29 June 2017

U.S. About to Fall Further Behind in Long Range Weather Forecasting

Posted by Dan Satterfield

Europe is testing an improved version of its already superior global numerical weather model.

The Europeans are now testing a new version of their ECMWF model with a resolution of around 9 km, and so far it looks very good. Not that the present model isn’t since we forecasters now depend on it heavily for long range forecasts beyond 2-3 days. The NOAA GFS model is almost always the least accurate and that is not my subjective opinion. Look at the skill scores below between both models.

Below is the NOAA GFS model, and you can see that it scores consistently lower.

The proposed federal budget cuts funding for modelling at NOAA severely, but on the bright side we now have GOES-16 up and it is amazing. That said, Japan and Europe beat us by several years in launching a modern 21st century satellite. One really bright spot is the short range modelling, where NOAA’s High-Resolution Rapid Refresh Model is almost stunningly good, at least here on the Delmarva, but it only goes out to 18 hours.

When it comes to long-range global modelling though, we can and MUST do better.

Note: The ECMWF model is something many TV stations now pay extra for. The data is proprietary which is why I posted an image that has already verified.