Extended-Range Severe Weather Guidance Using a Global Convection-Permitting Model (Invited Presentation)

Operational NWP models configured with convection-permitting horizontal resolutions are used to produce forecasts only over relatively short periods, typically 2 days or less. The rationale for the short forecasts are their significant costs, limitations in the technology used to produce regional forecasts (e.g. problems with grid nesting), and questions concerning what is predictable at convective scales and at forecast times greater than a few days. In order to examine the possibilities for extended-range convective guidance, we are employing the Model for Prediction Across Scales (MPAS) that uses variable-resolution meshes to economically span hydrostatic to nonhydrostatic scales for global forecasts. Importantly, the MPAS unstructured Voronoi horizontal mesh (nominally hexagons) is smoothly varying, thus eliminating many downscaling problems encountered using one- or two-way traditional grid nesting. In contrast to the one-way nesting used in most operational NWP applications, the global variable-resolution MPAS configurations allow for dynamical upscaling, i.e. the global forecast accuracy may benefit from the increased accuracy in the high-resolution region. In order to begin examining the utility of extended convection-permitting forecasts within a global framework, we have produced five-day forecasts for the 2015 Spring Forecast Experiment conducted at the NOAA Storm Prediction Center (SPC) using MPAS employing a variable mesh with 3 km mesh spacing over North America. These forecasts were conducted over the month of May, and we are evaluating them as part of SPC's ongoing efforts to provide extended range guidance for severe weather out to day five. Preliminary results from the MPAS forecasts indicate that there is some skill in capturing convection in the CONUS region in both the near (days 0-2) and extended (days 3-5) range. We also see indications that the skillfulness of the central US convective forecasts is linked to specific flow regimes. There were two significant periods of strong convection, including tornadic storms and heavy rainfall leading to widespread flooding in the south central United States (primarily in Oklahoma and Texas), that occurred during this period, and the MPAS forecasts initialized from the GFS analyses were able to capture the convective events in most cases, even at the longer (3-5 day) lead times. We will present results highlighting these forecast findings along with an overview of the key features of the MPAS model.