Sixteen years ago this week, nearly the entire eastern third of the nation was ravaged March 12-14, 1993, by a massive storm often called "The Storm of the Century" (20th century, that is). More modestly -- on account of other storms that have rivaled in strength and impact, like the the "Blizzard of '78" -- it's called the "Superstorm of March '93."
For me, though, it will forever remain THE Big One -- and not just because of its intensity, size and associated extreme weather, including significant snow in the D.C. area. In a former life, my research was largely focused on the development, evolution and especially the predictability of extratropical cyclones (a fancy term for the typical U.S. storm system that usually has a cold front and warm front extending out from a low-pressure center), especially those occurring during the cold season along the East Coast.
Perhaps the most remarkable aspect of this storm was that it was highly predictable.
Keep reading for the skinny on the '93 Superstorm...
I'll get back to the storm's predictability in a moment, but first please note there are many excellent accounts of the storm, its impacts and aftermath (e.g., here, here, and here). The main highlights?... the storm was characterized by hurricane-force winds, record snowfalls, and record low temperatures and barometric pressure readings. It may not have been the most severe blizzard on record, but it was the largest in terms of the area affected. Snowfall in the D.C. area ranged from 8 to 13 inches, with 18 inches north and west of the city in Frederick County. Tornadoes, thunderstorms and flooding occurred as the storm's trailing cold from barreled across the Southeast. The storm was blamed for 270 deaths and damages topping $6 billion.
Just how well was the storm predicted? This was the first time the National Weather Service (NWS) was able to accurately forecast a storm of this magnitude 5 to 6 days in advance -- and do so with a high level of confidence. Since then, however, similar forecasting successes in the medium range (3-10 days in advance) have been few and far between, even with the tremendous advances in atmospheric science, forecast models and computer power.
The only decent snowstorm locally this winter, which generally dumped 5-8" across the area March 1-2, most certainly was not included in the "few and far between," as confidence in the potential for significant snow didn't really come until just a couple days before, and confidence was fluctuating even as the storm got underway.
The key to understanding the exceptionally accurate medium-range forecasts of the Superstorm was, in fact, its size. Larger-scale weather systems are intrinsically "easier" to forecast. Thus, for example, large winter storms are more predictable than, for example, a summer line of thunderstorms. And, the Superstorm -- to the best of my knowledge -- was broader in areal extent than any other storm in recorded history.
In this regard, the measure relevant to the predictability of a weather system is its wavelength, or in other words the distance between the two high pressure centers on either side (west and east) of the low-pressure center. The wavelength of the Superstorm was about 3,500 miles. In comparison, the wavelength of the storm here earlier this month was approximately 2,100 miles. Not unusually, in both cases there were smaller-scale features (wavelength less than about 200 miles), such as snow bands and thunderstorms, embedded in the larger-scale circulations.
Recall from above that the Superstorm was not only predicted 5-6 days in advance, but done so with an unprecedented degree of confidence. There was skepticism in some quarters about this, including reluctance by many TV stations to buy into these early warnings; but, NWS stuck to its guns and proved correct. I, personally, played an important role in this. At the time I was among those at the NWS National Centers for Environmental Prediction (NCEP) who was pioneering the development and applications of an operational ensemble prediction system (background here; latest model output here).
Ensemble forecasting involves a single model being run multiple (nowadays up to 50) times, each time with a set of slightly different initial conditions (temperatures, winds, pressure, etc.) in order to account for possible inaccuracies in the measured initial conditions. Each of these runs is known as an ensemble member, and together ensemble members provide a reasonably good portrayal of possible outcomes.
Although the system was still in its experimental phase in March 1993, output was made available to forecasters at NCEP's Hydrometeorological Prediction Center who provide guidance to all NWS local forecast offices. The long and short of this was that beginning 5-6 days in advance the ensembles consistently showed virtual unanimity amongst ensemble members in predicting the nature and severity of the impending Superstorm.
The rest is history with regard to the NWS's successful forecasting of this particular storm, and more generally a recognition of the importance of continued development and use of ensemble prediction for assessing uncertainties and levels of confidence in all weather forecasts. This led a few years ago to NCEP implementing its operational Short Range Ensemble Forecast System.
ASIDE: Last Friday night Erin Burnett, a CNBC anchor, was on HBO's Real Time With Bill Maher. In regard to the current economic crisis, Maher asked her, "Why didn't anybody there (CNBC) predict what was going to happen?" Her response: "CNBC did accurately forecast some elements of the crisis: The question of timing and magnitude, nobody got."
Sound familiar? Just replace the words "the current economic crisis" with "almost every D.C. snowstorm."