The supercomputers in Virginia and Arizona that forecast weather around the world, nicknamed “Dogwood” and “Cactus” respectively, received a huge upgrade in late June – to the tune of three times as fast as the previous computers used.
The new computers operate at a speed of 12.1 petaflops each – which means 1.21 x 1016 floating-point operations per second. That also illustrates how incredibly complex the atmosphere is that a computer with that type of speed is needed to make approximations of the atmosphere into the future in a short enough time for the forecasts to be useful. These are now the 49th and 50th fastest supercomputers in the world.
This increase in computing speed and power should allow the National Weather Service (NWS) and other entities of the National Oceanic and Atmospheric Administration (NOAA) to provide more detailed, and potentially somewhat more accurate, weather forecasts further in advance. It should also allow them to run more ensemble and high-resolution forecast models to capture full ranges of potential weather and small-scale features such as thunderstorms and mechanisms that can cause tornadoes on mesoscale levels.
The Global Forecast System (GFS) is planned to be upgraded this fall as a result, and the new Hurricane Analysis and Forecast System (HAFS) is expected to be in full operation for the 2023 hurricane season. Of course, the experimental HAFS already appeared to forecast Hurricane Ian decently well in late September.
NOAA also has research and development supercomputers in Colorado, Tennessee, Mississippi, and West Virginia, bringing its total supercomputer capacity for research and forecasting to 42 petaflops.
See the full press release from NOAA here. At ForecastWatch we will be on the watch for potential improvement in weather forecasts starting later this year if all goes as hoped!
ForecastWatch analyzes accuracy of wind and temperature forecasts that can have impacts on energy demand and supply. Our data offers the ability to request customizable collections – by location, time period, and forecast parameters, such as wind and temperature – and have them conveniently delivered as easy-to-use CSV files. Contact us for more information.