Modeling rice evapotranspiration under water-saving irrigation condition: Improved canopy-resistance-based

Models capable of accurately calculating crop evapotranspiration (ET) are essential to improving water management at both the hydrological cycle and the agriculture irrigation project scale. Penman-Monteith (PM) and Shuttleworth-Wallace (SW) models based on the Jarvis-type canopy resistance (rc) model were used to calculate the hourly and daily ET of rice grown in East China under water-saving irrigation (WSI). The model parameters in rc and soil resistance ( r s s ) were recalibrated to improve the accuracy of rice ET estimations. Notable, three improvements were made to the rc model: (i) model parameters were calibrated at the canopy scale, and a term of effective leaf area index was used to reflect the influence of canopy coverage; (ii) the coefficient a4 was used to reflect the influence of the specific soil moisture patterns under WSI practice; and (iii) for rice crops, soil moisture at saturation, rather than at field capacity for upland crops, served as a key parameter in the soil water response functions. Both PM and SW models generated similar rice ET outputs due to their common theoretical basis. Surprisingly, the PM model, with an appropriately adjusted rc value, performed better than the SW model in simulating hourly and daily rice ET under WSI practice, particularly during periods when canopy cover was either dense or sparse. Sensitivity analyses showed both models to be highly sensitive to changes in rc, and soil moisture (θ), the latter being one of the dominant contributors to variations in rc and r s s . Accordingly, the PM model is recommended for the prediction of rice ET grown under WSI conditions in East China, with the caveat that both the rc and θ should be accurately quantified when used as inputs.