High-resolution models that are able to explicitly resolve convection have the potential to greatly aid forecasting for aviation concerns, where thunderstorms can have a huge impact. Current high-confidence thunderstorm forecasts are generally restricted to the range covered by extrapolation of a given thunderstorm complex, which can extend predictability in some cases up to an hour or so into the future. At this time operational models are not run at sufficient resolution to explicitly resolve convection, and are thus generally not able to add forecast confidence to a deterministic prediction of convection, although they certainly contribute to probability forecasts (such as a chance or Prob30 group in a TAF). Over the last few years, convection-resolving high-resolution models have shown potential to increase the predictability of thunderstorms. A number of these models have been run in association with forecast experiments, such as the NOAA NSSL/SPC Hazardous Weather Testbed (HWT) Spring Experiments. There are different methods used to initialize these high-resolution models, including the use of a coarser domain model (as is currently done for the experimental High-Resolution Rapid Refresh Model (HRRR), run at NOAA/GSD). Also at NOAA/GSD, the Forecast Application Branch has a long history of using a locally-run analysis to initialize a high-resolution model, using a system known as LAPS (Local Analysis and Prediction System). With increasing computing power, efforts have expanded to expand LAPS to a regional and even CONUS domain at high resolution, as well as at high temporal frequency. Such LAPS analyses were tested for the 2011 HWT experiment at 15-min temporal resolution and at a horizontal grid resolution of 1 km on the Regional Scale and 3 km on the CONUS scale. The analyses were used to initialize 3-km horizontal grid resolution WRF model runs every 2-h on the Regional Scale and once per day on the CONUS scale. The focus of the regional-scale runs was on the short-range period, with the runs extending out to 6-h. For the NOAA/AWC Aviation Weather Testbed (AWT) Summer 2011 experiment the domain was shifted to include the focus area of the Golden Triangle. In addition, an ensemble of runs (using different physics configurations as well as perturbations to create the ensemble members) was tried for each domain (these are discussed in a separate abstract). Also, a different analysis method known as STMAS (for Space and Time Multiscale Analysis System) was run at time and space resolutions similar to LAPS, and also used to initialize regional-scale WRF model runs for both the HWT and AWT domains. In this paper we will focus on the various deterministic runs that were made for each domain, with an emphasis on demonstrating whether an advantage in short-term forecasting of convection is obtained by initializing the model with a high-resolution analysis. Forecasts of simulated radar reflectivity will be compared with observed radar and with simulated radar reflectivity from the HRRR, which is currently run at hourly intervals on the CONUS scale and at a 3-km horizontal grid resolution. We will illustrate our results with several case studies from each experiment to demonstrate the potential for thunderstorm forecasting for aviation concerns.
This publication was presented at the following: