Improving the simulation of maize growth using WRF-Crop model based on data assimilation and local maize characteristics

Abstract

Global climate change presents a significant challenge to the sustainable development goal of eradicating hunger. Accurate assessment or projection of crop yields is crucial for ensuring food security at both global and regional levels in a changing environment. However, traditional crop models may introduce significant uncertainties due to lack of the intensified feedbacks between crop vegetation and climate systems. In this study, we coupled dynamic crop model (Noah-MP-Crop) with the Weather Research and Forecasting (WRF) model (WRF-Crop) based on data assimilation and local maize characteristics to simulate dynamic maize growth and subsequent yield at Jilin Province, China. We utilized in-site phenological observation data to refine the model's cumulative growing degree days (GDDs), and employed leaf mass assimilation to enhance the accuracy of crop phenology cycles. Our findings suggest that refining the model's GDDs thresholds and incorporating data assimilation leads to better alignment with MODIS-observed Leaf area index (LAI), evapotranspiration (ET), and gross primary productivity (GPP), with a reduction in the mean absolute error of 41.2 %, 14.1 %, and 27.5 %, respectively. The in-site eddy covariance flux observation data on soil moisture (layer 1 R = 0.9) and GPP (R = 0.82) also support our results. With the improvement of the maize growth cycles, the adjusted WRF-Crop model exhibited significantly improved accuracy in simulating maize yield, averaging 10,140 kg/ha in Jilin Province. This represents an approximate 9.26 % increase in accuracy compared to the default model configuration. Therefore, the dynamic crop-coupled WRF-Crop model showcases substantial potential for regional crop yield estimation and predictions, featuring dynamic downscaling capabilities through the incorporation of interactions between crops and the atmosphere.