Quantifying variability in field-scale evapotranspiration measurements in an irrigated agricultural region under advection

This study compares evapotranspiration (ET) measurements from eddy covariance (EC), lysimetry (LY), and water balance using a network of neutron probe (NP) sensors and investigates the role of within-field variability in the vegetation density in explaining the differences among the various techniques. Measurements were collected over irrigated cotton fields during a period of rapid crop growth under advective conditions. Using NP-based ET estimates as reference, differences in cumulative ET measurements from the EC systems and NP ranged between 2 and 14 %, while differences between LY and NP ranged from 22 to 25 %. The discrepancy in the ET between the three methods was largely attributed to variations in vegetation cover within the source areas of the sensors, which was reliably assessed using high-resolution remote sensing imagery. This analysis indicates that the source area contributing to the measurements must be considered, even in instances where one might consider field conditions uniform. Consequently, differences in measured ET require accounting for variability of vegetation cover conditions in measurement source areas, particularly when used for model validation. This point concerning model validation is exemplified by the difference in performance of a thermal-based energy balance model in estimating ET evaluated using LY versus EC measurements.