Approximating FAO-56 Penman–Monteith reference evapotranspiration with a wind-augmented Hargreaves–Samani model across the Czech Republic

Study Region: The Czech Republic, represented by 1 301 722 daily observations from 171 meteorological stations during 2000–2025. Study Focus: FAO-56 Penman–Monteith reference evapotranspiration (FAO-PM) requires inputs that are often unavailable. We developed a four-parameter wind-augmented Hargreaves–Samani model (HS-WA) combining a calibrated temperature–radiation baseline with an additive temperature-derived demand term and a bounded, concave wind response. HS-WA was compared with the original and calibrated Hargreaves–Samani equations and a recalibrated Cobaner wind correction using full-data calibration, 100 random hold-outs, 11 rolling-origin temporal tests, 30 station-blocked tests, and station-level regional analyses. New Hydrological Insights for the Region: HS-WA achieved full-data RMSE, MAE, and correlation of 0.409 mm day − 1 , 0.299 mm day − 1 , and 0.965, outperforming the recalibrated Cobaner model ( 0.435 mm day − 1 , 0.312 mm day − 1 , and 0.960). It retained lower RMSE and MAE and higher correlation in every random, temporal, and station-blocked split, and reduced station-specific RMSE at 164 of 171 stations. Its normalized wind response closely reproduced the data-conditioned aerodynamic wind-response shape embedded in FAO-PM, whereas the linear Cobaner multiplier matched it less closely. Cobaner nevertheless often produced smaller absolute bias, revealing a criterion-dependent trade-off between daily accuracy and mean offset. HS-WA therefore provides a robust regional reduced-input surrogate for daily FAO-PM approximation where temperature and wind observations are available.