APRIL 2004 CORTINAS ET AL. 377

An Analysis of Freezing , Freezing , and Pellets across the United States and Canada: 1976±90

JOHN V. C ORTINAS JR.* Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma

BEN C. BERNSTEIN National Center for Atmospheric Research, Boulder, Colorado

CHRISTOPHER C. ROBBINS NOAA/National Service, Fort Worth, Texas

J. WALTER STRAPP Meteorological Service of Canada, Downsview, Ontario, Canada

(Manuscript received 20 May 2003, in ®nal form 14 October 2003)

ABSTRACT A comprehensive analysis of , , and was conducted using data from surface observations across the United States and Canada. This study complements other studies of freezing in the United States and Canada, and provides additional information about the temporal charac- teristics of the distribution. In particular, it was found that during this period 1) spatial variability in the annual frequency of freezing precipitation and ice pellets is large across the United States and Canada, and these occur most frequently across the central and eastern portions of the United States and Canada, much of Alaska, and the northern shores of Canada; 2) freezing precipitation and ice pellets occur most often from December to March, except in northern Canada and Alaska where it occurs during the warm , as well; 3) freezing rain and freezing drizzle appear to be in¯uenced by the diurnal solar cycle; 4) freezing precipitation is often short lived; 5) most freezing rain and freezing drizzle are not mixed with other precipitation types, whereas most reports of ice pellets included other types of precipitation; 6) freezing precipitation and ice pellets occur most frequently with a surface (2 m) temperature slightly less than 0ЊC; and 7) following most freezing rain events, the surface temperature remains at or below freezing for up to 10 h, and for up to 25 h for freezing drizzle.

1. Introduction as well, since these precipitation types, primarily freezing Freezing precipitation, de®ned by the American Me- rain, have produced fatalities, injuries, and a signi®cant teorological Society's Glossary of as freez- loss of property (Bendel and Paton 1981; Forbes et al. ing rain, freezing drizzle, and freezing (Glickman 1987; Toth 1988; Rauber et al. 1994; Martner et al. 1992; 2000), can have a devastating effect on many industries, Marwitz et al. 1997; Jones and Mulherin 1998; De- including transportation, energy, agriculture, and com- Gaetano 2000; Irland 2000; Changnon 2003). An ice merce. In addition to freezing precipitation, ice pellets at that occurred during January 1998 in the north- the ground often are an indication of freezing precipi- eastern United States and southeastern Canada illustrates tation aloft, which can have a detrimental effect on air- the effect of these on society and underscores the craft (Zerr 1997; Bernstein et al. 1998). The effects of need for climatological information to understand the fre- these events on human life and activities are signi®cant quency of these hazards, which may lead to better mit- igation practices. [See Gyakum and Roebber (2001) and Roebber and Gyakum (2003) and for a complete mete- * Additional af®liation: NOAA/OAR/National Severe Storms Lab- orological description of this storm.] oratory, Norman, Oklahoma. a. Impacts of freezing precipitation Corresponding author address: Dr. John Cortinas Jr., NOAA Re- search, R/OSSX5, 1315 East±West Hgwy., Silver , MD 20910. The 1998 affected more than three million E-mail: [email protected] people in four states and two Canadian provinces, caus-

᭧ 2004 American Meteorological Society

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC 378 WEATHER AND FORECASTING VOLUME 19 ing 44 fatalities and damage estimated at $3 billion U.S. the ground, several authors have documented icing con- dollars (USD) in Canada and $1.4B USD in the United ditions caused by freezing drizzle and freezing rain aloft States (NOAA 1998). In addition to the obvious total that signi®cantly impacted aircraft performance (Sand disruption of transportation due to ice accumulation on et al. 1984; Pike 1995; Ashenden and Marwitz 1997; the roads, vehicles, and airport runways, Jones and Mul- Bernstein et al. 1999). Although a of su- herin (1998) report that the impact of this storm on the percooled large droplet (SLD) icing conditions aloft cur- public utilities industry and the communications system rently is very dif®cult to create because of the lack of was signi®cant. Heavy ice loads on distribution and historical, direct measurements of these phenomena transmission power lines, as well as on trees and branch- throughout the atmosphere, ground reports of freezing es that eventually broke off and fell onto power lines, precipitation can be used to infer partially the clima- caused a loss of electricity to roughly 1.5 million people tology of SLD icing conditions. Bernstein et al. (1997, during the storm. The communication system in this area 1998) and Bernstein and McDonough (2000) examined also was affected when broadcast and two-way towers pilot reports of in-¯ight icing and research aircraft mi- collapsed; nonwireless phone service to some residential crophysics measurements and found that when aircraft customers was disrupted when trees and branches fell ¯ew in close proximity to surface observations of freez- on phone drops. ing rain, freezing drizzle, and/or ice pellets, signi®cant State and federal forests also were damaged signi®- icing was often found in the lower atmosphere. Surface cantly by the storm. Irland (2000) and DeGaetano reports of ice pellets provide a strong indication of the (2000) report that the storm impacted 25 million acres presence of freezing rain aloft that is formed by the of forest in the United States and Canada and the level classical melting mechanism (Hanesiak and Stewart of tree damage was compared to that from a 1938 hur- 1995). Thus, it is important to include them in an anal- ricane that affected the area (Jones and Mulherin 1998). ysis of freezing precipitation. Most damage occurred to hardwood stands; evergreen stands were able to withstand the ice loading. The b. Previous studies storm's impact on the forest not only destroyed plant life and the habitats of many animals, but it also affected One of the most extensive reviews of freezing pre- industries that rely on the forests. The storm interrupted cipitation studies appears in a U.S. Army report by Ben- logging and hauling for a week or more in some places, nett (1959). Bennett shows that freezing precipitation and losses to the region's maple syrup industry exceeded in the United States occurs east of the Rockies primarily $10 million USD (DeGaetano 2000). between November and March. An area of relatively Although the total elimination of these impacts is not high occurrence (greater than 6 days annually) extends possible for every storm, the development and imple- from northwestern Texas northeastward to New Eng- mentation of mitigation procedures based on a clima- land. Bennett estimates that within this `` belt'' tology of these events may have reduced the amount of most areas will experience storms with ice accumula- damage to property, secondary economic losses due to tions between 0.64 and 1.27 cm (0.25 and 0.50 in.) once power outages, as well as the number of injuries and every 3 yr. In Canada and Alaska, limited data show fatalities. Losses to utilities may be reduced by vigilant that freezing precipitation is reported in up to 10% of tree trimming programs around transmission lines, de- the annual hourly observations across southern Alaska signing transmission lines for higher ice loads, provid- and west-central and southeastern Canada. ing electrical line redundancies in areas prone to ice Since Bennett's report, many regional and national storms, and designing procedures to melt and manually on freezing precipitation for the United remove accreted ice. Damage to forests may be reduced States and Canada have been published (McKay and by planting trees that can withstand heavy ice loading. Thompson 1969; Baldwin 1973; Gay and Davis 1993; Injuries and deaths may be reduced by an effective pub- Strapp et al. 1996; Robbins and Cortinas 1996; Branick lic education program in areas susceptible to storms that 1997; Bernstein and Brown 1997; Zerr 1997; La¯amme involve freezing precipitation. and PeÂriard 1998; Stuart and Isaac 1999; Cortinas 2000; The amount of property losses and injuries associated Cortinas et al. 2000; Bernstein 2000; Robbins and Cor- with severe ice storms is large. Changnon (2003) tinas 2002; Changnon and Karl 2003). These studies showed that 87 freezing rainstorms in the United States usually have focused on particular regions of North during 1949±2000 caused more than $16.3 billion (in America and/or one type of freezing precipitation. Mc- 2000 USD) in property losses. The highest frequency Kay and Thompson (1969), Strapp et al. (1996), Laf- of these storms occurred in the northeast United States. lamme and PeÂriard (1998), and Stuart and Isaac (1999) The Deep South, however, had the greatest percentage have produced climatographies of freezing rain and driz- of ice storms that produced insured property losses in zle in Canada. Their results show that freezing rain and excess of $1 million USD. Ice storms were responsible drizzle occur at all times of the year, including June, for 60% of the -related losses during 1988±95 and July, and August, when freezing drizzle can occur in 20% of all weather-caused injuries (Kocin 1997). the far northern regions. Throughout the rest of the year, In addition to the effects of freezing precipitation near freezing drizzle occurs more frequently than freezing

