Evapotranspiration Estimates from SSEBop using Synthetic Land Cover Data from the Continuous Change Detection and Classification Algorithm

Gridded actual evapotranspiration datasets have been successfully developed from Landsat data using methods such as the operational Simplified Surface Energy Balance (SSEBop) model. SSEBop has been validated against flux tower data and successfully incorporated into OpenET to generate accurate evapotranspiration estimates for the United States, however SSEBop can only be applied to regions of cloud free Landsat data. Here the Continuous Change Detection and Classification (CCDC) algorithm is utilized to generate synthetic Landsat data that is then used to initialize SSEBop to produce evapotranspiration estimates for six target areas in the western United States with irrigated agriculture, forest, desert and shrubland land cover types. CCDC fits a harmonic model to Landsat observations to estimate surface reflectance and emissivity that are used to produce the land surface temperatures and NDVI values required for SSEBop evapotranspiration estimates. Spatial means of all six target areas of CCDC evapotranspiration estimates mirror the phenological tendencies observed in the Landsat data and typically underestimate the average evapotranspiration by less than 1 mm/day. At the annual timescale, CCDC derived evapotranspiration estimates generally underestimate average evapotranspiration by 0.5 to 1.5 mm/day, as compared to the Landsat observations. The CCDC evapotranspiration estimates best fit the Landsat data for the forested target areas in Oregon and Colorado, while the croplands in California and Arizona that are irrigated shortly before Landsat flyovers show elevated levels of evapotranspiration that are not detected in the CCDC model. Future work will include running CCDC on Landsat derived surface temperatures and trying new harmonic fitting algorithms to correct for the underestimation of evapotranspiration from the CCDC algorithm. Additional future work will seek to determine the ability of CCDC to forecast regional evapotranspiration at seasonal timescales based on projected land cover change as determined through predictions in the Normalized Difference Vegetation Index. Seasonal CCDC derived evapotranspiration forecasts will contribute to enhanced drought prediction and aid decision makers in managing increasingly scarce water supplies in the western United States.