The recently developed Model for Prediction Across Scales (MPAS) was used to produce global forecasts in real time during August and September of 2013. MPAS is a non-hydrostatic code, integrated on an unstructured Voronoi mesh using a C-grid discretization appropriate for mesoscale flows. The unstructured mesh allows for a gradual transition between relatively coarse resolution over most of the globe, to higher resolution in a region of interest. In this talk, we will examine the overall performance of, and the relative differences between three sets of forecasts. One set integrates 10-day forecasts on a quasi-uniform 15-km grid. The other integrates forecasts of the same length on a variable mesh that ranges from 60-km grid spacing over most of the globe to 15 km over the tropical Pacific. The third set of forecasts places the high-resolution only over the Atlantic Ocean. Statistics of typhoon prediction are compared with the NCEP GFS model, and it is shown that MPAS exhibits comparable or greater skill over the Western Pacific basin. Analysis of environmental wind, temperature and moisture fields indicates a high correlation between the fixed and variable resolution over the Western Pacific, and overall lower biases compared with the coarse resolution MPAS.
Western Pacific Tropical Cyclone Prediction in the Model for Prediction Across Scales (MPAS)