Optimizing the Temporal Scale in the Assimilation of Remote Sensing and WOFOST Model for Dynamically Monitoring Heavy Metal Stress in Rice

Obtaining precise information regarding the levels of heavy metal stress in crops is vital for food security. The assimilation of remote sensing into the World Food Study (WOFOST) model provides a method for achieving the spatial-temporal evaluation of crop growth status, while the optimization of the temporal scale in assimilation framework has rarely been considered. In this study, the temporal scale was optimized based on a wavelet transform of the leaf area index (LAI) curves. The accurate simulation of LAI laid the foundation for high precision. As the dry weight of rice roots (WRT) was demonstrated to be the most stress-sensitive indicator, the measured WRT values were assimilated into the improved WOFOST model to realize the dynamic simulation of LAI. Finally, four optimal time points were determined based on the extreme areas in the d4 wavelet coefficient, providing a reference for the selection of remote sensing images. The verification in the two sample plots indicated that the assimilation with optimized temporal scale could significantly improve the efficiency on the basis of guaranteeing the accuracy, shortening the run time of model operation by more than 30%. Based on the optimized temporal scale, the RS-WOFOST assimilation framework was driven for each pixel in the study areas, achieving the spatial-temporal evaluation of heavy metal stress in rice. This study suggests that the wavelet transform to LAI is applicable for optimizing the temporal scale in assimilation, providing a reference for the improvement of assimilation results under the premise of balancing accuracy and efficiency.