7(2)-07.fm

Transcription

1 w» wz, 7«2y(2005) Korean Journal of Agricultural and Forest Meteorology, Vol. 7, No. 2, (2005), pp. 171~183 MODIS :» w ³ 1 Á½ 2 Á½ 1 1 w y w, 2 w y w ( ; ) Errors of MODIS product of Gross Primary Productivity byg using Data Assimilation Office Meteorological Data Sinkyu Kang 1, Youngil Kim 2 and Youngjin Kim 1 1 Department of Environmental Sciences, Kangwon National University 2 School of Environment, Seoul National University (Received May 5, 2005; Accepted June 8, 2005) ABSTRACT In order to monitor the global terrestrial carbon cycle, NASA (National Aeronautics and Space Administration) provides 8-day GPP images by use of satellite remote-sensing reflectance data from MODIS (Moderate Resolution Imaging Spectroradiometer) at 1-km nadir spatial resolution since December, MODIS GPP algorithm adopts DAO (Data Assimilation Office) meteorological data to calculate daily GPP. By evaluating reliability of DAO data with respect to surface weather station data, we examined the effect of errors from DAO data on MODIS GPP estimation in the Korean Peninsula from 2001 to Our analyses showed that DAO data underestimated daily average temperature, daily minimum temperature, and vapor pressure deficit (VPD), but overestimated short wave radiation during the study period. Each meteorological variable resulted in different spatial patterns of error distribution across the Korean Peninsula. In MODIS GPP estimation, DAO data resulted in overestimation of GPP by 25% for all biome types but up to 40% for forest biomes, the major biome type in the Korean Peninsula. MODIS GPP was more sensitive to errors in solar radiation and VPD than for temperatures. Our results indicate that more reliable gridded meteorological data than DAO data are necessary for satisfactory estimation of MODIS GPP in the Korean Peninsula. Key words : Gross primary product, Satellite remote sensing, Daily meteorological data I.»z y š» yk k w š, k ü w k y» ³ wš, k y y wš yw š (Sellers et al., 1994; Prince and Goward, 1995; Running et al., 2000). k k y l w x x», œ y w œ k w», š k x w w» š (Running et al., 1999)., œ w k» ù 20» wš, w w x ³ k y w d l(earth Corresponding Author : Sinkyu Kang (kangsk@kangwon.ac.kr)

2 172 Korean Journal of Agricultural and Forest Meteorology, Vol. 7, No. 2 Observing System, EOS) w» (Running et al., 1994). wœ d l w Terra Aqua œ k Moderate Resolution Image Spectrometer (MODIS), 36 Ÿ 1~ m, 500 m, 250 m œ w w (Running et al., 2000). MODIS Ÿ»w(geometric), (radiometric) e z, w š w», w, k w y w š. p, k y w, MODIS 8 (Leaf Area Index, LAI), (Gross Primary Productivity, GPP), (Net Primary Productivity, NPP) œw, x w š. MODIS GPP š y z ƒw Penman» š (Heinsch et al., 2003). GPP w», (LAI), Ÿw z (Fraction of absorbed Photosynthetically Active Radiation, FPAR) w m v x x(biome type) w w ƒ.» wœ w Data Assimilation Office(DAO)» w, LAI, FPAR, m v MODIS w š (Kang et al., 2005). x MODIS sƒ w» w w w š (Cohen and Justice, 1999; Morisette et al., 2002; Myneni et al., 2002; Kang et al., 2003; Turner et al., 2003). MODIS š ƒ d w wš», w ³ wš, w wr ª š w ý v ƒ (Kang et al., 2005; Zhao et al., 2005). j l w š l w ù ƒw. š sƒ w wš, w w v ƒ. p MODIS DAO, e w ƒ ¾, w w š, š w xp ƒ œ» w œ w» w š q.» y MODIS GPP w e, MODIS GPP w k w» w, DAO» y w MODIS GPP w, wz w x» w DAO» ew v ƒ sƒw w. 72 ³» d d w, w DAO» sƒwš, d» w MODIS GPP DAO» w MODIS GPP w, wœ œw MODIS GPP sƒwš w.» w w», MODIS GPP w w sƒ ü» w, w wz v k w x d, w k x w f» z ww ù z. II MODIS š wœ œw MODIS 4 : 8 (GPP, g C m 2 per 8- days), 8 Ÿw (Net Photosynthetic Productivity, PSNnet, g C m 2 per 8-days), (NPP, g C m 2 y 1 ), Quality Control Flag. ƒƒ š w w MODIS GPP š w kù w Numerical Terradynamic Simulation Group(NTSG) œw MOD17 GPP/NPP User's Guidedp ù (Heinsch et al., 2003). MODIS w, ƒ» w w (GPP)

