The medium-range global weather model run by the European Center for Medium-Range Forecasts (ECMWF) is slated to be upgraded to Cycle 48r1 late June 2023. Upgrades or replacements of weather models are common as improvements in computing power, observation data, and other tools are implimented. Cycle 48r1 is scheduled to replace Cycle 47r3 in late June, the latter having come out in October 2021.
There are two significant and exciting improvements to the ECMWF model coming in this update which will bring strong forecast improvements. First is the coverage of satellite data that is injested into the system. Among its vast array of inputs, the model pulls in microwave satellite data. This microwave radio band is able to penetrate clouds to show the model what is happening at the surface.
Cycle 48r1 is able to pull data from up to 30% more of the planet’s surface (depending on the specific channel) than previous versions as modelers have developed ways to have the computer make sense of data from sea ice and some land cover. Noatably, this opens up a lot of new data from the polar regions. ECMWF observed significant forecast improvements in the Arctic during testing because of this. Better forecasting of Arctic conditions will give greater understanding of weather patterns in the mid-latitudes as well.
ECMWF’s model uses a system they call 4D-Var to input current weather conditions when the model begins running. It creates a gridded output for current (and recent) weather that is being improved from a 50 km grid to a 40 km grid. Of particular note is how this is applied to tropical cyclones, which have conditions that vary significantly over relatively small areas.
The example pictured above shows a tropical cyclone with a minimum central pressure of 946 hPa. In the 50 km version of 4D-Var, the lowest air pressure value on the grid was 967 hPa while the 40 km version had a grid point much closer – 951 hPa. Starting with more accurate data will drive a better model output and help meteorologists to better forecast the timing, path, and strength of high impact weather events like tropical cyclones. Further improvements to 4D-Var’s resolution will require better data, more efficent computing, and stronger supercomputers.
Storm Vista reports that the accuracy of Cycle 48r1 shows more improvement than is usually seen with model updates like this. Their figures show it has especially improved in the ensemble guidance across the entire forecast period, which runs 360 hours into the future.
NOAA is also planning a major model update – a new high-resolution mesoscale model for North America called the Rapid Refresh Forecast System (RRFS) expected late 2023. RRFS is currently running in its test phase and will replace NOAA’s existing mesoscale models (HRRR, HREF, and NAM nest) when it becomes operational.
The European Center for Medium-Range Forecasting is an independent organization for research and modeling supported by 35 nations in and near Europe. More details about the 2023 model upgrade, including specific forecast improvement information, can be found in a recent article from the center.
The featured image is from the ECMWF.
