily stored after they are inserted at the
11. Section 3 and the entire BPLUT is shown in Table 2 1 The two parameters for TMIN and the two parameters for VPD are used to calculate the scalars that attenuate max to produce the final e kg C MJ used to predict GPP such that Emax TMIN scalar VPD scalar 1 1 The attenuation scalars are simple linear ramp functions of daily TMIN and VPD as illustrated for TMIN in Figure 1 2 Values of TMIN and VPD are obtained from the DAO dataset while the value of Emax is obtained from the BPLUT The resulting radiation use efficiency coefficient Version 2 0 12 2 2003 Page 11 of 57 MODI7 User s Guide MODIS Land Team 1 0 m gt TMIN Scalar VPD Scalar 2 0 0 TMIN TMIN ax VPD VPDmax Figure 1 2 The TMIN and VPD attenuation scalars are simple linear ramp functions of daily TMIN and VPD Table 1 2 BPLUT parameters for daily maintenance respiration Parameter Units Description SLA m kg c Projected leaf area per unit mass of leaf carbon froot_leaf_ratio None Ratio of fine root carbon to leaf carbon leaf_mr_base kg C kg C7 day Maintenance respiration per unit leaf carbon per day at 20 C froot_mr_base kg Ckg C day Maintenance respiration per unit fine root carbon per day at 20 C Q10_mr None Exponent shape parameter controlling respiration as a function of temperature e is combined with estimates of APAR to calculate GPP kg C day as GPP e APAR
12. Terra MODIS metadata scheme from a providers standpoint it is quite useful for interested readers who want more in depth coverage on this topic Version 2 0 12 2 2003 Page 25 of 57 MODI7 User s Guide MODIS Land Team 4 3 Data set characteristics As indicated in Figure 1 1 and Table 4 2 the MODIS vegetation productivity data stream consists of three biophysical products 1 8 day summation GPP 2 8 day summation PSNnet 3 annual NPP MOD17A2 Equation 1 2 MOD17A2 Equation 1 8 MOD17A3 Equation 1 15 To properly visualize and interpret any of these products it is necessary to convert them from scaled digital images to a biophysical quantity This can be accomplished using the equation Biophysical_pixel scale_factor digital_value 4 1 where Biophysical_pixel is sequestered carbon kg C m scale factor is the gain for the MODIS productivity products and digital value is the numeric value of a file pixel For example if we obtain a mid summer digital value of 421 for Gpp_1km from an HDF file an 8 day summation of Gpp_1km would be Biophysical pixel scale factor digital value 0 0001 421 0 0421 kg C m In order to obtain a daily estimate of Gpp_1km we must divide this number by 8 so that we get 0 0421 kg C m 8 0 00526 kg C m d The information contained in Table 4 2 can also be found within an HDF EOS data set and can be viewed using the various tools found at http hdfeos gsfc nasa gov
13. annual NPP non terrestrial fill value code definitions Code Definition 32767 Fill value conventional HDF EOS fill value assigned to non modeled pixels not falling into other categories below 32766 Perennial salt or inland fresh water body cover type 32765 Barren sparsely vegetated rock tundra desert cover type 32764 Perennial snow or ice cover type 32763 Permanent wetlands inundated marshland type 32762 Urban built up cover type 32761 Unclassified pixel 7 Missing Data There are several reasons for missing data in the MOD17A2 product stream identified as fill values Code 32767 sensor malfunction and cloud cover appear to be the primary causes The MODIS satellite has been very stable and there has been only one period of time during which the sensor malfunctioned As a result there were no MODIS products produced for the 8 day summation days 169 and 177 in 2001 Reconstruction of the data is possible but it is not done at the EDC Cloudiness and darkness also deleteriously affect MODIS measurements in the visible portion of the electromagnetic spectrum There is nothing to be done regarding darkness which fortunately is an issue primarily at the poles where it is dark during the winter Several methods for dealing with missing data resulting from cloud cover are discussed in Section 9 1 Figure 9 1 If there is at least 1 day of quality LAI FPAR data taken during any given 8 day period that data is used in the MOD17
14. global land cover Hansen et al 2000 Version 2 0 12 2 2003 Page 18 of 57 MODI7 User s Guide MODIS Land Team 30 35 25 F E 30r 20 o E 3 3 S 2 lt 207 Q a a 8 8 25 E E amp B p 15r b 2 3 S s a S 20b E E o o E top E E amp E B AS E z 5 Arizona 19r O Arizona O California O California w North Carolina w North Carolina V North Dakota V North Dakota 0 1 10 i 0 5 10 15 20 25 30 10 15 20 25 30 35 Observed Minimum Temperature deg C Observed Average Temperature deg C 4000 35 EZ a E 30r amp 3000 o o lt lt a a o E kz 5 g 25 E g g E E s 2 Q 2000 4 a 2 E E 5 20 E Z a E M o E E ants 9 1000F E lt e 15H Arizona gt Arizona O California 3 O California w North Carolina a Y North Carolina V North Dakota V North Dakota o s 10 I 0 1000 2000 3000 4000 10 15 20 25 30 35 Observed Average Daytime VPD Pa Observed Daily Total Shortwave Radiation MJ m d Figure 2 2 Comparisons of DAO and observed meteorological data Version 2 0 12 2 2003 Page 19 of 57 MODI7 User s Guide MODIS Land Team Table 3 1 The land cover types used in the MOD17 Algorithm UMD Land Cover Types Class Value Class Description 0 Water 1 Evergreen Needleleaf Forest 2 Evergreen Broadleaf
15. hdfeos index html A Remember the result is an 8 day summation In order to obtain daily estimates of Gpp_lkm or PsnNet_1km it is necessary to divide your Biophysical pixel value by 4 4 Links to MODIS friendly tools 4 4a HDFLook HDF and HDF EOS viewer This product is available for Solaris Alpha VMS HP UX IRIX AIX and Linux It is a handy little tool available at http www loa univ lillel fr Hdflook Table 4 2 Summary of output variables from the MODIS vegetation productivity algorithm Summary of MODI7 output variables Variable Data Units Fill Scale Valid Range Product Type Value Factor Gpp_lkm Intl6 KgC m 32766 0 0001 0 30000 MOD17A2 PsnNet_1km Int16 Kg Cm 32766 0 0001 30000 30000 MOD17A2 Npp_1km Int16 Kg Cm 32766 0 0001 30000 30000 MOD17A3 Psn QC Ikm Uint8 N A 255 N A 0 254 MOD17A2 Npp QC _1km Uint8 N A 255 N A 0 254 MOD17A3 Version 2 0 12 2 2003 Page 26 of 57 MODI7 User s Guide MODIS Land Team H 012 3 4 5 6 7 8 9 1011 12 13 14 15 16 1718 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 V 0 10 t R 0 NHK C Figure 4 1 MODIS tiling system Any location on the earth can be spatially referenced using the horizontal H and vertical V designators Each tile is 1200 x 1200 kilometers 4 4b Msphinx This free utility can read HDF and HDFEOS and it has some visualization and other capabilities Supported platforms include HP DE
16. in a more complete and more accurate product Version 2 0 12 2 2003 Page 33 of 57 MODI7 User s Guide MODIS Land Team 9 Considerations for MOD17A2 Product Improvement Based on studies conducted at NTSG several areas of research have been identified as possible improvements for future implementations for the MOD17 algorithm and output including 1 filling model values for cloudy pixels 2 changing the method of data compositing 3 landcover 9 1 Filling model values for cloudy pixels Under cloudy conditions MOD17A2 GPP has two sources of contamination 1 MOD15A2 products 2 DAO meteorological data Accurate retrieval of LAI and FPAR is not possible under cloudy conditions because the reflectances are distorted resulting in poor QA values and inaccurate calculations of GPP for the contaminated pixels DAO data are affected by cloud contamination because of the resolution difference as compared to MODIS 1 x 1 25 vs 1 km x 1 km As a result DAO data cannot MOD12Q1 Landcover Data from Previous Week i us MOD15A2 QC MOD15 17A2 MOD15 17A2 Figure 9 1 A schematic diagram illustrating the process of spatial and temporal interpolation using information from land cover and QA flags In this example the landcover map has only two values dark and dashed pixels In the bottom windows dark pixels are cloudy pixels and white pixels are those with the best QA conditions The thick bordered pixels are the pixels sel
17. m i imc Oud d odere arme dmm ie es Es CER4 A ee ee ee m Geer a ce bisg snme p ee pmo ee em Em eee Tim mimm eel ee pirea giy preg pecu fa num uis le Puta ey andy aom dal eed SALLE cM a Jr ia iska mbi akipona minean tse eer ruri pl ar vos nnt e Ms JL IT a a 3 mu i j rm b a at Sena ir Cee e Pem iri im Bi Jan en ae A Ademas Dent iran fap AB HE Lal r eA ide Pag bhinn Lie binkie mn NU n Fria ibid Las i eiia P AO AAA Aiii Dei berge famem at y m F ART Lal diri ri ie Pag Li Li i EI a AAN MIRA O inim iiias SCN EN EEE T Pul aiee Det bem Penn 3858 Lal r aem aian rr arer i ghinn Peig biiain Linn Liria NU mal u Fms im AT Les ao ene r E Ra occ Adios De irum fames call sera N i 18787 Lai SRSA ATES rie Tra A Li ee 4 _ TER a Lee Tt NADA r SCIES LEZ Joie Dar a um o Bate m ETAT I TE JPE PEA PA ATAN AR Pg da Li ia SO messes one eee P DOMO AEPD A lle as a a r 32527 Lai STATA REMOTE Mein Pema la Lulu iii FL po IM ET Les r DO ae aed ie Dub ine ors are Mite is AUN Lai QMOCETRZAINETMNHTVESHADETTEIUEFIFSINTPA P Mbdchi Lie r alin TENA ms _ MAT Lee muc DTI RI TII Corer irr m Comer Sores mime ma Fui er Capas je DID Dyar Cas camina 18 gee A a IRR assess NX Eee aban Ma E AA s ee o Figure 3 4 The page listing the granules you have requested Version 2 0 12 2 2003 Page 48 of 57 MODI7 User s Guide MODIS Land Team EET amem
18. missingandbad Periods otal 100 4 1 where Peri0dSmissingandbad i the number of times linearly interpolated and Periods is the total number of 8 day composite periods during the growing season The pixel with more periods of filled MODISA2 has a less reliable annual total for both GPP and NPP 5 Final BPLUT applied to Collection 4 5 MOD17 The standard Collection 4 MOD17A2 A3 product is calculated using a BPLUT that was calibrated to the GEOS3 0 DAO data set and Collection 3 MOD15A2 two primary inputs to the MOD17 algorithm The latest DAO GEOS402 and MOD15A2 Collection 4 have been updated and improved Given these enhanced inputs the BPLUT Table 2 1 Chapter I has been updated to improve global MOD17 outputs This BPLUT for Collection 4 5 is based on recent work by Nemani et al 2003 observed GPP data from 13 flux towers in 2001 Heinsch et al in prep Ecosystem Model Data Intercomparison EMDI NPP data Olson et al 2001 a recent book summarizing global NPP Roy et al 2001 and additional publications White et al 2000 Poorter et al 2001 and Hoffmann et al 2003 6 Results Improved MOD15A2 and improved DAO inputs together will enhance the MOD17 products For example even for North Dakota grasslands which have a lower frequency of cloud cover and smoother climatic gradients r from Collection 4 to Collection 4 5 increased from 0 54 to 0 77 in 2001 and from 0 50 to 0 57 in 2002 for the relationship between c
19. retrieval For example QA 4 has a binary equivalent 100 Version 2 0 12 2 2003 Page 29 of 57 MODI7 User s Guide MODIS Land Team 00000100 Leftmost bit 8 Rightmost Hi 1 aja Q bei A 5 A o 3 8 3 z d 2 B lt 5 E a E o i g O o A O o o ec 2 o ua 000 Main RT used with best results 1001 Main RT used with 00 Significant 00 Detectors 00 Best possible clouds NOT present apparently fine 01 OK but not the clear for up to 50 of best 01 Significant channels 1 2 10 Not produced clouds WERE 01 Dead due to clouds present detectors caused 00 Not produced 10 Mixed cloud gt 50 adjacent due to other reasons present on pixel detector retrieval 11 cloud state not defined assume clear problems empirical method used 100 Couldn t retrieve pixe 111 NOT PRODUCED AT ALL non terrestrial biome Figure 6 1 A diagram for a hypothetical MOD17A2 quality assurance value of 4 6 1 GPP and NPP Quality Assurance Variable Scheme The definitions of the bitfields within a given 8 bit GPP QA variable denoted as Psn_1km_0QC are shown in Tables 6 1 and 6 2 Recall that the quality of the precedent input data product FPAR LAI 8 day composite exerts a very direct influence on the quality of the GPP variable and for this reason we inherit the FPAR LAI QA scoring for a given pixel and pass this through as the GPP quality variable At this time the quality bits for MOD17A3 Ta
