Tailoring the surface energy balance algorithm for land-improved (SEBALI) model using high-resolution land/use land cover for monitoring actual evapotranspiration

Access to accurate near real-time actual evapotranspiration (ETa) data is crucial for monitoring and planning water uses, contributing to the sustainable management and development of water resources. However, ongoing ETa modeling efforts need to be improved when applied to small-sized farmlands in rugged landscapes like in the upper Blue Nile Basin. This study aimed to customize the Google Earth Engine-based Surface Energy Balance for the Algorithm Land–Improved (SEBALIGEE) model using high-resolution land use and land cover data. Additionally, the model, originally based on JavaScript, was updated to Python (SEBALIGEEpy), thereby enhancing its accessibility to a broader range of users and modelers. The ETa simulated with an updated version of the model was validated over croplands using publicly available AmeriFlux eddy covariance data. Furthermore, the SEBALIGEEpy-based ETa was evaluated over the Koga smallholder irrigation scheme in the Upper Blue Nile Basin, Ethiopia, by comparing it with the ETa obtained from the remote sensing-based Water Productivity Open-Access Portal (WaPOR) over seven seasons (2015/16–2022/23). The validation results show that the updated model provides more accurate ETa estimates with fewer missing records, with 9–11 % using the updated model compared to 14–29 % using the original SEBALIGEE model. The customized model has the potential to be used for agricultural water management in areas with fragmented land parcels like the Upper Blue Nile basin and beyond.