Research On Evaluation Of MODIS Gross Primary Production Using Eddy Covariance Flux Data

Gross primary productivity(GPP),which is defined as the overall photosynthetic fixation of carbon by green vegetations.GPP is the initial energy of carbon cycle in terrestrial ecosystem,and it measures the carbon balance of the biosphere and the atmosphere.The estimation precision of GPP not only directly determines the estimation accurancy of other subsequent vegetation productivity factors,but also relates to the evaluation of the affects of land ecological system on human sustainable development.In this study,MOD 17A2 and MOD17A2H were evaluated using eddy covariance flux measurements at five various biome types(8 sites)across China,and the factors whichs affected the accuracy of MOD 17A2H estimation were analyzed.From these researches,some major conclusions are drawn as follows:(1)Firstly,MOD17A2 and MOD17A2H were evaluated using eddy covariance flux measurements.In general,the estimation precision of MOD17A2 GPP 8d was slightly higher than the MOD17A2H.From the perspective of a single site,for DX and HB,MOD17A2 and MOD17A2H most effectively estimated GPP,for XSBN and DHS,the estimate precision are minimum;MOD17A2 and MOD17A2H most overestimated estimated GPP for Winter,and they most underestimated GPP for Summer,there is no obvious difference between them for Autumn,for Spring,the estimation accuracy of MOD17A2 was higher than MOD17A2H;For all site,the annual average value of MOD17A2 was more effecticely estimated than MOD17A2H.In terms of vegetation types,the estimation precision of MOD17A2 were higher than MOD 17A2H.(2)Secondly,the sensitivity of MOD17A2H to meteorological data and fractional photosynthetically active radiation(FPAR)products were examined by introducting site meteorological measurements and improvement Global Land Surface Satellite(GLASS)FPAR products.The 8d average value of MOD_Tower GPP(R2=0.54)?GMAO_GLASS GPP(R2=0.60)and Tower_GLASS GPP(R2=0.65)more effecticely estimated than MOD_GMAO GPP(R2=0.52);All of these four GPP accurately estimated the measurements for Spring,Autumn and Winter,however all four algorithma underestimated the Flux_Tower GPP for Summer;in terms of the estimateing effectiveness,the annual average value of MOD_GMAO GPP(R2=0.72)had the greatest performance.Generally speaking,replacing the meteorological and FPAR datas only slightly improved the correlation with tower GPP over eight days,and for seasonal and annual average value,all of these four GPP had no improvements.(3)Lastly,this paper analyzed the input factors of light energy utilization model,such as meteorological data,FPAR,light energy utilization as well as land classification accuracy.It turned out that the meteorological data used in MOD17A2H overestimated the site measured vapper pressure deficits(VPD)and photosyntheticall-y active radiation(PAR);Replacing the existing FPAR data with GLASS FPAR improved the estimation accuracy of MOD17A2H;Though land cover presents the fewest errors,smax prescribed in MOD17A2H were much lower than inferred ?max calculated from flux tower.