On the precipitation and related features of the 1998 Ice Storm in the Montréal area

Atmospheric Research 83 (2007) 36 – 54 www.elsevier.com/locate/atmos

On the precipitation and related features of the 1998 Ice Storm in the Montréal area William Henson a,⁎, Ronald Stewart a,1 , Bohdan Kochtubajda b,2 a

Department of Atmospheric and Oceanic Sciences, McGill University, Montréal, Québec, Canada b Hydrometeorology and Arctic Laboratory, Edmonton, Alberta, Canada

Received 30 October 2005; received in revised form 31 December 2005; accepted 28 March 2006

Abstract A major freezing rain storm causing catastrophic losses occurred in early January 1998 over eastern Canada and the northeastern United States. The types of precipitation and associated precipitation structures of this storm are described and discussed with a particular emphasis on the Montréal area. Using operational Doppler radar as well as other information, it is shown that a variety of precipitation types (snow, freezing rain, ice pellets and rain) occurred. The characteristics and overall organization of precipitation structures varied considerably and, in some instances, these were linked with topographic features. Hydrometeor growth often occurred within the storm almost down to the surface. The two periods of significant freezing accumulation were characterized by more organized precipitation than at other times. Other major storms over the region have illustrated a similar pattern of precipitation types (snow, freezing rain, ice pellets and rain), implying comparable organizational processes. In addition, pronounced lightning activity occurred of both negative and positive polarity in association with different phases of precipitation. © 2006 Published by Elsevier B.V. Keywords: 1998 Ice Storm; Radar observations; Freezing rain; Precipitation structures; Lightning

1. Introduction Winter storms commonly occur over Canada and the United States. Such storms bring precipitation in the form of snow, rain, freezing rain and ice pellets, and they often cause major problems for society. There continues to be a substantial amount of research

⁎ Corresponding author. Fax: +1 514 398 6115. E-mail addresses: [email protected] (W. Henson), [email protected] (R. Stewart), [email protected] (B. Kochtubajda). 1 Fax: +1 514 398 6115. 0169-8095/$ - see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.atmosres.2006.03.006

conducted on these storms, but the amount is far less than that focused on severe summer weather. However, there is a growing recognition that winter storm issues need to be studied in more detail (Cortinas et al., 2004). For example, the United States is in the process of developing a winter storm research program (Ralph et al., 2005). This program has identified some of the key American priorities spanning scales from the storms' large-scale environment to their small-scale internal structure and precipitation. In Canada, there is an awareness of the impact of winter storms and the types of storms that occur (Stewart et al., 1995), but there is not yet an organized program to examine these events.

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

The importance of winter storms was certainly clear in the wake of the January 1998 Ice Storm that caused huge problems to eastern Canada and the northeastern United States. Freezing precipitation fell between January 4 and January 10, 1998 and brought particular hardship to the Montréal area. The Ice Storm seriously affected the electricity supply to 3.5 million people, shut down transportation, restricted emergency services, and damaged farms, trees and personal property. The longest residential blackouts lasted 33 days and 80% of the trees on Montréal's Mount Royal were damaged. The mobilization of the Canadian Armed Forces for relief effort was the largest humanitarian assistance mobilization in the country's history (Abley, 1998). There were an estimated 28 deaths in Canada and 19 in the United States (Klaassen et al., 2003) and $4 billion US of damage caused by the Ice Storm in Canada and the United States (Cortinas, 2000) with $1 billion US of damage in New York and Maine (DeGaetano, 2000). In excess of 840,000 insurance claims in Canada and the United States from the damage in the Ice Storm, this is 20% more than for Hurricane Andrew, which, until 2005, was considered by the re-insurance industry to be the worst natural disaster in United States history (Lecomte et al., 1998). According to a recent issue of MacLeans magazine, the Ice Storm was “the biggest thing that's happened in Canada” in terms of property loss (MacQueen, 2005). For a more detailed description of the amount of damage caused by the Ice Storm in Canada and the northeastern United States, please refer to Lecomte et al. (1998), Kerry et al. (1999), Milton and Bourque (1999), DeGaetano (2000), Gyakum and Roebber (2001), Roebber and Gyakum (2003) or Cortinas et al. (2004). Even though there have been many ice storms during the 20th century over North America, the quantity of ice accumulation and the persistence of the 1998 Ice Storm is often considered unprecedented (Kerry et al., 1999; DeGaetano, 2000; Gyakum and Roebber, 2001), or the worst in living memory (Milton and Bourque, 1999). It has been estimated that there have been 25 significant freezing rain events over southern and eastern Ontario since the 1880s and 22 in the northern U.S. states bordering southern and eastern Ontario during the period 1909–2002 (Klaassen et al., 2003). Therefore, major ice storms are a relatively common event occurring approximately every 5 years in the northeastern U.S states and southern and eastern Ontario. Given that ice storms are a relatively common event and that calls had been made after the ice storm in 1972 to remember the lessons learned after that particular event

37

(Roger, 1999), it can be appreciated how disruptive the 1998 Ice Storm was. There is no one factor that determines the severity of an ice storm. Mahaffy (1961) noted that an ice storm occurred in December 1942 in the Montréal area and had a greater ice accumulation than an event in February 1961, but the later event was more severe due to the high winds that also occurred. Four factors are considered necessary for a damaging ice storm event. These are the total ice accumulation, duration of the event, wind speed and area affected. As such, the determination of classifying severity can at times be subjective. Despite the devastation caused by the Ice Storm, little research has been conducted on it. The few papers that have been published have mainly focused on the planetary and synoptic scales with some discussion of topographic effects (Gyakum and Roebber, 2001; Roebber and Gyakum, 2003) and on climatologies of ice storms including the 1998 event (DeGaetano, 2000; Higuchi et al., 2000). None of these articles focused on the smaller scale features of the storm. Given the significance of the Ice Storm and the lack of research on its smaller scale features, the main objective of this study is to document and better understand its internal structure, in particular the organization of its precipitation, although considerable attention is also paid to the occurrence of lightning. 2. Description of the datasets This study will mainly utilize observational datasets. Many of these (including Doppler radar and lightning network information) have not previously been examined in connection with this storm. The primary dataset was 1 km by 1 km S-band radar data from the McGill system at the J.S. Marshall Radar Observatory (SainteAnne-de-Bellevue). The radar was upgraded to make Doppler measurements in 1992 and dual-polarization measurements in 1999 (after the 1998 Ice Storm). With a 700 kW Klystron transmitter and a 9 m dish, the McGill S-band radar is the largest weather radar in Canada. For more information on the McGill radar, see Fabry (1993). Operational surface measurements were also utilized in this study. Hourly surface weather observations were obtained for many operational sites but the primary ones used here were Dorval International Airport in Montréal (now Montréal—Pierre Elliot Trudeau International Airport), Mirabel International Airport just outside Montréal, Jean Lesage International Airport in Québec City and MacDonald—Cartier International Airport in Ottawa. These sites were chosen as they were all major airports at the time and therefore had manned observers.

