Site-level Evaluation of Remote Sensing Based Terrestrial Gross Primary Production from New MODIS fAPARchl Product

Thursday, 18 December 2014: 10:35 AM
Tian Yao1, Qingyuan Zhang1, Alexei Lyapustin2 and Yujie Wang3, (1)Universities Space Research Association Greenbelt, Greenbelt, MD, United States, (2)NASA Goddard Space Flight Cen., Greenbelt, MD, United States, (3)University of Maryland Baltimore County, Baltimore, MD, United States
Gross primary production (GPP) is a measure of photosynthesis, which can be estimated from the fraction of photosynthetically radiation (PAR) absorbed by vegetation (FPAR) and leaf area index (LAI). The MOD15A2 LAI/FPAR product is one of the standard products derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. However, only the PAR absorbed by chlorophyll of the canopy, not the PAR absorbed by the foliage or by the entire canopy, is used for photosynthesis. Therefore, we use a canopy-leaf radiative transfer model PROSAIL2 with MAIAC retrieved surface reflectance to derive the fraction of PAR absorbed by the foliage (fAPARfoliage), the fraction of PAR absorbed by chlorophyll of the canopy (fAPARchl), LAI, and the photosynthetic section of LAI (LAIchl=LAI*fAPARchl/fAPARfoliage). We compare six years(2001-2006) of Community Land Model version 4.5 (CLM4.5) simulations of GPP in selected AmeriFlux tower sites across croplands and forests. Three LAI data sets of each site have been implemented into CLM to simulate GPP: (1) default LAI with CLM, (2) MOD15A2 LAI, and (3) MODIS LAIchl from fAPARchl. Differences among measured GPP and simulated GPP derived from CLM default LAI, MOD15A2 LAI and LAIchl have been analyzed. The results show that during the growing season, spring estimates of MODIS LAIchl/fAPARchl GPP are much closer to tower data than CLM default GPP and MOD15A2 GPP, with R-squared values of 0.71, -0.06 and -0.09, respectively. Thus, while continued efforts are needed in more sites and vegetation types, the new MODIS LAIchl/fAPARchl product will likely be able to increase accuracy of GPP simulations and improve vegetation phenology dynamics response to climate variability.