A remote sensing-based scheme to improve regional crop model calibration at sub-model component level

Parameter calibration is an importantly preliminary step before using a crop model to simulate crop growth and final yield. Compared with the traditionally accepted calibration method parameterizing the whole model simultaneously (called as "Global Scheme"), the Sub-Model Component (SMC) Scheme emphasizes on parameterizing different functional modules in a crop model sequentially. However, the SMC Scheme receives less attention, especially at regional scales. Therefore, this study led a performance evaluation of the two calibration schemes through using them to incorporate remote sensing data into a crop model (MCWLA-Rice) independently in Northeast China. We found the SMC Scheme reduced root mean square error (RMSE) on average by 4 days for heading date and 2 days for harvest date. Using the Pearson correlation coefficient (R) to assess the similarity between time series of modelled LAI and remotely-sensed LAI, the SMC Scheme decreased LAI estimation error by 0.04. Finally, the SMC Scheme greatly decreased relative RMSE (RRMSE) for yield by 11%. In addition, temperature and topography could affect the performance of SMC Scheme. Our findings demonstrated that the SMC Scheme calibrated the crop model more effectively and reliably, suggesting its potentially wide application in other regions and crops.