Sensitivity studies of developing convection in a cloud-resolving model

Cloud-resolving models (CRMs) remain an important tool for providing detailed process information about convection. In this short paper I focus on the development of deep convection and consider what can be considered a minimum expense benchmark simulation for comparison with a numerical weather-prediction model. To decide this a range of sensitivity studies are presented to aspects of the experimental set-up which strongly impact the computational expense. Many of the sensitivities shown in these CRM experiments are quite different to those seen in previous papers which have tended to focus more on deep active convection. Here it is shown that for the case-study presented a minimum expense benchmark simulation must be a 3D simulation. A 200 m horizontal grid length and a domain of 25 km are also required to capture the most important processes.