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC APRIL 2004 CORTINAS ET AL. 379 rain, and there is a general decrease in the frequency of ada. Strapp et al. combined their Canadian analysis with freezing rain and drizzle toward the north. Eastern New- a U.S. analysis from Vilcans and Burnham (1989), foundland receives the most freezing rain and drizzle, which undersampled freezing precipitation in the west- exceeding 100 h annually. ern United States. Ice pellets were also not included in Baldwin (1973) included separate maps of the annual their results. The purpose of our study is to show the distribution of freezing precipitation and ice pellets in characteristics of freezing rain, freezing drizzle, and ice the United States. A comparison of his freezing rain pellets that occurred in the United States and Canada map with Bennett's (1959) reveals a general agreement, between 1976 and 1990 using surface observations from even though the data periods were different (1939±48 both countries. In this paper, we do not attempt to pro- for Bennett and 1950±69 for Baldwin) and it appears vide thorough explanations for the distributions. In- that, based upon later climatologies of freezing rain and stead, this paper is meant to serve as a reference for drizzle (e.g., Bernstein and Brown 1997; Robbins and weather forecasters and as a starting point for additional Cortinas 1996), Baldwin (1973) included freezing driz- climatological research as it relates to freezing precip- zle in his freezing rain statistics. Baldwin's results for itation and ice pellets. Section 2 describes the data used freezing rain also agree generally with Changnon and in this study, section 3 describes the analysis, and sec- Karl (2003), who showed national maximums in a por- tion 4 lists the conclusions from this study. tion of New York and Pennsylvania, an east±west zone across the Midwest, along the eastern Appalachians, and 2. Data in the Paci®c Northwest. Additional support for this assertion comes from studies (e.g., Bernstein and Brown Hourly surface observations for the United States and 1997; Robbins and Cortinas 1996; Changnon and Karl Canada were obtained from the National Climatic Data 2003) that show the rarity of freezing rain in the Inter- Center (NCDC) for the period 1976±90. This time pe- mountain West, where freezing drizzle occurs most of- riod was selected because of the data availability at ten. Although they occur about half as frequently as NCDC. Data prior to 1976 were eliminated because freezing rain, Baldwin (1973) showed that ice pellets some station data were only digitized for 1 out of every occur over the same general area as freezing rain, as 3 h, and we wanted to document the duration and evo- well as throughout the Rockies, north of 35ЊN. New lution of these events using the highest observation fre- England experiences the greatest number of days with quency available, namely every hour. Data after 1990 ice pellets, where it occurs on an average of 12 days were also eliminated to assure that present weather ob- annually. servations were taken by human observers and not by The horizontal distributions of freezing rain, freezing the Automated Surface Observing System (ASOS) in drizzle, and ice pellets in the United States have been the United States.1 The surface dataset was created by studied by Gay and Davis (1993), Branick (1997), merging surface data in two different formats (DAT- Strapp et al. (1996), Robbins and Cortinas (1996), Bern- SAV2 and TD-3280) into one consistent format. The stein and Brown (1997), Bernstein (2000), Cortinas quality control procedures for each format are docu- (2000), Cortinas et al. (2000), and Changnon and Karl mented in reports published by the United States Air (2003). All of these studies generally agree with earlier Force (1986) and the United States Department of Com- freezing precipitation studies, but each has studied dif- merce (1994). ferent aspects of the climatology, such as average du- After examining these data, some reports of freezing ration, frequency of large-scale events, proportion of precipitation and ice pellets with a surface temperature freezing to frozen precipitation, and average meteoro- greater than 0ЊC were found. Most of these observations logical conditions during these precipitation types. In were associated with ice pellet observations. We spec- the United States, freezing precipitation and ice pellets ulate that a portion of these were actually obser- occur most frequently from November through March vations erroneously reported as ice pellets, especially at (Robbins and Cortinas 1996) and are usually short lived very warm temperatures. Ice pellets can survive falls (Cortinas et al. 2000). The greatest frequency of freezing through layers of above-freezing air, but their survival precipitation and ice pellets combined is found in the becomes increasingly unlikely with increasing surface Northeast, although the greatest frequency of freezing temperature. A temperature threshold was chosen to iso- drizzle occurs through the western portion of the central late the ice pellet observations from the hail observa- plains (Bernstein and Brown 1997; Bernstein 2000). The tions, since the focus of this study is not hail. Surface temporal and spatial scales associated with these events temperatures associated with each observation of freez- indicate that they are typically mesoscale events (Bran- ing precipitation were examined by plotting a cumu- ick 1997; Gay and Davis 1993; Cortinas 2000). lative frequency distribution of the temperature data (not While these studies provide some information on the shown). Almost none (Ͻ0.1%) of the freezing rain and distribution of freezing precipitation, with the exception freezing drizzle reports occurred at temperatures ex- of Strapp et al. (1996), none have provided extensive information on freezing precipitation across the entire 1 Widespread deployment of the ASOS network did not occur until area of the contiguous United States, Alaska, and Can- the mid-1990s.

