Performance assessment of empirical reference evapotranspiration estimation models on a seasonal basis in Northern Ethiopia

Abstract

Accurate estimation of reference evapotranspiration (ETo) is critical for sustainable water resource management, agricultural planning, and climate adaptation in Ethiopia, where strong agroecological diversity and pronounced seasonality challenge hydrological modeling. This study evaluates the seasonal performance of twelve empirical ETo models across seven representative meteorological stations spanning humid highland, sub-humid, and semi-arid to arid zones, using the FAO Penman–Monteith method as a reference. Results revealed that Ethiopia’s complex climate requires season-specific model selection rather than a single universal approach. In humid highlands, temperature-based Enku and Thornthwaite models showed the most consistent performance (e.g., R² up to 0.84 and NRMSE as low as 11.8%), while arid regions required summer–winter switching, with the Abtew model performing best in summer (NRMSE ≈ 17.8%) and the WMO model excelling in winter (IoA up to 0.80). Priestley–Taylor and Turc models consistently failed across all stations and seasons (NRMSE > 60%, NSE < 0), indicating their unsuitability for northern Ethiopia. The proposed seasonal framework improves localized ETo estimation and advances understanding of evapotranspiration modeling in tropical highland environments.