NSSL Operational WRF Background
Since 2003, SPoRT has provided real-time WRF model runs to the Huntsville NWS. In 2006, a Congressional earmark established a collaborative effort between SPoRT and the Hazardous Weather Testbed (HWT) in Norman, Oklahoma to support real-time modeling. This seed money helped procure a high-powered computer (a 64-processor SGI Altix 4700) at the National Severe Storms Laboratory (NSSL) on which a real-time model could be run, and SPoRT provided staffing to support the endeavor. The real-time model forecasts--performed at NSSL--are supplied to the Norman and Huntsville NWS offices to supplement model output from the National Centers for Environmental Prediction (NCEP). The forecasts are produced out to 36 hours once daily with a 4-km domain that covers the continental United States (see figure below). Specific details on the model configuration are provided in the table below.
To view real-time WRF model output, visit NSSL's website.
WRF model output on NSSL's website
| Version | 2.2 |
| Dynamic core | Advanced Research WRF (ARW) |
| Horizontal grid size | 980 x 750 |
| Vertical levels | 35 |
| Horizontal grid resolution | 4 km |
| Initial and lateral boundary conditions | NCEP Eta 212 grid |
| Computational platform | SGI Altix 4700 (64 procs) |
| Simulation length | 36 hours |
| Time step | 24 seconds |
| Cloud microphysics | WSM6 Scheme |
| Shortwave radiation | Dudhia Scheme |
| Longwave radiation | RRTM Scheme |
| Land surface physics | Noah Land-Surface Model |
| PBL physics | MYJ scheme |
In addition to aiding in the transition of real-time model forecasts to the NWS, SPoRT has been an integral part in producing new diagnostic variables that have been applied to the explicit convective forecasts within the NSSL WRF. These new variables compute the maximum values of selected diagnostic fields at every model time step and grid point in order to evaluate the evolution of convection at time-scales smaller than the standard model output intervals.
These diagnostic fields consist of:
- Maximum updraft and downdraft velocities between the surface and 400 hPa, which have obvious implications regarding the intensity of convective overturning.
- Maximum simulated reflectivity at the lowest model vertical level (a composite could also be generated taking information from other levels). As with updraft and downdraft velocities, this field is related to the intensity of convection and may provide some guidance for forecasting large hail.
- Maximum 10-m wind speed, which may be helpful in predicting the magnitude of convectively-induced wind gusts.
- Maximum updraft helicity, which can be used to detect mid-level mesocyclones in simulated convection (Kain et al. 2008a).
These fields are useful for diagnosing maximum values, as well as providing valuable clues about storm tracks, wind swaths, and the longevity of individual features. For example, the 4-km model can produce long-lived (multi-hour), strongly rotating mesocyclones revealed in hourly output as downstream-marching segments of high maximum updraft helicity. With relatively simple post-processing tools, these segments can be concatenated to reveal features such as storm tracks or wind swaths (Kain et al. 2008b).
Finally, SPoRT is testing unique NASA capabilities using the framework of the operational NSSL WRF. Specifically, SPoRT is conducting a sensitivity study for specific severe weather cases using the new NASA Goddard Space Flight Center (GSFC) microphysics, GSFC short/longwave radiation schemes, NASA Land Information System (LIS; SPoRT WRF LIS Page), and SPoRT MODIS sea surface temperature composites (SPoRT WRF SST Page). Preliminary results from a 28 March 2007 tornado outbreak over the U.S. Central Plains revealed that the updated Goddard shortwave radiation scheme improved surface heating and reduced low clouds, which led to improved timing and intensity of the severe convection compared to the operational NSSL WRF configuration. In this case study, adding the LIS land surface initialization and Goddard longwave radiation further improved the location and intensity of predicted convection (Case et al. 2008). These schemes are still being tested and have not yet been implemented into the operational version of the NSSL WRF.
