Natural Hazards

, Volume 48, Issue 1, pp 101–113

Characteristics of the top ten snowstorms at First-Order Stations in the U.S.

Authors

    • NOAA’s National Climatic Data Center
  • Stanley A. Changnon
Original Paper

DOI: 10.1007/s11069-008-9251-5

Cite this article as:
Houston, T.G. & Changnon, S.A. Nat Hazards (2009) 48: 101. doi:10.1007/s11069-008-9251-5
  • 110 Views

Abstract

Snowstorms can produce varying degrees of damage depending on the amount and intensity of the snowfall over a given amount of time. Concurrent weather conditions such as freezing rain and high winds often exacerbate the amount of damage received. In order to assess the frequency of potentially damaging conditions during climatologically significant snowstorms, the top ten snowstorms (TTS) at individual First-Order Stations in the eastern two-thirds of the conterminous U.S. were determined, and the hourly weather conditions during each event were analyzed. The results show that TTS have occurred as early as September and as late as June, with January being the peak month of occurrence. Hourly precipitation totals during TTS were 2.3 mm or less 88% of the time. Seven percent of TTS were classified as a blizzard with over half of the blizzards occurring in the West North Central region. The most common concurrent weather condition during a TTS was fog followed by blowing snow. Regionally, heavy snow events in the Northeast had relatively higher precipitation amounts, colder temperatures, higher winds, and more fog and blowing snow than any other region.

Keywords

SnowSnowstormsSnowfallConcurrent weather conditions

Abbreviations

FOS

First-Order Stations

NOAA

National Oceanic and Atmospheric Administration

NCDC

National Climatic Data Center

NWS

National Weather Service

NESIS

Northeast Snowfall Impact Scale

TTS

Top ten snowstorms

U.S.

United States

USD

United States dollars

1 Introduction

Snowstorms are one of the most damaging weather extremes (Changnon and Hewings 2001). Concurrent weather conditions, such as freezing rain and high winds often exacerbate these damages. Four percent of all weather related insured property losses in the U.S. are a result of snowstorms, averaging $408 million (USD) in insured property losses per year (Changnon and Changnon 2006). A report by Adams et al. (2004) found that the economic impact of snowfall on the U.S. economy, both positive and negative, range from $50 to $400 billion (USD) per year.

Previous snowstorm studies have typically focused on the synoptic conditions associated with snowstorms, the development of snowstorm climatologies, and case studies of particular events. In addition, studies by Branick (1997), Schwartz and Schmidlin (2002), Changnon and Changnon (2006), and Changnon et al. (2006) examined significant snowstorms and identified the spatial and temporal characteristics of damaging snowstorms and blizzards. This study builds on these previous studies by identifying the top ten 1- or 2-day snowfall totals at individual First-Order Stations (FOS) in the eastern two-thirds of the conterminous U.S. from 1948 to 2001. Concurrent weather conditions, such as low temperatures, high winds, and other weather conditions reported during top ten events were analyzed to examine the frequency of their occurrence. How these conditions contribute to the amount of damage produced by the storm are also noted. By knowing the types of conditions that can occur during a heavy snowstorm at a particular location, those impacted by these storms can be better prepared the next time a significant snowstorm threatens.

2 Data and analysis

In order to assess the frequency and types of potentially damaging conditions associated with locally significant snowfall events, hourly weather conditions during top ten snowstorms (TTS) at select FOS were analyzed. Daily observations of snowfall from FOS were used to identify the TTS at each station, and hourly observations of temperature, precipitation, wind speed, and other concurrent weather conditions were used to assess the characteristics of these storms.

Figure 1 shows the 121 FOS that were selected based on the quality of their snowfall measurements (see Changnon et al. 2006 for more information on the selection process). Inhomogeneities in the snowfall record can occur as a result of station moves, the change from manual observations to automated observations, from changes in observation practices over time, and from changing from one observer to another (Robinson 1989; Changnon 2006; Doesken and Judson 1997; Kunkel et al. 2007). Stations were chosen if there were no significant changes in snowstorm frequency which occurred as a result of these inhomogeneities or if the frequency was not significantly different from neighboring stations.
https://static-content.springer.com/image/art%3A10.1007%2Fs11069-008-9251-5/MediaObjects/11069_2008_9251_Fig1_HTML.gif
Fig. 1

Map of the First-Order Stations analyzed in this study

A snowstorm was defined as an event in which snow fell over a 1- or 2-day period. This definition is consistent with the findings of Changnon (1969) and Changnon et al. (2006) who found that 84% of all 2-day snow events lasted less than 24 h but, due to a fixed once-a-day observation time, were reported over two observational days. Research has also shown that snowstorms begin to produce significant damages when heavy snowfall occurs within 48 h or less (Changnon 1969; Branick 1997).

