Comparative analysis of different temporal up-scaling methods for evapotranspiration applied to water-saving irrigated paddy fields

Abstract: The temporal up-scaling has been the most popular to estimate the regional daily evapotranspiration (ET) using remote sensing and in-situ observation data. Several estimation methods of daily ET are usually assumed that the diurnal course of ET is similar to that of solar irradiance, including the evaporative fraction, crop coefficient, canopy resistance, and daily sine function. However, such an assumption is scarce for humid paddy fields. Taking the Water-Saving Irrigation (WSI) paddy fields as the research object, this study aims to compare the differences between the four ET estimations in the daily scale and the measured values using the Eddy Covariance (EC) system in the 2015 and 2016 rice season. The results showed that the average diurnal variation in the three scale conversion factors presented the special under WSI condition, including the Evaporative Fraction (EF), crop coefficient (Kc), and canopy resistance (rc) during the growth period. It infers that the outstanding variation was responded to the changes in the crop species, soil, and meteorological conditions. The diurnal and seasonal average EF and Kc were considered as an approximately concave-down shape, with a gentle variation during the 10:00-14:00 period, and the minimum around 10:00-11:00 AM. The diurnal variation of rc increased after sunrises, further to be almost constant from 09:00 to 12:00, and then tended to increase rapidly. By contrast, the ratio of the total daily solar irradiance to the instantaneous irradiance (J) was just determined using the latitude of the study area, indicating a typical concave-down shape. Furthermore, there was a better performance on the accuracy, correlation, and consistency of the crop coefficient and canopy resistance using 10:00-11:00 hours of the estimated daily evapotranspiration, compared with the other hours. In addition, a rough ET up-scaling was comparatively stable using the sine function. More importantly, the hourly EF estimation during the 10:00-14:00 period achieved fewer errors in the daily ET, indicating the accuracy and consistency were the best from 10:00-11:00. This finding was consistent with the time of satellite overpass of Terra/Modis, although the estimated values of evaporative fraction were relatively underestimated in each period. Such a stable and considerable gap was brought forward a high demand to establish a rational relationship between the effective daily ETEC and the ETEF calculated by the best time-of-day representativeness (10:00-11:00) of EF. Therefore, a correction procedure was also conducted to incorporate the variable Vapor Pressure Deficit (VPD) with the ET value following the EF up-scaling as an independent variable. Among them, the VPD was the key factor affecting the ET at both daytime and night. As such, a linear regression equation was achieved, where a low RMSE of 0.205 mm/d, the R2 of 0.987, and IOA of 0.996. Consequently, the results demonstrated that the daily ET calculated by the best time-of-day representativeness (10:00-11:00) of EF, Kc, and rc performed better, while, the improved evaporative fraction was optimal to the temporal up-scaling of daily scale ET estimation for the WSI paddy fields in the middle and lower reaches of the Yangtze River. This finding can also provide a potential guiding for farmland irrigation and agricultural water management.