Application of Evapotranspiration Models Coupled With Single and Dual‐Source Hypotheses in Maize ( Zea mays L.) Fields in Northwest and Northeast China

ABSTRACT Crop evapotranspiration ( ET c ) is a critical component of crop growth, yield formation, and the water cycle, and its accurate estimation is essential for agricultural management decisions. However, uncertainties remain regarding the applicability of existing ET c models under different climatic and planting conditions. In this study, the Penman‐Monteith (PM) model coupled with the Katerji‐Perrier (KP) and the Farias (FA) canopy resistance methods, the Priestley‐Taylor (PT), the Shuttleworth‐Wallace (SW), and the revised SW (R‐SW) models were applied for ET c calculation for rainfed maize in Harbin City, Heilongjiang Province (NE‐H) from 2023 to 2024, and film‐mulched maize in Ordos City, Inner Mongolia (NW‐I) from 2020 to 2021. All model outputs were calibrated and validated using measured ET c with the Bowen Ratio Energy Balance method. The PM‐KP, R‐SW, and PT models in NW‐I and the PM‐KP and PT models in NE‐H performed better overall, with the coefficient of determination ( R 2 ) close to 0.8. The PT model achieved the highest simulation accuracy in NW‐I, with an R 2 of 0.98, a mean absolute error (MAE) of 0.01 mm/h, and a root mean square error (RMSE) of 0.02 mm/h. In NE‐H, the PM‐KP model outperformed the other models, with an R 2 of 0.87, an MAE of 0.02 mm/h, and an RMSE of 0.08 mm/h. Additionally, model performance exhibited substantial variability across different regions and temporal scales. In NW‐I, model errors responded strongly to vapour pressure deficit ( VPD ) and net radiation ( R n ), with significant interactive effects with leaf area index ( LAI ). Whereas in NE‐H, errors of most models showed no consistent variation trends along VPD and R n gradients, except for the PM‐FA model, with only specific models exhibiting systematic responses to extreme conditions. This study provides targeted technical support for water‐efficient irrigation management under specific climates. Future research should enhance the integration of meteorological and crop physiological factors into ET c model parameterisation and develop region‐specific correction terms for canopy resistance and energy partitioning processes to improve model adaptability for broader agroclimatic contexts.