The Impact of High-Resolution Soil Moisture States on Short-Term Numerical Weather Prediction of Convective Initiation over South Africa

TL;DR

This study investigates how high-resolution soil moisture data improves short-term weather predictions of convective initiation over South Africa, showing modest improvements in storm location accuracy but limited impact on timing.

Contribution

It demonstrates that high-resolution soil moisture initial conditions can enhance the spatial accuracy of convective storm forecasts in a regional weather model.

Findings

Higher resolution soil moisture reduces storm centroid distance by 9%.

Most improvements occur at shorter lead times.

Convective initiation is more likely over dry and moderate soils.

Abstract

The interaction between the Earths surface and the atmosphere plays a key role in the initiation of cumulus convection. Over the land surface, a necessary boundary condition to consider for resolving land-atmosphere interactions is soil moisture. The aim in the study is twofold. One, through object oriented and traditional verification techniques determine how higher resolution soil moisture initial conditions influences the prediction of the location and timing of convective initiation (CI) within a convective permitting, operational NWP model over South Africa. Two, to study the modelled CI-soil moisture relationship during real afternoon thunderstorm events. The study reports the results from 66 Unified Model simulations (at 4.4km grid resolution) for nine summer afternoon CI events during synoptically benign conditions over South Africa. The higher resolution soil moisture conditions reduce centroid distance between observed and forecast storms on average by 7km (9 percent improvement), with the most decrease in centroid distance occurring at the shortest lead times, by 12km. Most improvement in location error occurs in the zonal directional. However, little to no difference is found in the timing of CI, most likely attributable to the dominant effect of model grid size on CI timing, overshadowing the influence from soil moisture anomalies. Probability of CI is highest over dry and moderate soils and areas along distinct soil moisture gradients. The conclusion is that modelled CI over South Africa preferentially occurs on the periphery of wet soil moisture patches, where there is increased surface convergence of wind and higher sensible heat flux.