20. uh pe d 8 80 8 GG l0 0 0 0 0 0 jww me nn E A kr Dum E a ca NA Rod Es a fem Pim sme a pile ard il Hoi bape WE DATEL 2 SE HIDEKR PTS 4T TER EET Te CENTES Duality cee Od zz am era par cd the PORCI lans preberie Tie nhimr gim mi H im far ema lei mad smcameni bm aaa quello ol tke ES piscis with aspect La thais Lute partos Rasrs aoe sprpareged io check tee yeas lee am nr DCENN bedi vides probris towers ther sere eee d stileert error mr ertiionhe Dia HODIZ Lamd grobara hams diddenear dete varices sod levels ub meiuritw rei lecitina alaecriibm col sns khe terii a weed iis prembasbien E eerste promise comin sa a C Ck Lede Dei CMrigibal bee dil Lara hasi urn Bay r proa wach sa Walri lermi prccesming CE AA PA rara ealids s dena sets Der bw dre thes and vm s brand bere er three sieges nd lacio mb Tli tira ee ee preter e Purcell ae d Eh hito c arddkl mewta pass aer DU FEN paufum rmi MOOD DTa aL GoDDirD LPRCE FLT CENTER pwrurLE METI ALL HODIE dais procemee sith HODIE LIE Warmim 8 cr miga in cummiderwd eal eed Em dui Dee Lawl iB Dara Qualius maakas ee CATIA This rales rf idm NODIS Lei 19 Talibrut iam Error eevee da ike rele mi pm per ma A AA SA AS pra Pid Ladle p ms a 11 depa acti then hass ian cia bc optimize SENEI mar orsencw Cia are docuwerim in ia aria amj cr They Links sem idol om bim QUE Era ie pai upper Tess CCST nase pogs kiipi ser post dl bizeier
21. 00 leaf gr base kgC kgC day 20C 03 03 03 0 3 0 3 froot leaf gr ratio kgC kgC 1 8 1 0 1 5 20 20 lrvewood leaf gr ratio kgC kgC 0 05 0 22 0 11 0 00 0 00 deadwood leaf gr ratio kgC kgC 0 5 0 0 0 0 0 0 0 0 ann turnover prop unitless 0 25 0 25 0 25 1 00 1 00 Corresponding UMD Land Cover 9 6 7 10 12 MF mixed forest Grass Grasslands Crop Croplands Version 2 0 12 2 2003 Page 17 of 57 MODI7 User s Guide MODIS Land Team period Compositing of LAI and FPAR is required to provide an accurate depiction of global leaf area dynamics with consideration of spectral cloud contamination particularly in the tropics 2 3 DAO daily meteorological data The MOD17 algorithm computes productivity at a daily time step This is made possible by the daily meteorological data including average and minimum air temperature incident PAR and specific humidity provided by the Data Assimilation Office DAO a branch of NASA Schubert et al 1993 These data produced every six hours are derived using a global circulation model GCM which incorporates both ground and satellite based observations These data are distributed at a resolution of 1 by 1 25 originally 1 x 1 in contrast to the 1 km gridded MOD17 outputs It is assumed that the coarse resolution meteorological data provide an accurate depiction of ground conditions and are homogeneous within the spatial extent of each cell Preliminary studies done by Numerical Terradynamic Simulati
22. 1 2 where APAR IPAR FPAR IPAR PAR incident on the vegetative surface must be estimated from incident shortwave radiation SWRad provided in the DAO dataset as IPAR SWRad 0 45 1 3 While GPP Equation 1 2 is calculated on a daily basis 8 day summations of GPP are created and these summations are available to the public The summations are named for the first day included in the 8 day period M Each summation consists of 8 consecutive days of data and there are 46 such summations created for each calendar year of data collection To obtain an estimate of daily GPP for this 8 day period it is necessary to divide the value obtained during a data download by eight for the first 45 values year and by five or six in a leap year for the final period 1 3b Daily maintenance respiration and net photosynthesis Maintenance respiration costs MR for leaves and fine roots summarized in the lower half of Figure 2 1a are also calculated on a daily basis There are five parameters within the BPLUT Table 2 2 needed to Version 2 0 12 2 2003 Page 12 of 57 MODI7 User s Guide MODIS Land Team calculate daily MR which is dependent upon leaf or fine root mass base MR at 20 C and daily average temperature Leaf mass kg is calculated as Leaf Mass LAI SLA 1 4 where LAI the leaf area index m leaf m ground area is obtained from MOD15 and the specific leaf area SLA projected leaf area kg leaf C for a given pixel
23. 4 0121 1 Significant clouds WERE present 1022 2 Mixed cloud present 1123 3 cloud state not defined assumed clear SCF QC 00020 0 Main RT method used with best possible Bits 5 7 00121 results NTSG Science 01022 1 Main RT method used with saturation Compute Facility 01123 2 Main RT method failed due to geometry Quality Control 100 4 problems empirical method used 11127 3 Main RT method failed due to problems other than geometry empirical method used 4 Couldn t retrieve pixel 7 NOT PRODUCED AT ALL Non terrestrial biome Table 6 3 NPP 8 bit Quality Assurance Variable bit field definitions Collection 4 Variable Bitfield Binary Decimal Description of bitfield s Values Npp QC ikm MODLAND QC 00 20 Highest overall quality Bits 0 1 0121 Good quality 1022 Not produced due to cloud 1123 Not produced due to other reasons NOT YET ASSIGNED Bits 2 4 SCF QC 00020 O Pixel produced best quality Bits 5 7 001 1 1 Pixel produced good quality NTSG Science 010 2 saturation in FPAR LAI algorithm Compute Facility 01123 2 Pixel produced poor quality due Quality Control 100 4 to geometry problems 11127 3 Pixel produced poor quality due to problems other than geometry 4 Couldn t retrieve pixel 7 NOT PRODUCED AT ALL non terrestrial biome Version 2 0 12 2 2003 Page 32 of 57 MODI7 User s Guide MODIS Land Team Table 6 4 GPP 8 day summation and
24. Algorithm and then converted into an 8 day summation but if no LAI FPAR data are available then the MOD17 pixel will not be calculated A All pixels without a GPP calculation will have a value greater than 30 000 regardless of the reason for the missing values 8 Usefulness of Data for Answering Research Questions One of the most important questions to ask before beginning a project is whether MODIS data are applicable to your research This really depends upon the questions you are asking and the scale of that research Spatially MODIS has a much coarser resolution than some other satellite sensors 1 km x 1 km Given the assumptions associated with the data a pixel to pixel comparison is not possible On the other hand MODIS data are well suited to large regional or global analyses Temporally MODIS is much better than many satellite sensors with its daily overpasses and 8 day compositing of the data which can be used to look at annual productivity and interannual variability of both GPP and NPP There is no other satellite that can provide a global 8 day look at vegetative productivity and carbon balance on an annual basis In addition these data are available in near real time which will allow users to make comparisons with their own research data during the growing season often within weeks of the actual data collection As mentioned previously periodic reprocessing of the data will allow for interpolation of missing data resulting
25. C Silicon Graphics IBM Sun and Linux For more information go to http www loa univ lillel fr Msphinx 4 4c Webwinds Webwinds is written in Java enabling it to run on any platform that supports Java It is a science data visualization system capable of reading several data formats For more information please see the Webwinds home page at http www openchannelsoftware com projects WebWinds 4 4d LDOPE Tools LDOPE tools were created to assist in quality assessment of MODIS Land products Look at the overview of this toolset at http ededaac usgs gov tools ldope There are several additional tools available some free and others not which support the HDF EOS data format in which MODIS data are stored For a larger listing see the N C S A s National Center for Super Computing Applications tool page at http hdf ncsa uiuc edu tools himl util Version 2 0 12 2 2003 Page 27 of 57 MODI7 User s Guide MODIS Land Team 5 Data Collection History As with any new product there have been modifications and improvements to the MODIS algorithm and outputs In fact there have been three such collections Collections 2 3 and 4 that may currently be in use It is important for the user to know which collection they have and furthermore what assumptions were made in the calculation of that collection It is also wise to periodically check the EDC website for any updates on the data The MODI7 product at the end of the process
26. Conca D rue in LE SS A A a FA E A A PE Em ue WE foo Tet Figure 3 1 The EDG search page your research restricts you to an older version Be sure to document which version you are using In this case it is version 3 V003 Next you must choose your search area You can do this in several ways by typing in latitude and longitude choosing a global search or by selecting the area you wish to search from Version 2 0 12 2 2003 Page 45 of 57 MODI7 User s Guide MODIS Land Team zmecce Qi 3 Deer emm dpem Qo ab d E E YF a De LAE LLL ER D NN iu Flee LLLI III Tu oe gos ee de ad s r Fw Clonar de ad sas i imn ig ami il Cb pp i Ln n ed Chusas a DataiTiree Range isst required Baba Pia 50505 MU IDIUNT TUR A SS AA IET Vara eee lice SLL Lies Sete TEA EE ELE PEINE I m ds irr niga air t anam na ce DRITTE St pag ime ato Eid pum Tem AFT Diar Tomar Fosisti asra Fn O sa a bun Fara Ti inma iapa amg Choose Additional Options pol resuled A of ER im pauses erga Peg ODE a T Qin drin grum sais A M sgbrduil Nel eR HUE N CEI rr _ Dea Oih ld e af id Mares ai cany ia Eu od rol ui i sop ui sat gg Ka pw St t Se ch SnvaPlestore Beaich C lera rot Perales Get Cea CE f PESTORE Bror Lita A Deere DIET Be Ce ti ei a ALIE a AA Lil ea e mi mi ai Esla dy AXEL Dale army ee eee A Reman acord ho pede ote em Mp kai Da Me E RS CR 4I e SPEC doseape MD Bend
27. DAAC What is stored in this area depends upon usage The primary benefit of using the DataPool is that data can be more quickly retrieved For more information please see http Ipdaac2 usgs gov datapool datapool asp A Please see the DataPool located at the DAAC you are using for specific access instructions as they may differ among DAAC s Please note The above URL will take you to the Land Processes DAAC DataPool page Version 2 0 12 2 2003 Page 54 of 57 MODI7 User s Guide MODIS Land Team MODIS FAQ s There are many MODIS related FAQ s available on the internet Perhaps the most comprehensive of these is located at http daac gsfc nasa gov MODIS FAQ MOD17A2 A3 FAQ 1 How much do MODIS data cost Nothing MODIS data are free 2 What units are my data in The MOD17A2 A3 data are in units of kg C m day However you must apply a scale factor of 0 0001 to each pixel to obtain these units 3 What is the difference between PSN_1km and GPP There is no difference 4 What is the difference between Gpp_1km and PSNnet 1km PSNnet_1km is equal to Gpp_1km MINUS the maintenance respiration from leaves and fine roots In other words Gpp 1km should always be 2 PSNnet_1km at any given pixel 5 What criteria do you use to create quality assurance QA for the annual net primary productivity NPP product The QA values are inherited from the final 8 day composite period in any given year from the MOD15 LAI FPAR input da
28. DI7 User s Guide MODIS Land Team AAA En ams IJ d emn amem Gee Rura d 0 0 1 kim DAA mia nia emm jum ho mum Pt D niet m Bei ui rerin rd enira Tar Step 1 Choose Ordering Options TR Hii Kn Wo page Ther man nip eani D ern in dp bn mind o Bain nea ru go io Eia E ee l an eso ue rta TA A usar Fei Lini pli ird rad le na m Fa eee E LL RITE EETECH DEW S y dcl Jara ROO to ee ee CIC Crees iow Greco Das Pipes do ril min ee cand ur rbe premi e Bom ek iem a arai spin dd uiis pog cede eee ed eile eed ee eh pees etre bre dae ada lima baas nido Ls maa led Deine cras ophcss Lineman GOMOD ari aT d Elk E r traps ike Lene EE AAA bie Desk m Chema i rs mc MODUNIINTTONTUNANSA d I3 R EEC F Chema AS E I EE li Ltacim Dpscms Leave DD 4 IETROO EDC DCD r Chama Graces Lirik GOOD a EA r Chee Comes Linea MOD AAA Ee T3 Ler cd irm aida commie reread ic UU A nacre ERUIT Red UE eee DO E m mim uus io peer Jar y oe Sey Pe Dams Be Dod An ey s kirimane Cira ihs Tim i sehmi r Tenha mala og ee i Saco d Dpi ADL GASP epee SC RIT AN A gt Figure 4 1 Choosing ordering options This will take you once again back to the Ordering Options page You will notice a change however Fig 4 3 The words FtpPull and Change Options will appear next to each row displayed on the page This is telling you that these granules are ready to be ordered Be careful here Tf you are
