Modeling rice evapotranspiration under water-saving irrigation by calibrating canopy resistance model parameters in the Penman-Monteith equation

A canopy-resistance-based Penman-Monteith (PM) model's performance in estimating rice evapotranspiration (ET) under water-saving irrigation (WSI) condition at hourly and daily intervals was evaluated. To improve the performance of Jarvis-type canopy-resistance model in calculating rice ET under water-saving irrigation condition with Penman-Monteith (PM) model, a term of effective leaf area index was used to reflect the influence of canopy coverage condition, and field capacity was replace by saturated soil moisture (θs) in soil water response functions due to the specific field moisture condition in WSI rice fields. Two years of hourly rice ET measured using eddy covariance (EThEC) with energy balance closure adjustment served to calibrate (2014 data) and validate (2015 data) a canopy-resistance-based PM model. Results indicate that the PM model's output (EThsim) with calibrated canopy resistance model parameters closely matched measured EThEC, with a regression slope, intercept, coefficient of determination (R2), root mean squared error (RMSE) and index of agreement (IOA) of 1.001, 0.010 mm h−1, 0.942, 0.057 mm h−1 and 0.985, respectively for validation data set in 2015, better than the results calculated by model with parameters calibrated for general paddy field. The improved PM model performed well not only in a dense canopy condition (LAI > 3) but also for a sparse canopy (LAI < 3). The model also performed well both in sunny and cloudy days of each rice growth stages. A more detailed analysis indicated that EThsim tended to overestimate or underestimate EThEC in varied degrees at noon time or during the transition period from day to night. These differences between EThsim and EThEC associated with the variation of net radiation (Rn), soil heat flux (G0), and soil moisture (θ). Additionally, the PM model calibrated on hourly dataset, performed well in calculating the daily ET directly under WSI for 2014 (R2 = 0.973, RMSE = 0.237 mm d−1, IOA = 0.993) and 2015 (R2 = 0.961, RMSE = 0.302 mm d−1, IOA = 0.990). Therefore, the PM model calibrated using the hourly data was capable of accurately estimating both hourly and daily ET for rice under WSI condition.