20th Conference on Weather Analysis and Forecasting/16th Conference on Numerical Weather Prediction

12.3

Implementing the Weather Research and Forecast (WRF) model with local data assimilation in a National Weather Service Weather Forecast Office (WFO)

Patrick T. Welsh, NOAA/NWSFO, Jacksonville, FL; and A. Wildman, B. Shaw, J. Smart, P. Ruscher, J. McGinley, B. N. Meisner, and P. Bogenschutz

As a single project in the multi-faceted NOAA Coastal Storms Initiative (CSI), the NWS Office of Science and Technology initiated a collaborative effort between the NOAA Forecast Systems Laboratory(FSL), the NWS WFO Jacksonville, Florida, and Florida State University (FSU) to implement a project to test a quasi-operational (real-time) local modeling effort using the WRF model with local data assimilation in the Jacksonville, Florida WFO. The model was run on a small Commercial Of The Shelf (COTS) computer cluster using an open source operating system and multiprocessor software. The high performance multiprocessor compilers were also COTS items.

NOAA FSL provided the model construction and integration expertise, WFO JAX staff provided the cluster maintenance and upgrade effort as well as the forecast use of the model, and the FSL/FSU team will present the modeling overview and verification effort in a pair of companion presentations. This presentation will discuss this effort from the WFO point of view, from inception as a NOAA COASTS project in 1998 to funding under the NOAA CSI through the WFO implementation. As in most projects, there were both pitfalls and serendipity along the way. The project development proceeded with fits and starts but rapidly coalesced when the participants were able to be physically collocated. The feasibility of the project has been clearly demonstrated.

Ultimately, the WRF model and data assimilation implementation effort succeeded, with the resulting WRF model output available to the forecaster via the Advance Weather Interactive Processing System (AWIPS) workstation within seven months of the cluster delivery. This included all software development work and automated script generation. This project can serve as a prototype WRF model implementation for a future National local-modeling and local mesoscale data assimilation effort within the NWS. While the need for such an effort can still be debated, the current effort shows that it can be physically accomplished, and at low cost, with a combination of COTS and open source hardware and software.

Some specific cases of WRF model success will be included in the presentation, as well as discussion of the impact of the WRF on local forecasts and forecaster acceptance of the model.

.

Session 12, WRF Model Development and Applications (ROOM 605/606)
Tuesday, 13 January 2004, 3:30 PM-5:45 PM, Room 605/606

Previous paper  Next paper

Browse or search entire meeting

AMS Home Page