This week I was in Boulder, Colorado at the National Center for Atmospheric research (NCAR) for the annual WRF model workshop. WRF stands for the Weather Research and Forecasting Model (pronounced WORF--like the security officer on the U.S.S. starship Enterprise). The WRF model is the main high-resolution weather simulation/forecast model used by the research community and most of the local weather predictions you see on the Atmospheric Sciences department web site (and which I show on this blog) are from WRF. It was designed to simulate down to ultra high resolution (e.g., the small turbulence eddies near the surface), but can handle systems thousands of miles across as well. Thousands of researchers and operational centers use WRF...an open, community system. Sounds good so far, right?
Now here is the problem. When WRF was developed in the late 1990's, ghe central idea was that WRF would represent a change in the way in the way numerical weather prediction was done in the U.S. In the the 1970s, 1980s, and 1990s the research community and the National Weather Service were using different computer models. So the insights of the research community were not improving the National Weather Service computer models...which were not as good as they could or should have been. In the mid-1990s...the idea that the NWS and the research community, centered at NCAR, would develop a next generation model and everyone would use it. Research results would flow into operations and students at universities could move to the NWS already experienced with the modeling system. It all started well but during the last few years it has all collapsed. Everyone talks of the divorce. The NWS has essentially decided that it could not use a modeling system developed partially or totally somewhere else and have gone ahead with developing a separate model and modeling system infrastructure. In the meetings this week, it became clear they also wanted to go it alone on their data assimilation system (the software used to analyze all the observations used for weather prediction). There were about 300 people at the meeting, only one was from the National Weather Service. Really sad.
The isolation of the National Weather Service has increasingly resulted in American operational numerical weather prediction falling below the standards of the rest of the world. We are not number one. We used to be number two in global prediction. Now we are maybe fourth or fifth. Don't get me wrong. The local weather forecasters are great and experienced. But they are crippled using models and software tools that are hardly state-of-the-art. And this is driven by an isolated, not-invented-here we don't want it, we know better attitude at the Environmental Model Center (part of the NWS) in Washington DC. The same attitude the delayed the coastal radar for ten years. Perhaps one day, enough people will understand what has happened and demand better, or perhaps our nations congressmen and senators will demand better. I hope so. The U.S. has the best and deepest meteorological research community in the world. We should have the best numerical weather predictions...and we don't. Not even close. With state-of-the-art numerical models and software weather predictions could be much better...saving lives and property. This needs to be changed.