Parameter Optimization for Improvement of MODIS Gross Primary Production over East Asia

Abstract

The satellite-based GPP product is crucial to build and validate various environmental models as an input of carbon absorption. However, the accuracy of MOD17A2H especially in East Asia is still uncertain because of the coarse resolution of land cover and reanalysis meteorological dataset, and the consistent value to calculate light use efficiency (LUE) in MOD17 GPP algorithm. For that reason, the objective of this study is to improve the MOD17A2H GPP product by using a high resolution land cover map and the LUE parameter optimization. Target area is East Asia that includes Eastern China, Korean Peninsula, and Japan in extent of 30-48°N and 106-149°E. The LUEmax in Biome Property Look-Up-Table(BPLUT) was modified by the land cover percentage map of FROM-GLC at 30 m spatial resolution. The TMINmax and VPDmin which make the daily LUE saturate to 1 in the scalars are optimized using RMSE that the GPP compared with AsiaFlux database. Top and bottom 5-40% of TMIN and VPD, respectively, from Japanese 55-year Reanalysis (JRA55) on the flux tower site are tested refining TMINmax and VPDmin thresholds. As the result, the RMSE between optimized FROM-MCD GPP (FROM-MCD-opt) and flux tower GPP decreased than that of between MOD17A2H and flux tower from 21.83 (gC/m2/8days) to 16.11 (gC/m2/8days). The bias also diminished from –12.11 (gC/m2/8days) to –2.78. The results of area sum of annual GPP through climate classes and others were 973.75, 1377.79, and 1385.87 (gC/m2/yr) for MOD17A2H, FROM-MCD-opt, and BESS, respectively. Consequently, showing more positive agreement with the GPP of BESS and flux tower than the MOD17A2H, the improvement of MOD17A2H was successfully achieved in this study.