Locally forced convection in sub‐kilometre scale simulations with the Unified Model and WRF

This study evaluates the performance and benefits of kilometre and sub‐kilometre scale convection permitting simulations over tropical Australia. Focusing on an extended Monsoon break period we can directly compare Unified Model (UM) and Weather Research and Forecasting model (WRF) simulations to CPOL radar observations and soundings. We show that the two models have different behaviour, and both are different to observations. Whereas WRF produces daily squall lines whether or not they occurred in observations, the UM primarily generates small but intense storms. The UM and WRF produce qualitatively different surface density currents at different times in the diurnal cycle. Once the density currents are present, the models also show different behaviour in relation to convective initiation. While higher resolution helps in the distribution of total precipitation over the domain, most characteristics do not change with higher resolutions, and model difference are always larger than resolution differences. While CAPE/CIN does not seem to be important to explain model differences, our findings point to the evolution of density currents in the boundary layer as most important source of model errors and differences