Influence of almond's canopy-induced shadows on actual evapotranspiration estimation by the TSEB model using sUAS multispectral and thermal imagery

Over the past decade, the estimation of water requirements in almond orchards has improved through the application of remote sensing models like the Two-Source Energy Balance (TSEB) model using various remote sensing platforms. However, there is limited understanding of how canopy-induced shadows influence surface reflectance and thermal infrared (TIR) signals particularly from small Unmanned Aircraft System (sUAS) imagery in energy balance models, and the effect on Latent Heat Flux (LE) estimations. This study evaluates LE estimates from the Priestley-Taylor TSEB model (TSEB-PT) with and without shadow filtering using sUAS-based multispectral and TIR imagery. It establishes a baseline for the impact of shadow exclusion on model inputs and performance. Datasets were collected in 2021 and 2022, as part of the USDA led Tree-crop Remote sensing of Evapotranspiration eXperiment (T-REX) in almonds orchards across California. LAI-2200C Plant Canopy Analyzer measurements facilitated the calibration of an empirical Leaf Area Index (LAI) model based on canopy fractional cover (FC) and NDVI (R ²  = 0.68). Shadow filtering caused land surface temperature (LST) differences up to 5°C in young to semi-mature orchards (FC 0.40–0.80). In contrast, mature orchards (FC > 0.80) showed minimal influence due to the limited shadow occurrence on the imagery. Shadows appeared to reduce surface albedo (α _alb ), mainly in interrow areas, thereby affecting the absorption of radiation and the partitioning of energy balance components. Their presence in sUAS imagery also hindered canopy delineation, impacting the accuracy of key TSEB inputs derived from canopy physical characteristics. Thus, the influence of shadow on TSEB estimated LE was more significant in lower fractional tree covers. While LE estimated by TSEB-PT without shadow filtering showed better agreement with observations, combining instantaneous TIR imagery with solar-noon shortwave data is recommended for accurate ETa assessment using sUAS datasets. These baseline results can be improved with more advanced formulations, supporting continued research on E/T partitioning and water stress in almond orchards under varying environmental conditions, particularly when there is advection of hot dry air.