29. Figure 1 we see that the local granule ID see the EDG web user s guide for a definition Attp edcimswww cr usgs gov pub imswelcome contains much information about the data stored therein The product short name tells us that we are looking at MODIS GPP data from the Terra Satellite MOD17A2 Naming Convention Product version Data format number identifier MOD17A2 A2001009 h10v04 004 20032 17234300 hdf Tile location on grid Product short Processing year name date and time Acquisition Date Figure 1 1 The MOD17A2 Standard Product naming convention The acquisition date is simply the year and yearday indicating when the data was collected The next field indicates the horizontal and vertical positions associated with the data granule these numbers are related to the map projection of the data see Fig 5 1 The product version number shows which version of the production software was used to generate the data The processing information tells you the year date and time when the processing was run on the data present in the granule The format identifier simply tells you what type of file format the data is stored in 2 Logging into the EDG When you first bring up the EDG web site http edcimswww cr usgs gov pub imswelcome in your browser you will be greeted by a page similar to that in Figure 2 1 You can either enter as a guest or as a registered user If you wish to become a registered user you may click on the appr
30. MCF files and each granule or tile HDF EOS file ODL enables the internal software used in MODIS production to access data defined within the Metadata Control File MCF with a unique MCF file defined for each ESDT that is archived such as MOD17A2 or MOD17A3 Users interested in quickly viewing the metadata contents of a HDF EOS file may wish to use the commonly available HDF utility called ncdump The ncdump utility for most computer platforms may be obtained from the National Center for Supercomputer Applications NCSA HDF web site http hdf ncsa uiuc edu hdftools html as well as from common NASA HDF EOS tool URL sites To produce a listing at the console of Science Data Set SDS properties as well as ECS metadata enter a command such as ncdump h MOD17A2 A2002353 h08v05 003 2003008095623 hdf Other interactive graphical user interface based software tools users may employ to view the original ECS metadata information in a HDF EOS tile are HDFLook on Unix Linux and the Java based WebWinds tool available on most platforms Additional information on these tools can be found in Section 4 4 Principal Investigators who wish to define additional attributes specific to their data product may also use the Product Specific Attribute PSA mechanism wherein a limited number of attributes not covered in the standard metadata may be included in a granule Note that the size of a global file attribute in HDF v4 x and therefore in an HDF EOS f
31. O meteorology This dataset is available upon request from NTSG http www ntsg umt edu It can be differentiated from the standard product by looking at the Product version number Figure 1 1 Chapter 3 The standard product version number will always begin with a 0 e g 004 for Collection 4 as in MOD17A2 A2003177 h10v04 004 2003201102319 hdf while any product created at NTSG will begin with a 1 e g 105 for Collection 4 5 as in MOD17A2 A2003177 h10v04 105 2003201102319 hdf For further information about Collection 4 5 including naming conventions granules imagery and analysis please visit http images ntsg umt edu A Users are strongly encouraged to obtain and use Collection 4 5 of the MOD17A2 A3 datasets when available 6 Quality Assurance Quality assurance QA measures are produced at both the file e g 10 deg tile level and at the pixel level At the tile level these appear as a set of EOSDIS core system ECS metadata fields described later in this document At the pixel level quality assurance information is represented by a separate data layer in the HDFEOS file whose pixel values correspond to specific quality scoring schemes that vary by product Earth Science Data Type ESDT The QC organization of MOD17A2 and MOD17A3 files generated from Collection 4 and higher is Version 2 0 12 2 2003 Page 28 of 57 MODI7 User s Guide MODIS Land Team summarized in Tables 6 1 and 6 2 Significant changes in
32. OR Dum en Pe 215 Lore SALLE IP ee ee AAA Figure 4 3 Choosing ordering options the Ready page 4 3 Order form In this step you will fill out the order form telling the specific D A A C Distributed Active Archive Center how to get in touch with you via e mail regarding your request and data availability If you have any questions concerning this form read the online tutorial or contact the EDG s help desk When you are done click on the button which will take you to step 3 of the ordering process Fig 4 4 Version 2 0 12 2 2003 Page 52 of 57 MODI7 User s Guide MODIS Land Team A emeei ea rudna A A A AAA ae AO imb AA redi gp Ue ie T i int ret A A A AO EE adici Dm SS iain To arma reme rd onam a AEERZCMM SiO NP ed ENNLCENENE 500 a na Es ii Le Mamm iMi bapa co L d e hen Die E 0 re kiss i hi HOP Veg Tie ADELA TT TELAS Tm Erase LL TT CU NOT E 1FIEMILITIE parias rur mj wea hry gom LITEI7ARE E CHE dn a er Pec Cae parr FL umpumd HGB is o ad LOW musei Lave LL es Sarena sa rd or LIT s iem Siete hem Ab me Se a ee ee M 1 Sr ar ner de eene Bete ure tm 3 Tire Up i Figure 4 4 The order form 4 4 Reviewing your order Step 3 This step gives you a summary of what you have accomplished thus far Since MODIS data are available at no cost to the public the total cost should amount to US 0 00 If everything is satisfact
33. User s Guide GPP and NPP MOD17A2 A3 Products NASA MODIS Land Algorithm Faith Ann Heinsch Matt Reeves Petr Votava Sinkyu Kang Cristina Milesi Maosheng Zhao Joseph Glassy William M Jolly Rachel Loehman Chad F Bowker John S Kimball Ramakrishna R Nemani Steven W Running Gross Primary Production GPP 1 km MODIS image MOD17A2 v105 NTSG Enhanced GPP over the Globe June 26 July 3 2002 D177 Average Daily GPP gC m day 3 6 9 12 2003 NTSG The University of Montana Global GPP image created by Andrew Neuschwander Version 2 0 December 2 2003 MODI7 User s Guide MODIS Land Team This page intentionally left blank Version 2 0 12 2 2003 Page 2 of 57 MODI7 User s Guide MODIS Land Team Table of Contents Synopsis 8 CHAPTER I THE MODIS ALGORITHM 1 The Algorithm Background and Overview 8 1 1 Estimating vegetative productivity from absorbed radiation 8 1 2 The Biophysical Variability of 9 1 3 The MOD17A2 MOD17A3 algorithm logic 11 2 Simplifying Assumptions for Global Applicability 16 2 1 The BPLUT and constant biome properties 16 2 2 Leaf area index and fraction of absorbed photosynthetically active radiation 16 2 3 DAO daily meteorological data 18 3 Dependence on MODIS Land Cover Classification MOD12Q1 18 4 Practical Considerations for Processing and Use of MODIS Data 20 4 1 MODIS tile projection characteristics 20 4 2 File format of MOD17 end products 21 4 3 Data set characterist
34. aries widely with different vegetation types Field et al 1995 Prince and Goward 1985 Turner et al 2003 There are two principle sources of this variability First with any vegetation some photosynthesis is immediately used for maintenance respiration For the annual crop plants from the original theory of Monteith 1972 these respiration costs were minimal so was typically around 2 gC MJ Respiration costs however increase with the size of perennial plants Hunt 1994 found published e values for woody vegetation were much lower from about 0 2 to 1 5 gC MJ and hypothesized that this was the result of respiration from the 6 27 of living cells in the sapwood of woody stems Waring and Running 1998 The second source of variability in is attributed to suboptimal climatic conditions To extrapolate Monteith s original theory designed for well watered crops only during the growing season to perennial plants living year around certain severe climatic constraints must be recognized Evergreen vegetation such as conifer trees or schlerophyllous shrubs absorb PAR during the non growing season yet sub freezing temperatures stop photosynthesis because leaf stomata are forced to close Waring and Running 1998 As a global generalization we truncate GPP on days when the minimum temperature is below 0 C Additionally high vapor pressure deficits 2000Pa have been shown to induce stomatal closure in many species This level of daily at
35. ation significantly improves DAO inputs for most stations although for a few stations interpolated DAO accuracy may be reduced due to the inaccuracy of DAO in these regions and local conditions as noted above Version 2 0 12 2 2003 Page 40 of 57 MODI7 User s Guide MODIS Land Team prd los Sec re pert E PO PIN po E Pad a a 1 1 Figure 3 1 Distribution of more than 5 000 WMO stations for 2001 and 2002 Breduced 96 stations with changed RMSE stations with changed COR increased 6 o percent a o E S A percent o 2 3 4 Tmin RMSE Tavg RMSE AVPRMSE VPD RMSE Tmin R Tavg R AVPR VPDR variables variables Figure3 2 Percent of WMO stations with changes in RMSE and COR between spatially interpolated and non interpolated DAO For most stations DAO accuracies are improved reduced RMSE and increased COR as a result of spatial interpolation Version 2 0 12 2 2003 Page 41 of 57 MODI7 User s Guide MODIS Land Team 4 Addition of annual GPP and QC to Collection 4 5 MOD17A3 In an effort to make the MOD17A3 product more complete we have added annual GPP summation of GPP and a meaningful QC flag for NPP values Currently Collection 4 MOD17A3 has two layers in an HDFEOS file NPP and NPP_QC although this NPP QC is meaningless We therefore define the MOD17A3 QC for a given pixel as QC Periods
36. ay PsnNet however will depend on the changes in both FPAR and LAI because improved MOD15A2 leads to increases in not only GPP but also respiration Equation 1 2 Chapter I We found that in most regions PsnNet increased in Collection 4 5 relative to Collection 4 But for some small portions of the globe PsnNet may not change or may even be reduced as shown in Fig 2 2 For the second problem arising from coarse spatial resolution daily DAO data we use spatial interpolation to enhance meteorological inputs The four DAO cells nearest to a given 1 km MODIS pixel are used in the interpolation algorithm There are two reasons for choosing four DAO cells per 1 km MODIS pixel 1 this will not slow down the computational efficiency of the MOD17 datastream which is a global product and 2 it is more reasonable to assume no elevation variation within four DAO cells than any greater number of DAO cells We first attempted to use linear spatial interpolation similar to the inverse distance weighting IDW function commonly found in most GIS software However it failed because DAO boundary lines remained Instead we used non linear interpolation Although there are many formulae for non linear spatial interpolation for simplicity we use a cosine function because the output value can be constrained between 0 and 1 This function still could not effectively Version 2 0 12 2 2003 Page 38 of 57 MODI7 User s Guide MODIS Land Team