3 Errors of MODIS product of Gross Primary Productivity by using ~ 173 w. MODIS GPP z» š [ 1]. GPP = [ε]ü[fparápar] (1) ε = [ε maxáf(t min )Áf(VPD)] (2) PAR = [0.45 Ü R s ]» ε z (radiation use efficiency), max z, T ε min», VPD s, R s q, FPAR w PAR 0 1., w Ÿw z ƒ, z, ƒw. wr MODIS GPP š z x max, ε T min VPD w š ƒ w [ 2].» PAR R s 45 % w w š ƒ w [ 3]. wr, T min VPD»œ sƒ w, ww ƒ x w,» z max) 0¾ yw w. [ 1] (ε FPAR MOD15 FPAR/LAI œ, [ 2] T min, VPD, [ 3] R s ƒƒ w ƒ DAO» w. [ 2] max T ε min, VPD»œ s w ww e BPLUT(Biome Property Look-Up Table) x w (Heinsch et al., 2003). [ 1] š w xk w x ƒ š. ƒ j x wš, ƒ» x w., ƒ w ƒ, ³ w. MODIS GPP š ³ GPP (Running et al., 1994). x ƒ w w y, w p w ƒ š. w GPP w BPLUT ³ sƒwš, w w GPP w v ƒ, š v w. MODIS GPP š yw, m z w e w y š w. m ƒw w, m w w x š v w. 2.2.» DAO» y sƒwš MODIS GPP e w sƒw» w, w sw 72 ³» d w (Fig. 1). Fig ³» d e DAO w.» yr œw ƒ ³» d 2001 l 2003», s³» (T avg ), š» (T max ),» (T min ), (PRCP) w. Fig. 1. Spatial distribution of National Weather Stations (points) and boundary of DAO pixels (solid lines). Spatial resolution of DAO pixel is 1.25 degree in longitude and 1 degree in latitude. Background map is MODIS Land cover.

4 174 Korean Journal of Agricultural and Forest Meteorology, Vol. 7, No. 2 w MTCLIM x(running et al., 1987) w, VPD R s w, MODIS GPP š w ƒ ³» d w GPP w. MTCLIM ³» d T max, T min, š PRCP w Rs w x, w x d (Running et al., 1987; Thornton et al., 1997; Kang et al., 2002). MODIS GPP t y z, ƒ ³» d w GPP GIS y ƒ w x wœ œw MODIS GPP w», DAO» w z, d w GPP w w» w GPP w z, w. DAO» wœ 6 œwù, kù w NTSG yw DAO w. Zhao et al.(2005). -180, -90 l ƒƒ DAO l, ƒ ³» d w w w 2001 l 2003 ¾» w, w ³» d w DAO» w, y w MODIS š MODIS GPP š w w» wì, FPAR, LAI š m v w. m v MODIS m v (MOD12), FPAR, LAI MODIS FPAR/LAI (MOD15) w w. MODIS (USGS) EROS» w. MOD12, MOD15 sww, MODIS Land Product SIN (sinusoidal projection) t, ƒ ƒƒ 1200 v k Hierarchical Data Format(HDF) œ. MODIS Reprojection Tools(MRT) vp w, SIN UTM(Zone 52) t, HDF TIFF s w z, Arc/Info (v ) vp Arc/Info y w. yw Land cover, FPAR, LAI ƒ ³» d e w ƒ 5Ü5 ü ü, ascii q w. w 46 ƒ 8 FPAR, LAI w ww, 2001 l 2003 ¾ 138 w, 72 ³» d w 5Ü5 ascii q. ƒƒ v 1 km w ƒ», ƒ ³» d w 5 km j». w ƒ ³» d w 5Ü5 j» m v w. ù m v 2001 m v wù w ww š, m v y š w. wr 8 FPAR yw wœ MODIS GPP, 8 w FPAR w w. 72 ³» d 25 v w m v 3 FPAR ƒ. ƒ ³» d 25 v w ƒƒ MODIS GPP š w GPP w. ƒ v ƒ GPP ƒ,» w w. r wz d» w w GPP NWSGPP, DAO» w GPP DAOGPP š w m w s³r (mean bias, MB) s³ (Mean Absolute Error, MAE) w. w d w DAO (Relative Error, RE) w. ƒ ³» d d DAO» w, DAO w. GPP w ƒ ³» d NWSGPP DAOGPP w š, DAO ww w š ƒƒ m v x w. m q w» w t ( d ) ww (p)

