Comparative study of crop growth models in field water and heat dynamics and maize growth under different irrigation conditions with film mulching
Аннотация
Crop growth models are powerful tools for evaluating crop yield and water productivity under specific environmental and agricultural management systems. With the widespread adoption of plastic film mulching, multiple crop growth models have been tailored to simulate film mulched field conditions. However, comprehensive comparative research on the simulation performance and functional characteristics of these mulching-adapted models remain limited. In this study, three film mulching-compatible models-AquaCrop, SWAP, and CropSPAC were compared in their capabilities to simulate field water and heat dynamics and crop growth processes. Model evaluation was conducted based on field maize experiments in Shiyang River Basin of the arid region of Northwest China, which included WF (full irrigation), WM (medium irrigation, 70%WF), WL (low irrigation, 40%WF) treatments under film mulching. The models were further applied to different irrigation schedules to identify optimized irrigation strategies using the entropy weight method. The results showed that the AquaCrop, SWAP, and CropSPAC models all performed well in simulating soil water and heat dynamics and crop growth, except that the AquaCrop model was unable to account for thermal effects. For the CropSPAC model, the normalized root mean square error (NRMSE) values were 9.53% for soil water storage (SW), 7.20% for crop evapotranspiration (ET), 8.84% for canopy coverage (CC), 13.86% for aboveground dry biomass (ADB), 4.31% for yield. For the AquaCrop model, the corresponding values were 11.02%, 9.30%, 7.52%, 15.53%, 7.59%, respectively. For the SWAP model, the NRMSE values were 9.62%, 11.20%, 8.56%, 14.02%, 6.82%, respectively. These results indicated that the CropSPAC model outperformed the AquaCrop and SWAP models in simulating SW, ET, ADB, yield for plastic-mulched maize field in the Shiyang River Basin of the arid region of Northwest China, while the AquaCrop model performed better in simulating CC than the SWAP and CropSPAC models. The CropSPAC model also outperformed the SWAP model in simulating soil temperature (ST), achieving a determination coefficient (R 2 ) of 0.84 and NRMSE of 5.05% compared to SWAP’s R 2 of 0.80 and NRMSE of 5.63% (note that the AquaCrop model could not simulate ST variations). Additionally, the CropSPAC model could characterize net radiation, energy partitioning and soil heat flux, whereas the AquaCrop and SWAP models could not. Under the WF treatment, all three models generally achieved better simulation performance for SW, ADB, yield, ET than under the WM and WL treatments, with only a few exceptions: ADB and ET simulated by the SWAP model under the WM treatment, and yield simulated by the CropSPAC model under the WM treatment. The AquaCrop and CropSPAC models performed better in CC under the WF treatment than under the WM and WL treatments, whereas the SWAP model showed the opposite trend. We propose that for studies focusing on layered soil, the AquaCrop and SWAP models are suitable choices; for studies focusing on soil solute migration or grasses-related research, the SWAP model is recommended; and for studies focusing on thermal aspects or different film mulching control factors, the CropSPAC model is the best choice. Considering yield and water productivity, all three models indicated that the irrigation quota of the WF treatment, coupled with a 14-d irrigation interval, performed well for drip-irrigated maize fields with film mulching in the Shiyang River Basin of the arid region of Northwest China under the conditions of this study (i.e., specific climate and soil types).