In the 1970s and 1980s, numerical weather prediction had reached a point where it was being increasingly used for operational forecasting, but would often fail to predict high impact weather events. Of particular note were nor'easters, strong extratropical cyclones that develop along the U.S. east coast and can bring crippling snowfall to metropolitan areas.
A look back at the literature during that time reveals many papers describing forecasts busts. Lance Bosart wrote an especially important paper on the President's Day Snowstorm of 18-19 February 1979, which affected the mid-Atlantic states. He noted that "the storm is noteworthy because of the failure of operational prediction models to signal the intensity of the event."
 |
| Source: https://journals.ametsoc.org/view/journals/mwre/109/7/1520-0493_1981_109_1542_tpdsof_2_0_co_2.xml |
We've come a long ways since then, but challenges remain today, including the details concerning the track of the system, distribution and intensity of precipitation, and in some storms, the type of precipitation (rain, snow, mixed).
Below is the GFS forecast for 0000 UTC 30 January showing an intense extratropical marine cyclone off the cost of New England. This storm has been forecast by our numerical modeling systems for several days now. It will develop off the southeast and mid-Atlantic coast beginning tomorrow and deepen explosively as it moves northeastward. It has all the markings of a very dangerous marine cyclone, including a bent-back occluded front (purple front in the image below) that wraps around the low center and features a highly concentrated pressure gradient where strong winds and extreme seas are often found.
Norwegian meteorologists refer to this region as "the poisonous tail of the bent-back occlusion." It's a good place to avoid if you're on the high seas!
The challenge in this forecast is the band of precipitation north and west of the system. Precipitation rates in the center of the band are quite high, but drop off quickly as one moves westward (color fill above is 6-hour accumulated water equivalent precipitation). Thus, the location is absolutely critical for snowfall prediction as the system moves up the west coast and there are a wide range of possibilities depending on track.
The latest messaging from the National Weather Service Weather Prediction Center illustrates the probabilities of snowfall in excess of 4 and 8 inches and shows the sharp drop off in the likelihood of such amounts as one moves inland.
If you live in the mid Atlantic and northeast coastal areas, monitor official forecasts as forecasts evolve and the storm approaches.
Despite the challenges that remain in forecasting nor'easters, of all the synoptic-scale weather that impacts the U.S., this is the one that the models have the most skill in predicting...more effort has been expended during model development on this one problem than all others, I believe even more than hurricane track forecasts. These storms have the potential to impact more U.S. citizens than perhaps any other type of storm that occurs in the U.S., thus, the huge effort, and the significant progress over the years.
ReplyDeleteI remember the Presidents Day 1979 storm well, as a 7th grader, already decided on a career in meteorology, living in the NYC area. Coming back from Radio City Music Hall Sunday afternoon the AM News radio forecast was for about 1-3 inches across the area, with heavier amounts across southern NJ, about 100 miles further south. Woke up the next morning with a foot of windswept snow, falling at the rate of over 2"/hr...totaling about a foot, with 2ft totals across Srn NJ to the DC area. About a decade later, my Synoptic Meteorology professor distributed Lance Bosarts paper on the storm, it helped piece together the complexities of the air/sea interactions and model limitations of the day, into why the initial forecast of that storm (and many other high impact events before that) would be such a bust. Thanks for bringing baack the memories. :)
ReplyDelete