Current issues with modelling convective storms in HARMONIE

Abstract

The performance of HARMONIE on simulating convective storms has been studied by comparing simulated and observed three-dimensional radar reflectivity fields and satellite images. This comparison was performed over two domains; one over the Netherlands and one over the Central US. Case studies lead to a list of five (potential) issues. Most important are the severe underestimation of the number of simulated shallow showers, a potential problem with initiating severe thunderstorms under the turbulence scheme HARATU, and a large underestimation (overestimation) of the area covered by weak (intense) precipitation. The underestimation of the number of shallow showers is related to the shallow convection scheme, and it has been shown that this scheme adjusts vertical temperature and dew point profiles in a way that is unfavourable for initiating showers. For the potential issue with initiating severe thunderstorms it has been shown that using HARATU leads to a drier boundary layer and steeper temperature lapse rates just above the boundary layer, effects that influence initiation of storms. Results of a detailed analysis of distributions of observed and simulated radar reflectivity confirm that the model greatly underestimates the area covered by relatively weak stratiform precipitation, and greatly overestimates the area covered by the more intense convective precipitation. It is shown that the model’s horizontal resolution has very little effect on these distributions, suggesting that improvement should mainly come from improvement of the microphysics. Literature suggests that a better representation of convective regions within storms is critical for improving on this issue, as is switching to a double-moment microphysics scheme.