Intercomparison and Uncertainty Assessment of Nine Evapotranspiration Estimates Over South America

This study examines the uncertainties and the representations of anomalies of a set of evapotranspiration products over climatologically distinct regions of South America. The products, coming from land surface models, reanalysis, and remote sensing, are chosen from sources that are readily available to the community of users. The results show that the spatial patterns of maximum uncertainty differ among metrics, with dry regions showing maximum relative uncertainties of annual mean evapotranspiration, while energy‐limited regions present maximum uncertainties in the representation of the annual cycle and monsoon regions in the representation of anomalous conditions. Furthermore, it is found that land surface models driven by observed atmospheric fields detect meteorological and agricultural droughts in dry regions unequivocally. The remote sensing products employed do not distinguish all agricultural droughts and this could be attributed to the forcing net radiation. The study also highlights important characteristics of individual data sets and recommends users to include assessments of sensitivity to evapotranspiration data sets in their studies, depending on region and nature of study to be conducted. Key Points In wet, energy‐limited regions, evapotranspiration uncertainty is high regarding both representation of annual cycles and anomalies Dry, water‐limited regions show maximum uncertainty of annual mean evapotranspiration, but products agree better on anomaly occurrence Land surface models detect all agricultural droughts reported in literature while the remote sensing products do not always detect them