Integration of prognostic sowing and harvesting schemes to enhance crop dynamic growth simulation in Noah-MP-Crop model

Detailed sowing and harvesting (S&H) information is crucial for climate-coupled crop models to accurately simulate dynamic crop growth. The timing of intra-annual crop growth not only reflects adaptation to climate change but also significantly influences terrestrial biophysical and biochemical processes, as well as the local climate. However, accurately providing sowing and harvesting dates in crop models is challenging due to the limited availability of S&H observations worldwide. In this study, we integrated a prognostic S&H scheme into the Noah-MP-Crop model to eliminate the need for prescribed S&H dates and optimised key crop-related parameters to better reproduce dynamic maize and soybean growth in the U.S. Corn Belt. Results indicated that the bias in estimating site-level S&H dates was within one week. The prognostic S&H schemes, along with optimised crop-related parameters, effectively captured maize and soybean growth at the site scale, as evidenced by leaf area index (LAI) and gross primary production (GPP) simulations. The determination coefficient (R2) for GPP ranged from 0.88 to 0.91 for maize and from 0.70 to 0.82 for soybean at two flux stations. Moreover, the prognostic schemes exhibited better regional LAI and GPP simulations at the beginning and end of the growing season compared to those using state-level median S&H dates, with significant improvements in correlation coefficients ranging from 0 to 0.6, particularly in maize-dominated regions. However, the accuracy in reproducing latent heat flux and sensible heat flux was less satisfactory and showed little association with crop growth status. This work provides an alternative approach to obtaining crop sowing and harvesting information in the Noah-MP-Crop model and facilitates studies on interactions between dynamic crop growth and the climate system, particularly when coupled with the widely used Weather Research and Forecasting (WRF) model.