This study examines the sensitivity of the resolution of the soil moisture initialization in the cloud-resolving grid in a real-data simulation of a mesoscale convective system (MCS) and its implication in soil moisture initialization in high resolution models (e.g., regional climate models, numerical weather prediction models). The quasi-stationary MCS of interest initiated in the Texas/Oklahoma border near a quasi-stationary front around 2200 UTC 26 July and dissipated around 0700 UTC 27 July 1998.

The soil moisture for the finest nested grid was derived from the Antecedent Precipitation Index (API) using a 4-km grid spacing precipitation data for a three-month period from the GCIP database. The resolution of the soil moisture initialization was coarsened by Barnes objective analysis with a response amplitude of 0.5 at cutoff wavelengths of 20 km (API20), 40 km (API40), and 80 km (API80). The simulation with the original API-derived soil moisture initialization is served as control (CTL). An additional experiment with homogeneous soil moisture, derived by horizontally averaging the original API-derived soil moisture, was also performed (HOM). The smoothing of the soil moisture by the Barnes objectively analysis resulted in a decrease in amplitude of the soil moisture anomalies, as well as the soil moisture gradient. As a result, the surface latent heat flux (as well as its gradient) is smaller as the smoothing increases, although the domain-averaged surface latent heat flux is not altered significantly.

All of the numerical experiments produced a quasi-circular cloud shield similar to satellite observations. The soil moisture anomalies in the experiments with heterogeneous soil moisture initialization produced non-classical mesoscale circulations (NCMC), similar to sea breeze, due to differential surface heating. Qualitatively, the experiments with heterogeneous soil moisture initialization produced similar results, but in general, the coarsening of the soil moisture resulted in a higher domain-averaged precipitation rate and the total amount of accumulated precipitation in the domain as well. The NCMCs from different soil moisture anomalies collided to produce zones of convergence, initiating convection. In addition, convection initiated by the synoptic-scale convergence also interacted with the NCMC as well as the convection initiated by the NCMC convergence zones. The precipitation pattern in HOM was different than those with heterogeneous soil moisture initialization due to the lack of NCMC and its interaction with convection produced by synoptic-scale convergence.