Drone-Based Ultra-High-Resolution Evapotranspiration Estimation using Synchronized Thermal and Multispectral Sensors and the HSEB Model

Accurate estimation of evapotranspiration (ET) holds paramount importance in optimizing irrigation strategies and managing crop water utilization in agriculture. Enhancing water use efficiency through precise monitoring of crop water requirements and irrigation scheduling is a critical goal. To tackle these imperatives, the precise estimation of ET at suitable spatial and temporal scales becomes imperative. This study investigates the potential of unmanned aerial vehicles (UAVs), commonly known as drones, for achieving ultra-high-resolution ET mapping. We assess the efficacy of the MicaSense Altum sensor in conjunction with the Hybrid Single Source Energy Balance Model (HSEB) to estimate ET over a 4.5 ha sprinkle-irrigated potato field at the American University of Beirut Agricultural Research and Education Center (AREC), situated in Lebanon's Bekaa Valley. Our methodology is validated against the established eddy covariance technique. The Altum MicaSense sensor offers multispectral and high-resolution imagery (2.6 cm), synchronized with thermal imagery captured at 33 cm when flown at 60 m above the ground. In alignment with local overpasses of Landsat and MODIS, eleven drone flights were conducted throughout the potato's active growing season. An Eddy Covariance system was deployed, equipped with three soil heat flux sensors, soil moisture sensors for soil heat flux measurements, a net-radiometer for net radiation measurements, and three infra-red radiometers to gauge land and canopy surface temperatures. Using the HSEB model, ET was computed from Altum Imagery after Pix4D software processing. Additionally, a comparative analysis is performed between drone-based ET estimations and those derived from Landsat 8, Landsat 9, and Sentinel-2. This analysis operates within a consistent modeling framework, utilizing ERA-5 Land meteorological data. The study demonstrates a remarkable agreement between instantaneous and daily latent heat flux HSEB ET estimates and those obtained through Eddy Covariance. At the daily scale, Altum ET averaged 12% higher than Eddy ET, while Sentinel-2 and Landsat HSEB ET were 12% and 2% lower than Eddy ET, respectively. Land surface temperatures derived from the Altum sensor exhibited a close alignment with field measurements (R2 = 0.9, RMSE = 1.09 degrees Kelvin), albeit slightly lower by one degree. In contrast, T-Sharp sharpened LST from Landsat and MODIS (sharpened with Sentinel-2) were consistently higher than measured LST, potentially due to field size heterogeneity, aridity, and the sparse density of irrigated fields surrounding AREC. These variations impact both ET calculations and the temperature differential between land surface and air, a pivotal aspect of the HSEB energy balance model. In summary, this research underscores the dependability of drone-based thermal imagery coupled with HSEB surface energy balance modeling as a robust approach for precise evapotranspiration estimation. The study's implications are substantial for advancing agricultural water management practices through precision farming techniques. Simultaneously, this research offers insights into the prerequisites for accurate ET estimation. Ultimately, the fusion of drone technology and advanced modeling techniques emerges as a promising avenue to address contemporary water scarcity challenges in agriculture.