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC 380 WEATHER AND FORECASTING VOLUME 19

FIG. 1. Locations used in this study (ALB ϭ Albany, NY; AMA ϭ Amarillo, TX; BHM ϭ Birmingham, AL; BOS ϭ Boston, MA; BRW ϭ Barrow, AK; COU ϭ Columbia, MO; DLH ϭ Duluth, MN; DSM ϭ Des Moines, IA; GEG ϭ Spokane, WA; GSO ϭ Greensville, SC; SHV ϭ Shreveport, LA; YCB ϭ Cambridge Bay, NT; YDA ϭ Dawson, YK; YET ϭ Edson, AB; YFC ϭ Fredericton, NB; YQB ϭ QueÂbec, QC; YRB ϭ Resolute, NT; YWG ϭ Winnipeg, MB; YYE ϭ Fort Nelson, BC; YYQ ϭ Churchill, MB; YYT ϭ St. John's, NF). Plus signs indicate surface station locations, and shaded circles in- dicate locations referenced in the text. ceeding 4ЊC, while ϳ1% of ice pellet reports did. There- fore, it was decided that only those with a temperature Յ4ЊC would be retained. No additional quality control procedures were performed on the data. To ensure reliable statistical results and that the an- nual cycle at each location was sampled adequately, a subjective choice was made to include locations that had at least 80% of the present weather reports for at least 10 of the 15 yr. Using these criteria, 609 stations in were available for analysis (Fig. 1). Moreover, robust statistical descriptions (e.g., median, median absolute deviation, etc.) of the data have been used, when possible, to avoid the in¯uence of occasional data errors on the results. All spatial plots were drawn by hand in order to preserve as much of the detail in the distributions as possible. We estimate that because of this procedure, some of the contoured maps may be in error by no more than 10% of the actual value in areas of large gradients.

3. Climatology a. Horizontal distribution

Freezing precipitation and ice pellets have been ob- FIG. 2. Median annual hours of (a) freezing rain, (b) ice pellets, served throughout most of the United States and Canada and (c) freezing drizzle from 1976 to 1990. (Figs. 2 and 3). They occur most frequently across most of central and eastern Canada, most of Alaska, and north of 32ЊN in the central and eastern contiguous United types of precipitation we examined in this study, freez- States. Moreover, this distribution shows synoptic-scale ing drizzle is the most widespread and occurs most fre- and some mesoscale variability (this variability will be quently across the central part of each country and the discussed in the following subsections). Of the three eastern part of Canada and the western and northern

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC APRIL 2004 CORTINAS ET AL. 381 coasts of Alaska. Ice pellets and freezing rain occur most frequently in the eastern part of each country.

1) FREEZING RAIN Freezing rain occurs when frozen precipitation melts in an elevated warm (Ͼ0ЊC) layer, then becomes su- percooled before impacting objects at the surface. Nu- merous investigators have shown that this thermal strat- i®cation can occur primarily as a result of air¯ow as- sociated with extratropical and secondarily by the proximity of water bodies and topographical effects (Stewart and King 1987; Stewart 1992; Bernstein 2000; Rauber et al. 2000; Robbins and Cortinas 2002; Roebber and Gyakum 2003). These conditions combine to pro- duce several dominant areas of freezing rain. Most of the United States and Canada receive less than 10 h of freezing rain most years, with the highest frequencies near the St. Lawrence River valley and Newfoundland, where usually 30 or more hours of freez- ing rain are observed annually (Fig. 2a). These results are similar to those of Stuart and Isaac (1999), who found in excess of 25 h of freezing rain annually in several regions across eastern Canada. Regional maxima also occurred over the Columbia basin, across much of the Midwest, as well as along the east slope of and within the Appalachians. Freezing rain also occurs less frequently (5±10 h annually) across the central plains of the United States and Canada, portions of central and western Canada, and a small part of Alaska. The results are consistent with studies by Bernstein and Brown (1997), Robbins and Cortinas (2002), Stuart and Isaac (1999), and Changnon and Karl (2003). A distinct min- imum in freezing rain frequencies was found along the western slope of the Appalachians, which was attributed to a rain shadow effect by Bernstein (2000). The variability associated with the annual distribution can be large at many locations, particularly in areas of frequency maxima (Figs. 2a and 3a). The freezing rain observations at some locations in eastern Canada and the United States varied from near 0 to 70 h annually (Fig. 4a). Since these events often occur due to warm air above the surface on the cold side of warm and stationary fronts, annual variations in their frequency is likely associated with changes in the storm track and location of the surface freezing line. A good example site is Greensboro, North Carolina (GSO), where the track of low pressure centers relative to the southern end of the Appalachians and the location of the strength of high pressure over New England is im- FIG. 3. Median annual precipitation-type days of (a) freezing rain, portant to creating a classical setup of midlevel warm (b) ice pellets, and (c) freezing drizzle from 1976 to 1990. (See text air advection over surface cold air dammed along the for description of a precipitation-type day.) mountains. The variability can also be greatly in¯uenced by single events, such as the ice storm of 1998 (de- scribed in the introduction), which produced a major An analysis of freezing rain days (Fig. 3a), days dur- fraction of the typical freezing rain hours per year in ing which at least one hourly observation of freezing one storm. (That event is not included in the dataset rain occurred, shows that the general spatial pattern is used here.) similar to the median annual frequency of hourly ob-

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC 382 WEATHER AND FORECASTING VOLUME 19

servations (Fig. 2a). For most of central and north- western United States, the median number of freezing rain days is between 1 and 4. While in the northeast United States and extreme Canada, the median exceeds 7 days annually. Although Changnon and Karl (2003) show more mesoscale variability in their maps of freez- ing rain days, the general spatial pattern agrees with their results, particularly during their latter data period (1981±96). Based on this distribution and on results of other freezing rain studies (Strapp et al. 1996; Stuart and Isaac 1999; Bernstein 2000; Cortinas 2000), we speculate that the geographic maxima occur because of their proximity to winter tracks, large bodies of water, and to- pographical features. For example, the Atlantic Ocean and the provide rich sources of warm moist air that can be advected over subfreezing surface layers, such as those that form in valley locations (e.g., the St. Lawrence River valley) or cold continental air masses that are in place as an pass- es. Although extratropical cyclones also impact the western coasts, subfreezing surface air is often missing because the dominant synoptic circulation usually does not advect subfreezing surface air into the region. There are exceptions, however, such as the Columbia basin, where cold continental air becomes trapped (Whiteman et al. 2001).

2) ICE PELLETS Freezing rain and ice pellets usually require an ele- vated warm layer and a subfreezing surface layer, al- though Strapp et al. (1996) have shown that during a small fraction of the freezing rain observations at St. John's, Newfoundland, no warm layer was present aloft. The most likely mechanism for ice pellet formation is through partial or complete melting of ¯akes in an elevated warm layer and then refreezing as they descend through the subfreezing layer. Both Hanesiak and Stew- art (1995) and Zerr (1997) found that incomplete melt- ing of snow¯akes in the warm layer was the primary factor in the production of ice pellets. The strength and depth of the cold layer were considered to be secondary factors. The geographic distribution of ice pellets is sim- ilar to that of freezing rain. Ice pellets occur most often in the eastern Canadian provinces, where several areas usually experience more than 20 h (Fig. 2b) or 10 ice pellet days (Fig. 3b) per year.