38

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

The surface weather observations were obtained from the National Climate Data and Information Archive, operated and maintained by Environment Canada (http://www.climate.weatheroffice.ec.gc.ca). Reanalysis products were used to estimate the height of the melting layer, specifically, the National Centers for Environmental Prediction (NCEP) North American Regional Reanalysis (NARR). The NARR dataset assimilates operational data, has a grid spacing of 32 km and spans from 1978 to the present. Thirty pressure levels are available from which temperature profiles of the atmosphere can be calculated. More information about NARR can be found in Luo et al. (2005) or at the NARR homepage (http://wwwt.emc. ncep.noaa.gov/mmb/rreanl/index.html). Lastly, lightning strike information was obtained from the Canadian Lightning Detection Network (CLDN). The CLDN includes both LPATS-IV (Lightning Positioning and Tracking Sensors, Series 4) and IMPACT/ES (Improved Accuracy from Combined Technology, ES version) type sensors. The LPATS-IV sensors detect time-of arrival of radio pulses generated by lightning, while the IMAPCT/ES sensors combine the magnetic direction-finding and time-of-arrival methods. Further information on sensors and detection methods are discussed in Holle and Lopez (1993) and Cummins et al. (1998). The CLDN is fully integrated

with the United States National Lightning Detection Network (Orville et al., 2002) and provides continuous lightning coverage over most of Canada and offshore to about 300 km. Over the region of interest to this study, the location accuracy of the CLDN is better than 500 m and cloud-to-ground (CG) flash detection efficiency is above 90%. 3. Background 3.1. Large-scale and overall features The 1998 Ice Storm struck the Montréal region late in the evening on January 4 1998 and continued until early on January 10 (local time). A series of low pressure systems had developed in the southern United States, and warm, moist air within these systems was advected from the Gulf of Mexico into southern Ontario and Québec. A large, stationary Arctic high pressure system was centered over central Québec and persisted from January 5 to January 9, and this caused cold air to flow into the St. Lawrence and Ottawa River valley regions (Fig. 1), which the warm advected air could not dislodge. This situation resulted in the precipitation mainly falling as freezing rain or ice pellets although rain and snow also occurred (Fig. 2, adapted from Milton and

Fig. 1. Topographic map of the St. Lawrence Valley region. Contours shown are at 500 m above sea level.

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

39

Fig. 2. The distribution of total accumulation of (a) rain (mm), (b) freezing rain (mm) and (c) ice pellets and snow (cm) for the January 4–10, 1998 Ice Storm.

Bourque, 1999). Some areas, especially those south of Montréal and south of the St. Lawrence River, experienced in excess of 100 mm of freezing precipitation during the event and other areas experienced significant accumulations of ice pellets and snow.

By comparing the various distributions in Fig. 2, it is evident that there were areas of preference for rain, freezing rain and ice pellet/snow accumulation. Rain fell largely in the Appalachian Mountains. The area just south of Montréal received a large amount of freezing

40

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

rain but little solid precipitation. Montréal received a mixture of freezing rain and solid precipitation, whereas the region north of Montréal and Québec City (in the St. Lawrence River valley) had considerable amounts of solid winter precipitation but little rain or freezing rain. It should be noted that the values shown in Fig. 2c) are the depth of the solid precipitation on the ground. It was felt in Milton and Bourque (1999) that a simple conversion to the equivalent liquid water accumulation would not give an accurate reflection of accumulated mass of solid precipitation. The standard snow depth to liquid water conversion of 10:1 used by the Environment Canada Manual of Climatological Observations (Environment Canada, 1992) would give an underestimation of the liquid water equivalent as the solid precipitation was probably wet or composed of solid ice (i.e., ice pellets). The duration of the various precipitation types occurring over the region is shown in Fig. 3; this information is based on the hourly observations. Note that observations such as drizzle and snow grains were not included. All of the stations in the region experienced all the types of winter precipitation. The stations in Mirabel, Montréal and Ottawa were similar in the durations of the various types of precipitation, but Montréal experienced far fewer hours of snow (6 h total) and Ottawa far fewer hours of ice pellets (22 h total). In contrast, Québec City experienced many hours of snow (106 h total) and much less freezing rain and ice pellets (6 h and 27 h total, respectively), whereas Sherbrooke

had many hours of rain (36 h total). Given that rain, freezing rain and ice pellets normally all involve a melting process and, in the case of freezing rain and ice pellets a sub-freezing layer, this duration information indicates that the occurrence of melting and refreezing layers was highly variable. For example, even though they are only 33 km apart, there were more hours of snow experienced at Mirabel (21 h total) compared to Montréal (6 h total). A melting layer was therefore often present over Montréal but not necessarily over Mirabel. The evolution between precipitation types for several locations is shown in Fig. 4. The figure illustrates a number of critical features. These are: 1. The precipitation did not occur continuously at any location over the duration of the event. The longest continuous occurrence of precipitation was at Montréal for 60 h, and Québec City experienced the longest duration of any single precipitation type, snow (40 h). 2. Much of the precipitation reported from hourly observations fell as combinations rather than as a single type. Overall, 15% of the time at all stations was associated with mixed precipitation, with Montréal being the one receiving the highest fraction of time experiencing combinations of precipitation type (25%). 3. Freezing rain/ice pellets was the most common combination overall. This combination was most common at Montréal (83% of the combinations),

Fig. 3. Number of hours when each precipitation type was observed at selected sites over the period 0000 UTC January 4, 1998 to 0000 UTC January 11, 1998. The total number of hours of precipitation at each site is shown in the insert.

