Spatio‐Temporal Representation of Mesoscale Convective Systems in Convection‐Permitting Simulations Over Northwestern South America: Insights Into Rainfall Overestimation

Abstract We perform km‐scale convection‐permitting simulations (Δ x = 4 km) to represent Mesoscale Convective Systems (MCSs) in Northwestern South America using the Weather Research and Forecasting (WRF) model. An algorithm to identify and track MCSs was used to characterize simulated MCS features and compare with satellite observations. Our results show that WRF adequately represents the seasonal occurrence of MCSs, although large discrepancies arise during periods in which the Intertropical Convergence Zone migrates northward and southward. The complex diurnal cycles of MCS initiation and occurrence are well reproduced by the model, including in‐land nocturnal convection, oceanic hotspots during the morning, and an afternoon maximum in the flatlands. WRF reproduces the evolution of MCS size, minimum brightness temperature, and maximum rainfall through the lifecycle, but especially for short‐lived systems. In terms of MCS features, simulated and observed storms travel comparable distances, and both propagate westward with the trade winds. Notably, the model is able to capture higher MCS speeds on the Amazon‐Savannas flatlands and lower velocities over and to the west of the Andes. Nevertheless, WRF simulates smaller and shorter‐lived systems, while the rainfall intensity and the fraction of heavy rainfall are largely overestimated. A more detailed analysis showed that rainfall overestimation is especially marked in smaller MCS objects and during the development stage. After separating the results by region, we notice a marked overestimation of rainfall intensity in MCSs occurring east of the Andes, while the model underestimates the area of those occurring west of the Andes and in the ocean.