Sensitivity analysis of an aerosol aware microphysics scheme in WRF during case studies of fog in Namibia

Aerosol aware microphysics parameterisation schemes are increasingly being introduced into numerical weather prediction models, allowing for regional and case specific parameterisation of cloud condensation nuclei (CCN) and cloud droplet interactions. In this paper, the Thompson aerosol aware microphysics scheme, within the Weather, Research and Forecasting (WRF) model, is parameterised for two fog cases during September 2017 over Namibia. Measurements of CCN and fog microphysics were undertaken during the Aerosol, Radiation and Clouds in southern Africa (AEROCLO-sA) field campaign at Henties Bay on the coast of Namibia during September 2017. A key concept of the microphysics scheme is the conversion of water friendly aerosols to cloud droplets (hereafter referred to as CCN activation), which could be estimated from the observations. A fog monitor 100 (FM100) provided cloud droplet size distribution, number concentration (Nt), liquid water content (LWC) and mean volumetric diameter (MVD). These measurements are used to evaluate and parameterise WRF model simulations of Nt, LWC and MVD. A sensitivity analysis was conducted through variations to the initial CCN concentration, CCN radius and the minimum updraft speed, important factors that influence droplet activation in the microphysics scheme of the model. The first model scenario made use of the default settings with a constant initial CCN number concentration of 300 cm-3 and underestimated the cloud droplet number concentration while the LWC was in good agreement with the observations. This resulted in droplet size being larger than the observations. Another scenario used modelled data as CCN initial conditions which were an order of magnitude higher than other scenarios. However, these provided the most realistic values of Nt, LWC, MVD and droplet size distribution. From this it was concluded that CCN activation of around 10 % in the simulations is too low, while the observed appears to be higher reaching between with a mean (median) of 0.55 (0.56) during fog events. To achieve this level of activation in the model, the minimum updraft speed for CCN activation was increased from 0.01 to 0.1 ms-1. This scenario provided Nt, LWC, MVD and droplet size distribution in the range of the observations with the added benefit of a realistic initial CCN concentration. These results demonstrate the benefits of a dynamic aerosol aware scheme when parameterised with observations.