5 Errors of MODIS product of Gross Primary Productivity by using ~ 175 w. ƒ» ƒ GPP s w ƒ w» w, xz ww (coefficient of determination, R 2 ) w. III. š 3.1. DAO», d» DAO» w, DAO»» VPD d w sƒ, R s sƒ (Fig 2). t (p) T avg VPD, R s p< m w ù, T min p=0.69 w. w 72œ ³» d w, T avg smb 0.6 o C, MAE 2.0 o C, T min ƒƒ 0.1 C o 2.8 o C, VPD ƒƒ Pa Pa, š R s ƒƒ -1.0 MJ m 2 d MJ m 2 d 1 ùkû (Table 1). ƒ ³» d VPD R s w DAO» RE ƒƒ 31~65%(s³ 44%), 17~50%(s³ 22%), VPD R s ƒƒ, û (65%) (50%) ù kû. Fig. 2 yw ùk ù w, e w DAO» s (Fig. 3). MAE s,» ùk û. T avg T min w ûw j ü. VPD û ü ƒ j ùkû, R s Fig. 2. Scatter plots of annual means of meteorological variables between weather station (NWS) and DAO data from 2001 to Meteorological variables are (a) daily mean temperature (T avg, o C), (b) daily minimum temperature (T min, o C), (c) daily mean vapor pressure deficit (VPD, Pa), and (d) daily shortwave radiation (R s, MJ m 2 d 1 ).

6 176 Korean Journal of Agricultural and Forest Meteorology, Vol. 7, No. 2 Table 1. Comparison of DAO meteorological data with weather station data for each DAO pixel: mean bias (MB) and mean absolute error (MAE) of daily mean temperature (T avg, o C), daily minimum temperature (T min, o C), vapor pressure deficit (VPD, Pa), and shortwave radiation (R s, MJ m 2 d 1 ) from 2001 to Here, MB is subtraction of weather station data by DAO data and DAO ID is combination of numbers of row and column in FigU 1. T avg T min VPD R s DAO ID MB MAE MB MAE MB MAE MB MAE ³ Table 2. Monthly mean bias (MB) and mean absolute error (MAE) of DAO meteorological data with respect to weather station data for each meteorological variable: daily mean temperature (T avg, o C), daily minimum temperature (T min, o C), vapor pressure deficit (VPD, Pa), and shortwave radiation (R s, MJ m 2 d 1 ) from 2001 to Month MB MAE T avg T min VPD R s T avg T min VPD R s Mean w w j 6.1 MJ m 2 d 1 ¾ sƒ. Fig. 2(d) d sw w d. Table 1 w DAO DAO» s³, û, š T avg T min sƒ,

7 Errors of MODIS product of Gross Primary Productivity by using ~ 177 VPD sƒ š, R s w sƒ. 72 ³» d w DAO» s ³ Table 2. DAO T min sƒ š, sƒ w. VPD sƒ š, ƒwš. R s 7 w sƒ š, ƒ š w ùkû. wr DAO, T avg T min š û w, VPD R s š ƒ Table 3. Mean bias (MB, gc m 2 d 1 ) and mean absolute error (MAE, gc m 2 d 1 ) of GPPs, which were estimated with weather station and DAO data, respectively. ( ) and [ ] show MB and MAE during for each weather station and DAO pixel, respectively. N/A indicates that MODIS GPP was not estimated for all of 5-by-5 pixels around the weather station because of no vegetation land cover types. Lon. Lat. 37.5~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ [-1.9, 1.9] (-1.9,1.9) [-0.2, 0.4] y (-0.2,0.4) [-0.2, 0.3] (-0.2,0.3) [-0.2, 0.3] (0.0,0.1) (-0.1,0.2) (-0.2,0.3) [-0.2, 0.4] s (-0.1,0.4) wû (-0.3,0.4) [-2.0, 2.0] (-2.0,2.0) [-1.1, 1.1] š (-0.9,0.9) (-1.3,1.3) [-0.5, 0.6] ( ) y (-0.9,0.9) [-0.3, 0.4] (-0.1,0.2) (-0.5,0.5) (-0.1,0.3) s (-0.2,0.3) (-0.2,0.3) (-0.4,0.4) [-0.4, 0.4] t (N/A) (-0.3,0.3) (-0.3,0.3) (-0.5,0.6) (-0.5,0.5) (-0.5,0.6) [-0.5, 0.7] Ÿ (-0.6,0.6) (-0.6,0.7) û (-0.4,0.8) (-1.0,1.1) (-0.4,0.8) š (N/A) (-0.1,0.5) ûw (-0.5,0.6) [-1.7, 1.7] (-1.7,1.7) [-0.1, 0.5] (0.1,0.4) (0.1,0.4) (N/A) w (-0.4,0.7) (-0.2,0.5) [-0.6, 0.6] (-0.7,0.8) (-0.7,0.7) k (-0.5,0.6) y (-0.9,1.0) (-0.4,0.5) (-0.4,0.5) [-0.6, 0.7] (N/A) (-0.7,0.7) (-0.8,0.9) (-0.4,0.4) w (-0.6,0.7) [-0.5, 0.9] (-0.3,0.8) m (-0.2,0.9) (-1.1,1.1) (-0.4,0.9) [-1.5, 1.5] (-1.1,1.2) (-1.8,1.8) [-0.5, 0.5] sw (-0.5,0.5) [-1.6, 1.6] (-1.6,1.6)