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

41

Fig. 4. Precipitation type as a function of time for five sites in the St. Lawrence Valley region. Data are based on hourly observations from Environment Canada. Vertical grey lines represent the four high peaks identified in Section 4.1.

whereas snow and ice pellets combined was most common at Québec City (88% of the combinations). 4. There was never a simple transition between rain and snow. At many sites, the first precipitation to occur was snow, followed by ice pellets and then ice pellets/freezing rain, but there was no significant period (and therefore quantity) of rain. However, this was not the case for Sherbrooke and the immediate surrounding region to the south, which as can be seen in Fig. 2, experienced significant amounts of rain. These observations indicate that when the precipitation structures passed through the Sherbrooke area an above-freezing warm layer of air (greater than 0 °C) was able to reach the ground. 5. The transitions between precipitation types sometimes followed patterns similar to those predicted for a warm frontal passage with a 5-step process (Stewart and King, 1987). This pattern is snow, snow/ice pellets, snow/ice pellets/freezing rain, ice pellets/ freezing rain and freezing rain. Such a pattern occurred once at Montréal and twice at Mirabel and Ottawa. This implies that there was a systematic evolution of the inversion layer at Mirabel, Montréal and Ottawa. Based on the work in Stewart and King (1987), that would suggest that the temperature and/ or height of the inversion increased or decreased steadily as opposed to fluctuating randomly. Another implication of the work in Stewart and King (1987) is that the presence of combinations (such as snow/ice pellets or freezing rain/ice pellets) implies that, with a

small difference in the height and temperature of the melting and/or freezing layer, the precipitation could have shifted to one of the constituent types or to another combination. That is, certain combinations illustrate that the atmosphere is close to a threshold between types or other combinations. Additionally, it should be noted that the occurrence of the transitions from freezing rain to rain or snow, or from ice pellets to ice pellets/freezing rain imply that the evolving precipitation aloft will also be modifying the atmospheric temperature through melting and freezing processes (Szeto and Stewart, 1997). 6. The time of occurrence of different precipitation types was at times comparable between sites. In particular, the timing at Mirabel, Montréal and Ottawa was very similar, undoubtedly due to their geographical proximity. However, there were differences between these three sites. For example, in the early hours (UTC) of January 8, 1998 when Montréal experienced combinations of freezing rain and ice pellets, Ottawa and Mirabel experienced only freezing rain. 4. Radar-observed structure of the storm 4.1. Overall perspective There were many precipitation bands and single cell structures that evolved and/or passed through the range of the McGill radar during the 1998 Ice Storm. In order

42

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

to examine these precipitation bands and structures, all available radar images during the event were examined. A radar reflectivity threshold was set and the number of adjacent 1 km × 1 km pixels above this threshold was computed from a Maximum Radar Reflectivity plot. A larger value of adjacent pixels is indicative of a (relatively) high reflectivity structure, whereas a lower value is indicative of lesser reflectivity structure. The number of adjacent pixels will be called the “organization” factor. Using a radar reflectivity threshold in this manor is similar to using iso-echo displays as described in Raghavan (2003). For these calculations, a Maximum Radar Reflectivity plot was chosen rather than a Constant Altitude Plan Position Indicator (CAPPI). A Maximum Radar Reflectivity plot displays the maximum reflectivity at each location regardless of height, and this technique is more likely to indicate areas of melting and therefore allows a more ready delineation of the structure of the precipitation region. When using CAPPIs, it is recognized that some may be under or over the melting layer or cut through melting layers at one location but not at others. This can result in misleading radar depictions. Using a Maximum Radar Reflectivity plot in this fashion is comparable to the use of NOAA Long Range Radar plots (http://www.nws.noaa.gov/radar/radinfo/radinfo. html). The “organization” factors were calculated for minimum radar reflectivity values of 37 and 45 dBZ over the lifetime of the storm, normalized relative to their maximum values, and shown in Fig. 5. These

values were chosen to highlight periods with substantial reflectivity during the Ice Storm. Three observations can be made on the basis of Fig. 5. First, there is a great deal of variation in the “organization” factor, and peaks in this parameter occurred throughout the Ice Storm. Second, the maximum values of this parameter occurred at the beginning and the end of the Ice Storm. Third, as the radar reflectivity threshold value is increased, four peaks in the “organization” factor become increasingly prominent. These occurred at the beginning and end as well as the middle of the event. Maximum Radar Reflectivity plots taken at the time of the four prominent peaks can be seen in Fig. 6. Detailed characteristics of the precipitation features giving rise to the peaks found in Fig. 5 are summarized in Table 1. The values for the “fraction covered” in Table 1 were calculated by determining the number of pixels that were above a radar reflectivity threshold value of 37 dBZ within a 120 km radius of the McGill radar. These values are then normalizing relative to the total possible area. One consequence of a large fractional coverage percentage indicates a large precipitation structure passing over many sites during its lifetime. Therefore, it is possible to associate the fraction covered with the widespread severity of a structure. Table 1 also lists the shape and orientation of reflectivity structures associated with each of the peaks as well as the speed and direction of motion of the reflectivity structure at the time of the peak “organization” factor. A “reflectivity structure” here is defined as any one of the peaks in the “organization”

Fig. 5. Normalized organization factor for a minimum radar reflectivity of 37 dBZ (grey) and 45 dBZ (black). Each of the peaks are associated with a reflectivity structure and identified by a number. Arrowed sections near the top of the figure indicate prolonged periods of radar reflectivity (>18 dBZ), and the blank sections at the bottom indicate when the McGill radar was inoperative.

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

43

Fig. 6. Maximum radar reflectivity plots produced from the McGill radar at (a) 0933 UTC January 5, 1998, (b) 1846 UTC January 7, 1998, (c) 1643 UTC January 9, 1998 and (d) 2018 UTC January 9, 1998. White range rings are at intervals of 40 km.

