Doppler weather radars observe air flows and precipitation intensity within thunderstorms. Computer models run for the HWT Spring Experiment incorporated observational data from more than 120 weather radars enabling the most realistic storm prediction to date.

"The powerful Cray XT3 system was designed for the highest scalability and reliability, which allows researchers to incorporate a greater and greater number of variables into their models to create increasingly accurate computer simulations and shed light on scientific phenomena that we can't explain today", stated Ian Miller, senior vice president of sales and marketing for Cray. "When advanced high-performance computing meets sophisticated field research, the result is this kind of scientific breakthrough that has the potential to advance science and provide tremendous benefit to society."

This milestone marks the second consecutive year of "firsts" in the HWT Spring Experiment runs. Last year, researchers from CAPS ran the Weather Research and Forecasting (WRF) model faster than ever before on the Cray XT3 supercomputer at PSC. This year, researchers were able to incorporate radar data into the WRF model using software developed by CAPS for the first time ever at such resolutions and scale using the same Cray system.

"The Cray XT3 system at PSC provided much greater computing power than the allocation available at the National Weather Service operational forecasting centre, allowing us to test and refine storm-scale 'ensemble' forecasting, which was a key goal of the experiment", stated Kelvin Droegemeier, vice president for research at the University of Oklahoma and director emeritus of CAPS. "Ensemble forecasting involves running a forecast model multiple times to assess the degree of uncertainty inherent in the forecast. Since it requires running a model many times within a short time period, ensemble forecasts demand large amounts of computational power, and Cray answered that call."

Each day during the seven weeks, CAPS scientists transmitted weather data to the Cray supercomputer. Running a 10-member ensemble - 10 different configurations of the forecast model - covering nearly the entire continental United States with one grid point every 4 kilometers, the Cray system produced forecasts for the next day and transmitted the data to researchers as they were produced. In addition to the 10-member ensemble runs, CAPS also ran on the Cray XT3 higher-resolution forecasts with one grid point every 2 kilometers covering the same area.

"The finer resolution of this run better captures the structure of thunderstorms", stated Ming Xue, director of CAPS. "The result is improved accuracy in the forecasting of severe storms and tornado potential. The Spring Experiment is an iterative process and each improvement equates, hopefully, to diminished loss of life and economic damage. The goal is to build each year on our growing knowledge of storm prediction and forecasting. Our efforts should help accelerate the implementation of advanced prediction capabilities at the national operational weather forecasting centre."