37. ble 6 3 are taken from the last full 8 day period of MOD15A2 and therefore caution should be used in interpreting these QA values 6 2 Identifying non terrestrial fill values in the GPP NPP data products We recognize that many users will want to use GPP and NPP data products in combination with a geographical information system GIS and remote sensing analysis software To facilitate production of single layer MODIS data product maps we now classify non terrestrial e g non modeled pixels with special identification codes to allow for quick masking and exclusion from quantitative ecological analysis A dual encoding scheme is followed whereby pixels whose values lie within the valid range for the biophysical variable may Version 2 0 12 2 2003 Page 30 of 57 MODI7 User s Guide MODIS Land Team Table 6 1 GPP 8 bit Quality Assurance Variable bit field definitions Coll 3 Variable Bitfield Binary Decimal Description of bitfield s Values Psn lkm QC MODLAND 00 0 0 Highest overall quality Bits 0 1 0121 1 Good quality 1022 2 Not produced cloud 1123 3 Not able to produce ALGOR PATH 0020 O Empirical FPAR method used Bits 2 2 0121 1 FPAR R T Main method used DEAD DETECTOR 0020 O Detectors acceptable for up to Bits 3 3 0121 50 of channel 1 2 Dead detectors affected gt 50 of adjacent detectors retrieval CLOUDSTATE 0020 0 Cloud free Bits 4 5 0121 1 Significant cloud covered pixel 1022 2 Mixed clouds pr
38. chable by the system Both the Core and Archive metadata elements are defined in the Metadata Control File MCF that accompany each process generation executable PGE in the system 4 2b ii StructMetadata Attributes For gridded Level 3 4 products such as the PSN NPP products a GridStructure object is defined no swatch structure is defined The StructMetadata 0 block contains the physical e g non science attributes of the dataset These are the minimum attributes that a software reader utility would need to correctly read and interpret the data at a physical level These include the grid name MOD Grid MODI17A2 the data set dimensions 1200x1200 the grid upper left origin coordinates the General Cartographic Transform Package GCTP map projection conversion parameters and a list of the science data set SDS names An abbreviated list of StructMetadata attributes is GridName MOD Grid MOD17A2 XDim1200 YDim1200 UpperLeftPointMtrs 8895604 158132 5559752 598833 LowerRightMtrs 7783653 638366 4447802 079066 ProjectionGCTP_ISINUS ProjParams 6371007 181000 0 0 0 0 0 0 0 86400 0 1 0 0 SphereCode 1 PixelRegistration HDFE_CENTER DimensionName YDim Size 1200 DimensionName XDim Size 1200 DataFieldName Gpp_1km DataType DENT INTI16 DimList YDim XDim DataFieldName PsnNet_1km DataType DENT INTI16 DimList YDim XDim DataFieldName Psn QC 1km DataType DENT UINT8 DimList YDim XDim ECS MODIS data are generally prod
39. clude the new 3 bit scheme of the SCF bits in MOD17A2 and overall change in the QC of the MOD17A3 In general two broad types of quality assurance activities are performed at the SCF and by the Land Data Operations Processing Environment LDOPE group at Goddard Space Flight Center 1 foutine QA 2 Broblem triggered QA The primary quality assurance activity routinely conducted at the SCF is the post processing assignment of the tile level SCIENCEQUALITYFLAG and accompanying SCIENCEQUALITYFLAGEXPLANATION Due to the volume of data this activity is performed on only a small percentage of product tiles Valid components for this field include 1 PASSED 2 FAILED 3 SUSPECT BEING INVESTIGATED 4 NOT BEING INVESTIGATED Routine QA involves periodic sampling of the product tiles using visual and statistical methods Problem triggered QA follows from a report of an inconsistency or other problem in the data which has been discovered by the Land Data Operational Product Evaluation LDOPE front line QA personnel the SCF staff or users In this case an effort is usually made to duplicate the problem under controlled conditions to resolve it During the design phase the MODIS team chose to provide a Quality Assurance measure for each pixel The quality assurance flags make it possible for the user to match data sets to their applications The user is encouraged to make use of the quality assurance information associated with each
40. d then estimating growth respiration costs for leaves fine roots and woody tissue using values defined in Table 1 3 Finally these components are subtracted from the accumulated daily PSNnet to produce an estimate of annual NPP Version 2 0 12 2 2003 Page 13 of 57 MODI7 User s Guide MODIS Land Team Table 1 3 BPLUT parameters for annual maintenance and growth respiration Parameter Units Description livewood leaf ratio None Ratio of live wood carbon to annual maximum leaf carbon livewood mr base kg C kg C7 day Maintenance respiration per unit live wood carbon per day at 20 C leaf longevity yrs Average leaf lifespan leaf gr base kg C kg Cy Respiration cost to grow a unit of leaf carbon froot_leaf_gr_ratio None Ratio of live wood to leaf annual growth respiration livewood_leaf_gr_ratio None Ratio of live wood to leaf annual growth respiration deadwood_leaf_gr_ratio None Ratio of dead wood to leaf annual growth respiration ann turnover proportion None Annual proportion of leaf turnover Annual maximum leaf mass the maximum value of daily leaf mass is the primary input for both live wood maintenance respiration Livewood MR and whole plant growth respiration GR To account for Livewood MR it is assumed that the amount of live woody tissue is 1 constant throughout the year and 2 related to annual maximum leaf mass Once the live woody tissue mass has been determined
41. e logic behind the MOD17 Algorithm in calculating both a 8 day average GPP and b annual NPP 10 1 2 The TMIN and VPD attenuation scalars are simple linear ramp functions of daily TMIN and VPD 12 2 The linkages among MODIS land products 16 2 2 Comparisons of DAO and observed meteorological data 19 4 1 MODIS tiling system Any location on the earth can be spatially referenced using the horizontal H and vertical V designators Each tile is 1200 x 1200 kilometers 27 6 1 A diagram for a hypothetical MOD17A2 quality assurance value of 4 30 9 1 A schematic diagram illustrating the process of spatial and temporal interpolation using information from land cover and QA flags In this example the landcover map has only two values dark and dashed ones In the bottom windows dark pixels are cloudy pixels and white pixels are those with the best QA conditions The thick bordered pixels are the pixels selected after filtering In temporal filling data from the previous week is used to fill MOD15 or MOD17A2 34 9 5 Merging MODIS productivity data with high resolution LandSat TM Data 36 CHAPTER II 2 1 Comparison of temporal profiles of 2001 Collection 4 MOD15A2 with original values FPAR_noQc LAI noQc and temporally linearly filled FPAR and LAI FPAR filling LAI filling and of temporal profiles of MOD17A2 with original MOD15A2 inputs GPP noQc PSN_noQc and MOD17A2 with filled MOD15A2 GPP filling PSN filling The pixel i
42. e with the ecological constraint of the amount of leaf area capable of absorbing that solar energy Such a combination avoids many of the complexities of carbon balance theory The radiation use efficiency logic requires an estimate of APAR while the more typical application of remote sensing data is to provide an estimate of the fraction of incident PAR absorbed by the surface FPAR Measurements or estimates of PAR are therefore required in addition to the remotely sensed FPAR Fortunately for studies over small spatial domains with in situ measurements of PAR at the surface the derivation of APAR from satellite derived FPAR is straightforward APAR PAR FPAR Implementation of radiation use efficiency for the MODIS productivity algorithm depends on global daily estimates of PAR ideally at the same spatial resolution as the remote sensing inputs a challenging problem Currently large scale meteorological data are provided by the NASA Data Assimilation Office DAO http polar gsfc nasa gov index php Atlas and Lucchesi 2000 at a resolution of 1 x 1 25 In Version 2 0 12 2 2003 Page 8 of 57 MODI7 User s Guide MODIS Land Team spite of the strong theoretical and empirical relationship between remotely sensed surface reflectance and FPAR accurate estimates of vegetative productivity GPP NPP will depend strongly on the quality of the radiation inputs 1 2 The Biophysical Variability of The PAR conversion efficiency v
43. ected after filtering In temporal filling data from the previous week is used to fill MOD15 or MOD17A2 Temporal filling When po cloudy free pixels of same landcover 1 Version 2 0 12 2 2003 Page 34 of 57 MODI7 User s Guide MODIS Land Team capture the effect of clouds on the local meteorology of any given pixel as DAO data is averaged across the spatial domain of the data There is nothing to be done at this point to account for cloudiness in the DAO data However three interpolation methods for filling the GPP or LAI FPAR of cloudy pixels are suggested 1 fill the GPP of a cloudy pixel with GPP values from surrounding cloud free pixels 2 fill the FPAR of a cloudy pixel with FPAR values from surrounding cloud free pixels and then recompute the GPP of the cloudy pixel using the filled FPAR 3 fill the FPAR and LAI of a cloudy pixel with FPAR and LAI values from surrounding cloud free pixels and recalculate the MOD17A2 algorithm The process of spatial and temporal interpolation is illustrated in Figure 9 1 When the central pixel of a 5x5 moving window is cloudy nearby cloud free pixels with the same landcover are used to interpolate the value of the central pixel If there is no cloud free pixel with a same landcover within the moving window the central pixel inherits the value from the preceding week 9 2 Data compositing Current
44. el et al 1995 Different photosynthesis nitrogen relations in deciduous hardwood and evergreen coniferous tree species Oecologia 104 1 24 30 Version 2 0 12 2 2003 Page 56 of 57 MODI7 User s Guide MODIS Land Team Reich P M Walters et al 1994 Photosynthesis nitrogen relations in Amazonian tree species I Patterns among species and communities Oecologia 97 1 24 30 Roy J B Saugier and H A Mooney ed 2001 Terrestrial global productivity Academic Press San Diego CA Running S and E J Hunt 1993 Generalization of a forest ecosystem process model for other biomes BIOME BGC and an application for global scale models Modeling Sustainable Forest Ecosystems D C L a R A Sedjo Washington DC American Forests 1 15 Running S W P E Thornton et al 2000 Global terrestrial gross and net primary productivity from the Earth Observing System Methods in Ecosystem Science O Sala R Jackson and H Mooney New York Springer Verlag 44 57 Running S W Scepan J 1999 Validation of High Resolution Global Land Cover Data Sets Photogrammetric Engineering amp Remote Sensing 65 1051 1060 Schubert S R Rood et al 1993 An assimilated dataset for earth science applications Bulletin of the American Meteorological Society 74 2331 2342 Schwarz P T Fahey et al 1997 Seasonal air and soil temperature effects on photosynthesis in red spruce Picea rub
45. ens saplings Tree Physiology 17 3 187 194 Thornton P E B E Law et al 2002 Modeling and measuring the effects of disturbance history and climate on carbon and water budgets in evergreen needleleaf forests Agricultural and Forest Meteorology 113 185 222 Turner D P S Urbanski et al 2003 A cross biome comparison of daily light use efficiency for gross primary production Global Change Biology 9 383 395 Waring R H and S W Running 1998 Forest Ecosystems Analysis at Multiple Scales Academic Press San Deigo CA White M P Thornton et al 2000 Parameterization and Sensitivity Analysis of the BIOME BGC Terrestrial Ecosystem Model Net Primary Production Controls Earth Interactions 4 Wolfe R E M Nishihama A J Fleig J R Kuyper D P Roy J C Storey F S Patt 2002 Achieving sub pixel geolocation accuracy in support of MODIS land science Remote Sensing of Environment 83 1 2 31 49 Version 2 0 12 2 2003 Page 57 of 57
