I still don’t get it. If you follow the meteorological community or weather enthusiasts on Twitter, there is a level of hyperventilation (and vitriol) at times when it comes to the debate about what weather model is better: European “Euro” or American. It is well documented that the European model (run by the European Center for Medium-Range Weather Forecasts or ECMWF) has historically performed better than the American model, Global Forecast System (GFS), run by NOAA. In 2013, Jason Samenow wrote in the Washington Post after Hurricane Sandy:
Last year, criticism began to emerge concerning the inferior accuracy of the NWS’s Global Forecast System (GFS) model – run on earlier versions of the supercomputers – compared to the model run at the European Centre for Medium Range Forecasting (ECMWF) based in the United Kingdom. The GFS and ECMWF models are, by far, the most heavily relied on by meteorologists around the world for forecasting.
Backlash from Sandy heightened the urgency of U.S. policymakers to improve U.S. model capabilities. Supplemental funding from Congress and continued model development by NOAA has delivered the “new” American model called FV3-GFS. This week NOAA announced that the new model, FV3-GFS, is scheduled to be operational on March 20th. The “operational date” was actually delayed by the recent government shutdown. What does this the new operational GFS model mean for U.S. weather prediction and the “debate”?
The debate will rage on for certain. The weather community, particularly its most vocal components, loves to debate. I wrote a piece several years ago in Forbes comparing the performance of the European and American models. The reality is that all models have their strengths and weaknesses in certain situations. It is borderline comical to watch the banter about weather models. I often wonder when I missed the invitation to join a model fan club. I take a more objective view of “the model wars” because I understand that the best forecasters consider all of the available models and are not locked in a “Numerical Weather Predication Arms Race.” While President of the American Meteorological Society, I had dinner with the Director of ECMWF while visiting Reading, U.K. He literally chuckled at how animated the debate in the U.S. is concerning the “Euro” vs “GFS.” He even pointed out that the GFS beats the European model often.
The European model has an different way of integrating data into its model. That integration very much depends upon U.S. and other weather-climate satellite assets. I discuss their methodology at this link and why the Euro model has generally performed better. The more relevant point that I am trying to make is that there is collaboration not animosity between the modeling centers.
So what’s the deal with the new American GFS model going operational in March? In a previous NOAA press release, National Weather Service Director Louis Uccelini pointed out,
The new dynamic core, Finite-Volume on a Cubed-Sphere (FV3), was developed by NOAA’s Geophysical Fluid Dynamics Laboratory in Princeton, New Jersey. The FV3 core brings a new level of accuracy and numeric efficiency to the model’s representation of atmospheric processes such as air motions. This makes possible simulations of clouds and storms, at resolutions not yet used in an operational global model.
Just like a phone camera with more Megapixels resolves a picture with better clarity, higher resolution in a model can resolve more weather processes that may have been missed or poorly sampled. A good overview of the new model can be found here.
Even though the model is going operational in March, NOAA has actually been running the “old” and “new” models for the past year or so. Eric Berger wrote this in ArsTechnica last November:
The top-line finding is that yes, the FV3 core generally performs a bit better than the GFS model….. it is nice to see some modest improvement from the FV3, but it still is not the best in the world. For the entire globe, the FV3 model would still rank behind the best-in-class European model (0.910) and the United Kingdom Met Office (0.887).
Berger pointed out that the upgrade brings the U.S. from fourth to third place. Improvements were observed in how GFS intensifies tropical cyclones and represents track within the first five days, according to Berger. I highly recommend Berger’s article because he provides details on the metrics used in the comparison. He also highlights weaknesses in the “new GFS” such as a dry bias for high-impact precipitation events and a tendency to be too cold with nocturnal temperatures.
I look forward to seeing where the new GFS model goes. I also like that unlike the Euro model, it will remain freely available to the public rather than behind paywalls. Some will whine that third place is not good enough. Frankly, pushback (without the snark factor) from the scientific and user community is valuable because it keeps the pressure on everybody to improve the models. Emerging model capabilities from private companies like IBM and Panasonic will also keep the pressure on.
As an optimist, I see the “American model” trending in the right direction and am willing to see how it further evolves. NOAA also has other weather models like the HRRR and RAP that are used daily so the concept of a “singular” American model is flawed. At the end of the day, if all of the models get better, the public wins.