Abstract
This study focuses on Patagonia, where Foehn events observed in the lee of the Andes mountains are not yet well simulated by state‐of‐the‐art climate models. It has been agreed that one source of this shortcoming is related to the poor relief representation in models. To resolve this need, a common method used is to enhance the spatial resolution of the model to retrieve a more complex surface elevation, at the expense of calculation time or surface area covered. This paper tackles the problem from a different angle by addressing the Digital Elevation Model (DEM) generalization, that is, the altitudes generalization from a high‐resolution DEM to a coarser resolution grid model. Most current climate models use DEM generalization methods that smooth the relief, a key controlling factor in Foehn events modelling. The aim of this study is to compare three original methods of DEM generalization (percentile 90 [P90], envelope maximum [EM], and thalweg and crests [TC]) and to evaluate their impact on simulated precipitation and temperature fields on the eastern part of Patagonia, where warm and dry air masses are expected. Thanks to MAR, a Regional Climate Model, we validate the models at 10 and 5 km resolutions against the Climate Research Unit and perform three sensitivity experiments involving a change in the DEM generalization. Our results show that (a) a finer spatial resolution can slightly improve the temperature biases, however, it cannot resolve the precipitation biases and (b) a more appropriate use of DEM generalization induces a significant decrease in precipitation for the P90 and EM methods and an increase in mean temperature for all three methods in the study area. This study serves as a recommendation for a better use of DEM generalization in climate models performing in Patagonia, but also regions sharing the same orographic features as the Patagonian relief.