FIG. 4. Distribution of (a) freezing rain, (b) ice pellet, and (c) freezing drizzle observations at selected locations (see Fig. 1 for names of locations). Each box encloses 50% of the data with the median value of the variable displayed as a line within the box. The lines extending above and below each box indicate the maximum and minimum values in between the upper and lower inner fences (1.5 times the distance of the interquartile range away from the quartiles). Open circles are far-out points (outside of the inner fences).

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC APRIL 2004 CORTINAS ET AL. 383

In the United States, ice pellets usually occur for more than 10 h annually in the Great Lakes, the Northeast, and the mid-Atlantic states. Along the mid-Atlantic and northeast coasts, the gradient in the occurrence of ice pellets is much less than that of freezing rain. Bernstein (2000) noted that this was due to ice pellets surviving to reach the surface at above freezing temperatures along the coast, where freezing rain would have changed to rain. Ice pellets also extended westward across Can- ada, and reached the Paci®c coasts of British Columbia and extreme southeastern Alaska. In eastern Canada and the United States, annual ice pellet totals were highly variable from year to year, rang- ing from near 0 to 70 in some locations (Fig. 4b). Since the mechanism for ice pellets is similar to that for freez- ing rain, variability in the location of the storm track and surface freezing line are again likely to explain FIG. 5. Median annual hours of freezing rain and freezing drizzle much of the annual changes in ice pellet frequency. combined from 1976 to 1990.

3) FREEZING DRIZZLE with upslope ¯ow along the slopes of major mountain There are two established mechanisms for freezing ranges and in valleys where stagnant cold and moist air drizzle formation: collision±coalescence (or warm rain) pools were present. We speculate that these locations and the ``classical'' melting processes (Bocchieri 1980; are favored because most freezing drizzle in these re- Huffman and Norman 1988; Ohtake 1963). The warm gions occurs as a result of collision and coalescence rain process typically occurs in composed pri- within warm or subfreezing clouds (Strapp et al. 1996; marily of liquid water and with temperatures greater Bernstein 2000; Rauber et al. 2000), with the latter en- than about Ϫ10ЊC. This thermodynamic environment vironment present in 75%, 35%±42%, and 0%±83% of supports supercooled droplet production due to the very the cases described by Rauber et al. 2000, Strapp et al. low concentrations of active ice nuclei. Previous studies 1996, and Jeck (1996), respectively. suggest that this is the primary process for the formation Gentle synoptic forcing associated with warm fronts of freezing drizzle (Bernstein 2000; Rauber et al. 2000; and stationary fronts has also been shown to be asso- Kajikawa et al. 2000). ciated with freezing drizzle production, and likely ex- Freezing drizzle occurs over a large portion of Can- plains much of its occurrence across the middle of the ada; the central United States; western and northern continent (Bernstein et al. 1998; Bernstein 2000). Rel- Alaska (Fig. 2c), where maximum annual frequencies atively clean air is often present within the formation greater than 40 h occur on the southwest side of Hudson zones in all of the situations described above. Rasmus- Bay from Manitoba to western Ontario; and across east- sen et al. (2002) have shown that this can be important ern Labrador and Newfoundland. Freezing drizzle oc- to the production of freezing drizzle. Strapp et al. (1996) curs most frequently in northeastern Newfoundland, speculated that the presence of sea ice along the coast where the median number of freezing drizzle hours is of Newfoundland might have been important in keeping ϳ100 annually. It is the primary type of freezing pre- surface conditions cold enough to generate freezing cipitation to extend to the east slope of the Rocky Moun- drizzle instead of above-freezing drizzle. Fluctuations tains, where upslope clouds with warm tops produce an in the amount of sea ice present, as well as the frequency excellent situation for freezing drizzle development. The of warm-topped clouds may explain the large variability number of freezing drizzle days (Fig. 3c) is also greater in the annual frequency of freezing drizzle there. than those for the other types of precipitation. At many A plot of the two types of freezing precipitation com- locations, the variability associated with the annual bined shows a very broad distribution that covers all of freezing drizzle distribution is usually larger than other the central and eastern United States and Canada (Fig. types of precipitation (Fig. 4c). 5). In general, a broad swath extends from the western The widespread geographic distribution of freezing high plains through the Great Lakes region, into most drizzle suggests that several factors may contribute to of eastern Canada, the Maritime Provinces, New Eng- its formation, including water source proximity, topog- land, and the eastern slope of the Appalachians. New- raphy, and synoptic forcing. Several studies have noted foundland receives the most freezing precipitation an- that a large portion of freezing drizzle events near bodies nually in the United States and Canada, with an annual of water were associated with onshore ¯ow (Strapp et frequency that is almost a factor of 2 larger than the al. 1996; Stuart and Isaac 1999; Bernstein 2000). Bern- highest frequency observed elsewhere. stein (2000) also found freezing drizzle was associated It is also of interest to examine what fraction of winter

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC 384 WEATHER AND FORECASTING VOLUME 19

FIG. 6. Percentage of hourly winter weather observations (freezing precipitation, ice pellets, and snow) during which freezing precipi- tation occurred between 1976 and 1990.

FIG. 7. Monthly distribution (%) of all freezing rain, ice pellet, and observations (i.e., freezing precipitation, ice pellets, and freezing drizzle observations from 1976 to 1990. Monthly data have snow) are freezing precipitation observations (Fig. 6). been normalized to a 30-day month. Because of the dominance of snow in Canada and the northern United States during the winter, the percentage January, with January averages highest in the East and of freezing precipitation observations is extremely small December highest in the West. Across most of Canada, there, whereas the percentage over the southern and Stuart and Isaac (1999) found that freezing precipitation southeastern United States is relatively high, exceeding occurs primarily between October and May, with March 70% in southern Texas.2 We believe that this pattern in being the month with the highest frequency. the southern United States occurs as a result of cold air The data also show that a seasonal dependence of layers near the surface that are insuf®ciently deep to freezing precipitation and ice pellet occurrence varies displace warm air aloft or cold enough to cause ice with latitude and, to a lesser degree, longitude. Since nucleation. This creates a thermodynamic pro®le that is the spatial dependence is similar for all types of freezing conducive only to freezing precipitation, as shown by precipitation and ice pellets, we only show histograms Strapp et al. (1996) and Rauber et al. (2000). b. Temporal distribution In the United States and Canada, freezing precipita- tion and ice pellets occur most frequently during the winter months (December, January, and February; Fig. 7), except along the Arctic coast, where they occur most- ly during the warm season (May±October, Figs. 7 and 8). When the data are normalized to a 30-day month, the months of maximum occurrence for freezing rain, ice pellets, and freezing drizzle are January, February, and December, respectively. In the fall, there is a rapid increase in the frequency of freezing precipitation be- tween October and December and a similar decrease between March and April, although the decrease of ice pellet frequency is more gradual in the spring. The results are consistent with other national studies. Changnon and Karl (2003) showed that peak months of freezing rain in the United States were December and FIG. 8. Median annual hours of freezing drizzle by month at se- lected locations. The x axis of each plot is the month and the y axis is the median number of observations between 1976 and 1990 for 2 Results in the extreme southern United States should be used that month. Numbers shown below each station name indicate the cautiously since there are only a few observations in this region with median number of freezing drizzle observations for the 15-yr period, a temperature Յ4ЊC. shown in Fig. 2c.

