Abstract: To find a solution to the mismatch between remote sensing observation at regional scale and crop modeling at local scale and to improve the accuracy of crop growth modeling, we simulated the growth of winter wheat in Hebi, Henan during 2013-2017 based on the WheatSM crop model integrated with retrieval of leaf area index (LAI) derived from remote sensing data. The remote sensing data included MODIS and Landsat 8 Operational Land Image (OLI) data. Research methods included the reconstruction of LAI process line during the winter wheat growth period, extraction of sub-pixel information, and Ensemble Kalman Filter. Results showed that LAI values after the reconstruction of MODIS LAI process line and extraction of sub-pixel information from pure pixels with winter wheat over 80% were close to the measured LAI with RMSE of 0.69 during two key growing seasons. Compared with actual regional yield during 2013 to 2017, RMSE was 6.73×10⁸ kg for simulation without any assimilation process, 8.24×10⁸ kg for simulation assimilating original LAI, and lowered to 3.48×10⁸ kg for simulation with assimilating MODIS LAI processed with process line reconstruction and sub-pixel information extraction. Our results suggested that the accuracy of crop model at regional scale can be improved when LAI process line is reconstructed and sub-pixel information is extracted to match the spatial scale of crop model.

Key words: AM fungal diversity, community structure, phospholipid fatty acids (PLFAs), high-throughput sequencing, straw returning