The 1- or 2-day snowstorms were determined based on the following criteria. A subset of daily snowfall totals of 25.4 mm or greater at the chosen 121 National Weather Service (NWS) FOS from 1948 to 2001 was obtained from the National Climatic Data Center’s (NCDC) Cooperative Summary of the Day dataset (NOAA 2004a). From this subset, if one day reported snow with no snow reported on the day before or after, this snowstorm was considered for the TTS list. If two adjacent days reported snowfall, this 2-day total was considered for the TTS list. If three consecutive days reported snowfall, the two adjacent days with the greatest snowfall total were considered as a 2-day event. The remaining day was also considered, but as a separate event. If four or more consecutive days reported snowfall, the three consecutive days criterion was followed to identify the first snowstorm for consideration. Then if the next two adjacent days reported snowfall, or if only one day remained, then this snowstorm was also considered for the TTS list. Additional snowstorms were identified until snow was no longer reported. The snowstorms at each station were then ranked to determine the TTS at that station.

Once the TTS were determined for each station, hourly data for each event were obtained from NCDC’s Surface Airways Hourly and Airways Solar Radiation dataset (NOAA 2004b). This data included dry bulb, wet bulb, and dew point temperatures, wind speed and direction, visibility, and other concurrent weather conditions such as fog and freezing rain. Hourly precipitation was also obtained from NCDC’s Hourly Precipitation dataset (NOAA 2004c) in order to examine water equivalent totals during each event.

3 Characteristics of top ten snowstorms

Overall a total of 1,182 TTS were analyzed (28 TTS did not have hourly data available). The following sections describe the general characteristics of TTS, such as the frequency and timing of events, precipitation types and temperatures during TTS, the occurrence of blizzards and blowing snow during TTS, as well as the occurrence of other significant weather conditions.

3.1 Frequency and timing of TTS

The snowfall totals of the number one TTS at each station varied considerably depending on where the storm occurred. In general, there was a latitudinal increase in totals from south to north with higher totals downwind of the Great Lakes and in mountainous areas. The greatest TTS total was 1168.4 mm at Sault Ste Marie, MI on 9–10 December 1995. This event helped break the station’s all time record monthly snowfall total with a monthly total of 2507.0 mm. The smallest snowfall total for a number one storm on a TTS list was 139.7 mm at Jackson, MS from a storm that occurred on 13 January 1982.

Some snowstorms that occurred over a large geographic area were in the top ten list of multiple FOS. The snowstorm that generated the most TTS was the “Storm of the Century” that occurred on 12–13 March 1993. This storm was ranked among the top ten at 22 FOS stations throughout 15 states and was listed as the number one storm at seven stations. Overall, this storm produced maximum snowfall totals of over 1270.0 mm and caused over $1.8 billion (USD) in insured property losses throughout 21 states from Louisiana to Maine and was rated as a Category 5 storm (extreme, the largest value) on the Northeast Snowfall Impact Scale (NESIS) (Kocin and Uccellini 2004; Changnon and Changnon 2006). Other damaging conditions associated with this storm included high winds, tornadoes, flooding, and record low temperatures (Lott 1993; Kocin et al. 1995).

Studies have shown that snowstorms have occurred as early as September, are most frequent in January, and end as late as June (Doesken and Judson 1997; Hirsch et al. 2001; Changnon et al. 2006). TTS are no exception. Five TTS occurred during the month of September and one occurred in the month of June. All six of these snowstorms occurred at stations along the Rocky Mountains and had 1- or 2-day snowfall totals ranging from 279.4 to 584.2 mm. The month with the most TTS was January (26.8% of all TTS) followed by March (19.6%) and February (18.8%). Figure 2 shows that TTS occurred most frequently in January in most regions assessed, but peaked in February in the Northeast and in March in the West North Central region.
https://static-content.springer.com/image/art%3A10.1007%2Fs11069-008-9251-5/MediaObjects/11069_2008_9251_Fig2_HTML.gif
Fig. 2

Map of the nine U.S. climate regions. Within each region (excluding Northwest and West) is the frequency of TTS in percent for the months of December through March (D, J, F, M)