factor. Furthermore, the reflectivity structures associated with the four prominent peaks in the “organization” factor will be referred to as the high reflectivity structures and the other seven will be referred to as the moderate reflectivity structures. The orientation of each reflectivity structure's most prominent reflectivity band is also shown if there was one. The determination of shape was a subjective judgment based on an examination of a Maximum Radar Reflectivity plot, whereas the orientation was determined from an analysis of the plot. The speed and direction of the structures were calculated based on comparing sequential Maximum Radar Reflectivity plots and methodically shifting one plot relative to another and finding the least squared difference between the reflectivity values of the shifted plots. The speed of the primary band (if there was one) is not listed in Table 1 because, in the cases examined, the primary band was embedded within a larger structure, and traveling at the same speed as the larger structure, or there was no band present. To a large extent, Table 1 shows that the direction of the reflectivity structures was from the southwest to the northeast. However, the direction of travel of the reflectivity structures was not perpendicular to the

orientation of the primary band and the difference in angle shows no correlation to the speed of travel of the precipitation structure (not shown). Only the reflectivity structure associated with peak 2 was near perpendicular (within 20°) to the direction of travel. This indicates that the internal structure of the precipitation structures associated with the various peaks is complex and likely affected, for example, by the local topography. The four high peaks indicated in Fig. 5 can be studied separately. The fact that they all are prominent at a higher reflectivity threshold indicates that these reflectivity structures had a greater areal coverage of large reflectivity values than the other peaks. The timing of the four high peaks is indicated in Fig. 4 (grey vertical lines). Based on Fig. 4, it is evident that the first two peaks occurred close to the start of the freezing precipitation (ice pellets and freezing rain). The timing of the precipitation in relation to the third and fourth peaks is not so evident mainly due to the fact that the Ice Storm caused the McGill radar to be inoperative for certain periods. However, there are periods of freezing rain and/or ice pellets that began in Mirabel, Montréal and Ottawa not long after the third

44

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

Table 1 Fractional coverage, associated precipitation, shape of structure, structure speed, direction of travel of structure and orientation for identified structures Peak number

Date and time (UTC)

Fraction covered (%)

Precipitation associated with structure

Shape of structure(s)

Structure speed (km h –1)/ direction of motion (°) a

Orientation of main band (°) b

1 2 3

5 January, 0933 5 January, 1703 5 January, 2358

37 (5.4) 8 9

No organized structure Band, W–E Band, WNW–ESE

73/81 85/82 60/90

N/A 93 115

4

6 January, 0423

15

No organized structure

76/72

N/A

5

6 January, 1307

13

Band, NW–SE

73/81

130

6

7 January, 1846

13 (1.7)

No organized structure

81/63

N/A

7

8 January, 0546

10

Three bands

51/45

110

8

8 January, 1251

12

Three bands

87/56

N/A

9

9 January, 0109

16

No organized structure

68/45

N/A

10

9 January, 1643

20 (4.0)

Two bands

65/22

132

11

9 January, 2018

7 (1.7)

Ice pellets (Montreal ´ ) Rain and mist (Burlington) Freezing rain and fog (Montreal) ´ Freezing rain and fog (Montreal) ´ Freezing rain and fog (Montreal) ´ Freezing rain and mist (Burlington) Freezing rain, ice pellets and fog (Montreal) ´ Freezing rain, ice pellets and fog (Montreal) ´ Freezing rain, ice pellets and fog (Montreal) ´ Freezing rain and ice pellets (Montreal) ´ Freezing rain, ice pellets and fog (Mirabel)

Band, NWN–SES

70/31

170

Applicable values were calculated using a radar reflectivity threshold of 37 dBZ. The four major peaks at a minimum maximum reflectivity value of 45 dBZ are highlighted and the values of fractional coverage are shown in brackets for comparison. a Direction of travel, clockwise relative to north. b Orientation, clockwise relative to north.

peak and it is likely that the fourth peak occurred too soon after the third peak to allow any differentiation between them. 5. Precipitation types and precipitation structures 5.1. Background Some of the key issues associated with the Ice Storm involve the occurrence of precipitation structures and the various types of precipitation that are observed at the surface. To this end, the following questions were posed: 1. How does the radar reflectivity information relate to the various types of precipitation? 2. Which structures produced the greatest accumulation of freezing precipitation? 3. To what extent did topography influence the precipitation structures and types? 4. What were the electrical effects associated with the storm? Each of these questions will now be examined in a systematic manner.

5.1.1. Radar reflectivity and the type of precipitation As can be seen in Fig. 2 there was a large variation in the amounts recorded of the different types of precipitation in the St. Lawrence River valley region, especially for freezing rain. Therefore, the question naturally can be asked whether the radar reflectivity was a different function of height over the various sites for the same form of precipitation or during the passing of the same precipitating structure. To answer this question, radar reflectivity profiles were calculated for the storm (Fig. 7). To illustrate the results of this analysis, the period between 2100 UTC January 7 to 1800 UTC January 8 was chosen since it covers much of a period of large freezing accumulation (Milton and Bourque, 1999). The sites chosen were Montréal, Mirabel and St. Hubert because they were close to the McGill radar and therefore good vertical resolution of radar reflectivity at low heights was achievable and they also experienced continuous precipitation. Hourly observations indicate that freezing rain and ice pellets were commonly observed at all three sites during this period; however, there were several changes in the type of precipitation or combination of precipitation types. Plots of the average reflectivity profiles over this period can be

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

45

Fig. 7. Average radar reflectivity as a function of height above sea level during the period 2100 UTC January 7 to 1800 UTC January 8, 1998. The figure also indicates the sites over which the reflectivity profile was calculated. The horizontal bars indicate the height of the upper 0 °C level as calculated from NARR products.

seen in Fig. 7; many of the individual radar profiles exhibited similar patterns. The heights of the upper 0 °C isotherms, calculated using NARR products are also shown in Fig. 7. When an inversion occurred, there were normally two 0 °C isotherm levels. If ice phase precipitation was occurring aloft, the upper level should be linked to a significant increase in radar reflectivity associated with the bright band. In general, the NARR-calculated upper 0 °C isotherm is not at the same height of the major inflection point in the radar reflectivity or the maximum reflectivity aloft. However, differences (∼ 200–500 m) are relatively small and, therefore, it is reasonable to assume that the melting layer is near the height of the major inflection point in the radar reflectivity. The radar reflectivity profiles in Fig. 7 show similarities as well as differences. In all three reflectivity profiles, there is a bright band aloft, a region of constant or decreasing reflectivity below, and an increase in reflectivity toward the surface. The radar reflectivity near the surface at all three sites is greater than the reflectivity at the bright band, even though the peak reflectivities are all relatively low (< 28 dBZ). There are at least two mechanisms that could explain the increase in reflectivity toward the surface. The first is that, if the cloud base is low, the hydrometeors can still grow by accretion and/or coalescence below the melting layer. Unfortunately, cloud base information was not available for any of the observation sites used in this study. The second is an increase in the number of hydrometeors, which could occur from the warm rain process.

