Assessment of the representation of West African storm lifecycles in convection‐permitting simulations

Convection-permitting models perform better at representing the diurnal cycle and the intermittency of convective rainfall over land than parameterized-convection models. However, most of the previous model assessments have been from an Eulerian point of view, while key impacts of the rainfall depend on a storm-relative perspective of the system lifecycle. Here a storm-tracking algorithm is used to generate storm-centered Lagrangian lifecycle statistics of precipitation over West Africa from regional climate model simulations and observations. Two versions of the Met Office Unified Model with and without convection parameterization at 4-, 12-, and 25-km resolution were analyzed. In both of the parameterized-convection simulations, storm lifetimes are too short compared to observations, and storms have no preferred propagation direction; the diurnal cycle of initiations and dissipations and the spatial distribution of storms are also inaccurate. The storms in the convection-permitting simulations have more realistic diurnal cycles and lifetimes but are not as large as the largest observed storms. The convection-permitting model storms propagate in the correct direction, although not as fast as observed storms, and they have a much improved spatial distribution. The rainfall rate of convection-permitting storms is likely too intense compared to observations. The improved representation of the statistics of organized convective lifecycles shows that convection-permitting models provide better simulation of a number of aspects of high-impact weather, which are critical to climate impacts in this important geographic region, providing the high rainfall rates can be taken into account.