Estimating Daily Reference Evapotranspiration in a Semi-Arid Region Using Remote Sensing Data

Estimating daily evapotranspiration is challenging when ground observation data are not available or scarce. Remote sensing can be used to estimate the meteorological data necessary for calculating reference evapotranspiration ETo. Here, we assessed the accuracy of daily ETo estimates derived from remote sensing (ETo-RS) compared with those derived from four ground-based stations (ETo-G) in Kurdistan (Iraq) over the period 2010–2014. Near surface air temperature, relative humidity and cloud cover fraction were derived from the Atmospheric Infrared Sounder/Advanced Microwave Sounding Unit (AIRS/AMSU), and wind speed at 10 m height from MERRA (Modern-Era Retrospective Analysis for Research and Application). Four methods were used to estimate ETo: Hargreaves–Samani (HS), Jensen–Haise (JH), McGuinness–Bordne (MB) and the FAO Penman Monteith equation (PM). ETo-G (PM) was adopted as the main benchmark. HS underestimated ETo by 2%–3% (R2 = 0.86 to 0.90; RMSE = 0.95 to 1.2 mm day−1 at different stations). JH and MB overestimated ETo by 8% to 40% (R2 = 0.85 to 0.92; RMSE from 1.18 to 2.18 mm day−1). The annual average values of ETo estimated using RS data and ground-based data were similar to one another reflecting low bias in daily estimates. They ranged between 1153 and 1893 mm year−1 for ETo-G and between 1176 and 1859 mm year−1 for ETo-RS for the different stations. Our results suggest that ETo-RS (HS) can yield accurate and unbiased ETo estimates for semi-arid regions which can be usefully employed in water resources management.