Downscaling heavy rainfall in the subtropics – a simple approach for dynamical nesting

Abstract. The simulation of local scale precipitation with nested models often suffers from large errors in the boundary rows. Advection of precipitation into the model domain of the small scale model can lead to an overestimation of precipitation in the boundary grid cells of the nested model and a drying of the interior grid area. Consequently, the finer scale structure of rainfall events of the small scale model can not evolve. These errors result from three main sources: "dynamical", "scale", and "parameterization" problems. As a first step to reduce the "parameterization" boundary errors, we propose a nesting procedure where rainwater from the driving larger scale model is converted to cloud water in the smaller scale model. The nesting method is applied to a case study of heavy rainfall in semi-arid southern Morocco. The results show the elimination of erroneous excessive rainfall in the boundary rows and slightly enhanced rainfall in the interior of the nested model domain. Additionally, fine scale structures in the precipitation patterns develop. The excessive surface runoff is clearly diminished in comparison to the standard nesting procedure. The proposed approach enables scale consistent precipitation patterns resulting from model physics and grid-resolution of the smaller scale model for the complete model domain.