Of the three sites examined, Mirabel experienced the most changes in the type of precipitation over the selected period. The observed precipitation combinations were freezing rain, freezing rain/ice pellets and snow/ice pellets. A plot of reflectivity against height at Mirabel with each of these combinations of precipitation type can be seen in Fig. 8. This figure indicates that, for the same precipitation type or combination of precipitation types, there were different radar reflectivity profiles. The cases of freezing rain, freezing rain/ice pellets and snow/ice pellets show differences that are must due to variations in the number of hydrometeors, melting layer processes and/or the size of hydrometeors that nonetheless do not affect the actual precipitation type(s) at the surface. This section has shown that there were no unique profiles associated with any precipitation type or combination of precipitation types. Aloft, the reflectivity profiles varied between those with a distinct bright band and those with only an inflection point, although both suggest that ice phase processes were occurring. As well, many profiles exhibited an increase in reflectivity near the surface, implying growth of the hydrometeors at low levels. 5.1.2. Precipitation structures and the periods of significant freezing accumulation Milton and Bourque (1999) states there were two periods of significant freezing accumulation in the 1998 Ice Storm (Fig. 9). These periods were between 2300 UTC 5 January and 1300 UTC 6 January (14 h), and 2300 UTC 7 January and 2300 UTC 9 January (48 h). St. Hubert received the most freezing accumulation (80 mm) during

46

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

Fig. 8. Average radar reflectivity as a function of height above sea level for Mirabel for (a) freezing rain, (b) freezing rain/ice pellets and (c) snow/ice pellets. The times (UTC) and types of precipitation are indicated in the figure. The horizontal bars indicate the height of the upper 0 °C level as calculated from NARR products.

the entire storm; 18 mm and 42 mm fell in these two periods, respectively. This corresponds to an average precipitation rate of 1.3 mm h− 1 and 0.9 mm h− 1 over the respective periods.

The periods of significant freezing accumulation were associated with continuous radar reflectivity greater than 18 dBZ (Fig. 9). However, larger reflectivity values also occurred with these mainly

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

47

Fig. 9. Freezing accumulation as a function of time during the 1998 Ice Storm. The periods of significant freezing accumulation according to Milton and Bourque (1999) are shown. The vertical lines indicate the occurrence times of the peak “organization” factors and the horizontal arrows when radar reflectivity (>18 dBZ) was present over half the McGill radar range. When several peak “organization” factors occurred in the same period, they are separated by a comma. The four high peak “organization” factors are circled.

being associated with moderate (37–45 dBZ) reflectivity structures although some high (> 45 dBZ) reflectivity structures also occurred. The moderate reflectivity structures (3, 4 and 7–9) occurred at various times during the periods of significant freezing accumulation whereas the high reflectivity structures (10 and 11) only marked the end of the period. The moderate and high reflectivity structures comprised 49% and 8% of the periods of freezing accumulation, respectively; 43% of the time reflectivities were 18–37 dBZ. Unfortunately, it is not possible to directly estimate the accumulated freezing accumulation from the various reflectivity structures due to the highly variable form of precipitation that was occurring. The periods of significant freezing accumulation were also associated with precipitation bands (Table 1). Of the 12 bands that had been identified, 10 occurred during these periods. They occurred in association with weak, moderate and high reflectivity values. Some of these were aligned in a manner that suggests that orographic factors were significantly affecting them (Fig. 10). Reflectivity profiles during the periods of significant accumulation were examined. The resulting profiles were very similar to those discussed in Section 5.1.1. That is, there was a bright band aloft, a region of constant or decreasing reflectivity below, and an

increase in reflectivity toward the surface. Such results imply near continual hydrometeor growth. Radar features within the periods of significant freezing precipitation showed some similarities to but also some differences from those outside these periods. First of all, approximately 145 h of precipitation occurred in the Montréal area with 62 h (43%) of this occurring within the periods of significant accumulation and 93 h (57%) outside these periods. In terms of similarities, widespread, moderate and high reflectivity features as well as precipitation bands all occurred in both domains. However, their relative occurrences were quite different. The total widespread precipitation (18– 37 dBZ) was greater inside the significant accumulation periods (25 h) than outside them (32 h). In contrast, moderate and high reflectivity structures were much more common within the significant accumulation periods (25 h) as opposed to outside (13 h). As well, 10 of the 12 precipitation bands occurred within the periods of significant freezing accumulation. Collectively, these observations illustrate that the periods of significant freezing accumulation accounted for a far greater portion of the organized structures than the periods outside. In summary, this section has shown that periods of significant freezing accumulation were associated with several precipitation features, not just one. Most of the

48

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

Fig. 10. Maximum reflectivity plot produced from McGill radar at 0845 UTC January 8, 1998 with the stationary precipitation band circled.

freezing accumulation was associated with continuous radar reflectivity greater than 18 dBZ, although moderate (37–45 dBZ) and high (> 45 dBZ) reflectivity structures were also present. The moderate reflectivity structures occurred throughout the periods of significant freezing accumulation, whereas the high reflectivity structures marked the end. The periods during which significant freezing accumulation occurred showed a general increase in reflectivity toward the surface, implying continual growth of hydrometeors. 5.1.3. Evidence that topography played a role in the production of precipitation structures or the distribution of the different types of precipitation It is apparent that the surrounding topography had a significant effect on the 1998 Ice Storm. For example, bands of precipitation were often affected by topography. In particular, one of the precipitation bands summarized in Table 1 interacted strongly with the topography. As structure #7 passed the Southern Laurentian Mountains a stationary precipitation band was induced. This stationary band grew to almost 200 km in length (Fig. 10). Associated precipitation was enhanced because of an across band jet from the southeast that pushed moisture laden air up through this band and over the Laurentian

Mountains (figure not shown). From Doppler information, the top of this region of ascending air rose from 2 to 3 km near the McGill radar to 5 km at the location of the stationary precipitation band. The radial velocity of the ascending air at the band location was up to 5 m/s and directed away from the radar. This situation is believed to be similar to cold air damming (Bailey et al., 2003; Brennan et al., 2003 for example). The various types of precipitation, especially in the Sherbrooke area, were all shown to be strongly influenced by the Burlington valley region. Based on the temperature observed at Sherbrooke and Burlington, an above freezing layer of warm air was present during most of the precipitation, which caused the precipitation to melt and fall as rain. It is likely that the topography in the narrow valley region in which Burlington lies had an influence on the presence of this above zero layer of warm air. From a historical view, the 1998 Ice Storm illustrated some features similar to those of previous ice storms over this region. For example, the 1961 Ice Storm had a similar pattern of precipitation to that of the 1998 Ice Storm. There was an area of heavy glaze around the Montréal area with snow recorded to the north in the Laurentian Mountains and rain in the Appalachian Mountains (Fig. 11). Choosing three sites (Maniwaki,

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

49

Fig. 11. The precipitation in the St. Lawrence Valley in the vicinity of Montréal Island for the 1961 Ice Storm. The area receiving a heavy glaze coating is encircled in small dots. Values in the figure are the melted precipitation in inches. Figure adapted from Mahaffy (1961).