46. er surface meteorological fields from DAO data and the subsequent estimation of maintenance and growth respiration terms that are subtracted from GPP to arrive at annual NPP The maintenance respiration MR and growth respiration GR components are derived from allometric relationships linking daily biomass and annual growth of plant tissues to satellite derived estimates of leaf area index LAI MOD15 These allometric relationships have been developed from an extensive literature review and incorporate the same parameters as those used in the BIOME BGC ecosystem process model Running and Hunt 1993 White et al 2000 Thornton et al 2002 For any given pixel within the global set of 1 km land pixels estimates of both GPP and NPP are calculated The calculations summarized in Figure 1 1 are a series of steps some of which e g GPP are calculated daily and others e g NPP on an annual basis Calculations of daily photosynthesis GPP are shown in the lower half of Figure 1 1a An 8 day estimate of FPAR from MODIS and daily estimated PAR from DAO are multiplied to produce daily APAR for the pixel Based on the at launch landcover product MOD12 a set of biome specific radiation use efficiency parameters are extracted from the Biome Properties Look Up Table BPLUT for each pixel There are five parameters used to calculate GPP as shown in Table 1 1 The actual biome specific values associated with these parameters will be discussed in
47. esent 1123 3 Not set assume clear SCF QC 0020 O Best model result Bits 6 7 0121 1 Good quality not the best NTSG Science 10 2 2 Use with caution see other QA Compute Facility 11 3 3 Could not retrieve with either Quality Control method be interpreted as biophysically relevant while non modeled pixels are given a special higher number integer code at the high end of the numeric integer range for the GPP and NPP variables Table 6 4 describes these non terrestrial land cover type codes which range from 32761 to 32767 Recall that valid GPP or NPP biophysical values are restricted to values less than or equal to 30 000 This value separation may thus be used to quickly separate subpopulations of pixels into two classes 1 valid modeled pixels lt 30000 and 2 non modeled pixels gt 30000 Version 2 0 12 2 2003 Page 31 of 57 MODI7 User s Guide MODIS Land Team Table 6 2 GPP 8 bit Quality Assurance Variable bit field definitions Coll 4 Variable Bitfield Binary Description of bitfield s Decimal Values Psn lkm QC MODLAND QC 0020 0 Best Possible Bits 0 1 0121 1 OK but not the best 1022 2 Not produced due to cloud 1123 3 Not produced due to other reasons DEADDETECTOR 0020 O Detectors acceptable for up to 50 of channel Bits 2 2 01 1 1 2 1 Dead detectors caused gt 50 adjacent detector retrieval CLOUDSTATE 00 0 O Significant clouds NOT present clear Bits 3
48. every square meter of the earth s land surface Unfortunately such a system does not exist and even if it did it would be impossible to derive vegetation productivity algorithms suited for all combinations of vegetation at such a fine resolution NASA s Earth Observing System and more specifically the MODIS instrument have been tasked with documenting and monitoring global biospheric health Running et al 2000 Thornton et al 2002 Among other things such a task requires timely and objective measures of vegetation productivity This requisite necessitates several noteworthy simplifying assumptions discussed below 2 1 The BPLUT and constant biome properties Arguably the most significant assumption made in the MOD17 logic is that biome specific physiological parameters do not vary with space or time These parameters are outlined in the BPLUT Table 2 1 within the MOD17 algorithm The BPLUT constitutes the physiological framework for controlling simulated carbon sequestration These biome specific properties are not differentiated for different expressions of a given biome nor are they varied at any time during the year In other words a semi desert grassland in Mongolia is treated the same as a tallgrass prairie in the Midwestern United States Likewise a sparsely vegetated boreal evergreen needleleaf forest in Canada is functionally equivalent to its coastal temperate evergreen needleleaf forest counterpart 2 2 Leaf area index LAI a
49. hdfeos gsfc nasa gov hdfeos index html In addition the Earth Observing System EOS Core System ECS Project Office developed the HDF EOS to GeoTIFF HEG The HDF EOS to GeoTIFF HEG tool provides conversion for HDF EOS formatted files granules converting HDF EOS swath and grid data to HDF EOS Grid GeoTIFF or a generic binary format The tool can be used to re project data from its original format to other standard projections as well as to subset data and to mosaic adjacent granules together The HEG packages are available for Sun and SGI systems in tar format and a User s Guide in Microsoft Word is available Download and installation instructions can be found at http eosweb larc nasa gov PRODOCS misr geotiff_tool html M Remember potential byte order problems can be avoided by unpacking the HDF files via the MRT or other means on the same computer with which they will be doing their analysis 4 2a Local Science Dataset SDS Attributes A complete updated description of each MODIS land product is found in the MODIS File Specification documents for MOD17A1 MOD17A2 and MOD17A3 ftp modular gsfc nasa gov pub LatestFilespecs With each SDS or HDF EOS gridfield a series of local SDS attributes are included 1 Scale factor and offset if appropriate 2 Data range minimum maximum 3 Fill value 4 Longname 4 2b Global Attributes All EOS Core System ECS data products are assigned a unique Earth Science Data T
50. ics 26 4 4 Links to MODIS friendly tools 26 5 Data Collection History 28 6 Quality Assurance 28 6 1 GPP and NPP Quality Assurance Variable Scheme 30 6 2 Identifying non terrestrial fill values in the GPP NPP data products 30 7 Missing Data 33 8 Usefulness of Data for Answering Research Questions 33 9 Considerations for MOD17A2 Product Improvement 34 9 1 Filling model values for cloudy pixels 34 9 2 Data compositing 35 9 3 Land cover 35 CHAPTER II PROPOSED IMPROVEMENTS TO THE COLLECTION 4 ALGORITHM Introduction 37 2 Problems with Collection 4 MOD17 37 3 Improvements from Collection 4 to Collection 4 5 38 4 Addition of Annual GPP and QC to Collection 4 5 MOD17A3 42 5 Final BPLUT applied to Collection 4 5 MOD17 42 6 Results 42 CHAPTER III ORDERING MOD17A2 DATA 1 Naming Conventions 43 2 Logging into the EDG 43 3 Searching the Data 44 3 1 EDG search page 44 3 2 Search in Progress page 46 3 3 Granule listing page 47 3 4 Disclaimer page 48 Version 2 0 12 2 2003 Page 3 of 57 MODI7 User s Guide MODIS Land Team Table of Contents cont 4 Ordering the Data 49 4 Ordering options page 49 4 2 Ordering options page part II 49 4 3 Order form 52 4 4 Reviewing your order Step 3 53 4 5 Submitting the order 53 5 The DataPool 54 MODIS FAQ s 55 REFERENCES 56 Version 2 0 12 2 2003 Page 4 of 57 MODI7 User s Guide MODIS Land Team List of Figures Fig Caption Page CHAPTER I 1 1 Flowcharts showing th
51. ie query Query medis tused Use of thas site constitutes an apsessent to US Goverzanent security policy and US Goversment prasy pokey Ow acerssibulty policy is alto evailable NASA Tosh Papresensonve Medora Macte Mail Code 413 NASA GSFC Creenbek MD 20771 R the back button on the browser and correct the issue Figure 3 3 The Search in progress page when your search is successful and the other fields will give you more information including number of granules returned Fig 3 3 Some errors on this page will be in the form of server errors caused by neglecting to fill in a required field on the search page If this happens just hit 3 3 Granule listing page The granule listing page lists the granules that were returned as a result of your search This page also gives you the choice of adding the granules to your shopping cart Using this Version 2 0 12 2 2003 Page 47 of 57 MODI7 User s Guide MODIS Land Team page you may select some or all of the granules by clicking the checkboxes in the desired granules row Fig 3 4 Once you have selected your granules you must add them to your cart 3 4 Disclaimer page After your data are found the EDG will display a disclaimer Fig 3 5 This page is displayed because of the current state of the MODIS data stream MODIS data are still being evaluated and validated If you wish to continue click the Accept button at the bottom of the page eae hs ere a
52. ile is limited to 64Kb The ECS tile level metadata sections are summarized in Table 4 1 Note that although the examples below refer to the MOD17A2 8 day photosynthesis ESDT these metadata also apply to the MOD17A3 annual NPP product 4 2b i Core and Archive Metadata What s The Difference The ECS Core metadata CoreMetadata 0 are granule level metadata that describe a number of useful tile level attributes for the granules held in a common Collection such as the current Collection 4 These metadata are also known as INVENTORY metadata reflecting the fact they constitute a baseline resource Table 4 1 ECS Metadata Summary for PSN PSNnet and NPP Data Products Block Organization Contents StructMetadata 0 Object Data Language ODL Geospatial data tile origin coordinates map projection attributes CoreMetadata 0 Object Data Language ODL Inventory attributes ArchiveMetadata 0 Object Data Language ODL Archive metadata attributes Version 2 0 12 2 2003 Page 22 of 57 MODI7 User s Guide MODIS Land Team describing the inventory of data available INVENTORY or Core metadata includes all granule level metadata that will reside in ECS inventory tables and will thus be searchable Archive metadata stored in each granule in the ArchiveMetadata O block on the other hand contain metadata fields that the producer wants to accompany the granule when it is delivered to end users but need not be sear
53. ing line because of its reliance on other MODIS products Fig 2 1 is one of the last products to be updated Collection 3 If your data were downloaded prior to December 13 2002 the algorithm uses Collection 3 inputs of Land Cover and LAI FPAR and an older less restrictive BPLUT To avoid potential confusion this version of the BPLUT will not appear in the User s Guide The dataset using the older BPLUT is no longer available from the EDC If your data were downloaded after December 13 2002 and before January 15 2003 then you have the most complete version of Collection 3 data The primary modification is a change in the VPD constraints on max in Equation 1 1 This product continues to rely upon Collection 3 inputs of Land Cover MOD12Q1 and LAI FPAR MOD15A2 Collection 4 Distribution of Collection 4 data began on January 15 2003 Collection 4 will continue to use the improved algorithm of Collection 3 but will employ Collection 4 inputs from both the Land Cover and LAI FPAR algorithms providing the most up to date calculations available Collection 4 5 While not available through the EOS Data Gateway as a standard product Collection 4 5 represents an improved MOD17A2 A3 dataset This dataset includes a revised BPLUT Table 3 1 based on the most current research and is described in Chapter II of this document This dataset includes temporal interpolation of cloud contaminated MOD15A2 LAI FPAR data and spatial smoothing of the DA
54. ipia hiel Tee HOE geste cep s Peeer piy J PEO sandre nomaj y mal uz ADE Libs iimas daba ce Lis Lie piel Jame 15 TIL L Kile i aXX Tha cuum m tha beilumm is comsimimpt si dn OFS ee Erbin ssi Likely ingtiebed by a hzgh meerwry TE nm erent that cr hbe Beeu idm dr Foel Eidos Tpommimire POMFET Wilhi Eh dra Gl far o Dabl Vip ida Tis Ll T tha ONE instrabaxt hui bassa amp cquiripg isti uming election mis as dim BODES Lindl aie real il de la Dd is mue dela wong Moear Suppim 5 amd alcoi slds bcuiamca dela col lected A thet OIE A Bremy othe ei ls bebben rel der DLE rei ci pillo dmLh es les lil tai eric side Dow lih E a Aur mwvvwhu rna PFryu diua Imi AL 1 mmm e ee e lala Figure 3 5 The disclaimer 4 Ordering the Data 4 1 Ordering options page The ordering options page is very similar to the granule listing page You are shown your granules 10 at a time and you may choose the ordering options for them one at a time or all at once For our purposes just click on Choose Options Fig 4 1 This will take you to the second part of the ordering options page 4 2 Ordering options page part II The second half of the ordering options page allows you to select the method of data transfer Fig 4 2 For our purposes select FtpPull Then mark the selection box which states you wish to use this option for all the granules of this data set Fig 4 2 Then press the OK button Version 2 0 12 2 2003 Page 49 of 57 MO