8 178 Korean Journal of Agricultural and Forest Meteorology, Vol. 7, No. 2 w. GPP w p» DAO» ƒ GPP j w e. T min sƒ GPP j, VPD sƒ Rs s ƒ GPP ƒ j ƒ. ù w DAO»» yp w, GPP ƒ w ƒ. DAO» w w GPPƒ sƒ w,» ƒ MODIS GPP š Fig. 3. Maps showing distribution of three-year mean absolute errors (MAE) between weather station and DAO data. Meteorological variables include (a) daily mean temperature (T avg, o C), (b) daily minimum temperature (T min, o C), (c) daily mean vapor pressure deficit (VPD, Pa), and (d) daily short wave radiation (R s, MJ m 2 d 1 ). Minimum, maximum, and mean values of MAE were provided at the lower-right corner of each map. Size of each circle scaled from the minimum to the maximum values for each map and background map is MODIS Land cover.

9 Errors of MODIS product of Gross Primary Productivity by using ~ 179 w wš ƒ q w ³» d NWSGPP DAOGPP d» w w GPP(NWSGPP) DAO» w w GPP (DAOGPP) ³» d Table 3. m v x v sww 11 ³» d GPP w. ù 61 ³» d w 2001~2003 NWSGPP s³ 790 gc m 2 y 1, DAOGPP s³ 990 gc m 2 y 1 DAO» GPP sƒ (RE=25%)w. MB gc m 2 d 1, MAE 0.67 gc m 2 d 1. t w m w (p<0.0001). MB -2.0~0.1 gc m 2 d 1, MAE 0.1~2.0 gc m 2 d 1. GPP y w ƒƒ -731~38 gc m 2 y 1, 40~734 gc mm 2 y w 1 w. MAE û, ùkû. DAOGPP NWSGPP 14%( ) 65%( )¾ s ƒ. ù DAOGPPƒ NWSGPP w sƒ, DAO» w MODIS GPP T min R s VPD j w š q w. Table 3 DAO ƒ ³» d NWSGPP w DAOGPP, w w ü, û, š w ƒ j,,,» ƒ ql ùkû. Fig. 4 ƒ ³» d txw, Fig. 3» s w q l. w ƒ» w DAO» w s, MODIS GPP» ƒ w w». ƒ ³» d GPP» (R ) w 2 GPP w e» q w. MB T min (R 2 =0.50, p<0.05) R s (R 2 =0.42, p<0.05)ƒ, Fig. 4. Maps showing distribution of three-year mean absolute error (MAE, gc m 2 y 1 ) between NWSGPP and DAOGPP for each station. Size of each circle scaled from minimum (0.1 gc m 2 y 1 ) to maximum (2.0 gc m 2 y 1 ) values of MAE and background map is MODIS Land cover. š MAE R s (R 2 =0.40, p<0.05)ƒ GPP œ s ƒ w. R s (RE=22%) w VPD (RE=44%) š w»e w. w ü j VPD w j R s s» w. sww w R s GPP s w» w w ü GPP s w VPDƒ» w. w w ù w GPP» z ww, VPD R (p<0.05)» w R 2 ƒ ùkû (Fig. 5) m v x NWSGPP DAOGPP GPP ww 61 ³» d ƒ

