Comparison of cropping system models for simulation of soybean evapotranspiration with eddy covariance measurements in a humid subtropical environment

Authors: Chatterjee, Amitava; Anapalli, Saseendran

Submitted to: Water
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/23/2023
Publication Date: 8/28/2023
Citation: Chatterjee, A., Anapalli, S.S. 2023. Comparison of cropping system models for simulation of soybean evapotranspiration with eddy covariance measurements in a humid subtropical environment. Water. 15(17). Article 3078. https://doi.org/10.3390/w15173078.
DOI: https://doi.org/10.3390/w15173078

Interpretive Summary: Evaporation from soil and plants is the main loss of water in crop production, and its quantity is important for managing irrigation water. Direct measurement of evaporation requires expensive and complex instrumentation. Crop simulation models use local weather data to better predict optimal irrigation water management, but these models require validation before adoption. Performance of three widely used models were compared against field observations of soybean yield and standard measurement of evaporation. Our results showed that all three models showed a wide range of variation over three growing seasons. We demonstrated that these models need to incorporate modern soybean hybrid characteristics to improve the prediction of grain yield and evaporation. Our results provide constructive evaluations of three models and will be of interest to modelers for prediction of crop water demand and management of irrigation water.

Technical Abstract: Prediction of crop yield and evapotranspiration (ET) are critical to achieve the goal of sustainable use of water and crop productivity. This study compared soybean growth and evapotranspiration (ET) using the eddy covariance (EC) against simulations from three models for the soybean production at the Mississippi Delta during 2017-19 growing seasons. For three seasons, soybean grain yield was 4,979, 5,157, and 5,665 kg/ha; three models, APSIM, DSSAT, and RZWQM, simulated yields showed a variation of 4.35%, -10.8%, and 15.4%, 1.51%, -24.0%, and 1.56%, 6.00%, -6.22%, and 9.98%, respectively for three consecutive seasons. Daily ET deviated from EC estimates with root mean square errors (RMSEs) of 2.60 mm, 3.10 mm, 3.20 mm for APSIM, 1.96 mm, 1.82 mm, and 2.14 mm for DSSAT, 1.39 mm, 2.13 mm, and 2.56 mm for RZWQM, for three seasons, respectively. The results demonstrated that model performance varied with particular simulation model and growing seasons. Errors of these magnitude can be an obstacle in using simulation outputs of ET for computing irrigation demand. Improvements in simulating crop growth parameters incorporating characteristics of modern soybean hybrids are necessary to improve ET simulation process.