Uncertainty in estimating reference evapotranspiration using remotely sensed and forecasted weather data under the climatic conditions of Southern Spain

Abstract

ABSTRACTIn this study, a detailed spatial analysis of the reference evapotranspiration (ETo) estimates provided by the MAK‐Adv approach and the well‐known Penman‐Monteith equation (PM‐FAO56) has been carried out. MAK‐Adv approach considers a regionally calibrated Makkink equation based on a combination of remotely sensed solar radiation, Rs, provided by the Land Surface Analysis Satellite Applications Facility (LSA SAF), and numerical weather forecasts of near‐surface air temperature, T2m, provided by the European Centre for Medium‐range Weather Forecasts (ECMWF). This analysis evaluated the performance of these approaches with data measured by 57 weather stations under different weather conditions within a semi‐arid region, Andalusia, located at Southern Spain, and determined the role of certain physical and climate variables, such as wind speed and sensible heat advection, on the quality of the ETo estimation.For three years (2007–2009), daily Rs and T2m provided by LSA SAF and ECMWF were compared with measured data showing very good estimates, with RMSD equal to 1.47 W m−2 and 1.53 °C, respectively. Equally, ETo using MAK‐Adv approach with data provided by LSA SAF and ECMWF was compared with ETo provided by the PM‐FAO56 approach with measured data, obtaining RMSD = 0.69 mm, linear regression slope = 0.94, and R2 = 0.92. Although ETo estimates using both approaches were similar between them, the small differences in terms of estimation ETo were mainly correlated with the wind speed (u) and the advection index (Ia), depicting the critical role of these factors in the ETo estimations under semi‐arid conditions, and increasing the uncertainty in ETo estimation when wind speed and/or Ia are high.The comparison of MAK‐Adv and PM‐FAO56 approaches has proved the usefulness in providing ETo values at a regional scale and detecting those areas where the ETo estimation had higher uncertainty, contributing to an improvement in water management.