The Weather, Research and Forecasting (WRF) model and Self-Organizing Maps (SOM) artificial neural network (ANN) have been used to classify regional evapotranspiration (ETo) 'weather'. Here, this concept is expanded to develop a pilot ETo 'climate' classification system in a regional subset of the Andes and the Amazon. ETo 'climate' is defined as the frequency of ETo 'weather' classes passing through a geographical location (i.e., a pixel in the WRF model), allowing the construction of a histogram for each pixel. The histogram is then used as a one-dimensional signal for another SOM classification, offering a regional perspective on how ETo behaves and specifically what a geographic location can expect in terms of ETo variables, similar to the Köppen climate classification system. This research presents the results of several classification comparisons to produce a repeatable pilot system for Eto 'climate' classification and a map that offers the potential to improve irrigation decision-making.