A novel dynamic zero-plane displacement height approach to improve remote sensing-based estimation of heat fluxes over corn fields

Abstract

Accurate estimation of latent heat flux (LE) and sensible heat flux (H) is critical for determining actual crop evapotranspiration (ETa) rates and optimizing irrigation. However, heat fluxes over cropped systems are sensitive to surface aerodynamic properties, including the zero-plane displacement height (do), which may introduce uncertainties in the ETa characterization. This study presents a novel approach to better characterize do through a dynamic fractional vegetation cover (fveg) and new canopy porosity (Φdp) term, derived from an unmanned aerial system (UAS) imagery. Field experiments were conducted in 2024 at the USDA-ARS Limited Irrigation Research Farm in Greeley, Colorado, across two corn fields, one fully irrigated (FI) and the other deficit irrigated (DI). Soil water content sensors, soil heat flux plates, soil temperature sensors, net radiometers, infra-red radiometers, and eddy covariance (EC) systems were deployed to measure all the components of the land surface energy balance. Weekly crop development was monitored using a multispectral radiometer (MSR5) to capture the surface reflectance and canopy temperature, and crop height (Hc) was measured manually using a measuring tape. Near-daily calibrated mini-satellite imagery (PlanetScope) was used to derive continuous estimates of spatially distributed net radiation (Rn) across both fields. The upwind fetch area sampled by the EC tower varies with Hc, weather conditions, wind speed (uz), and wind direction (θ) were used to dynamically determine the heat flux footprint area and to align it with the contributing field zones. The newly developed and existing do models were applied to estimate H and LE; the results were evaluated with EC-based corrected H and LE values. Both the Φdp-based and fveg-based do models demonstrated improved accuracy over existing models in estimating H, reducing NRMSE by up to 15.6% (DI) and 21.1% (FI), and 16.9% (DI) and 21.9% (FI), respectively. Similarly, both models achieved a higher agreement index, dr (0.7 in DI, 0.74 in FI) reflecting a stronger model-observation correlation. These results underscore the potential of incorporating Φdp and fveg in dynamically characterizing do to improve H and LE estimation, thereby enhancing energy balance modeling and advancing water management strategies in irrigated agricultural systems.