Forecasting the 12-14 March 1993 Superstorm
Total Page:16
File Type:pdf, Size:1020Kb
Louis W. Uccellini,* Paul J. Kocin,* Russell S. Schneider,* Paul M. Stokols,+ Forecasting the + 12-14 March 1993 and Russell A. Dorr Superstorm Abstract damaged homes along the Florida, Cuba, and New York coastlines; spawned tornadoes and damaging This paper describes the decision-making process used by the wind squalls throughout Florida, Cuba, and other forecasters in the National Meteorological Center's Meteorological Operations Division and in Weather Forecast Offices of the National Caribbean nations; sank several large container ships Weather Service to provide the successful forecasts of the superstorm and numerous smaller vessels; produced record-break- of 12-14 March 1993. This review illustrates 1) the difficult decisions ing snowfall amounts along the spine of the Appala- forecasters faced when using sometimes conflicting model guid- chians from Birmingham, Alabama, to Syracuse, New ance, 2) the forecasters' success in recognizing the mesoscale York; and created blizzard conditions that closed aspects of the storm as it began to develop and move along the Gulf and East Coasts of the United States, and 3) their ability to produce major roadways and stranded thousands of people at one of the most successful heavy snow and blizzard forecasts ever airports throughout the eastern United States. The for a major winter storm that affected the eastern third of the United severity of the storm was demonstrated by a report States. from the Department of Commerce that attributed The successful aspects of the forecasts include the following. much of the economic downturn for the first quarter of 1) Cyclogenesis was predicted up to 5 days prior to its onset. 2) The 1993 to the occurrence of this storm and its impact unusual intensity of the storm was predicted three days in advance, allowing forecasters, government officials, and the media ample over such a large geographical domain. time to prepare the public, marine, and aviation interests to take The 12-14 March 1993 superstorm ranks with, or precautions for the protection of life and property. 3) The excessive perhaps exceeds, the magnitude of other storms that amounts and areal distribution of snowfall were predicted two days have buried the eastern United States with wind- in advance of its onset. 4) An extensive number of blizzard watches and warnings were issued throughout the eastern United States with driven heavy snows, such as the March blizzard of unprecedented lead times. 5) The coordination of forecasts within 1888 (Kocin 1983), the East Coast blizzard of Febru- the National Weather Service and between the National Weather ary 1899 (Kocin et al. 1988), the New York City Big Service, private forecasters, and media meteorologists was per- Snow of December 1947 and the Lindsay Storm of haps the most extensive in recent history. February 1969 (Kocin and Uccellini 1990), the New England snowstorm of February 1978 (Brown and Olson 1978), the Presidents' Day storm of February 1979 (Bosart 1981; Uccellini et al. 1984), and the 1. Introduction Megalopolitan snowstorm of February 1983 (Sanders On 12-14 March 1993, the eastern third of the and Bosart 1985). With the exceptions of the 1899, United States was hit by a major winter storm of 1978, and 1983 storms, these and other major East historic proportions. The storm produced the most Coast storms have become notorious for their lack of extensive distribution of heavy snow across the east- predictability, even as heavy snows enveloped the ern United States in modern times (Kocin et al. 1995); major cities along the coast (see, e.g., Kocin and established numerous low pressure records along the Uccellini 1990). The generally poor forecasts have East Coast; generated coastal flooding that severely contributed to a public perception that these weather systems are either unpredictable and/or the forecast- ers do not know what they are doing, a perception that *Meteorological Operations Division, National Weather Service, National Meteorological Center, Camp Springs, Maryland (Current survives even the good forecasts. affiliation: Office of Meteorology, National Weather Service, Silver In contrast to many of the previous major storm Spring, Maryland). events, the overall forecast for the March 1993 +Scientific Services Division, National Weather Service, Eastern superstorm was very successful, as illustrated by the Region, Bohemia, New York. Corresponding author address: Paul J. Kocin, Meteorological front page of the Boston Herald newspaper (Fig. 1) Operations Division, NWS/NMC, 5200 Auth Rd., Camp Springs, MD published on the morning prior to the storm's arrival. 