Optimizing Latent Heat Flux Calculation via Composited Thermal Infrared Temperatures

Latent heat flux (LE) is pivotal in the regional water-energy nexus, exemplifying complex interplays between atmosphere and land surface. Thermal infrared (TIR) land surface temperature (LST) offers direct and vital information for estimating LE through the single-source energy balance method. Nevertheless, variations in the viewing angles of remote sensing sensors can introduce angular effects in the retrieval of LST, potentially causing significant incompatibility issues in estimating LE. To alleviate this uncertainty, we adopt a viable approach by using two composited LSTs derived from the integration of soil and vegetation component temperatures from Sentinel-3 SLSTR, combined with fraction vegetation coverage (FVC) obtained from both the GEOV2 FVC product and MODIS LAI-derived estimates. This composited LST was subsequently used as one of the inputs of a single-source energy balance system (SEBS) model driven by measured meteorological and ERA5 reanalysis data in Heihe River Basin in China during 2016-2022, respectively. The results demonstrate that two types of composited LST reduced the root mean square error (RMSE) of estimated LE by 4.8 W/m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and 8.8 W/m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>, respectively, by using measured meteorological data; and using ERA5 meteorological data, the RMSE was reduced by 6.8 W/m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and 11.0 W/m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>, respectively. Regardless of the meteorological data and FVC used, the RMSE for all stations assessed in the study decreased. This indicates that by partially mitigating the angular effects of TIR LST, improvements in TIR-based surface LE estimation can be achieved over regional scales.