Montréal and Sherbrooke), hourly observations and daily accumulations of the five major ice storms mentioned in Milton and Bourque (1999) were examined. Maniwaki was chosen instead of Mirabel as the historical data from Maniwaki extends further back than for Mirabel. Details of the precipitation accumulation that fell at these three sites can be seen in Table 2. Table 2 Amounts of liquid and solid accumulations at Maniwaki, Montréal and Sherbrooke during the five major ice storms prior to the 1998 Ice Storm mentioned in Milton and Bourque (1999) Site/ice storm date

23–25 22–23 27–30 December February March 1942 1961 1972

12–14 4–6 December January 1983 1997

Maniwaki

24.1 cm solid

4.4 mm liquid 16.4 cm solid 38.2 mm liquid 6.6 cm solid 83.2 mm liquid

Montréal

42.2 mm liquid 46.3 cm solid Sherbrooke 32.8 mm liquid

11.9 cm solid

9.4 mm liquid 21.8 cm solid 37.3 mm 26.4 mm liquid liquid 3.0 cm 3.3 cm solid solid 48.8 mm 6.9 mm liquid liquid 0.5 cm solid

39 mm total

18.6 mm liquid 1.2 cm solid 21 mm total

Liquid precipitation is defined as rain, drizzle, freezing rain and solid precipitation is defined as snow and ice pellets.

The Environment Canada archive has two categories for precipitation, solid and liquid. In this case, solid precipitation is considered to be ice pellets, snow, and liquid precipitation is considered to be rain, freezing rain, drizzle. From Table 2, it is evident that, in these major ice storms, solid types of precipitation dominate in Maniwaki, mixed types occur in the Montréal region and liquid precipitation dominates in the Sherbrooke region. Hourly observations of Montréal and Sherbrooke precipitation during these five events are listed in Table 3. Maniwaki does not have hourly observations of precipitation. Therefore, there is some evidence that, from a historical point of view, the precipitation patterns that were observed in the 1998 Ice Storm were similar to all five major ice storms mentioned in Milton and Bourque (1999). This suggests that, at least on the scale of ∼200 km from Montréal, there are common factors affecting the distribution of precipitation types between the five storms and the 1998 Ice Storm and one of these factors must be the surrounding topography. Two explanations for these results include the following. First, the local topography in the Burlington valley region allowed air above freezing temperature to reach locations such as Sherbrooke, therefore causing the precipitation to fall preferentially as rain. Second, to the north in the Laurentian Mountains, the general direction of flow (southwest to northeast) would cause the air to

50

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

Table 3 Hourly observations of rain (R), freezing rain (ZR), ice pellets (IP) and snow (SN) during the five major ice storms prior to the 1998 Ice Storm mentioned in Milton and Bourque (1999)

27–30 December 1942 23–25 February 1961 22–23 March 1972 12–14 December 1983 4–6 January 1997

Sherbrooke

Montréal

No hourly observations No hourly observations 9hR 1 h ZR 0 h IP 42 h RN 1 h ZR 1 h IP 10 h R 3 h ZR 0 h IP

No hourly observations 0hR 13 h ZR 5 h IP 14 h R 9 h ZR 1 h IP 15 h R 21 h ZR 8 h IP 13 h R 7 h ZR 0 h IP

17 h SN 0 h SN 21 h SN

ascend, thereby leading to adiabatic cooling which would decrease near surface and inversion temperatures and force more of the precipitation to fall in solid types. In summary, it is evident that topography played an important role in the determination of fine scale aspects of the Ice Storm. These effects include the development of at least one precipitation band and the recurrent distribution of precipitation types in such severe storms. 5.1.4. Electrical effects associated with the precipitation structures and the relation to the form of precipitation During the 1998 Ice Storm, on January 9, there were approximately 900 cloud-to-ground (CG) lightning strikes from 0958 UTC to 2245 UTC between the

Maniwaki 1 h SN 20 h SN 1 h SN 18 h SN

No hourly observations No hourly observations No hourly observations No hourly observations No hourly observations

latitudes of 43 and 47°N and longitudes of 72 to 77°W. The positions of the lightning strikes, also indicating polarity, can be seen in Fig. 12. The majority of these lightning strikes were negative; however, the percentage of positive CG strikes varied during the event and the overall average percentage of positive strikes was 21.8%. The average positive CG strike peak current (39.34 kA) is almost 5 kA more than the average negative CG strike peak current (− 34.50 kA). In general, the average positive CG strike peak current exceeded the negative CG strike peak current on an hourly basis. All of these results and the fact that the lightning occurred on January 9, late into the Ice Storm, are similar to values or trends seen in other winter lightning cases (Orville et al., 1987). Kitagawa and

Fig. 12. Cloud to ground lightning strikes on January 9, 1998 during the Ice Storm.