10 180 Korean Journal of Agricultural and Forest Meteorology, Vol. 7, No. 2 ig. 5. Scatter plots between meteorological errors and GPP errors: (a) and (d) for daily minimum temperature (T min, οc); (b) and (e) for daily mean vapor pressure deficit (VPD, Pa); (c) and (f) for daily short wave radiation (R s, MJ m 2 d 1 ). (a), (b), and (c) are scatter plots for all of 61 weather stations, while (d), (e), and (f) are scatter only for inland weather stations except coastal and island areas. Solid lines are linear regressions and R 2 is coefficient of determination (p<0.05). ƒ 25 v w m v x w w Table 4 Fig. 6. ƒ m v x w Table 4» w. q v w ( v 42%). Table v Savanna, Savanna ù wì ùkù v x w w. Savanna ³ y ƒ wì ùkù, wz x mw y w. m v x GPP NWSGPP 582(Crop)~1488(Deciduous Needleleaf Forest, DNF)

11 Errors of MODIS product of Gross Primary Productivity by using ~ 181 Table 4. Mean GPPs and error statistics, mean bias (MB) and mean absolute error (MAE), between NWSGPP and DAOGPP by biome types. Value in parenthesis is number of pixels of each biome type for all of 1-km pixels around weather stations. NWS GPP DAO GPP MB MAE ENF (63) EBF (39) DNF (3) DBF (66) MF (924) Shrub (72) Savanna (303) Grass (45) Crop (1119) Total Average (2634) ENF, Evergreen Needleleaf Forests; DNF, Deciduous Broadleaf Forests; DNF, Deciduous Needleleaf Forests; DBF, Deciduous Broadleaf Forests; MF, Mixed Forests; Shrub, Shrublands; Savanna, Savannas; Grass, Grasslands; Crop, Croplands gc m 2 y 1, DAOGPP 628(Grass)~2195(DNF) gc m 2 y 1 (Table 4). m v x w DAOGPP NWSGPP w, GPP x m v x j sƒw (0.68~1.96 gc m 2 d 1 ). w 65% w ù DAO» w GPP w, w ƒ w., Table 4 ù ùkù e (Evergreen Needleleaf Forest, ENF) yz (Mixed Forest, MF) DAOGPP NESGPP ƒƒ 40% 32% sƒ w. IV. DAO» w w, T avg T min VPD w sƒw, R s sƒ Fig. 6. Biome-specific scatter plots between three-year mean NWSGPP and DAOGPP (gc m 2 y 1 ) for pixels of 61 weather stations. Figures show scatter plots for (a) Mixed Forests (MF); (b) Evergreen Needleleaf Forests (ENF), Evergreen Needleleaf Forests (EBF), and Deciduous Broadleaf Forests (DBF); (c) Croplands; and (d) all of vegetation biome types.

