NOAA Improves Weather Forecasting Model
The National Oceanic and Atmospheric Administration (NOAA) is making strides toward improving the predictive capability of its weather forecasting software. On June 12, 2019, NOAA upgraded its proprietary Global Forecast System (GFS), which is expected to increase the accuracy of forecasts as far out as one week into the future. This is obviously fantastic news – a better weather prediction model is yet another bullet in the chamber for utility emergency planners.
Nuts and Bolts of the New & Improved NOAA Weather Forecasting Model
The NOAA model is called Finite-Volume Cubed-Sphere (FV3). The strange ‘squared-circle’ part of the name aside, the enhancement involves the use of satellite imagery to improve the model’s predictive accuracy by factoring in variables like jet stream patterns, atmospheric moisture, cyclone activity, and physical properties of snow and rain.
The impetus of the upgrade was Hurricane Sandy, which the NOAA model predicted at the time would be relatively harmless. We all know how that turned out, and the hope is that the new model will prevent such grossly inaccurate forecasts going forward.
In order to develop the new model, NOAA deployed a system upgrade that bolstered both performance and storage capacity by over 50%. The upgraded model will run in conjunction with the old model for approximately 3 months to enable the collection of additional testing data.
At first, I was shocked to learn that this is the first weather model update NOAA has deployed in 40 years! But after I learned more about the process, it started making more sense. The June 2019 upgrade involved a whopping 100 workers from all across the US and encompassed the comparison of real time modelling with 3 years’ worth of historical weather models.
Yes, it’s been a massive amount of work, and it’s not even close to being done. The June deployment can best be viewed as phase 1 of “x.” NOAA plans to continue to tweak, refine and improve the process of both collecting relevant data and analyzing it, to facilitate gradual improvements over time. But even this first upgrade should make the lives of emergency planners at least slightly better!