55. is obtained from the BPLUT Fine root mass Fine Root Mass kg is then estimated as Fine Root Mass Leaf Mass froot leaf ratio 1 5 where froot leaf ratio is the ratio of fine root to leaf mass unitless as obtained from the BPLUT Leaf maintenance respiration Leaf MR kg C day is calculated as Leaf MR Leaf Mass leaf mr base Q10 mr 290 100 1 6 where leaf mr base is the maintenance respiration of leaves kg C kg ct day as Obtained from the BPLUT and Tavg is the average daily temperature C as estimated from the DAO meteorological data The maintenance respiration of the fine root mass Froot_MR kg C day 1s calculated as Froot MR Fine Root Mass froot mr base Q10 mr 20 0 10 0 1 7 where froot mr base is the maintenance respiration per unit of fine roots kg C kg C day at 20 C as obtained from the BPLUT Finally PSNnet kg C day can be calculated from GPP Equation 2 2 and maintenance respiration Equations 2 5 2 6 as PSNnet GPP Leaf MR Froot MR 1 8 As with GPP PSNnet is summed over an 8 day period M This product does not include the maintenance respiration associated with live wood Livewood MR nor does it include growth respiration GR 1 3c Annual maintenance respiration Given a calendar year s worth of outputs from the daily algorithm the annual algorithm Fig 1 1b estimates annual NPP by first calculating live woody tissue maintenance respiration an
56. itations of suboptimal environmental conditions on the related conversion efficiency for different biome types Detailed structure and processes of the algorithm can be found in Chapter I The objective of this Chapter is to provide users with a brief outline of the maturation of MOD17 products from Collection 4 to Collection 4 5 available upon request from NTSG and information on how Collection 4 5 improves the quality of MOD17 products First we provide a retrospective view of the Collection 4 MOD17 algorithms and its problems second we show how we have resolved these issues and improved MOD17 Finally some results from Collection 4 5 are provided 2 Problems with Collection 4 MOD17 To investigate the problems with the MOD17 algorithm it is necessary to understand how MOD17 operates Fig 2 1 Chapter I For a given pixel MOD17 requires two upstream MODIS data inputs MOD12Q1 and MOD15A2 The algorithm reads MOD12Q1 to obtain land cover type to match the corresponding parameters in the Biome Parameter Look Up Table BPLUT Table 2 1 Chapter I The LAI FPAR algorithm MOD15A2 contains an 8 day MVC Maximum Value Composite Fraction of Photosynthetic Active Radiation absorbed by the green vegetation canopy FPAR and a corresponding Leaf Area Index LAI The MOD17 algorithm assumes that there is no variation of FPAR and LAI within a given 8 day period Outputs from MOD12Q1 and MOD15A2 provide real time ground vegetation conditions The
57. le MBR defined by the tile rather than the typically trapezoidal shaped polygon represented by the GRINGPOINT coordinates 4 2b iii ArchiveMetadata Attributes The following Archive metadata attributes are designed to assist end users in using and effectively interpreting the Terra MODIS PSN NPP data These attributes are not considered essential as searchable metadata in the overall ECS metadata partially because some of this information is overlapped by almost equivalent elements in the CoreMetadata 0 attributes which as INVENTORY metadata are searchable ALGORITHMPACKAGEACCEPTANCEDATE ALGORITHMPACKAGEMATURITYCODE ALGORITHMPACKAGENAME ALGORITHMPACKAGEVERSION CHARACTERISTICBINANGULARSIZE CHARACTERISTICBINSIZE DATACOLUMNS DATAROWS DESCRREVISION GEOANYABNORMAL GEOESTMAXRMSERROR GLOBALGRIDCOLUMNS GLOBALGRIDROWS GRANULEBEGINNINGDATETIME GRANULEDAYNIGHTFLAG GRANULEENDINGDATETIME INSTRUMENTNAME LOCALINPUTGRANULEID LONGNAME MAXIMUMOBSERVATIONS NADIRDATARESOLUTION NUMBEROFGRANULES PLATFORMSHORTNAMEB PROCESSINGCENTER PROCESSINGDATETIME PROCESSINGENVIRONMENT SPSOPARAMETERS NORTHBOUNDINGCOORDINATE EASTBOUNDINGCOORDINATE SOUTHBOUNDINGCOORDINATE WESTBOUNDINGCOORDINATE 4 2b iv Other Helpful ECS Metadata References A number of detailed ECS related web pages and Adobe Postscript documents may be found at http observer gsfc nasa gov including the document ECS ProvidersGuideToMetadata pdf Although document describes the
58. lipped herbaceous biomass in July and integrated MOD17 PsnNet from Composite Period 1 through 193 Reeves et al in prep indicating Collection 4 5 significantly improves NPP compared to Collection 4 Several of the corrections discussed in this Chapter cannot be performed in a forward processing mode Therefore at the end of each year the data from that year will be reprocessed to include the corrections such as linear interpolation of the MOD15A2 input data While we feel that Collection 4 5 data are the most accurate data users should balance their research needs with data availability to determine the product that fits their needs The standard Collection 4 0 product will continue to be available in near real time Currently Collection 4 5 data are available for 2001 2002 Additional data will be released as they become available A The global Collection 4 5 image gallery can be found at http www ntsg umt edu Data from Collection 4 5 2001 2002 are available upon request from NTSG Version 2 0 12 2 2003 Page 42 of 57 MODI7 User s Guide MODIS Land Team CHAPTER III ORDERING MOD17A2 DATA 1 Naming Conventions In order to efficiently take advantage of the Earth Observing System Data Gateway EDG it is useful to understand the naming convention for the MODIS granule ID The granule names are a combination of several pieces of key information which will help you to discern if the granule in question is what you desire From
59. ly the MOD17A2 output is an 8 day summation product However in cloudy areas such as the tropics this scheme is not always sufficient as there are times during the year for which there are no cloud free 8 day periods As a result researchers at NTSG are looking into a 16 day summation which might be more useful for exploring interannual differences in GPP This conversion will only occur if it provides an improved data stream For those areas of the earth s land surface which are reasonably cloud free an 8 day summation may be continued 9 3 Land cover The land cover classification scheme ingested by the MODI7 Algorithm is at a 1 km resolution as are all MODIS products There are areas of the world however for which improved finer resolution land cover data sets are available Given the importance of accurate land cover for the MOD17 algorithm research is needed to determine if such data sets would improve the MOD17 end products For example consideration is being given to using LandSat derived land cover with 15 m resolution in place of the MOD12Q1 Land Cover product in some areas Fig 9 5 In this way there would be a multi resolution product that could increase the accuracy of the MOD17 Algorithm LandSat land cover data could provide enhanced spatial resolution while MODIS data LAI FPAR provide the temporal resolution needed for the goals set by the MODIS Science Team LandSat derived land cover hasn t been used often for productivi
60. mospheric water deficit is commonly reached in semi arid regions of the world for much of the growing season So our algorithm mimics this physiological control by progressively limiting daily GPP reducing when high vapor pressure deficits are computed from the surface meteorology We also assume nutrient constraints on vegetation growth to be quantified by limiting leaf area rather than attempting to compute a constraint through amp This assumption isn t entirely accurate as ranges of leaf nitrogen and photosynthetic capacity occur in all vegetation types Reich et al 1994 Reich et al 1995 Turner et al 2003 Spectral reflectances are somewhat sensitive to leaf chemistry so the MODIS derived FPAR and LAI may represent some differences in leaf nitrogen content but in an undetermined way To quantify these biome and climate induced ranges of we simulated global NPP in advance with a complex ecosystem model BIOME BGC and computed the or conversion efficiency from APAR to final NPP This Biome Parameter Look Up Table BPLUT contains parameters for temperature and VPD limits specific leaf area and respiration coefficients for representative vegetation in each biome type Running et al 2000 White et al 2000 The BPLUT also defines biome differences in carbon storage and turnover rates Since the relationships of environmental variables especially temperature to the processes controlling GPP and those controlling autotrophic res
61. nd fraction of absorbed photosynthetically active radiation FPAR As illustrated in Figure 2 1 the primary productivity at a pixel is dependent upon among other things LAI and FPAR calculated with the MOD15 algorithm The LAI FPAR product is an 8 day composite product The MOD15 compositing algorithm uses a simple selection rule whereby the maximum FPAR across the eight days is chosen for the inclusion as the output pixel The same day chosen to represent the FPAR measure also contributes the current pixel s LAI value This means that although primary productivity is calculated daily the MOD17 algorithm necessarily assumes that leaf area and FPAR do not vary during a given 8 day Version 2 0 12 2 2003 Page 16 of 57 Classification MOD120Q1 EOS biome class key ENF evergreen needleleaf forest EBF evergreen broadleaf forest DNF deciduous needleleaf forest DBF deciduous broadleaf forest WL grassy woodland Werass wooded grassland Cshrub closed shrubland Oshrub open shrubland MODI7 User s Guide MODIS Land Team Table 2 1 The Biome Properties Look Up Table BPLUT for MOD17 BIOME CLASSIFICATION PARAMETER ENF EBF DNF DBF MF WL Epsilon_max 0 001008 0 001159 0001103 0 001044 0001116 0 000300 Daily Tmin max C 3 31 9 09 10 44 794 3 50 11 39 Tmin min C amp 00 3 00 3 00 3 00 3 00 3 00 VPD_max Pa 2500 3900 3100 2500 2500 3100 VPD min Pa 650 1100 650 650 650 930 SLA projected m2 kg leaf C 21 1 23 3 31 0
62. ni m ii a MA Genero Dargadrr pirim Pirru are aalhariird Ya uur thin i irr dr rth aw ieee ee n nati cide dala aal ee Ait ta hii ai a ie iss id A ia Eee i er nri aa Bs a eed ic tds ser Th pe in rd and iaip be Mansi ii aikan than Cu mdr ned mod desig an ai A APR DA EE ILE Fon FRAIS ey PCS EOM AL LM b VR E II A Be FEMME A AE P mri is DT 17 Figure 2 1 The EDG home page a registered user you are still able to order data as a guest although you will not be able to save any searches or user information 3 Searching the Data 3 1 EDG search page The search page Fig 3 1 consists of several fields in which you enter the information on which you want to search Since the EDG has an extensive online help system and several tutorials here we will only cover what you need to know to order MOD17A2 data First you will have to specify which data set you wish to search for In this case you will want to enter MOD17A2 in the field labeled Method 1 Data Set Lookup Then just click on the button labeled GO The screen will change slightly to display the results of the data set search You should see something like MODIS TERRA NET PHOTOSYNTHESIS 8 DAY L4 GLOBAL 1KM ISIN GRID V003 You want to select the most recent version of the data unless Version 2 0 12 2 2003 Page 44 of 57 MODI7 User s Guide MODIS Land Team See ere FT LRL I bre eri Ain Ep T rex fm ESSE S nate Desi are Igi HM icu T m Cim T aston