12 182 Korean Journal of Agricultural and Forest Meteorology, Vol. 7, No ~2003 s³ VPD w 44%, R s 22%. t T avg VPD, R s p< m w. ƒ j ƒ, T avg T min w ûw, VPD û ü, R s w j. wr ƒ» w w. 11 ³» d w 61 ³» d w 2001~2003 NWSGPP s³ 790 gc m 2 y 1, DAOGPP s³ 990 gc m 2 y 1 DAO» d» w GPP 25% sƒw (p<0.0001). wr DAO» w ù MODIS GPP» VPD j w š q. w ûw š w ƒ j,»á «ƒ. sww ù w GPP s w w», w w ü GPP s w VPDƒ w». wr w GPP š» w mw» w w sƒw, wz û. m v x GPP, x m v x GPP j sƒw ùkû (MAE: 0.68~1.96 gc m 2 d 1 ). 65% w ù DAO» w GPP w, w ( e RE = 40%, yz 32%)ƒ w w. w, DAO» MODIS GPP š w ù GPP w ww, w d ƒ» w v ƒ.» y œ w w, p DAO» j w, ü VPD, w w œw,» w GPP y w w.» w w», MODIS w sƒ ü» w. w wz v k w x d, w k x w f» z ƒ w. y w y» (Eco- Techopia 21 Project) 21» v p y» ( y: 1-8-2) w. w w Ì. x Cohen, W.GB., and C.GO. Justice, 1999: Validating MODIS terrestrial ecology products: linking in situ and satellite measurements. Remote Sensing of Environment 70, 1-4. Heinsch, F.G A., M. Reeves, C.G F. Bowker, P. Votava, S. Kang, C. Milesi, M. Zhao, J. Glassy, W.G M. Jolly, J. S. Kimball, R.GR. Nemani, and S.GW. Running, 2003: User s guide: GPP and NPP (MOD17A2/A3) products, NASA MODIS Land Algorithm. umt.edu/ ntsg/. Kang, S., S. Kim, and D. Lee, 2002: Spatial and Temporal Patterns of Solar Radiation Based on Topography and Air Temperature. Canadian Journal of Forest Research 32, Kang, S., S.GW. Running, J. Lim, M. Zhao, C. Park, and R. Loehman, 2003: A MODIS-based climatological phenology model for detecting onset of growing seasons in temperate mixed forests in Korea. Remote Sensing of Environment 86, Kang, S., S.G W. Running, M. Zhao, J.G S. Kimball, and J. Glassy, 2005: Improving continuity of MODIS terrestrial photosynthesis products using an interpolation scheme for cloudy pixels. International Journal of Remote Sensing 26, Morisette, J.G T., J.G L. Privette, and C.G O. Justice, 2002: A framework for the validation of MODIS Land products. Remote Sensing of Environment 83, Myneni, R.G B., S. Hoffman, Y. Knyazikhin, J.G L. Privette, J. Glassy, Y. Tian, Y. Wang, X. Song, Y. Zhang, G.G R. Smith, A. Lotsch, M. Friedl, J.G T. Morisette, P. Votava, R. R. Nemani, and S.GW. Running, 2002: Global products of vegetation leaf area and fraction absorbed PAR from

13 Errors of MODIS product of Gross Primary Productivity by using ~ 183 year one of MODIS data. Remote Sensing of Environment 83, Prince, S.G D., and S.G N. Goward, 1995: Global primary production: A remote sensing approach. Journal of Biogeography 22, Running, S., R. Nemani, and R. Hungerford, 1987: Extrapolation of synoptic meteorological data in mountainous terrain and its use for simulating forest evapotranspiration and photosynthesis. Canadian Journal of Forest Research 17, Running, S.G W., C.G O. Justice, V. Salomonson, D. Hall, J. Barker, Y.GJ. Kaufmann, A.GH. Strahler, A.GR. Huete, J. P. Muller, J.GP., V. Vanderbilt, Z.GM. Wan, P. Teillet, and D. Carneggie, 1994: Terrestrial remote sensing science and algorithms planned for EOS/MODIS. International Journal of Remote Sensing 15, Running, S.G W., D. Baldocchi, D. Turner, S.G T. Gower, P. Bakwin, and K. Hibbard, 1999: A global terrestrial monitoring network integrating tower fluxes, flask sampling, ecosystem modeling and EOS satellite data. Remote Sensing of Environment 70, Running, S.GW., P.GE. Thornton, R.GR. Nemani, R.GR., and J.GM. Glassy, 2000: Global terrestrial gross and net primary productivity from the earth observing system. Methods in Ecosystem Science, O.GE. Sala, R.GB. Jackson, R.GB., H.G A. Mooney, and R.G W. Howarth (Eds.), Springer- Verlag. New York. Sellers, P.GJ., C.GJ. Tucker, G.GJ. Collatz, S.GO. Los, C.GO. Justice, D.GA. Dazlich, and D. A. Randall, 1994: A global 1 o by 1 o NDVI data set for climate studies: 2. The generation of global fields of terrestrial biophysical parameters from the NDVI. International Journal of Remote Sensing 15, Thornton, P.G E., H. Hasenauer, and M.G A. White, 2000: Simultaneous estimation of daily solar radiation and humidity from observed temperature and precipitation: an application over complex terrain in Austria. Agriculural and Forest Meteorology 104, Turner, D.G P., W. Ritts, W.G B. Cohen, S.G T. Gower, M. Zhao, and S.G W. Running, 2003: Scaling gross primary production (GPP) over boreal and deciduous forest landscapes in support of MODIS GPP product validation. Remote Sensing of Environment 88, Zhao, M.GF.GA. Heinsch, R.GR. Nemani, and S.GW. Running, S. W., 2005: Improvements of the MODIS terrestrial gross and net primary production global dataset. Remote Sensing of Environment 95,