Applying Different Vegetation Indices for Gross Primary Productivity Estimation in Soybean and Maize Based on a Modified Light-Use Efficiency Model

Crop gross primary productivity (GPP) represents a fundamental variable for the investigation of carbon exchange dynamics among elements within agroecosystems. The usage of light use efficiency (LUE) models based on satellite data to estimate regional field GPP is regarded as an efficacious approach. The accuracy of conventional LUE models in estimating GPP is influenced by soil temperature dynamics and subsurface moisture conditions, both of which are challenging to fully characterize. Consequently, there is a necessity to enhance the LUE for soil temperature and soil moisture. Vegetation indices have been shown to effectively capture canopy dynamics and improve the accuracy of GPP estimation, with different indices yielding varying outcomes in LUE-based models. Therefore, in this study, we proposed a modified light use efficiency (M-LUE) model and quantified the influence of soil temperature on GPP estimation. The model was evaluated in three cropland sites, each characterized by distinct crop rotation systems and irrigation strategies. In addition, we tested the performances of nine vegetation indices in estimating the GPP. The results showed that M-LUE was accurate in GPP estimation with R2 of 0.92±0.04 in maize and 0.81±0.05 in soybean. Compared to EC-LUE, M-LUE improved prediction accuracy in GPP estimation, especially for soybeans. For irrigated soybeans, the R2 of M-LUE with EVI improved by 24.4%, while for rainfed soybeans, the R2 of M-LUE with SR improved by 10.5%. In addition, the performance of the model is different between the irrigated and rainfed crops. It performed better in irrigated maize and rainfed soybean. The findings of this study demonstrate the considerable potential of the M-LUE model in the estimation of GPP for soybeans and maize.