JACKSONVILLE, Fla – The Nation's premiere forecast model will go live on July 12, with an upgrade intended to improve weather forecasts around the world.
NOAA’s “next-generation” weather model called the GFS-FV3, replaces the old Global Forecast System (GFS) with an overhaul to its core processing backbone.
The accuracy and efficiency in how the model represents future atmospheric conditions will be improved in a way that's similar to increasing camera pixels to improve photo details.
This higher resolution will allow finer simulations of clouds and storms across smaller portions of the planet.
Capturing weather processes at very small spatial scales, should provide a major leap forward in US weather prediction capabilities, leading to improved prediction of extreme storms.
Over the past three years the new FV3 has been running in evaluation mode and has shown equal or improved forecast skills in many areas including, forecasting the atmospheric pressure at the base of the jet stream and daily precipitation cycles over the United States (*among a list of other improvements).
The rollout was delayed by errors in overestimating the impacts of cold and snow but those bugs have until next winter to be resolved.
Euro Vs. GFS and now FV3 & fvGFS: The Winner is?
The GFS and Europe’s ECMWF (EURO) model are, by far, the most heavily relied on by meteorologists around the world for forecasting.
The successful European forecast model has paved the way for improvements in how the new FV3 needed to be retooled.
According to NOAA’s Geophysical Fluid Dynamics Laboratory which designed the FV3 core, it has the potential to produce better hurricane forecast skill than the world‐leading European model.
The Princeton, New Jersey based lab, runs a forked version of the FV3 for research evaluation called the fvGFS.
It showed better track forecasts for the 2017 Atlantic hurricane season compared to the best performing ECMWF operational model and it greatly improved the ECMWF’s poor track forecast for Hurricane Maria.
For Hurricane Irma, which was well predicted by the ECMWF model, the fvGFS still had slightly lower 5-day track forecast errors.
Forecasting TC intensity has remained a challenge for all global models, and the fvGFS also shows better model predicted storm intensities than both the old GFS and the ECMWF operational models.
Debate among meteorologists as to which model is better has swirled in the community for years but the EURO often comes out on top.
One of the main drivers in the ECMWF’s forecast skill is how it continuously assimilates current weather conditions in an analysis.
This representation is used as the starting point for forecasts. The quality of the analysis is thus crucial for successful weather forecasts.
The new FV3 framework will take a similar approach feeding new observations into the model almost right up to the start of the forecast calculations.
We will soon find out if the improvements will bring the reconfigured American model to the skill level of the EURO.
Real‐time 10‐day fvGFS forecasts are run every 6 hours at GFDL and are publicly available online and the GFS-FV3 here.
*A portion of the last hurricane season was tested from August 26, 2018, through October 31, 2018 with promising results. Benefits include less deepening of intense tropical cyclones which was overestimated with the old version. Multiple erroneous tropical cyclone centers generated by GFS not seen in new GFS and tropical cyclone track forecasts improved within 5 days. The nested HWRF runs also get new simulated composite reflectivity output as a nice addition.