MODIS and Sentinel-2 Data Fusion For 10-m Daily Evapotranspiration mapping

Downscaling methods seem to be a reasonable solution to solve the problem of having no simultaneous high spatial and temporal satellite data, and it is possible somehow to meet the requirement of having high spatial-temporal resolution satellite data for monitoring the natural phenomena such as evapotranspiration, through these methods. Sentinel-2 satellite launched in 2015 enables to provide 10-m spatial resolution data with a 5-day revisit time; however, its sensor does not acquire data in thermal infrared wavelength. This study aims to generate 10-m daily evapotranspiration maps based on Sentinel-2 and MODIS data fusion for Amir-Kabir Agroindustry farms. For this purpose, STARFM and improved TSHARP methods were applied for downscaling MODIS data to Sentinel-2 data. To achieve this goal, First, MODIS visible and near and middle infrared bands were downscaled by STARFM to 10-m spatial resolution. Then, improved TSHARP was applied for downscaling MODIS thermal band to 10-m spatial resolution and SEBAL algorithm fed by the downscaled bands, was used to produce daily evapotranspiration map with 10-m spatial resolution. Assessing downscaled evapotranspiration maps with those derived from FAO Penman-Monteith equation indicated a RMSE of 0.64 mm/day showing efficient performance of the downscaling framework proposed for 10-m daily evapotranspiration mapping in this study.