Date of this Version
Evapotranspiration (ET) is the second-largest component in the water balance equation, globally consuming 70% of the earth’s annual precipitation. Accurate and consistent estimation of ET is essential for ensuring water resources sustainability, proper management, planning, and regulations of water resources. Though a 100% accurate estimation of ET may not be feasible with the current technology, there are proven techniques that give us estimates of ET we can heavily rely on. Mapping Evapotranspiration at High Resolution with Internalized Calibration (METRIC) is a widely used surface energy balance model that produces relatively accurate ET maps utilizing remote sensing data and requires skilled personnel with background knowledge in various disciplines as well as an extended time to run the model when applied manually. The Earth Engine Evapotranspiration Flux (EEFlux) is a completely automated ET estimation application that resides on the Google Earth Engine (GEE) and is based on the METRIC model. EEFlux can produce ET maps for any Landsat 5, 7, and 8 image scenes in a matter of seconds. The eeMETRIC model is a special version of the automated application of METRIC, also on GEE. eeMETRIC is one of six operational ET models utilized in the US-wide OpenET program (Melton et al., 2021) and also uses the automated EEFlux model as the basis, but with enhanced calibration.
The performance of EEFlux has been compared with the manually operated METRIC model by Fooad et al (2018) showing the amount of difference between the products of the two models to be within the acceptable range. Since the automated ET estimation model is still evolving, this study compared recent products from eeMETRIC (version 0.20.15) with that from the METRIC model. Selected sites from central Nebraska focusing on the agricultural lands along the Platte River and in Southern Idaho representing a large, irrigated area in the Snake River Plain were utilized in comparing daily ETrf values, monthly ETrF values, and monthly ET values from both models for the focused areas. Daily ETrF values were produced for each Landsat overpass date. ETrF represents actual ET for an image pixel expressed as a fraction of the reference ET, which in this case represents ET from the alfalfa reference crop. Surprisingly, the results did not always show a good match between both models, and the possible reasons are discussed in this study.
Advisor: Ayse Kilic