Aiming to address the lack of a high-resolution spatial modeling method for effectively quantifying evapotranspiration in farmland, a temperature-corrected drone estimation farmland ET method was proposed. In order to obtain temperature data and multispectral data, M100 drone was equipped with an FLIR VueProR infrared camera and a Micasense RedEdge multispectral camera. The remote sensing evapotranspiration model was studied, the mapping evapotranspiration at high resolution with internalized calibration (METRIC) model and the remote sensing energy balance (RSEB) model were compared. The soil heat flux calculation method for the RSEB model was not applicable to the farmland environment. The model was improved based on multi-spectral data to raise the applicability of the model. Because equipment and environmental factors can easily produce errors when measuring temperature in the RSEB model, the relationship between remote sensing temperature data and the actual temperature was applied as a method to correct thermal imaging. The data was substituted to obtain the flux value and the evapotranspiration, and the model calculation results were compared with the open path eddy covariance (OPEC) measurement results. The results showed that the RSEB model combined with multispectral data can obtain accurate flux data after temperature correction. The root mean square error of the sensible heat flux was 20.013W/m2, and the mean absolute error was 15.835W/m2, the root mean square error of the latent flux was 42.202W/m2, and the mean absolute error was 26.017W/m2.The spatial distribution map of farmland evapotranspiration with resolution decimeter level was obtained. The method can effectively obtain the high-efficiency spatial pattern of field evapotranspiration and provide support for farmland irrigation.