Lessons learned from the DTC Winter Forecast Experiment

One of the most significant accomplishments of the Developmental Testbed Center (DTC) this past year was the establishment, conduct, and evaluation of the DTC Winter Forecast Experiment (DWFE). Two configurations of the Weather Research and Forecasting (WRF) model were run out to 48 hours on 5-km grids covering large (CONUS) domains. This collaborative effort involving the operational and research communities included: forecaster training; three-dimensional, full-resolution displays of model fields using FX-Net; dissemination of selected two-dimensional fields for display on AWIPS and the DTC website; statistical verification and online forecaster evaluations; and innovative forecast products.

The DWFE provided valuable experience for the DTC in running major multi-organizational predictive experiments, including how to more effectively present and examine the very detailed forecast fields from these experimental models. Lessons were also learned about how better to serve the needs of the research and operational WRF model communities, including balancing the desire of the model developers to make improvements to model configurations during the course of the experiment, without compromising the statistical results of the verification. Some of the more traditional ways of looking at model forecasts, both in terms of the kinds of fields displayed (e.g., unfiltered geopotential height and vertical velocity fields) and model diagnostic fields (such as frontogenesis and quasi-geostrophic forcing) were found to be of much less use compared to more innovative displays of the forecast fields – including derived radar reflectivity fields estimated from the model microphysics, surface temperature and wind fields at maximum resolution, precipitation type distributions, and precipitable water fields. We will present examples illustrating the kinds of exciting mesoscale phenomena that were predicted by these high-resolution models (run without a convective parameterization), as well as mesoscale verification of such forecasts (to the extent that was possible). Finally, we will summarize how lessons learned from DWFE might benefit future NWP field experiments from the planning stage all the way to improving the value of feedback to both forecasters and model developers.