Estimating evapotranspiration with thermal UAV data and two source energy balance models

Abstract

Abstract. Estimating evapotranspiration is important when managing water resources and cultivating crops. Evapotranspiration can be estimated using land surface heat flux models and remotely sensed land surface temperatures (LST) which recently have become obtainable in very high resolution using Unmanned Aerial Vehicles (UAVs). Very high resolution LST can give insight into e.g. distributed crop conditions within cultivated fields. In this study evapotranspiration is estimated using LST retrieved with a UAV and the physically-based, two source energy balance models: the Priestley–Taylor TSEB (TSEB-PT) and the Dual-Temperature-Difference (DTD). A fixed-wing UAV was flown over a barley field in western Denmark during the spring and summer in 2014 and retrieved images of LST is successfully processed into thermal mosaics which serve as model input for both TSEB-PT and DTD. The aim is to assess whether a lightweight thermal camera mounted on a UAV is able to provide data of sufficient quality to obtain high spatial and temporal resolution surface energy heat fluxes. Furthermore, this study evaluates the performance of the two source energy balance (TSEB) model scheme during cloudy and overcast weather conditions. This is feasible due to the low data retrieval altitude compared to satellite thermal data that are only available during clear skies and sunny conditions. Flux estimates from TSEB-PT and DTD are compared and validated against field data collected using an eddy covariance system located at same site at which the UAV flights were conducted. Furthermore, spatially distributed evapotranspiration patterns are evaluated using known irrigation patterns. Evapotranspiration is well estimated by both TSEB-PT and DTD with DTD as the best predictor. The DTD model provides results comparable to studies estimating evapotranspiration with satellite retrieved LST and physical land-surface models. This study shows that the UAV platform and the lightweight thermal camera provide high spatial and temporal resolution data valid for model input and for other potential applications requiring high resolution and consistent LST. Lastly, this study explicates thermal UAV data processing and the mosaicking of images into ortho-photos suited for model input.