WRF downscaling improves ERA-Interim representation of precipitation around a tropical Andean valley during El Nino: implications for GCM-scale simulation of precipitation over complex terrain

Precipitation in the tropical Andes is strongly influenced by the ENSO phases and orographic effects. In particular, precipitation can be drastically reduced during El Nino. Decision-making about water resources relies on modelling precipitation as the main source for water availability. Here we evaluate ERA-Interim ' s capacity to represent precipitation in the mountainous central Colombian Andes, a strategic region for water supply and hydropower generation, for different phases of ENSO during 1998-2012. Our results show that ERA-Interim fails to reproduce important features of precipitation spatial and temporal variability during different ENSO phases. Most critical in these results is how ERA-Interim overestimates precipitation during the dry season in El Nino years, which corresponds to the most critical condition for water supply. We show that ERA-Interim limitations are likely related to its simplified representation of the complex topography in the region, which excludes the inter-Andean Cauca river valley. To improve this, we implement a dynamical downscaling experiment using the WRF regional climate model, including a sensitivity analysis that considers three convective parameterization schemes and a convection-permitting simulation. WRF downscaling outperforms ERA-Interim in the representation of precipitation during the dry season of El Nino years, especially through correcting positive precipitation biases. This improvement is related to a better representation of orographic effects in WRF simulations. Our results suggest that ERA-Interim and, more generally, climate simulations with comparable coarse resolutions, may produce misleading precipitation overestimations in the tropical Andes if they do not adequately represent inter-Andean valleys, with important implications for water resources management.