20746. The paper's emphasis on the potential for severe In final form 28 October 1994. winter weather, expected impacts of the storm, and Bulletin of the American Meteorological Society 183 Unauthenticated | Downloaded 09/23/21 11:41 PM UTC ultimately represent an impor- tant milestone in what has been an ongoing long-term effort to transform weather forecasting from a subjective blend of ob- servations and interpretations to an intricate process based on mathematics and applied phys- ics. Over the past 40 years, the forecast process has evolved to a "forecaster-machine" mix that today involves more extensive observations; model-based data assimilation systems; global and regional numerical models; model output statistics (MOS); forecaster interpretations of nu- merical model forecasts; produc- tion of national guidance prod- ucts; coordination of the guid- ance products with local fore- casters; production of local fore- casts by government, private, and media meteorologists; and the timely coordination of watches and warnings with local officials prior to and during se- vere weather situations. The evo- lution of the forecast process toward a forecaster-machine mix has been based on the intro- duction and subsequent im- provements in numerical mod- els and model-based data as- similation schemes (see, e.g., Bonner 1989). It has also de- FIG. 1. Headline from the Boston Herald newspaper, Boston, Massachusetts, 13 March pended on the ability of fore- 1993. Reprinted with permission. casters tQ JNTERPRET the mode, output and the MOS, recogniz- ing the strengths and weak- the emergency preparedness taken in anticipation of nesses of each, and to then make a judgment on the the onset of a "meteorological bomb" indicates how most probable solutions (Stokols et al. 1991; Funk seriously the forecasts were taken by the media and 1991; Tracton 1993). The steady improvement in general public for this case. Furthermore, the general manual forecast scores, which have maintained their public and media generally acknowledged and praised edge over the model mean sea level pressure (MSLP) the public and private weather forecasters for their and quantitative precipitation forecasts (Stokols et al. accurate snow and blizzard forecasts and cited the 1991; Tracton 1993; Uccellini et al. 1994), is thus importance of these forecasts for facilitating emer- based on a number of factors, not the least of which is gency preparedness actions (NOAA 1994). The sue- the forecasters' ability to work with the new observa- cessful forecasts for the March 1993 superstorm could tions, model data, and derived products with increas- prove to be a turning point in dispelling the public ing skill and confidence. perception about the inability to forecast major coastal The purpose of this paper is to describe the suc- winter storms. More importantly, the overall perfor- cesses and shortcomings of the forecast process mance of the public and private forecasters could discussed above as it applied to the superstorm of 12- 184 Vol. 76, No. 2, February 1995 Unauthenticated | Downloaded 09/23/21 11:41 PM UTC 14 March 1993. A brief synoptic overview is presented panhandle. By 1200 UTC 13 March, the central sea in section 2. The performance of the forecasters in the level pressure had decreased to 971 mb as the low Meteorological Operations Division (MOD) of the Na- moved into southeastern Georgia. By this time, bliz- tional Meteorological Center (NMC) priorto and during zard conditions and very heavy snows were occurring this storm is described in section 3. This section to its north and west, severe convective storms were highlights the actions taken by the forecasters who occurring to its south and east, and a 3-4-m storm had to make important forecast decisions despite the surge was inundating the northwest Florida coast. sometimes conflicting numerical guidance available During the following 24 h, the storm moved rapidly for their use. In section 4, the actions of local National to the northeast, continued to deepen to 960 mb over Weather Service (NWS) forecasters are discussed to the Delmarva Peninsula at 0000 UTC 14 March, and illustrate the unprecedented effort that was made to then began to fill as it propagated toward the Maine provide forecasts, watches, and warnings with enough coast by 1200 UTC 14 March (Fig. 2). The storm track lead time for local officials to take appropriate action to was slightly inland of the coastline so that the heaviest protect life and property. The impact of the new WSR- snowfall fell west of the coastline from northern Ala- 88D radars on the forecast process at the local fore- bama to central New York (see Kocin et al. 1995; Fig. cast office in Washington, D.C., is also described in 1). Along the spine of the Appalachian Mountains, as section 4. A summary is provided in section 5. much as 30 to 50 in. (75 to 125 cm) of snow fell, accompanied in some instances by wind gusts of 30 to 40 m s_1. The major northeastern urban centers from Washington, D.C., to Boston generally received snow- 2. Overview of the 12-14 March fall accumulations of 10 to 12 in. (25 to 30 cm), with superstorm snowfall totals reduced somewhat by a changeover to The synoptic and mesoscale aspects of the ice pellets and a brief period of rain. superstorm of March 1993 are described in the com- panion paper by Kocin et al.