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

Michimoto (1994) suggest that high wind shear, low charge height above ground and weak convective updrafts tend to coincide with thunderstorms producing mostly positive CG lightning. However, not all winter storms produce large positive flash percentages. Biswas and Hobbs (1990) showed that most ground flashes over the Gulf Stream were negative and associated with convective rain bands. It was not noted if there was a strong wind shear or not. From Fig. 12, we can see that the clear majority of the positive and negative CG lightning flashes that struck in the range of the McGill radar occurred in the Appalachian Mountains to the south and east of Montréal. Clear “streaks” of negative strikes can be seen in Fig. 12 and from radar images these “streaks” originated from isolated cells. One of these cells produced 30 negative strikes between 2005 and 2120 UTC. There were no positive strikes from this particular cell before or after these times within the range of the McGill radar although there were other cells surrounding this one that produced sparse positive strikes. However, these cells were usually separated by 50 km or more. During the same period, a relatively widespread cloud produced a mixture of positive and negative strikes. Five positive and five negative strikes were produced between 2015 and 2100 UTC. Prior to 2015 UTC, there had not been a lightning strike in the same general area since 1940 UTC and there were no lightning strikes from this section of cloud after 2100 UTC in the range of the McGill radar. Vertical cross sections were taken of the cloud which produced solely negative strikes and the cloud which

51

produced mixed polarity strikes, and these vertical cross sections can be seen in Figs. 13 and 14, respectively. The vertical cross section through the mixed strike cloud was chosen at a time when there were two positive strikes and with an orientation such that it passed as close as possible through the points where the positive strikes were recorded. The height of the mixed cell (Fig. 14) is much lower compared to the negative cell, reaching up to only 5 km compared to the solely negative strike cell which reached up to 9 km (Fig. 13). Due to the widespread horizontal nature of the mixed polarity strike cloud and the lack of vertical development, this indicates that it is a stratiform precipitation structure whereas the negative strike cloud is a convective precipitation structure. It is possible that the lightning strikes from the mixed strike cloud are occurring more like the bipole hypothesis suggests, i.e., there is a charge separation due to wind shear. The widespread nature of the cloud would also explain, in part, the scattered nature of the mixed strikes. The strength of the average positive strike from the mixed strike cloud was 50.0 kA and the average negative strike was −60.7 kA, which is larger than the positive average mainly due to a single − 210.1 kA strike. The average strike from the solely negative strike cloud was −30.8kA. The low cell height of the mixed strike cloud (5 km) would suggest that the size of the lightning strikes would be smaller than the solely negative strike cloud (9 km), if charge transfer was based predominantly on collisions. This was definitely not the case, based on the values given above, indicating that charge transfer was not based solely on collisions.

Fig. 13. Vertical cross section through a precipitation structure producing negative lightning strikes. Indicated on the figure is the precipitation structure that produced the negative lightning strikes. Cross section was produced from McGill radar at 2008 UTC January 9, 1998.

52

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

Fig. 14. Vertical cross section through a precipitation structure producing negative and positive lightning strikes. Indicated on the figure is the precipitation structure that produced the mixed polarity lightning strikes. Cross section was produced from McGill radar at 2013 UTC January 9, 1998.

The mixed strike cloud passed over Montréal and the negative strike cloud passed over Burlington. Surface observations were obtained and freezing rain was recorded during the passing of the mixed strike cloud and rain during the passing of the negative strike cloud. It is possible that topography played a role in that Burlington lies in a valley and it is possible that a warm layer that extended to the ground, or sufficiently close to it, was trapped in the valley. This warm layer would have then caused the precipitation to fall as rain. Similar observations were found by Dinger and Gunn (1946) and also Drake (1968). There is some conflict between these observations and those found in Orville and Huffines (2001). They found lower number of lightning strikes per km in mountains compared to plains, but they noted that this observation was due to the fact that convective events do not dominate in the Appalachian Mountains in the summer and in this case many of the negative lightning strikes were caused by convective cells. It was noted in Orville and Huffines (2001) that Rakov et al. (1989) found the opposite in the North Caucasus region of Russia. It is possible that this disagreement is again due to the differences between summer and winter lightning events. Orville and Huffines (2001) did note that the existence of wind shear does facilitate the production of positive lightning strikes, and there was large wind shear on January 9 during the 1998 Ice Storm together with a large number of positive lightning strikes. Wind shear is common in the presence of temperature inversions. This could be the main reason why positive lightning strikes

are more common in winter events. Only a more detailed analysis of winter lightning events, over a much larger area and many more cases, will give a more conclusive answer as to the causes of charge transfer and the reasons why typically winter lightning events usually have a higher proportion of positive CG strikes. In summary, the bipole hypothesis and particle collisions appear to not have been the sole cause of charge transfer in the generation of lightning production. In this case, there is a connection between the polarity of the lightning strikes and the type of precipitation structure. Negative polarity strikes occurred from convective precipitation structures and a mixture of negative and positive polarity strikes occurred from stratiform precipitation structures. 6. Concluding remarks A study has been carried out of the structure of the 1998 Ice Storm that devastated parts of eastern Canada and the northeastern United States. The focus was on the Montréal area which experienced some of the greatest impacts and for which a number of datasets were available, including hourly observations, lightning strike data, and reflectivity and Doppler radar data. From the analysis of the available information, a number of observations and conclusions can be made, including: • Precipitation types (snow, freezing rain, ice pellets and rain) varied systematically across the region and

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

•

•

•

•

•

•

•

they often occurred in combinations. In the Montréal area, liquid precipitation preferentially occurred to the south, mixed precipitation near Montréal (where ∼ 25% of the precipitation occurred as a combination of precipitation types) and solid precipitation to the north. The precipitation was organized in different ways. The main ones were widespread precipitation (18– 37 dBZ), reflectivity structures (either 37–45 dBZ or > 45 dBZ) and precipitation bands. Some convective cells also developed towards the end of the event. The various types, and combinations, of precipitation were linked with both the same reflectivity structure and from different reflectivity structures. Additionally, certain reflectivity structures were linked with transitions between different types or combinations of precipitation. Many radar reflectivity profiles associated with surface precipitation (other than snow) exhibited a similar shape. Typical profiles showed a bright band or inflection point aloft, below this a layer of decreasing or near constant values, but increasing values near the surface. This implies growth of the hydrometeors even at low levels. However, the actual values of reflectivity and heights of features above the surface varied even for the same precipitation type or combination of precipitation types. This indicates that the atmospheric conditions can vary substantially for the same precipitation type or combination of precipitation types. When radar data was available, the periods of significant freezing accumulation were mainly associated with widespread and moderate (3, 4 and 7–9) reflectivity structures rather than structures with higher reflectivity (1 and 6). This indicates that the freezing accumulation was not due to bursts of very heavy freezing rain, but to longer durations of more moderate rates. The periods of significant freezing accumulation were characterized by more organization than the periods outside. Reflectivity structures and precipitation bands were far more common during significant freezing accumulation. Topography had a significant effect on the detailed nature of precipitation within the Ice Storm. This includes, for example, the development of a single persistent band, lasting approximately 6 h long, along the foot of the Laurentians and it was also linked with a change in the type of precipitation across it. The occurrence of liquid to the south, mixtures of liquid and solid near Montréal, and predominantly