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC APRIL 2004 CORTINAS ET AL. 385 for freezing drizzle (Fig. 8). At central longitudes, freez- ing precipitation moves from as far south as near the Gulf coast in midwinter to the Arctic coast in midsum- mer. The migratory nature of freezing drizzle is less evident, but still present in the western and eastern parts of the continent, though freezing drizzle is much less frequent at central and southern latitudes. Freezing driz- zle in coastal stations in the Canadian arctic and Hudson Bay areas is often maximized just before the sea freezes in the fall or after it thaws in the spring (Strapp et al. 1996; Stuart and Isaac 1999). Although freezing precipitation and ice pellets are rare across the United States and southern Canada dur- ing the months, they do occur at far northern sites, such as Barrow, Alaska, and Resolute Bay, North- west Territories (not shown). It is during these months that temperatures are moderate enough to allow liquid- phase clouds to exist and perhaps become prevalent at far northern latitudes. Winter clouds tend to be glaciated in this region (Stuart and Isaac 1999; Bernstein et al. 2003). Our examination of the surface observations indicates FIG. 9. Diurnal distribution (%) of all freezing rain, ice pellet, and that there is an association between the occurrence of freezing drizzle observations from 1976 to 1990 for locations in- cluding and south of 66ЊN(0NSTϭ 24 NST). Time is expressed freezing precipitation and ice pellets and the diurnal as normalized solar time (see text for description of this quantity). solar cycle. We investigate this relationship by using the local sunrise and sunset times for each site (south of, and including, 66ЊN) to normalize the observation times cipitation (or cessation of precipitation) following each [hereafter referred to as normalized solar time (NST)]. event. The relative frequency was computed by dividing This normalization sets the sunrise time to 1200 UTC each precipitation type at each hour by the total number and the sunset time to 0000 UTC, and divides the hours of events that ended in that hour. For example, if freez- of daylight and night between 12 NST hours of sunlight ing rain was observed between 0800 and 1100 NST, and 12 NST hours of darkness. The details of this cal- then no precipitation was reported at 1200 NST, we culation are discussed by Kelly et al. (1985), who per- would count one event that was followed by the ces- formed a similar analysis for severe . The sation of precipitation at 1200 NST. Figure 10a shows purpose of this normalization is to identify any rela- that for all freezing rain events that ended between 1100 tionship between the diurnal solar cycle and the occur- and 1200 NST, ϳ31% reported no precipitation and 27% rence of freezing precipitation. reported only rain in the following hour (1200 NST). Our NST analysis shows a relationship between the This analysis shows that freezing drizzle or freezing diurnal solar cycle and freezing rain and freezing driz- rain events most commonly end with a period of no zle, but not for ice pellets (Fig. 9). Freezing rain and precipitation (Figs. 10a,c), while ice pellets are most freezing drizzle occur most frequently before sunrise commonly followed by a change over to snow, rather (typically the coldest time of the day) then drop off than precipitation ending (Fig. 10b). The percentage of sharply during the morning, reaching a minimum during freezing drizzle events ending with a cessation of pre- the late afternoon. A similar diurnal trend was reported cipitation is elevated at night and peaks at sunrise (52%; by Strapp et al. (1996) for freezing precipitation at St. Fig. 10c). Those that end with a change in precipitation John's, Newfoundland. There is no apparent diurnal type are strongly dominated by snow, likely due to a trend for ice pellets, perhaps because their occurrence decrease in -top temperature or seeding from is less sensitive to the surface temperature. above (cf. Politovich and Bernstein 1995). Changes to The expected reason for the diurnal trend in freezing (nonfreezing) drizzle are markedly less frequent, but are precipitation is that the decrease in freezing precipitation maximized during daylight hours, when freezing drizzle occurs because of surface heating by insolation. How- events ending with a cessation of precipitation are at a ever, other factors responsible for the decrease after sun- relative minimum. At this time of day, slight increases rise can be identi®ed by examining the temporal fre- in surface temperature are likely responsible for the quency distribution of different types of precipitation change, since both drizzle and freezing drizzle are typ- (or cessation of precipitation) reported immediately fol- ically formed via the collision±coalescence process. lowing each freezing precipitation event (one or more When a freezing rain event ends with a change in sequential hours of freezing precipitation). We counted precipitation type, that type is most often rain during every event and noted the hour (NST) and type of pre- daylight hours, but evenly distributed among several

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC 386 WEATHER AND FORECASTING VOLUME 19

FIG. 11. Duration of freezing precipitation events. Each event is de®ned as a set of sequential observations of a particular type of freezing precipitation, including a mixture of precipitation types.

is also at its relative minimum during daylight hours. The same pattern holds true for freezing rain following an ice pellet event (Fig. 10b; day, ϳ8%; night, ϳ12%), while the reverse is true for subsequent rain, which reaches its relative maximum (ϳ20%) during the day, and its relative minimum (ϳ15%) at night. This makes sense, given expected diurnal temperature changes.

c. Duration The duration of freezing precipitation events was evaluated by counting the number of sequential hourly observations of each type at all stations. Given this def- inition, two reports of freezing rain with a 1-h break would be counted as two 1-h events. Although one could arbitrarily de®ne an event many ways, Cortinas (2000) showed that for freezing rain increasing the number of intervening hours did not signi®cantly change the dis- tribution since these events typically do not have long intervening periods during which precipitation was not reported. The duration of most freezing precipitation events is less than 2 h, with ice pellets making up the largest FIG. 10. Frequency (relative to total number of events that ended in that hour) of precipitation type (or no precipitation) following a percentage of these short-lived events (Fig. 11). The sequential set of (a) freezing rain, (b) ice pellet, and (c) freezing decrease in the frequency of events with increasing drizzle observations (0 NST ϭ 24 NST). Time is expressed as nor- event time is exponential and decreases to less than 1% malized solar time (see text for description of this quantity). for event durations in excess of 9 h. Although most freezing precipitation events are short lived, it is im- portant to note that 5%, 11%, and 15% of ice pellets, precipitation types, including snow, at night (Fig. 10a). freezing rain, and freezing drizzle events, respectively, During the day, freezing rain events ending in snow or lasted longer than 4 h. Freezing drizzle events tended a cessation in precipitation are at a relative minimum. to last the longest. Strapp et al. (1996) and Bernstein Though snow is the most common precipitation type to (2000) found several events that lasted for more than follow an ice pellet event at any time of day, this change 24 h at St. John's and Spokane. These long-lived freez-

