Precipitation sensitivity to autoconversion rate in a numerical weather-prediction model

Aerosols are known to significantly affect cloud and precipitation patterns and intensity, but these interactions are ignored or very simplistically handled in climate and numerical weather-prediction (NWP) models. A suite of one-way nested Met Office Unified Model (UM) runs, with a single-moment bulk microphysics scheme was used to study two convective cases with contrasting characteristics observed in southern England. The autoconversion process that converts cloud water to rain is directly controlled by the assumed droplet number. The impact of changing cloud droplet number concentration (CDNC) on cloud and precipitation evolution can be inferred through changes to the autoconversion rate. This was done for a range of resolutions ranging from regional NWP (1 km) to high resolution (up to 100 m grid spacing) to evaluate the uncertainties due to changing CDNC as a function of horizontal grid resolution. The first case is characterised by moderately intense convective showers forming below an upper-level potential vorticity anomaly, with a low freezing level. The second case, characterised by one persistent stronger storm, is warmer with a deeper boundary layer. The colder case is almost insensitive to even large changes in CDNC, while in the warmer case a change of a factor of 3 in assumed CDNC affects total surface rain rate by ~17%. In both cases the sensitivity to CDNC is similar at all grid spacings <1 km. The contrasting sensitivities of these cases are induced by their contrasting ice-phase proportion. The ice processes in this model damp the precipitation sensitivity to CDNC. For this model the convection is sensitive to CDNC when the accretion process is more significant than the melting process and vice versa.