The annual (January–December) number of TTS are shown in Fig. 3, revealing that 1978 had the most with 59 TTS from 23 separate storms, a majority of which occurred in the Central region. The year 1983 was second with 47 TTS from 19 storms followed by 1987 with 45 TTS from 15 storms and 1966 with 41 TTS from 16 storms. Since snowfall is a winter phenomenon that crosses from one year to the next, annual frequencies were also determined based on a July to June year. Using this criterion, the winter of 1977–1978 had 52 TTS from 19 separate storms followed by the winters of 1978–1979 and 1982–1983 which had 42 TTS each from 14 and 18 storms, respectively. The temporal distribution has a mid-period peak and does not have a long term upward or downward trend.
https://static-content.springer.com/image/art%3A10.1007%2Fs11069-008-9251-5/MediaObjects/11069_2008_9251_Fig3_HTML.gif
Fig. 3

Number of TTS per year

Call (2005) found that if two snowstorms occurred within several days of each other and the temperatures remained cold, the second storm tended to be more disruptive than the first due to the increased snow depth and snowfall mitigation equipment problems. Several TTS did occur within days of another TTS at several stations. One station reported three TTS which occurred over five consecutive days. During 24–28 December 2001, Buffalo, NY reported: (1) 645.2 mm of snow on 24–25 December which ranked fourth on the TTS list, (2) 759.5 mm on 26–27 December which ranked second, and (3) 665.5 mm on 28 December which ranked third. These events contributed to the 2100.6 mm of snow reported that month which became a new monthly snowfall record for Buffalo. While Buffalo is used to dealing with copious amounts of snow, help was needed to aid in the snow removal. Since these events occurred over a holiday, the impacts were less than what may have happened if they occurred even a week or two later (CNN 2007).

3.2 Damaging types of precipitation

The more water content snow contains the more damage the storm can cause. Light fluffy snow can easily be removed and driven on but heavy wet snow can strain snow removal equipment, break tree limbs, and cause building roofs to cave in under the added weight. Even in cities that have efficient snow mitigation capabilities, snow intensity can often have a greater impact than the amount of snowfall (Call 2005). At high intensities, snow falls too quickly for removal crews to keep up and visibility is reduced, as was the case during the Buffalo, NY events described in Sect. 3.1.

The most frequent hourly liquid equivalent precipitation total during a TTS was 0.3 mm which included melted snow as well as other forms of liquid precipitation such as rain and melted ice. Eighty-eight percent of the time, hourly liquid equivalent precipitation totals were 2.3 mm or less. The highest hourly liquid precipitation total was 17.0 mm at Bristol, TN on 21 November 1952. During that hour, heavy snow fell and contributed to a 1-day snowfall total of 411.5 mm which is the number one storm on Bristol’s top ten list. Regionally, hourly precipitation totals were slightly higher in the Northeast and Southeast than in the other regions as can be seen by the higher frequencies at higher thresholds in Table 1.
Table 1

Relative frequencies (in percent) of hourly precipitation amounts during TTS in each region and across the eastern two-thirds of the nation

Region

Trace

0.3 mm

0.5 mm

0.8 mm

1.0 mm

1.3 mm

1.5 mm

1.8 mm

2.0 mm

2.3 mm

2.5+ mm

C

4.6

20.6

16.2

11.9

10.0

8.5

5.9

4.6

3.8

2.8

11.2

ENC

6.9

24.9

16.7

11.8

9.0

7.6

5.2

4.0

2.9

2.3

8.8

NE

5.7

17.3

13.5

10.6

9.5

8.2

5.7

4.0

4.7

3.5

17.4

S

4.9

21.1

18.1

11.8

11.1

8.4

5.5

4.3

3.7

2.0

9.0

SE

1.0

16.2

12.6

11.2

11.0

8.1

6.8

6.9

6.9

3.9

15.7

SW

5.1

24.6

18.6

11.4

10.8

6.6

3.8

3.1

4.1

1.1

10.7

WNC

6.4

21.6

17.1

11.9

11.0

7.7

5.3

3.9

3.4

2.3

9.4

U.S.

5.3

20.9

16.1

11.6

10.2

8.0

5.5

4.3

3.9

2.6

11.6

Note: Precipitation amounts were measured in hundredths of inches and have been converted to mm. Therefore, the values listed represent actual precipitation totals and are not a range of values

C, Central; ENC, East North Central; NE, Northeast; S, South; SE, Southeast; SW, Southwest; WNC, West North Central

During TTS, heavy snowfall (i.e., visibility less than 400 m) occurred seven percent of the time while snowfall intensity was light (i.e., visibility greater than 805 m) approximately 75% of the time. These percentages were also true regionally except in the Northeast and Southeast where heavy intensities occurred 14% and 9% of the time, respectively, and light intensities occurred in both regions 65% of the time.