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC APRIL 2004 CORTINAS ET AL. 387

TABLE 1. Frequency (%) of concurrent precipitation observations. types of precipitation, whereas ice pellets are most often Columns may not add up to 100% since more than one type of mixed with snow, rain, or freezing rain (Table 1). Al- precipitation can be reported (FZRA, freezing rain; FZDZ, freezing drizzle; PE, ice pellets; DZ, drizzle; RA, rain; SN, snow; T, thunder; though freezing rain usually is not mixed with other none, no other precipitation type). precipitation types, the predominant mixtures are with ice pellets and snow. FZRA (%) FZDZ (%) PE (%) Like freezing rain, freezing drizzle occurs most fre- FZRA Ð 1 18 quently in the absence of other precipitation types. FZDZ 2 Ð 3 When a mixture does occur, it nearly always is with PE 18 2 Ð DZ 0 0 1 snow. Freezing drizzle has been shown by a number of RA 0 1 18 studies (Huffman and Norman 1988; Strapp et al. 1996; SN 14 24 37 Jeck 1996; Rauber et al. 2000) to most likely result from T 1 0 1 the collision±coalescence process, and often in the ab- None 69 73 30 sence of a warm layer. The existence of the ice phase represents a competing process that may eventually eliminate freezing drizzle. The predominance of freez- ing drizzle and especially freezing rain events, although ing-drizzle-only reports (73%) suggests that large con- uncommon, may produce the greatest accumulation of centrations of ice particles are usually not present in ice, most hazardous conditions, and most signi®cant dis- these clouds, and further supports that these are the ruptions, particularly for transportation. The ice storm favored conditions for freezing drizzle development. of 1998 in the northeastern United States and south- Ice pellet mixtures, when they occur, are usually with eastern QueÂbec and eastern Ontario is a case in point. snow, followed by ice pellet mixtures that include rain and freezing rain. Ice pellets typically fall within a nar- d. Concurrent weather conditions row transition zone between these precipitation types, caused by the narrowing of the melting zone in the Other surface conditions associated with freezing pre- direction of the cold . From the warm to the cipitation were analyzed in order to understand which cold side of the transition zone, snow from above falls conditions might provide useful information to fore- into an increasingly shallow melting zone, resulting in casters anticipating freezing precipitation and icing con- a transition from complete melting to incomplete melt- ditions at the surface and aloft. These surface conditions ing and, eventually, no melting. Martner et al. (1993) also provide some insight into the predominant pro- found that surface precipitation type changed from rain cesses that control the evolution of the freezing precip- and/or freezing rain to ice pellets, then snow across a itation. In this study we examined concurrent types of transition zone during the 1990 St. Valentine's Day ice precipitation and the near-surface (2 m) air temperature. storm. Robbins and Cortinas (2002) have noted, how- Surface observations reveal that freezing rain and freez- ever, that freezing rain does not always occur as part of ing drizzle occur most frequently in the absence of other this classical transition. The dominance of a snow±ice pellet mixture may be related to the size distribution of the frozen particles and the speed at which the thermodynamic environment evolves. In the absence of thermal advection, latent heat- ing associated with melting at the top of the elevated warm layer will ultimately cool the warm layer to 0ЊC (Stewart 1985; Kain et al. 2000). The frequency of a snow±ice pellet mixture as well as the short duration of most ice pellet events (Fig. 11) suggests that the snow- ¯ake size distribution may have a greater role in dis- tinguishing between snow and ice pellets, than between other types of precipitation. The lack of ice pellet mix- tures with drizzle and freezing drizzle may be due to the distinct difference in formation mechanisms, since partial melting is important to the formation of ice pel- lets and freezing drizzle is primarily formed in an all- liquid-water environment with no melting. The surface temperature for all reports of freezing precipitation and ice pellets were examined because of its in¯uence in determining precipitation type at the ground. The analysis shows that freezing precipitation FIG. 12. Air temperature (2 m AGL) associated with freezing and ice pellets are associated most frequently with sur- precipitation and ice pellet observations. face temperatures slightly less than 0ЊC (Fig. 12). Al-

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC 388 WEATHER AND FORECASTING VOLUME 19

FIG. 13. Distribution of the length of time that the air temperature (2 AGL) was Յ0ЊC, starting with the ®rst observation freezing pre- cipitation events, for freezing precipitation events.

though most observations of freezing precipitation oc- curred at T ϾϪ15ЊC, freezing rain, freezing drizzle, and ice pellets were observed at surface temperatures near the theoretical limit for the existence of super- cooled liquid water (Ϫ40ЊC). The high frequency (ϳ85%) of freezing rain and freezing drizzle reports with surface temperatures between Ϫ10Њ and 0ЊC sup- ports the theory that warmer surface temperatures also re¯ect warmer in-cloud temperatures, and that an in- suf®cient number of active ice nuclei within this tem- perature range prevents the formation of frozen precip- itation. This also applies to the freezing rain and freezing drizzle observations that occurred with a 2-m temper- ature greater than 0ЊC in our dataset. The temperature of the ground may have been ϳ0ЊC while the 2-m tem- perature was slightly higher than 0ЊC. Of particular interest to transportation, energy, and communication industries is the length of time that the near-surface (2 m) air temperature is less than or equal to 0ЊC after precipitation begins. The subfreezing period following the initial observation of freezing precipita- FIG. 14. (a) Median and (b) maximum number of hours during which the air temperature (2 AGL) was Յ0ЊC, starting with the ®rst tion shows that, generally, subfreezing conditions last observation of freezing precipitation events. (See Fig. 1 for location longer during freezing drizzle events than during freez- names.) ing rain events (Fig. 13), although there is a large var- iability among the sites (Fig. 14). As expected, the great- est subfreezing periods occur at sites in Alaska, Canada, 4. Conclusions and the northwest United States, where the period can In this study we investigated the distribution of freez- exceed 4800 h (200 days) for freezing drizzle and 2200 ing precipitation and ice pellets and these are the key h (92 days) for freezing rain in Alaska. In Canada these ®ndings. periods exceed 2 weeks (400 h or 17 days). Even in the southern United States, these periods can exceed 50 h, • There is large spatial variability in the annual fre- causing signi®cant problems for transportation. quency of freezing precipitation and ice pellets across

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC APRIL 2004 CORTINAS ET AL. 389

