An Assessment of Grid Resolution on Numerical Simulations of Precipitation

A common misconception in numerical weather prediction is that higher resolution produces a better simulation since smaller-scale processes are being considered. However, the fact that initial and boundary conditions often include observations at a much lower resolution than the model domain precludes the improvement of model forecasts at increasingly smaller grid spacing. This fact is especially true for variables based on mesoscale (or smaller) processes, such as precipitation, where observations are often scarce and the physical processes related to accurate prediction of the variable are complicated and/or not completely understood. This project will address the issue of numerical weather model grid resolution on accuracy of precipitation simulations by varying the resolution of a numerical grid over a static domain for a series of selected events characterized by convective and stratiform-type precipitation. The resolution will vary geometrically from 2-km to 32-km (maintaining a ratio of two), and the resulting statistical distribution of precipitation will be assessed to determine the effect on precipitation depth and extent. Additionally, the influence of interpolation from higher to lower resolution will also be tested by comparing the precipitation simulations at the varying resolutions with interpolated values from the next lowest resolution simulation. Results from this project will provide insight into the ideal resolution for precipitation forecasting using numerical weather prediction models, as well as the impact of interpolation on high-resolution model grids with respect to error and the underlying statistical distribution.