Comparison of the Automatically Calibrated Google Evapotranspiration Application—EEFlux and the Manually Calibrated METRIC Application

• DOI: 10.20944/preprints201807.0040.v1
• Publication Date: 2018-07-06T04:25:02Z
• Authors: Foad Foolad; Philip Blankenau; Ayse Kilic; Richard G. Allen; Justin L. Huntington; Tyler A. Erickson; Doruk Ozturk; Charles G. Morton; Samuel Ortega; Ian Ratcliffe; Trenton E. Franz; David Thau; Rebecca Moore; Noel Gorelick; Baburao Kamble; Peter Revelle; Ricardo Trezza; Wenguang Zhao; Clarence W. Robison
• Keywords: Albedo (alchemy); Water balance

Reliable evapotranspiration (ET) estimation is a key factor for water resources planning, attaining sustainable use, irrigation management, and regulation. During the past few decades, researchers have developed variety of remote sensing techniques to estimate ET. The Earth Engine Evapotranspiration Flux (EEFlux) application uses Landsat imagery archives on Google platform calculate daily at local field scale (30 m). Automatically calibrated each image, EEFlux design based widely vetted Mapping high Resolution with Internalized Calibration (METRIC) model produces ET maps any 5, 7 or 8 scene in matter seconds. In this research we evaluate consistency accuracy products that are produced when standard US global assets used. Processed METRIC 58 scenes distributed around western central United States were used as baseline comparison. goal paper compare results from applications illustrate utility they currently stand. Given derived METRIC, differences expected occur due differing calibration methods (automatic versus manual) input datasets. compared include fraction reference (ETrF), actual (ETa), surface energy balance components net radiation (Rn), ground heat flux (G), sensible (H), well Ts, albedo NDVI. product comparisons show intermediate Albedo, NDVI, also Rn similar values behavior both METRIC. Larger found H G. Despite more significant G, able ETrF ETa comparable trained expert users agricultural areas. For non-agricultural areas such semi-arid rangeland forests, automated algorithm needs be improved order reproduce manually products.