North America. These precipitation types occur most Baldwin, J., 1973: The of the United States. NOAA, 113 frequently across the central and eastern portions of pp. [NTIS COM-74-11708/6.] Bendel, W. B., and D. Paton, 1981: A review of the effect of ice the United States and of Canada, much of Alaska, and storms on the power industry. J. Appl. Meteor., 20, 1445±1449. the northern shores of Canada. Bennett, I., 1959: Glaze: Its meteorology and climatology, geograph- • The spatial distribution of freezing precipitation and ical distribution, and economic effects. Environmental Protection ice pellets suggests that topographical features, water Research Division Tech. Rep. EP-105, Headquarters, U.S. Army source proximity, and location relative to climatolog- Quartermaster, Research and Engineering Command, Natick, ical extratropical cyclone tracks may determine the MA, 217 pp. [NTIS AD-216668.] Bernstein, B., 2000: Regional and local in¯uences on freezing drizzle, climatology of freezing precipitation. freezing rain, and ice pellets. Wea. Forecasting, 15, 485±508. • Freezing precipitation and ice pellets occur most often ÐÐ, and B. Brown, 1997: A climatology of supercooled large drop from December to March, except in northern Canada conditions based upon surface observations and pilot reports of and Alaska where it occurs during the warmer . icing. Preprints, Seventh Conf. on Aviation, Range, and Aero- • Freezing rain and freezing drizzle appear to be in¯u- space Meteorology, Long Beach, CA, Amer. Meteor. Soc., 82± enced by the diurnal solar cycle. These precipitation 87. types occur most often just before sunrise and have ÐÐ, and F. McDonough, 2000: Environments associated with large droplet, small droplet, mixed-phase icing, and glaciated condi- the lowest frequency near sunset, whereas the fre- tions aloft. Preprints, Ninth Conf. on Aviation, Range, and Aero- quency of ice pellets is nearly constant throughout the space Meteorology, Orlando, FL, Amer. Meteor. Soc., 239±244. day. The rapid frequency decrease near sunrise for ÐÐ, T. Omeron, F. McDonough, and M. Politovich, 1997: The re- freezing rain is a result of precipitation cessation and lationship between aircraft icing and synoptic-scale weather con- conversion to rain. Freezing drizzle events either end- ditions. Wea. Forecasting, 12, 742±762. ed with a cessation of precipitation or a change to ÐÐ, ÐÐ, M. Politovich, and F.McDonough, 1998: Surface weather features associated with freezing precipitation and severe in- snow. ¯ight aircraft icing. Atmos. Res., 46, 57±73. • Freezing precipitation is often short lived; however, ÐÐ, T. Ratvasky, D. Miller, and F. McDonough, 1999: Freezing rain 5%, 11%, and 15% of ice pellets, freezing rain, and as an in-¯ight icing hazard. Preprints, Eighth Conf. on Aviation, freezing drizzle events, respectively, lasted for more Range, and Aerospace Meteorology, , TX, Amer. Meteor. than 4 h. Soc., 38±42. • Most freezing rain and freezing drizzle are not mixed ÐÐ, F. McDonough, and R. Bullock, 2003: An inferred climatology with other precipitation types, whereas most reports of supercooled large droplet icing conditions for North America. Proc. 39th Aerospace Science Meeting and Exhibit, Reno, NV, of ice pellets included other types of precipitation. American Institute of Aeronautics and Astronautics, Paper AIAA Snow is the most frequent type of precipitation re- 2003-0563. ported with freezing precipitation and ice pellets. Bocchieri, J. R., 1980: The objective use of upper air soundings to • Freezing precipitation and ice pellets occur most fre- specify precipitation type. Mon. Wea. Rev., 108, 596±603. quently with a surface (2 m) temperature slightly less Branick, M. L., 1997: A climatology of signi®cant winter-type weath- than 0ЊC. er events in the contiguous United States, 1982±94. Wea. Fore- • Subfreezing periods following the initiation of freez- casting, 12, 193±207. Changnon, S. A., 2003: Characteristics of ice storms in the United ing precipitation are usually longer for freezing drizzle Sates. J. Appl. Meteor., 42, 630±639. than for freezing rain events and can exceed 2 weeks ÐÐ, and T. R. Karl, 2003: Temporal and spatial variations of freezing at many locations across the northern United States, rain in the contiguous United States: 1948±2000. J. Appl. Me- Canada, and Alaska. teor., 42, 1302±1315. Cortinas, J. V., Jr., 2000: A climatology of freezing rain over the Acknowledgments. We thank Mr. Neal Lott at the Great Lakes region of North America. Mon. Wea. Rev., 128, NOAA/National Climatic Data Center for providing us 3574±3588. with the surface dataset; Ms. Kathleen Jones from the ÐÐ, C. C. Robbins, B. C. Bernstein, and J. W. Strapp, 2000: A climatography of freezing rain, freezing drizzle, and ice pellets Army's Cold Regions Research and Engineering Lab- across North America. Preprints, Ninth Conf. on Aviation, Range, oratory for providing information relevant to the energy and Aerospace Meteorology, Orlando, FL, Amer. Meteor. Soc., industry; and three anonymous reviewers who provided 292±297. comments that improved this paper. Partial funding for DeGaetano, A. T., 2000: Climatic perspective and impacts of the 1998 this research was provided under NOAA±OU Cooper- northern New York and New England ice storm. Bull. Amer. ative Agreement NA17RJ1227, the National Weather Meteor. Soc., 81, 237±254. Service, the National Center for Atmospheric Research, Forbes, G. S., R. A. Anthes, and D. W. Thompson, 1987: Synoptic and mesoscale aspects of an Appalachian ice storm associated and Transport Canada. A portion of this research was with cold-air damming. Mon. Wea. Rev., 115, 564±591. in response to requirements and funding by the Federal Gay, D. A., and R. E. Davis, 1993: Freezing rain and climatology Aviation Administration (FAA). The views expressed of the southeastern USA. Res., 3, 209±220. are those of the authors and do not necessarily represent Glickman, T. S., Ed., 2000: . 2d ed. Amer. the of®cial policy or position of the FAA or NOAA. Meteor. Soc., 855 pp. Gyakum, J. R., and P.J. Roebber, 2001: The 1998 ice stormÐAnalysis of a planetary-scale event. Mon. Wea. Rev., 129, 2983±2997. REFERENCES Hanesiak, J., and R. Stewart, 1995: The mesoscale and microscale Ashenden, R., and J. Marwitz, 1997: Turboprop aircraft performance structure of a severe ice pellet storm. Mon. Wea. Rev., 123, response to various environmental conditions. J. Aircraft, 34, 3144±3162. 278±287. Huffman, G. J., and G. A. Norman Jr., 1988: The supercooled warm

