The National Weather Service is testing a new weather model it hopes will replace a few older models. It hopes this new model with further increase weather model accuracy .
Over the last 40 years our weather computer models have been increasing in resolution. Back in the 1980s our weather models had a resolution of around 50 miles by 50 miles. This meant one 50 mile by 50 mile square could only have one type of weather in it- rain or snow or any other precipitation type. Of course we know there are much smaller zones of the various precipitation types in the winter storm.
NOAA’s newest model they hope to make operational in 2026 will have a resolution of 1.8 miles by 1.8 miles. This means there will be about 525 smaller squares in the 1980s 50 mile by 50 mile model. This is a good way to understand why resolution is so important to us meteorologists.
Of course model resolution has been increasing over the years. The most current, great short-term model we use, called the HRRR, has a resolution of 2.5 miles by 2.5 miles.
HRRR stands for High Resolution, Rapid Refresh.
The new model is called the Rapid Refresh Forecast System, or RRFS.
Jacob Carley, engineering and implementation branch chief at NOAA’s Environmental Modeling Center, also says the vertical resolution will be much higher. Right now the HRRR has 50 layers in the vertical. The RRFS will have 65 layers. More layers going from the surface to the very top of the atmosphere could be able to better forecast mixed precipitation of snow, sleet, freezing rain and rain. This is a key component of a winter storm forecast. More vertical layers should help with warm-season severe thunderstorm forecasting.
The new RRFS will have an architecture that allows any weather agency around the world or any university to have the chance of contributing a module in the model. For example, if the University of Oklahoma came up with a modeling scenario that forecasted severe thunderstorms better, the module could maybe be implemented in the RRFS. Carley says this brings the chance of more scientific collaboration. The architecture is called the Unified Forecast System.
The RRFS will cover all of North America, which is a larger coverage area than the current HRRR. Rapid refresh says just what it means. We will get a new model run every hour. Most of the runs will go out to 18 hours in advance. A run every six hours will forecast 60 hours in the future.
Carley targets one reason the new model should have increased accuracy: The RRFS is going to have the newest and best initial starting conditions. The current observations at all levels in the atmosphere will be more accurate. To have a great forecast hours later, we need an accurate depiction of the starting conditions.
In fact the accurate starting conditions are one of the main reasons the European Model is know to be the most accurate model out several days in the future.
The model will have data from the new GOES satellites imported into it. The model’s starting conditions will also incorporate ground radar data.
The new RRFS has to be tested thoroughly before becoming operational. Carley says the model will be tested by various users in spring 2025. If the accuracy looks good the new model could start to be used by the National Weather Service in 2026.
Carley hopes the new model will increase the accuracy of forecasting key weather elements like severe weather, hazardous winter weather, heavy rainfall and extreme weather. The model will also forecast smoke and dust in the air.