CP-MIP: One cold pool in five models, ten runs, and one campaign

Cold pools play an important role in determining the structure and properties of the marine boundary layer in the subtropics. Cold pools are precipitation-driven downdrafts that reach the surface and spread concentrically, leading to cloud suppression inside the cold pool and an active gust front at the perimeter, where converging winds often trigger new convection. Although their importance has long been recognised, the net effect on cloud amount, organisation and radiative effects is still not fully clear. Before we can trust large-eddy simulations to study these aspects, we need to better understand the dependency of simulated cold pools on the chosen model and set-up. Here we analyse outputs from ten runs from five models simulating a cold pool observed during the EUREC4A campaign. All runs were performed as part of the Cold Pool Model Intercomparison Project (CP-MIP). Our goal is to assess the trustworthiness of LES and understand the origin of model differences. We focus on differences related to the following research questions: What conditions lead to the formation of the cold pool? How does the cold pool expand and eventually recover? What is the internal moisture and temperature structure?

Our first results show that most runs produce a single strong cold pool.While large differences in the onset time are caused by forcing differences, the speed of moisture aggregation, and the microphysical model, the growth rate is primarily controlled by the mean buoyancy anomaly inside the cold pool. Furthermore, we show that the internal structure can differ greatly between runs that differ only in their microphysical schemes. While some runs produce a single cold pool that eventually spreads to a size exceeding 100 km, other runs initially create more than ten individual cold pools that all collide and form a super cold pool of comparable size. The EUREC4A observations show a growth rate and timing that fall within the inter-model spread. Measurements of the early stage of the cold pool reveal an almost homogeneous internal structure rather than multiple events. As demonstrated, this unique setup allows not only for a comparison of the runs but also for validation of relevant processes with observations.