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC 390 WEATHER AND FORECASTING VOLUME 19

rain process and the speci®cation of freezing precipitation. Mon. layer clouds: The role of radiative cooling of cloud droplets, Wea. Rev., 116, 2172±2182. cloud condensation nuclei, and ice initiation. J. Atmos. Sci., 59, Irland, L. C., 2000: Ice storm 1998 and the forests of the Northeast. 837±860. J. For., 96, 32±40. Rauber, R. M., M. K. Ramamurthy, and A. Tokay, 1994: Synoptic Jeck, R., 1996: Representative values of icing-related variables aloft and mesoscale structure of a severe freezing rain event: The St. in freezing rain and freezing drizzle. FAA Tech. Note DOT/FAA/ Valentine's Day ice storm. Wea. Forecasting, 9, 183±208. AR-TN95/119, FAA Technical Center, Atlantic City, NJ, 54 pp. ÐÐ, L. Olthoff, M. Ramamurthy, and K. Kunkel, 2000: The relative Jones, K. F., and N. D. Mulherin, 1998: An evaluation of the severity importance of warm rain and melting processes in freezing pre- of the January 1998 ice storm in northern New England. CRREL cipitation events. J. Appl. Meteor., 39, 1185±1195. Rep., Rep. for FEMA Region 1, 66 pp. [Available from U.S. Robbins, C., and J. V. Cortinas Jr., 1996: A climatology of freezing Army Cold Regions Research and Engineering Laboratory, 72 rain in the contiguous United States: Preliminary results. Pre- Lyme Rd., Hanover, NH 03755.] prints, 15th Conf. on Weather Analysis and Forecasting, Nor- Kain, J. S., S. M. Goss, and M. E. Baldwin, 2000: The melting effect folk, VA, Amer. Meteor. Soc., 124±126. as a factor in precipitation-type forecasting. Wea. Forecasting, ÐÐ, and ÐÐ, 2002: Local and synoptic environments associated 15, 700±714. with freezing rain in the contiguous United States. Wea. Fore- Kajikawa, M., K. Kikuchi, Y. Asuma, Y. Inoue, and N. Sato, 2000: casting, 17, 47±65. Supercooled drizzle formed by condensation±coalescence in the Roebber, P. J., and J. R. Gyakum, 2003: Orographical in¯uences on mid-winter season of the Canadian Arctic. Atmos. Res., 52, 293± the mesoscale structure of the 1998 ice storm. Mon. Wea. Rev., 301. 131, 27±50. Kelly, D., J. Schaefer, and C. Doswell III, 1985: The climatology of Sand, W., W. Cooper, M. Politovich, and D. Veal, 1984: Icing con- non-tornadic severe events in the United States. ditions encountered by a research aircraft. J. Climate Appl. Me- Mon. Wea. Rev., 113, 1997±2014. teor., 23, 1427±1440. Kocin, P. J., 1997: Some thoughts on the societal and economic im- Stewart, R. E., 1985: Precipitation types in winter storms. Pure Appl. pacts of winter storms. Proc. Workshop on Social and Economic Geophys., 123, 597±609. Impacts of Weather, Boulder, CO, NCAR, 55±60. ÐÐ, 1992: Precipitation types in the transition region of winter La¯amme, J. N., and G. PeÂriard, 1998: The climate of freezing rain storms. Bull. Amer. Meteor. Soc., 73, 287±296. over the province of QueÂbec in Canada: A preliminary analysis. ÐÐ, and P. King, 1987: Freezing precipitation in winter storms. Atmos. Res., 46, 99±111. Mon. Wea. Rev., 115, 1270±1279. Martner, B. E., R. Rauber, R. Rasmussen, E. Prater, and M. Rama- Strapp, J., R. Stewart, and G. Isaac, 1996: A Canadian climatology of murthy, 1992: Impacts of a destructive and well-observed cross- freezing precipitation and a detailed study using data from St. country . Bull. Amer. Meteor. Soc., 73, 169±173. John's, Newfoundland. Proc. FAA Int. Conf. on Aircraft In¯ight ÐÐ, J. B. Snider, R. J. Zamora, G. P. Byrd, T. A. Niziol, and P. I. Icing, Vol. 2, Spring®eld, VA, FAA, DOT/FAA/AR-96/81, 45± Joe, 1993: A remote-sensing view of a freezing-rain storm. Mon. 56. Wea. Rev., 121, 2562±2577. Stuart, R., and G. Isaac, 1999: Freezing precipitation in Canada. Marwitz, J., M. Politovich, B. Bernstein, F. Ralph, P. Neiman, R. Atmos.±Ocean, 37, 87±102. Ashenden, and J. Bresch, 1997: Meteorological conditions as- Toth, J. J., 1988: Comment on ``Synoptic and mesoscale aspects of sociated with the ATR72 aircraft accident near Roselawn, In- an Appalachian ice storm associated with cold-air damming.'' diana, on 31 October 1994. Bull. Amer. Meteor. Soc., 78, 41± Mon. Wea. Rev., 116, 2002±2002. 52. United States Air Force, 1986: DATSAV2 surface. Climatic database McKay, G. A., and H. A. Thompson, 1969: Estimating the hazard of users handbook 4, Rep. USAFETAC/UH-86/004, 52 pp. [Avail- ice in Canada from climatological data. J. Appl. Me- able from National Climatic Data Center, 151 Patton Ave., Ashe- teor., 8, 927±935. ville, NC 28801-5001.] NOAA, cited 1998: Eastern US ¯ooding and ice storm. [Available United States Department of Commerce, 1994: Hourly surface air- online at http://www.ncdc.noaa.gov/ol/reports/janstorm/janstorm. ways observations. NCDC TD-3280, 40 pp. [Available from html.] National Climatic Data Center, 151 Patton Ave., Asheville, NC Ohtake, T., 1963: Hemispheric investigation of warm rain by radio- 28801-5001.] sonde data. J. Appl. Meteor., 2, 594±607. Vilcans, J., and D. Burnham, 1989: Climatological study to determine Pike, W. S., 1995: Extreme warm frontal icing on 25 February 1994 the impact of icing on the Low Level Windshear Alert System. causes an aircraft accident near Uttoxeter. Meteor. Appl., 2, 273± Tech. Rep. DOT-TSC-FAA-89-2, 32 pp. [Available from the 279. National Technical Information Service, Spring®eld, VA 22161.] Politovich, M. K., and B. C. Bernstein, 1995: Production and deple- Whiteman, C. D., S. Zhong, W. J. Shaw, J. M. Hubbe, X. Bian, and tion of supercooled liquid water in a Colorado winter storm. J. J. Mittelstadt, 2001: Cold pools in the Columbia basin. Wea. Appl. Meteor., 34, 2631±2648. Forecasting, 16, 432±447. Rasmussen, R. M., I. Geresdi, G. Thompson, K. Manning, and E. Zerr, R. J., 1997: Freezing rain: An observational and theoretical Karplus, 2002: Freezing drizzle formation in stably strati®ed study. J. Appl. Meteor., 36, 1647±1661.

Unauthenticated | Downloaded 09/30/21 07:01 AM UTC