63. od GR Leaf GR livewood leaf gr ratio 1 13 where livewood leaf gr ratio is the ratio of livewood leaf growth respiration unitless as found in the BPLUT And lastly deadwood growth respiration Deadwood GR kg C day is calculated as Deadwood GR Leaf GR deadwood leaf gr ratio 1 14 where deadwood leaf gr ratio is the ratio of deadwood to leaf growth respiration unitless as found in the BPLUT As a final step the per pixel annual net primary productivity NPP kg C day is calculated as the sum of the cumulative daily PSNnet annsum daily PSNnet kg C day less the costs associated with annual maintenance and growth respiration such that NPP annsum dailyPSNnet Livewood MR Leaf GR Froot GR Livewood GR Deadwood_GR 1 15 where all terms have been previously defined Version 2 0 12 2 2003 Page 15 of 57 MODI7 User s Guide MODIS Land Team 1 km MODIS MODIS product suite Surface Reflectances MOD09 LAI FPAR GPP PSNnet 1 km MODIS 8 day summation 8 day summation Surface MOD15A2 MOD17A2 Reflectances Land MODAGAGG Cover Biome Designation MOD12 LAI FPAR GPP PSNnet Daily Intermediate Daily intermediate MOD15A1 MOD17A1 Figure 2 1 The linkages among MODIS land products 2 Simplifying Assumptions for Global Applicability In an ideal world remote sensing would render an infallibly accurate depiction of surface conditions and deliver the data in a timely cost effective manner for
64. of interest Few proprietary image processing or geographic information system GIS software have the capability to reproject MODIS data from an ISIN projection Fortunately however there are good tools which are simple to download and are freely available The primary tool currently used to reproject MODIS data in both formats is the MODIS Reprojection Tool MRT This tool and more information can be found at Attp edcdaac usgs gov tools modis Version 2 0 12 2 2003 Page 20 of 57 MODI7 User s Guide MODIS Land Team 4 2 File format of MOD17 end products All NASA biophysical products are archived in the NASA HDF EOS data format HDF EOS is based upon the Hierarchical Data Format pioneered by the National Center for Supercomputer Applications NCSA at the University of Illinois Champaign Urbana The HDF EOS format has the advantage of multiple layers of data and supporting ancillary information such as projection characteristics scaling factor time and date of production etc in a single file The drawback is that in order to use the actual vegetation productivity layer one must extract this layer from the data stack Therefore the MRT serves two purposes 1 reproject MODIS data from ISIN or SIN 2 extract the desired data layer from the stack Several tools and software systems allow the user to browse through the various data layers within a given HDF EOS file The growing body of HDF EOS tools can be found at http
65. on Group NTSG suggest that the relationship between surface observations and DAO data across the U S appears reasonable Fig 2 2 but comparisons have yet to be made on a global scale 3 Dependence on MODIS Land Cover Classification MOD12Q1 One of the first MODIS products used in the MOD17 algorithm is the Land Cover Product MOD12Q1 The importance of this product cannot be overstated as the MOD17 algorithm relies heavily on land cover type through use of the BPLUT Table 3 1 While the primary product created by MOD12 is a 17 class IGBP International Geosphere Biosphere Programme landcover classification map Belward et al 1999 Scepan 1999 the MOD17 algorithm employs Boston University s UMD classification scheme Table 3 1 More details on these and other schemes and their quality control considerations can be found at the Land Cover Product Team website hitp geography bu edu landcover userguidelc index html Given the global nature and daily time step of the MODIS project a broad classification scheme which retains the essence of land cover is necessary Since all MODIS products are designed at a 1 km grid scale it can be difficult to obtain accurate land cover in areas with complex vegetation and misclassification can occur However studies have suggested that the MODIS vegetation maps are accurate to within 65 80 with higher accuracies for pixels that are largely homogeneous and allow for consistent monitoring of the
66. opriate link If you do not wish to become Version 2 0 12 2 2003 Page 43 of 57 MODI7 User s Guide MODIS Land Team E f dm em Pueri Ti Pub m um the 2143 we utem Gee d oR du P Nui quac cun gr RR oe ms dcm AB HUNLE Fa ee LUN e LAA AME y A PP E Pl A Eb ide Pis eed eui dorem tS OB BRE P LP A NTE cee ee rth Observing System Data Gateway Eart iw and ondes dosTh ee Lan pe inca lr aad aed ete E a Enterita Data at gen aa Dala Comer Btalus Briss ds pons BEES Gear roue Tera Pragecta Fuge siiis aderat plenus vien Ha lara easily ALES Lets tent curfu 343 Abeg TRE Fani pacts pags DOM VITE SST ER ri Daighi priiis qe riw rider add Benn the EOD Fii 7 en md eben prias please vun DC Data Producti Forgot rey aT pb i Pind 4 Jon DEII Ee oe UT t2 Pha PS ey EXT 00100 UT Caria Jan og Tha Loddard Cab GCG mi ER dde tee tz Tz A nee D Doi bngyan hast Bade cin AER of Rash FO bea NAAA O 2128 eT A ieri SU Bale van nl de reed iberaj Fap iota FTC pets iret ba Emiame barmuk c rep Er tmo Peer dir coh soc ua uta kart contrasta La related femmt sddmieen Calc estan a barto Mera aed Venit rec Apri 30 2021 F a a E Hina ba WE mtii Paap A Futsal Saat out T h Mora CR ita rus wan recs tha ED e erred ee SF unes 1 are ET LIDO a ihata Eric Earth pont daba mur an amd aim held Earth E amas data Bian ded eias Enemy 5 b rrmabrr ine ihis arida Carei uiam flee mortie m
67. ordering more than one data set say MOD17A2 and MOD17A3 not all of your granules are ready The display shows 10 granules at a time Read the page carefully to find on which page the second data set starts and click on that button the numbered buttons on the bottom of the page 1 10 Fore each extra data set repeat the steps for data retrieval that were covered previously and then you proceed to the next section To move forward in the ordering process press the Go to Step 2 Order Form button Version 2 0 12 2 2003 Page 50 of 57 MODI7 User s Guide MODIS Land Team Derim Carn ME SC Y E peu mese kB MIS CHR A TEE OCA A HA e lc VARI AAN t i Eae Lp UIT ome muri dd AE UA HE 4 ITI ge de Ll A A Li Ed ee Lee dc A Carzhra dp OO Lar ay meras EN A A Figure 4 2 Choosing ordering options part II Version 2 0 12 2 2003 Page 51 of 57 MODI7 User s Guide MODIS Land Team Rugs dem ducem pue aom C e nen fm zzzi jhe Riches eel ol gt e Spa Cart Step 1 Choose Ordering Options ine q poa p ds arene Hai MG Kad Raga heb mens sd di de piss eg Te ey h A em Bon gun d pea cm e rp reque inem antc Eeri obe Finge t aea Lue SMO A 4JEIEED EDCEZS r Erst lii SOMO TEER LTB EDCECS r ERA eae SOMONE LITER DOES r nM Lake GH bie ii BR r Petri E A EDs r ited Vemi POODE ETE EDpos r EEDA Urea MCMODIMAIDIIENENUYS IER Det r LAR O A II eee eee Ce A i Lor nd m Sabre ma ar Cree bp C
68. ory just click the button that allows you to continue on to the next step Submit Order Fig 4 5 4 5 Submitting the order Once you have submitted the order you will see a screen similar to the search in progress page This will notify you if any errors occurred while contacting the data center which houses the data you requested Once this step is complete you will see a page stating that your order was submitted with a comprehensive listing of what you ordered You will receive e mail notification of your request at the address you provided in the order page You will also receive a notification via e mail when your order is being filled This last notification is important because it tells you from where you need to get your data ftp address directory where the data are stored file names If you have any other questions please utilize the materials at the EDG s web site Attp edcimswww cr usgs gov pub imswelcome Version 2 0 12 2 2003 Page 53 of 57 MODI7 User s Guide MODIS Land Team PH dibs E Pani reacio fout Lal Bert jasa Linear iib rj diresa 17 DLL aplica NT HEN WRTTED EMT Ficus ausis ESEE iffoloacoce GOTEENREKT EEIFFINS apDEEMe Enas E isba E Asiti w aasuaLz2a nas Lab Bane ETEF gee ci dirsas in ETAT Ti Anais FI UNNIN WETTED Te Poms PLL ETL LL i VI E De ie fe Joke Bo Bairi feigaainathen lsin Lab hara Read jmaztkipcarjab ony tiirsa rhe Anpuhara Mi roster KT mu WNITED Sore
69. piration have fundamentally different forms Schwarz et al 1997 Maier et al 1998 it seems likely that the empirical parameterization of the influence of temperature on production efficiency would be more robust if the gross production and autotrophic respiration processes were separated This is the approach employed in the MOD17 algorithm Version 2 0 12 2 2003 Page 9 of 57 MODI7 User s Guide MODIS Land Team Figure 1 1 Flowcharts showing the logic behind the MOD17 Algorithm in calculating both a 8 day average GPP and b annual NPP Version 2 0 12 2 2003 Page 10 of 57 MODI7 User s Guide MODIS Land Team Table 1 1 BPLUT parameters for daily gross primary productivity Parameter Units Description EUM kg C MJ 5 The maximum radiation conversion efficiency TMINmax C The daily minimum temperature at which y for optimal VPD TMINmin C The daily minimum temperature at which e 0 0 at any VPD VPDmax Pa The daylight average vapor pressure deficit at which Emax for optimal TMIN VPDmin Pa The daylight average vapor pressure deficit at which e 0 0 at any TMIN 1 3 The MOD17A2 MOD174A3 algorithm logic 1 3a Gross primary productivity The core science of the algorithm is an application of the described radiation conversion efficiency concept to predictions of daily GPP using satellite derived FPAR from MOD15 and independent estimates of PAR and oth
70. pixel because it permits quick objective and repeatable screening to filter out undesirable pixels The QA flags that users will find in the MOD17A2 products are summarized in Table 6 1 Each flag is divided into a series of bitfields which can be parsed to allow separate interpretation of each field for maximum control over the data set The EDC DAAC Earth Resources Observation Systems Data Center Distributed Active Archive Center is currently working on tools that will enable the user to automatically parse and process each bitfield The process of parsing bitfields may seem confusing at first As a general rule if the user does not wish to examine every bitfield independently a threshold value of zero should produce the best quality pixels for scientific analysis although this may reduce the number of pixels available for evaluation Fig 6 1 There are two steps to interpreting MOD17A2 QA values 1 Convert the file pixel QA number to its binary equivalent 2 Alter the binary equivalent to become 8 digits long For example if the binary equivalent is 100 you must add zeroes to the left hand side until there are a total of 8 binary digits since the MOD17A2 QA value is an 8 bit unsigned integer So 100 becomes 00000100 3 Parse individual bit fields from the 8 bit integer Figure 6 1 and interpret their meaning per Table 6 1 Therefore in this example the third bit 1 indicating that dead detectors caused gt 50 adjacent detector
71. ration 14 2 1 The Biome Properties Look Up Table BPLUT for MODI7 17 3 1 The land cover types used in the MODI7 Algorithm 20 4 1 ECS Metadata Summary for PSN PSNnet and NPP Data Products 22 4 2 Summary of output variables from the MODIS vegetation productivity algorithm 26 6 1 GPP 8 bit Quality Assurance Variable bit field definitions Collection 3 and earlier 31 6 2 GPP 8 bit Quality Assurance Variable bit field definitions Collection 4 32 6 3 NPP 8 bit Quality Assurance Variable bit field definitions Collection 4 32 6 4 GPP 8 day summation and annual NPP non terrestrial fill value code definitions 33 Version 2 0 12 2 2003 Page 7 of 57 MODI7 User s Guide MODIS Land Team Synopsis Vegetative productivity is the source of all food fiber and fuel available for human consumption and therefore defines the habitability of the earth The rate at which light energy is converted to plant biomass is termed primary productivity The sum total of the converted energy is called gross primary productivity GPP Net primary productivity NPP is the difference between GPP and energy lost during plant respiration Campbell 1990 Global productivity can be estimated by combining remote sensing with carbon cycle processing The U S National Aeronautics and Space Administration NASA Earth Observing System EOS currently produces a regular global estimate of gross primary productivity GPP and annual net primary productivity
