Real-time storm-scale data assimilation and forecasting experiments for NOAA's Warn-on-Forecast Project

The NOAA Warn-on-Forecast (WoF) Project is tasked with developing a regional 1-km storm-scale prediction system for the United States that assimilates radar, satellite, and conventional (e.g., surface) data.  The proposed WoF system, to become operational sometime in the next decade, will generate new 0-3 h probabilistic forecasts 3-4 times an hour, for the purpose of predicting hazardous weather phenomena, such as thunderstorm rotation, hail, high winds, and flash flooding.  A prototype system—known as the NSSL Experimental Warn-on-Forecast System for ensembles (NEWS-e)—has been developed in collaboration with the Global Systems Division (GSD), and is based on the Weather Research and Forecasting (WRF) model with 3-km (1-km) horizontal grid spacing for real-time (retrospective) experiments.

This year, the NEWS-e was run in realtime each day from 2 May – 3 June, which coincided with the NOAA Hazardous Weather Testbed 2016 Spring Forecast Experiment.  The starting point for each day’s experiment was a 3-km, hourly cycled HRRRE under development at GSD.  The 1500 UTC forecast cycle from the GSD ensemble provided initial and boundary conditions for the NEWS-e, a 36-member ensemble covering a 1000-km wide region with very frequent (sub-hourly) updates.  The daily NEWS-e domain location targeted the primary region where severe weather was anticipated.  Radar reflectivity and radial velocity, satellite (cloud water path retrievals), and surface data were assimilated every 15 min using the ensemble Kalman filter (EnKF) approach encoded in the Data Assimilation Research Testbed (DART).

This study is primarily focused on the short-term (i.e., 90-min) ensemble forecasts initialized from the resultant storm-scale analyses every 30 minutes during each hour of the real-time experiment.  Preliminary work will evaluate the ability of model-derived measures of low-level (e.g., vorticity) and mid-level (e.g., updraft helicity) storm rotation to anticipate tornadic supercell thunderstorms and mesoscale convective systems.  These results will be further compared to similar measures derived from the HRRRE, as well as the HRRR model (which is commonly used for operational storm-scale weather prediction).