Study on Soybean Yield Estimation Using the CoupledCASA and WOFOST Model

China is an agricultural country. Yield estimating on field scales rapidly and accurately is not only instructional to farmers’ field management, but also important for the response evaluation of farmland ecosystems to climate change, making scientific and rational food policies, external food trade and so on. The current primary estimation models include empirical statistical model, light use efficiency model, and crop growth model. Each type of model is relatively complete in its individual research filed, but all of them have certain amount of limitations. Remote sensing technology was used to estimate crop yield on a field scale within small regional areas. A farm of Heilongjiang Province was selected as the study area, and the soybean was as the research object. Based on the coupled CASA-WOFOST model and time-series HJ-1A/B remotely sensed data which covering the entire growing season of soybean to generate high temporal resolution Normalized Difference Vegetation Index (NDVI), we achieved daily continuous monitoring of crop and simulating crop yield by CASA model and CASA-WOFOST model respectively. The results indicated that the coupled model had a faster running speed of the light use efficiency model, it could also give full play to mechanism advantages of crop growth model and overcome the limitations of the CASA model applied to field scales. The R² of soybean yields increased from 0.668 53 to 0.844 72 and RMSE decreased from 51.41 to 29.52 kg/ha. It is indicated that the coupled mode of light use efficiency model and crop growth model could simultaneously consider the light utilization and the whole physiological and ecological process of crop growth. So that the coupled model could improve the precision, reliability, and stability of crop yield estimation, and provide theoretical support for the estimation of crop yields in regional field scales and better serve the development of precision agriculture.