72. ributes PSA if supplied by the Principle Investigator are contained within the ADDITIONALATTRIBUTE objects The spatial extent of the tile is described by the GRINGPOINT latitude and longitude attributes where these describe the four corner coordinates N E S W of the tile ADDITIONALATTRIBUTENAME ADDITIONALATTRIBUTESCONTAINER ASSOCIATEDINSTRUMENTSHORTNAME ASSOCIATEDPLATFORMINSTRUMENTSENSORCONTAINER ASSOCIATEDPLATFORMSHORTNAME ASSOCIATEDSENSORSHORTNAME AUTOMATICQUALITYFLAG AUTOMATICQUALITYFLAGEXPLANATION DAYNIGHTFLAG EXCLUSIONGRINGFLAG GPOLYGONCONTAINER GRINGPOINTLATITUDE GRINGPOINTLONGITUDE GRINGPOINTSEQUENCENO INPUTPOINTER LOCALGRANULEID LOCALVERSIONID MEASUREDPARAMETERCONTAINER PARAMETERNAME PARAMETERVALUE PGEVERSION PRODUCTIONDATETIME QAPERCENTCLOUDCOVER QAPERCENTINTERPOLATEDDATA QAPERCENTMISSINGDATA QAPERCENTOUTOFBOUNDSDATA RANGEBEGINNINGDATE RANGEBEGINNINGTIME RANGEENDINGDATE RANGEENDINGTIME REPROCESSINGACTUAL REPROCESSINGPLANNED SCIENCEQUALITYFLAG SCIENCEQUALITYFLAGEXPLANATION SHORTNAME VERSIONID Users may find the following distinction helpful to understand the difference between the GRINGPOINT coordinates in the CoreMetadata 0 and the NORTH SOUTH EAST WEST BOUND coordinates in the ArchiveMetadata 0 block In the ArchiveMetadata block the Version 2 0 12 2 2003 Page 24 of 57 MODI7 User s Guide MODIS Land Team NORTH SOUTH EAST and WEST BOUND coordinates represent a minimum bounding rectang
73. ristics of the nearest neighbor DAO cell As a result a DAO cell boundary line may appear in 1 km MOD17 images due to the relatively steep gradients between DAO cells Fig 2 2 Such treatment on a global or regional scale may be acceptable while at the local scale especially for topographically diverse terrain or sites located at relatively abrupt climatic gradient zones it may introduce inaccurate climatic predictions for some productivity calculations 3 Improvements from Collection 4 to Collection 4 5 We solved the first problem related to MOD15A2 inputs by removing poor quality FPAR and LAI data based on the QC label for every pixel If any LAI FPAR pixel does not meet the quality screening criteria its value is determined through linear interpolation between the previous period s value and that of the next good period Fig 2 1 illustrates how this temporal filling approach is applied to a MODIS pixel in the Amazon region where higher frequency and persistence of cloud cover exists As depicted in Fig 2 1 contaminated MOD15A2 was improved as the result of the filling process However there are some unusual 8 day periods with lower FPAR and LAI but good QC labels In spite of this we still depend on QC labeling as the only source of quality control Improved MOD15A2 leads to improvements of MOD17 Under most conditions 8 day composited GPP will increase because the temporal filling process generally acts to increase FPAR Changes in 8 d
74. s located in the Amazon rainforest lat 1 0 lon 60 with the MODIS land cover Evergreen Broadleaf Forest EBF 39 2 2 Comparison of Collection 4 and Collection 4 5 MOD17A2 GPP composite period 241 and MOD17A3 NPP for 2001 40 3 1 Distribution of more than 5 000 WMO stations for 2001 and 2002 41 3 2 Percent of WMO stations with changes in RMSE and COR between spatially interpolated and non interpolated DAO For most stations DAO accuracies are improved reduced RMSE and increased COR as a result of spatial interpolation 41 Version 2 0 12 2 2003 Page 5 of 57 MODI7 User s Guide MODIS Land Team List of Figures cont CHAPTER III 1 1 The MOD17A2 Standard Product naming convention 2 1 The EDG home page 3 1 The EDG search page 3 2 Choosing the time range 3 3 The Search in progress page 3 4 The page listing the granules you have requested 3 5 The disclaimer 4 1 Choosing ordering options 4 2 Choosing ordering options part II 4 3 Choosing ordering options the Ready page 4 4 The order form 4 5 Verifying and submitting the order Version 2 0 12 2 2003 Page 6 of 57 43 44 45 46 47 48 49 50 51 52 53 54 MODI7 User s Guide MODIS Land Team List of Tables Table Title Page CHAPTER I 1 1 BPLUT parameters for daily gross primary productivity 11 1 2 BPLUT parameters for daily maintenance respiration 12 1 5 BPLUT parameters for annual maintenance and growth respi
75. s the ratio of live wood mass to leaf mass unitless and is obtained from the BPLUT Once the mass of live wood has been determined it is possible to calculate the associated maintenance respiration Livewood MR kg C day as Livewood MR Livewood Mass livewood mr base annsum mrindex 1 10 Version 2 0 12 2 2003 Page 14 of 57 MODI7 User s Guide MODIS Land Team where livewood mr base kg C kg a day is the maintenance respiration per unit of live wood carbon per day from the BPLUT and annsum mrindex is the annual sum of the maintenance respiration term Q10 mr 1870 01001 1 3d Annual growth respiration and net primary productivity Annual growth respiration and maintenance costs are based on their relationship to leaf growth respiration Leaf GR kg C day which is calculated as Leaf GR ann leaf mass max ann turnover proportion leaf gr base 1 11 where ann turnover proportion unitless is the annual turnover proportion of leaves and leaf gr base is the base growth respiration kg C kg C day for leaves Both of these terms are acquired from the BPLUT Growth respiration for fine roots Froot_GR kg C day is calculated as Froot GR Leaf GR froot leaf gr ratio 1 12 where froot leaf gr ratio is the ratio of fine root growth respiration to leaf growth respiration unitless as found in the BPLUT Next the growth respiration of livewood Livewood GR kg C day can be calculated as Livewo
76. ta Version 2 0 12 2 2003 Page 55 of 57 MODI7 User s Guide MODIS Land Team REFERENCES Atlas R M Lucchesi R 2000 File Specific for GEOS DAS Celled Output Goddard Space Flight Center Greenbelt Maryland Belward A J Estes et al 1999 The IGBP DIS Global 1 km Land Cover Data Set DISCover A Project Overview Photogrammetric Engineering amp Remote Sensing 65 1013 1020 Campbell N A 1990 Biology Redwood City CA The Benjamin Cummings Publishing Company Inc Cannell M 1982 World Forest Biomass and Primary Production Data London Academic Press Field C B J T Randerson and C M Malmstrom 1995 Global net primary production Combining ecology and remote sensing Remote Sensing of the Environment 281 237 240 Goward S C Tucker et al 1985 North American vegetation patterns observed with the NOAA 7 advanced very high resolution radiometer Vegetatio 64 3 14 Hansen M R DeFries et al 2000 Global land cover classification at the 1km spatial resolution using a classification tree approach International Journal of Remote Sensing 21 1331 1364 Heinsch F A S W Running et al in prep Validation of the MOD17A2 GPP product using eddy covariance flux tower data Hoffmann W A and A C Franco 2003 Comparative growth analysis of tropical forest and savanna woody plants using phylogenetically independent contrasts Journal of Ecology 91 475 484 Hunt E J 1994
77. ty analysis in the past because of the long return time 16 days and cloudiness associated with these products but combining that land cover with the daily overpass of the MODIS data could improve the MOD17A2 product As with all MODIS products MOD17 would continue to be produced at a 1 km resolution Version 2 0 12 2 2003 Page 35 of 57 MOD 17 User s Guide MODIS Land Team MODIS 1km GPP Merged Image Figure 9 5 Merging MODIS productivity data with high resolution LandSat TM Data Version 2 0 12 2 2003 Page 36 of 57 MODI7 User s Guide MODIS Land Team CHAPTER II PROPOSED IMPROVEMENTS TO THE COLLECTION 4 ALGORITHM 1 Introduction MODIS is the primary global monitoring sensor on the two NASA EOS satellites and features improved geolocation atmospheric correction and cloud screening provided by the MODIS science team MOD17 is a near real time continuous consistent operational data set of global terrestrial gross primary productivity GPP and net primary productivity NPP at a 1 km spatial scale at both 8 day MOD17A2 and annual MOD17A3 time scales Current satellite data for MOD17 comes from the TERRA EOS AM platform which was launched on December 19 1999 MOD17 began to provide 8 day estimates of GPP in December 2000 There are currently almost 3 years of MODIS data available The MOD17 algorithm is based on the idea of the conservation ratio between APAR and NPP proposed by Monteith 1972 and the lim
78. uced and organized at a high level in terms of collections A collection may be considered a generation of data sharing the common property of having been produced with a latest milestone set of processing algorithms Within a given collection considerable effort is made to re process all ESDT products usually for entire period when raw Level 0 1A satellite data are available the period of record into consistent structured collections Since Terra launched in December 1999 there have currently been four 4 collections produced Collection 4 is currently being re processed as of March 2003 with each collection taking into account the latest algorithm improvements Each subsequent collection is therefore expected to represent incrementally higher quality science data than the previous Scientists using Terra MODIS data are therefore encouraged to base their science research and applications on the most recent collection of data available Recall that the Version 2 0 12 2 2003 Page 23 of 57 MODI7 User s Guide MODIS Land Team collection identifier is also contained in each production tile individual files name as in the following example tile name shown for MOD13A2 where the collection identifier 003 is shown highlighted MOD13A2 A2002353 h08v05 003 2003008095623 hdf A simplified list of CoreMetadata 0 attributes for the MOD17A2 ESDT are shown below The names are typically self descriptive Note that Product Specific Att
79. ype ESDT and are provided to users with several types of quality metadata Level 3 and 4 data products are gridded using the Integerized Sinusoidal ISINUS or Sinusoidal SIN rectangular map projection and supplied to users with several types of metadata Two broad types of metadata are defined collection level and granule level with the granule level metadata specific to a given granule or tile All ECS metadata entries are formally introduced to the system and are registered within the ESDT definition For complete details on ECS metadata issues interested readers are encouraged to visit URL http observer gsfc nasa gov A fairly complete ECS related glossary relevant to metadata may be found at Attp ecsinfo gsfc nasa gov sec2 glossary html Version 2 0 12 2 2003 Page 21 of 57 MODI7 User s Guide MODIS Land Team At the tile or granule level the standard ECS metadata are organized into three different sections each appearing in a given HDF EOS file as a global character attribute Table 4 1 A granule is the smallest unit of data that is produced inventoried and archived within the EOSDIS Within each of these large metadata blocks data are organized using the Object Data Language ODL conventions established by NASA with the data itself formatted as a series of name value pairs or Parameter Value Language PVL An example of PVL syntax is the GROUP END GROUP and OBJECT END OBJECT form commonly found in both the

Download Pdf Manuals

Related Search

Related Contents

38TRA Air Conditioning Unit Installation and Start  AOS AKVA Sensor Buoy - Log On  PERCEPTION Getting Started Guide  V2426 Series  Additif Remplace Raccordement des enceintes HiFi  ダイビングライト SOLA Nightsea 用 取扱説明書 第3版 安全上  Contact presse Corine BUSSON-BENHAMMOU  Trango Broadband 5010 User's Manual  

Copyright © All rights reserved. Failed to retrieve file