High resolution computer forecast models are increasingly becoming part of the forecast process at National Weather Service offices around the country. These sorts of forecasts and simulations can be valuable in a variety of different situations - from severe weather outbreaks to the local effects of terrain on temperature and wind.
When examining high-resolution model forecasts in severe weather situations, one of the things forecasters will usually examine is "simulated reflectivity". This field is basically the model simulating what the radar would look like at a specific level above the ground. In addition to helping forecast the timing and location of thunderstorms, this can also help forecasters determine the type of storms that will form - anything from a large squall line to isolated supercells. This can have important implications for the type of severe weather that is expected.
Here are several examples of images from a high-resolution forecast model for June 17, 2010. The images are from OU/CAPS. This was a 1-km WRF model forecast that started at 7pm the preceding evening. On the left is the "simulated reflectivity" composite for 5pm on June 17th, or basically what the radar would look like. On the right is a forecast of "updraft helicity" at the same time. This basically indicates areas where the model is generating strong rotation in storms that develop.
As you can see, this model forecasted an arc of thunderstorms to be in progress from eastern North Dakota into western Minnesota by 5pm. The brightly colored blips in the image on the right show that the model was also forecasting the potential for some strong rotation in some of the storms.
Discussion from Chris Franks and Tom Hultquist (NWS Chanhassen, MN):
Viewing June 17th retrospectively give meteorologists a chance to learn why events unfolded as they did, but it is also a great case to test new technology with the hope for better prediction down the road. High resolution numerical modeling shows great promise in showing when, where and what time severe weather can be expected. However, in order to test and continually refine a model's performance, we need to simulate weather events such as the June 17 tornado outbreak many times. This type of research is done in government offices, private companies and academic institutions all around the country. The National Weather Service in Twin Cities/Chanhassen, MN has been utilizing technology maintained by the Minnesota Supercomputing Institute (MSI) for Advanced Computational Research to do this type of research. Recently, we ran several dozen numerical model simulations of the weather on June 17 each time with a slightly different configuration. All of the simulations taken together are known as an ensemble, and it can be a valuable tool for forecasters. The forecasters can use the model information to aid in decision making for potential high impact events. It is important to note that this sort of numerical modeling is relatively new technology, because it is extremely computationally expensive to perform these simulations and computers have only been powerful enough in the last several years.