3.3 Temperatures

The temperature during a winter storm plays a significant role in determining whether a location will receive rain, freezing rain, or snow and also determines how dense the snow will be. Approximately 50% of the time, snowfall occurred when the dry bulb and wet bulb temperatures were between −4.4 and 0.6°C (Fig. 4). The warmest dry bulb and wet bulb temperatures during a TTS was 7.2 and 5.0°C, respectively at Albuquerque, NM on 29 March 1973 where a 2-day total of 271.8 mm of snow was reported. These temperatures occurred in the early afternoon during a brief thunderstorm and light snow. The coldest dry bulb and wet bulb temperature reported during a TTS was −27.2°C at Sault Ste Marie, MI on 16 January 1982, where a 2-day total of 368.3 mm of snow was reported. Snow continued into the next day (which was beyond the 2-day definition of an event for this study) where the dry bulb and wet bulb temperature dropped to −30.6°C with light snow. Overall, this storm produced $104.6 million (USD 2006) in insured property losses in 24 states throughout the Midwest, Southeast, and Northeast.
https://static-content.springer.com/image/art%3A10.1007%2Fs11069-008-9251-5/MediaObjects/11069_2008_9251_Fig4_HTML.gif
Fig. 4

Frequency in percent of (a) dry bulb, (b) wet bulb, and (c) dew point temperatures during TTS

Dew point temperatures ranged from −7.8 to 0°C 67% of the time during TTS (Fig. 4). The warmest dew point temperature of 3.3°C occurred during the “Storm of the Century” in Atlanta, GA on 13 March 1993 where 106.7 mm of snow was reported. The coldest dew point temperature was −33.3°C which occurred during a snowstorm in Helena, MT on 10 January 1971 where a total of 307.3 mm of snow was reported.

During TTS, the dry bulb temperature peaked at −1.1 to 0.6°C for each region except for the East North Central and Northeast regions (Fig. 5). The frequency distribution in both of these regions tended to be more uniform over a wide range of temperatures. In the other regions that had higher frequencies near the freezing point, the frequency distribution peaked more sharply. Dew point temperatures peaked at lower temperatures in the East North Central, Northeast, and Central regions than in the other regions (Fig. 6).
https://static-content.springer.com/image/art%3A10.1007%2Fs11069-008-9251-5/MediaObjects/11069_2008_9251_Fig5_HTML.gif
Fig. 5

Frequency in percent of dry bulb temperatures in select regions during TTS

https://static-content.springer.com/image/art%3A10.1007%2Fs11069-008-9251-5/MediaObjects/11069_2008_9251_Fig6_HTML.gif
Fig. 6

Frequency in percent of dew point temperatures in select regions during TTS

3.4 Blizzards and blowing snow

Winds also contribute to the amount of damage received during a snowstorm. According to Changnon and Kunkel (2006), winds greater than 13.4 m/s increase damages by 25% compared to heavy snow alone. High winds blow falling snow and even pick up accumulated snow which can cause whiteout conditions and significant drifting. The NWS defines a blizzard as falling or blowing snow that reduces visibility to less than 400 m with sustained winds or frequent wind gusts 15.6 m/s or greater for a period of 3 h or more (NOAA 2007). A blowing snow advisory is issued when visibility is less than 400 m and sustained winds or frequent wind gusts are 15.6 m/s or less (NOAA 2007).

Out of the 1,182 TTS, 80 TTS from 56 separate storms were considered blizzards by the NWS definition. Thirty TTS blizzards were from storms that caused catastrophic damage (based on insured property losses of $1 million (USD) or greater (Changnon 2005)). TTS blizzards were reported in each month between October and April and occurred most frequently (25% of the time) in the month of March. Fifty-five percent of the TTS blizzards occurred in the West North Central region which includes the nation’s maximum blizzard zone states of North Dakota and South Dakota, as well as the western part of Minnesota (Schwartz and Schmidlin 2002).

Ninety percent of the time hourly wind speeds were 11.2 m/s or less during TTS. Only three percent of the time did winds during a TTS exceed 15.6 m/s, the threshold for defining a blizzard. The highest hourly wind speed reported during a TTS was 26.4 m/s at Rapid City, SD on 30 April 1967 where a total of 345.4 mm of snow was reported. Approximately 54% of the time, the prevailing wind direction during a TTS was from the northern quadrant.

