Abstract
Mastering crop evapotranspiration (ET) and improving the accuracy of ET simulation is critical for optimizing the irrigation schedule and saving water resources, particularly for crops cultivated in a greenhouse. Taking greenhouse-grown tomato under drip irrigation as an example, two weighing lysimeters were used to monitor ET at two seasons (2019 and 2020), whilst meteorological factors inside the greenhouse were measured using an automatic weather station. Then the path analysis approach was employed to determine the main environmental control factors of ET. On this basis, an improved Priestley-Taylor (IPT) model was developed to simulate tomato ET at different growth stages by considering the influence of environmental changes on model parameters (e.g., leaf senescence coefficient, temperature constraint coefficient and soil evaporative water stress coefficient). Results showed that the average daily ET varied from 0.06 to 6.57 mm d−1, which were ~0.98, ~2.58, ~3.70 and ~3.32 mm/d at the initial, development, middle and late stages, respectively, with the total ET over the whole growth stage of ~333.0 mm. Net solar radiation (Rn) and vapor pressure deficit (VPD) were the direct influencing factors of ET, whereas air temperature (Ta) was the limiting factor and wind speed (u2) had a little influence on ET. The order of correlation coefficients between meteorological factors and ET at two seasons was Rn > VPD > Ta > u2. The IPT model can accurately simulate ET in hourly and daily scales. The root mean square error of hourly ET at four stages changed from 0.002 to 0.08 mm h−1 and daily ET varied from 0.54 to 0.57 mm d−1. The IPT coefficient was close to the recommended PT coefficient (1.26) when the average Ta approaches 26 °C and LAI approaches 2.5 cm2 cm−2 in greenhouse conditions. Our results can provide a theoretical basis for further optimization of greenhouse crop irrigation schedules and improvement of water use efficiency.