Regional-scale modeling of reference evapotranspiration: Intercomparison of two simplified temperature- and radiation-based approaches

Two regionalized models for the distributed estimation of daily reference evapotranspiration, the temperature-based Hargreaves (HE) and the radiation-based Makkink (MK) equations, are applied in Southern Italy during the years 2007 and 2008. Spatially distributed meteorological inputs, such as air temperature and incoming solar radiation, were derived from geostatistical interpolation of ground data and from the Land Surface Analysis Satellite Application Facility (LSA-SAF) surface radiation product, respectively. Comparison of the latter with 83,394 daily measurements provided by 128 weather stations shows a not negligible seasonal error in daily solar radiation that is corrected by means of a periodic equation. A preliminary local calibration of the MK coefficient highlights its strong dependency on the interactions between moist winds from the Atlantic Ocean and orographic obstructions, leading to two western and eastern subzones. Hence, a regionalization of the MK-adjusted coefficient was performed following a procedure similar to the one already used by the authors in a previous work for the HE coefficient. Application of both regionalized models and their comparison with Penman-Monteith estimates derived from the same stations leads to average whole-year absolute errors of approximately 0.5 mm d-1, further showing the great potential of the MK method in terms of informative content and less dependence on the site density of a ground-based monitoring network. Overall, discussing pros and cons of both regionalized models, it is shown that their integrated use in southern Italy is possible and advisable.