Stochastic Parameterization of Processes Leading to Convective Initiation in Kilometer-Scale Models

Stochastic Parameterization of Processes Leading to Convective Initiation in Kilometer-Scale Models

Predicting convective precipitation (e.g. typical summer thunderstorms) with numerical weather prediction models is limited by their ability to initiate convection. For this initiation small-scale processes in the atmospheric boundary layer are particularly important. Yet due to the limited grid sizes of the numerical models, these processes are often not adequately represented in the models. Two such processes are boundary-layer turbulence and lifting by small-scale orography.
In this work, the representation of these two processes is improved by approximating the variability of subgrid-scale turbulence and subgrid-scale orography by stochastic parameterizations.
The impact of the PSP2 scheme, which takes the variability of turbulence into account, is shown in the figure in comparison to an undisturbed reference run and precipitation measurements (radar) for 10 days in May/June 2016 over Germany. The first figure shows an improved precipitation amplitude for weather conditions with weak synoptic forcing.
The second and third figures depict two precipitation-related prediction scores. The Structure score evaluates the structure of the precipitate compared to the observations, with values near zero being desirable. An improvement by PSP2 can be seen especially for weather conditions with weak synoptic forcing. The Fraction Skill Score (FSS) considers the spatial distribution of precipitation with one as the optimal value. Again the PSP2 scheme mostly improves the score.

Hirt, M., S. Rasp, U. Blahak, and G.C. Craig, 2019: Stochastic Parameterization of Processes Leading to Convective Initiation in Kilometer-Scale Models. Mon. Wea. Rev., 147, 3917–3934, https://doi.org/10.1175/MWR-D-19-0060.1