Regionally, the Southwest had higher frequencies of lower wind speeds and lower frequencies of higher speeds than other regions (Fig. 7). Wind speeds were 15.6 m/s or greater 6% of the time in the blizzard prone West North Central region. In the other regions, wind speeds were 5.4–7.6 m/s approximately 40% of the time.
https://static-content.springer.com/image/art%3A10.1007%2Fs11069-008-9251-5/MediaObjects/11069_2008_9251_Fig7_HTML.gif
Fig. 7

Frequency in percent of wind speeds in select regions during TTS

3.5 Other significant weather conditions

Many types of weather conditions occurred at the same time as snowfall during the TTS. These conditions included: thunderstorms, rain, freezing rain, fog, smoke, haze, and blowing snow and included the various types of precipitation (rain, rain showers, drizzle, ice crystals, ice pellets, etc.) and intensities (light, moderate, and heavy). Sixty percent of hours reporting snow also reported one additional concurrent weather condition, 6% reported two additional concurrent weather conditions, and less than 0.5% reported three additional concurrent weather conditions. Regionally, the South had the highest percentage of additional concurrent weather conditions with one or more condition occurring 76% of the time, while in the Southwest, one or more concurrent weather conditions were reported 52% of the time.

The most common concurrent weather condition during a TTS was fog (52% of the time) followed by blowing snow (40% of the time). Table 2 shows the frequency of occurrence of each general concurrent weather type (i.e., fog includes fog, ice fog, and ground fog; ice includes freezing rain, freezing drizzle, and ice pellets) for all TTS. Fog was the primary concurrent condition in the Central, Southwest, South, and Southeast regions and blowing snow was the primary concurrent condition in the West North Central and East North Central regions. In the Northeast, fog and ice were dangerous additions to hours when heavy snowfall was occurring.
Table 2

Types and frequency of occurrence of concurrent weather conditions during TTS

Concurrent weather type

Frequency of occurrence (%)

Fog

52.1

Blowing snow

40.4

Ice

4.1

Rain

2.1

Smoke, haze, dust

1.3

4 Conclusions

This study assessed the frequency of various weather conditions that occurred during hours when heavy snowfalls occurred. Data from 121 FOS across the eastern two-thirds of the U.S. and for 1948–2001 were used to identify the TTS at each station. The results revealed that TTS have occurred as early as September and as late as June, with January being the peak month of occurrence. The year 1978 had the greatest number of TTS.

The most frequent hourly liquid equivalent precipitation total during a TTS was 0.3 mm with 88% of all hours having values of 2.3 mm or less. Seven percent of the time, snowfall intensity was heavy.

An assessment of temperatures showed that dry bulb temperatures during TTS ranged from −27.2 to 7.2°C and most frequently occurred near the freezing point. Fifty-one percent of all values were from −4.4 to 0.6°C. Sixty-seven percent of all snow hours during TTS had dew point values between −7.8 and 0°C.

Approximately, 54% of the snow hours had winds from the northern quadrant. Wind speeds less than 11.2 m/s prevailed 90% of the time. Seven percent of TTS were classified as a blizzard with over half of the blizzards occurring in the West North Central region.

Weather conditions occurring during the hours when heavy snow fell were assessed, and the most frequent conditions in the U.S. were fog (52% of the time) and blowing snow (40% of the hours), both of which reduce visibility. These two conditions were more frequent in the Northeast, and fog was also frequent with snows in the Central, Southwest, South, and Southeast regions. Freezing rain, rain, and haze also occurred during TTS, further increasing damages and limiting visibility.

The regional results revealed that heavy snow events in the Northeast had relatively higher precipitation amounts, colder temperatures, higher winds, and more fog and blowing snow than any other region. The East North Central region had relatively high frequencies of relatively low temperatures, high winds, and blowing snow. Thus, heavy snows are more dangerous in these two regions of the U.S.

A listing of the TTS at each station and their concurrent weather conditions along with additional snowfall data and information found in Changnon (2005), Changnon and Changnon (2006), and Changnon et al. (2006) can be obtained from two products available from NCDC. A CD-ROM titled “Snowstorm Data: Long-Term Data Sets about Snowstorms in the United States” and a publication titled “Snowstorms Across the Nation: An Atlas about Storms and Their Damages” can be obtained from the NCDC web site at http://www.ncdc.noaa.gov or by calling NCDC Customer Service at +1-828-271-4800.

Acknowledgments

The authors would like to thank Imke Durre and Mike Squires of NCDC and two anonymous reviewers for their helpful comments and suggestions. Portions of this research were funded by a grant from the National Oceanic and Atmospheric Administration (NOAA) and the National Aeronautics and Space Administration (NASA), as part of the Climate Change Enhanced Data Set Project, under Grant No. NA16GP1585 as well as by a grant from the Office of Biological and Environmental Research, U.S. Department of Energy, under Grant No. DE-AI02-96ER62276.

Copyright information

© Springer Science+Business Media B.V. 2008