Evaluation of the NASA POWER reanalysis dataset for estimating reference evapotranspiration in the Peruvian Altiplano
摘要
The Peruvian Altiplano (PA) exhibits high climatic sensitivity and limited availability of meteorological data, which hinders accurate estimation of reference evapotranspiration (ETo). This study evaluated the NASA POWER (NSP) reanalysis dataset for estimating ETo in the PA. Maximum (Tmax), minimum (Tmin), and mean temperature (Tmean), relative humidity (Rh), wind speed (U2), and solar radiation (Rs) data from eight SENAMHI stations and NSP were used. ETo was estimated using the Penman–Monteith (PM) and Hargreaves–Samani (HS) including a calibrated HS version, under ten experimental configurations for evaluation and sensitivity analysis. The results show that NSP adequately reproduces Tmean, Tmin, Rh, and Rs (R > 0.7; R2 = 0.75–0.81), with low bias in Rs. In contrast, U2 exhibited weak to moderate correlations (R = 0.13–0.58) and systematic underestimation. ETo estimation using PM with NSP data demonstrated strong performance (R = 0.87–0.93; R2 = 0.64–0.86; RMSE < 0.9 mm d⁻¹), indicating that replacing observed data with reanalysis does not compromise accuracy. In contrast, HS showed lower predictive capacity (R = 0.68–0.88; R2 = 0.46–0.78), and although calibration reduced bias (mean MBE ≈ 0.01 mm d⁻¹), it did not improve correlation or explanatory power. Sensitivity analysis identified Rh (30.0%), Rs (21.8%), and U2 (19.7%) as the main contributors to ETo uncertainty, while temperature variables had a marginal effect. NSP thus represents a reliable source for estimating ETo in high Andean regions with limited meteorological data, although limitations in U2 representation persist under complex topographic conditions.