19B.2 Probabilistic thunderstorm guidance from a time-lagged ensemble of High Resolution Rapid Refresh (HRRR) forecasts

Friday, 5 June 2009: 10:45 AM
Grand Ballroom West (DoubleTree Hotel & EMC - Downtown, Omaha)
Curtis R. Alexander, CIRES/Univ. of Colorado, Boulder, CO; and D. A. Koch, S. S. Weygandt, T. G. Smirnova, S. G. Benjamin, and H. Yuan

Since Sept. 2007, ESRL GSD has been running an hourly convection resolving model, driven by the radar reflectivity assimilating RUC, over a domain covering the NE U.S. aviation corridor. Known as the High Resolution Rapid Refresh (HRRR), the model utilizes a 3 km horizontal grid spacing configuration of the Weather Research and Forecasting (WRF) model with a diabatic digital filter initialization (DFI) radar reflectivity assimilation procedure. Plans for the 2009 convective season include an expansion of the HRRR domain to cover most of the eastern 2/3 of the U.S. and demonstration / evaluation of its utility in providing convective storm guidance for aviation and other applications.

We have recently begun using time-lagged ensemble output from the HRRR to produce experimental probabilistic thunderstorm guidance (thunderstorm likelihood and estimated echo top). Model reflectivity and convective available potential energy (CAPE) fields are used to identify regions of convection among the time-lagged ensemble members. Using the time-lagged ensemble members and information from adjacent model gridpoints, a HRRR convective probabilistic forecast (HCPF) has been developed. Verification of the HCPF is performed using the 4 km National Convective Weather Diagnostic (NCWD) product on an hourly basis. Performance of the HCPF (skill and reliability) at various lead times is being optimized by changing the weighted contribution of individual HRRR forecasts and fields and adjusting other parameters.

Additional probabilistic/ensemble fields are under development, including convective echo tops. We are also examining a variety of statistical quantities from the ensemble forecasts, including mean, standard deviation and composite fields. Standard deviation fields are being compared with mean and individual member forecast errors using observations, while composite fields provide a measure of the range of forecasts. Insights from this analysis will be used to guide the planned development of additional ensemble diagnostics, including estimates of storm mode, storm porosity, and likelihood of specific convective hazards.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner