Increased competition for water resources for agriculture calls for improved water productivities. Improved water productivities must start with accurate estimation of crop evapotranspiration (ET). Accurate estimation of ET has been a challenge for sparsely vegetated crops especially under varying soil water conditions. In this paper, we aim to improve estimation of ET using the Penman-Monteith (PM) method by (1) improved canopy resistance estimation by better representing soil water (2) improved surface resistance estimation by implementing a segmented approach of surface resistance calibration, and (3) investigating the performance of PM models over sparse vegetation by comparing four different solutions; the standard-PM model (PM-STD), the PM-Coupled (PM-CO) single-layer interactive model, the PM Shuttleworth-Wallace (PM-SW) dual-source interactive model and the PM Two-source Patch (PM-TSP) non-interactive model. The model results are evaluated against ET measurements from in-field Bowen ratio-energy balance observations. The results show that, under water-limiting conditions, correctly representing the soil moisture during the different crop development stages enhanced a PM model's ability to accurately estimate ET. The approach raised the Nash-Sutcliffe Efficiency (NSE) of a PM-STD from 0.72 to 0.75. Applying a segmented surface resistance further improved the PM-STD to an NSE of 0.77. When these improvements were implemented on the other PM models, the PM-SW and PM-CO performed superiorly with NSE of 0.83 and RMSE of 0.07 mm hr⁻¹. The PM-TSP followed with an NSE of 0.79 and RMSE of 0.09 mm hr⁻¹ while the PM-STD model trailed with an NSE of 0.77 and RMSE of 0.08 mm hr⁻¹. The performance of all models highlighted the need to separate canopy and soil resistance terms during surface resistance calibration and the value of implementing a segmented approach for resistance calibration. High performance of PM-SW especially demonstrated the need to implement surface energy partitioning and surface aerodynamics interaction for canopy and soil surfaces in PM models

农业水资源竞争日益激烈，要求提高水生产率。提高水生产率必须从准确估算农作物蒸散量（ET）开始。ET的准确估算一直是植被稀疏的农作物面临的挑战，尤其是在土壤水分条件不同的情况下。本文旨在通过Penman-Monteith（PM）方法来改善ET的估算，方法是：（1）通过更好地表示土壤水来改善冠层阻力，（2）通过实施分段的表面电阻校准方法来改善表面电阻， （3）通过比较四种不同的解决方案来研究PM模型在稀疏植被上的性能；标准PM模型（PM-STD），PM耦合（PM-CO）单层交互模型，PM Shuttleworth-Wallace（PM-SW）双源交互式模型和PM Two-source Patch（PM-TSP）非交互式模型。根据实地Bowen比率-能量平衡观测的ET测量评估模型结果。结果表明，在水限制条件下，正确表示不同作物生长阶段的土壤水分可以增强PM模型准确估算ET的能力。该方法将PM-STD的Nash-Sutcliffe效率（NSE）从0.72提高到0.75。施加分段的表面电阻可将PM-STD进一步提高至NSE为0.77。在其他PM型号上实施这些改进后，PM-SW和PM-CO的NSE为0.83，RMSE为0.07 mm hr，表现出色。根据实地Bowen比率-能量平衡观测的ET测量评估模型结果。结果表明，在水限制条件下，正确表示不同作物生长阶段的土壤水分可以增强PM模型准确估算ET的能力。该方法将PM-STD的Nash-Sutcliffe效率（NSE）从0.72提高到0.75。施加分段的表面电阻可将PM-STD进一步提高至NSE为0.77。在其他PM型号上实施这些改进后，PM-SW和PM-CO的NSE为0.83，RMSE为0.07 mm hr，表现出色。根据实地Bowen比率-能量平衡观测的ET测量评估模型结果。结果表明，在水限制条件下，正确表示不同作物生长阶段的土壤水分可以增强PM模型准确估算ET的能力。该方法将PM-STD的Nash-Sutcliffe效率（NSE）从0.72提高到0.75。施加分段的表面电阻可将PM-STD进一步提高至NSE为0.77。在其他PM型号上实施这些改进后，PM-SW和PM-CO的NSE为0.83，RMSE为0.07 mm hr，表现出色。正确表示不同作物生长阶段的土壤水分，增强了PM模型准确估算ET的能力。该方法将PM-STD的Nash-Sutcliffe效率（NSE）从0.72提高到0.75。施加分段的表面电阻可将PM-STD进一步提高至NSE为0.77。在其他PM型号上实施这些改进后，PM-SW和PM-CO的NSE为0.83，RMSE为0.07 mm hr，表现出色。正确表示不同作物生长阶段的土壤水分，增强了PM模型准确估算ET的能力。该方法将PM-STD的Nash-Sutcliffe效率（NSE）从0.72提高到0.75。施加分段的表面电阻可将PM-STD进一步提高至NSE为0.77。在其他PM型号上实施这些改进后，PM-SW和PM-CO的NSE为0.83，RMSE为0.07 mm hr，表现出色。^-1。PM-TSP的NSE为0.79，RMSE为0.09 mm hr ^-1，而PM-STD模型的NSE为0.77，RMSE为0.08 mm hr ^-1。所有模型的性能突显了在表面电阻校准期间需要将冠层和土壤电阻项分开的观点，以及实施分段的电阻校准方法的价值。PM-SW的高性能尤其证明了在PM模型中需要对冠层和土壤表面实施表面能分配和表面空气动力学相互作用