Influence of Spatial Heterogeneity in Evapotranspiration Modeling at Natural Areas Using sUAS High Resolution Data

Abstract:

Understanding the spatial variability in complex natural environments such as savannas and river corridors is an important issue in modeling energy fluxes, particularly for evapotranspiration (ET) estimates. The natural environment is usually characterized by heterogeneity in soil and plants, in addition to other biophysical processes. Different land surface and hydrological models can be applied to estimate ET in such environments; however, the capability of these models is limited due to lack of robust methods accounting for the complexity of the interaction between surface and atmosphere. Nowadays, remote-sensing-based surface energy balance (SEB) models are widely and routinely applied to obtain ET information on an operational basis for use in water resources management. In this research effort, sUAS data was used to study the influence of land surface spatial heterogeneity on the modeling of ET at high resolution, specifically at the San Rafael River corridor in Utah, which is part of the Upper Colorado River Basin and is dominated by a wide range of vegetation types, including Tamarisk, cotton, willow, grass, and others. The study area is also characterized by arid conditions and variations in soil moisture status, soil types, and tree heights. First, a spatial variability analysis was carried out using variogram analysis to identify a representative contextual spatial domain/model grid size for adequately solving energy balance components to derive ET. Then the physically based ET model, namely Two-Source Energy Balance (TSEB), was evaluated over different vegetation, soil conditions and times. Remote sensing data with multispectral images were acquired through multiple sUAS campaigns carried out over different seasons (June, July and October) by the AggieAir^TM sUAS Program at Utah State University (https://uwrl.usu.edu/aggieair/). Optical data including red, green, blue, and near infrared bands were acquired at 2.5-cm spatial resolution, while thermal data were acquired at 15 cm using a microbolometer camera. The procedure and the validated results for the different locations, times of day, and seasons are presented and discussed.