53

solid precipitation to the north occurred within this storm as well as the five major previous ice storms as listed in Milton and Bourque (1999). This suggests that topographic forcing and related factors help to consistently organize the environment conducive to the different precipitation types. • Negative and mixed (both negative and positive) polarity lightning strikes were produced from convective and stratiform precipitation structures, respectively. Collectively, this study has carried out an examination of the precipitation associated with the devastating 1998 Ice Storm. The internal features of this storm were characterized by complex relationships between precipitation structures, precipitation types, topography and lightning polarity. A complete understanding of these internal features is required in order to develop strategies for minimizing impacts. Acknowledgements The authors would like to thank Paul Joe, Norman Donaldson and Bradley Power at Environment Canada, and also Aldo Bellon and Alamelu Kilambi at the McGill radar for all the training, data, help and advice they have given. The authors would also like to thank the anonymous reviewers for their helpful comments. This research was carried out with the financial support of Environment Canada, the Institute for Catastrophic Loss Reduction, and the Natural Sciences and Engineering Research Council of Canada. References Abley, M., 1998. The Ice Storm: An Historical Record in Photographs of January 1998. McClelland & Stewart, p. 192. Bailey, C.M., Hartfield, G., Lackmann, G.M., Keeter, K., Sharp, S., 2003. An objective climatology, classification scheme, and assessment of sensible weather impacts for Appalachian cold-air damming. Weather Forecast. 18, 641–661. Biswas, K.R., Hobbs, P.V., 1990. Lightning over the Gulf Stream. Geophys. Res. Lett. 17, 941–943. Brennan, M.J., Lackmann, G.M., Koch, S.E., 2003. An analysis of the impact of a split-front rainband on Appalachian cold-air damming. Weather Forecast. 18, 712–731. Cortinas, J., 2000. A climatology of freezing rain in the Great Lakes region of North America. Mon. Weather Rev. 128, 3574–3588. Cortinas, J.V., Bernstein, B.C., Robbins, C.C., Strapp, J.W., 2004. An analysis of freezing rain, freezing drizzle, and ice pellets across the United States and Canada: 1976–90. Weather Forecast. 19, 377–390. Cummins, K.L., Murphy, M.L., Bardo, E.A., Hiscox, W.L., Pyle, R.B., Pifer, A.E., 1998. A combines TOA/MDF technology upgrade of the US National Lightning Detection Network. J. Geophys. Res. 103, 9035–9044.

54

W. Henson et al. / Atmospheric Research 83 (2007) 36–54

DeGaetano, A.T., 2000. Climate perspective and impacts of the 1998 northern New York and New England Ice Storm. Bull. Am. Meteorol. Soc. 81, 237–254. Dinger, J.E., Gunn, R., 1946. Electrical effects associated with a with a change of state with water. Terr. Magn. Atmos. Electr. 51, 477–494. Drake, J.C., 1968. Electrification accompanying the melting of ice particles. Q. J. R. Meteorol. Soc. 94, 176–191. Environment Canada, 1992. Manual of Climatological Observations (MANCLIM), 3rd ed. Meteorological Service of Canada, p. 61. Fabry, F., 1993. Wind-profile estimation by conventional radars. J. Appl. Meteorol. 32, 40–49. Gyakum, J.R., Roebber, P.J., 2001. The 1998 Ice Storm—analysis of a planetary-scale event. Mon. Weather Rev. 129, 2983–2997. Higuchi, K., Yuen, C.W., Shabbar, A., 2000. Ice Storm '98 in southcentral Canada and northeastern United States: a climatological perspective. Theor. Appl. Climatol. 66, 61–79. Holle, R.L., Lopez, R.E., 1993. Overview of Real-Time Lightning Detection Systems and Their Meteorological Uses. NOAA Technical Memorandum ERL NSSL-102, Norman, OK, p. 68. Kerry, M., Kelk, G., Etkin, D., Burton, I., Kalhok, S., 1999. Glazed over: Canada copes with the Ice Storm of 1998. Environment 41, 6–32. Kitagawa, N., Michimoto, K., 1994. Meteorological and electrical aspects of winter thunderclouds. J. Geophys. Res. 99, 10713–10722. Klaassen, J., Cheng, S., Auld, H., Li, Q., Ros, E., Geast, M., Li, G., Lee, R., 2003. Estimation of Severe Ice Storms Risks for SouthCentral Canada. Study and Report Prepared by MSC-Ontario Region for the Office of Critical Infrastructure Protection and Emergency Preparedness, p. 99. Lecomte, E.L., Pang, A.W., Russell, J.W., 1998. Ice Storm '98. Institute for Catastrophic Loss Reduction, p. 99. Luo, Y., Berbery, E.H., Mitchell, K.E., 2005. The operational Eta model precipitation and surface hydrologic cycle of the Columbia and Colorado Basins. J. Hydrometeorol. 6, 341–370. Mahaffy, F.J., 1961. The ice storm of 25–26 February 1961 at Montréal. Weatherwise 6, 241–244. MacQueen, K., 2005. When B.C. gets hit. MacLeans's 118 (18), 34–43.

Milton, J., Bourque, A., 1999. A climatological account of the January 1998 Ice Storm in Québec. Atmospheric Sciences and Environmental Issues Division, Environment Canada, Québec Region, p. 87. Orville, R.E., Huffines, G.R., 2001. Cloud-to-ground lightning in the United States: NLDN results in the first decade, 1989–98. Mon. Weather Rev. 129, 1179–1193. Orville, R.E., Weisman, A.R., Pyle, R.B., Henderson, R.W., Orville Jr., R.E., 1987. Cloud-to-ground lightning flash characteristics from June 1984 through May 1985. J. Geophys. Res. 92, 5640–5644. Orville, R.E., Huffines, G.R., Burrows, W.R., Holle, R.L., Cummins, K.L., 2002. The North American lightning detection network (NALDN)—first results: 1998–2000. Mon. Weather Rev. 130, 2098–2109. Raghavan, S., 2003. Radar Meteorology. Kluwer Academic Publishers. p. 549. Rakov, V.A., Adjiev, A.K., Akchurin, M.M., Shoivanov, Y.R., 1989. Study of spatial distribution of ground flash density using the “Ochag” lightning location system. Meteorol. Gidrol. 2, 48–53. Ralph, F.M., Rauber, R.M., Jewett, B.F., Kingsmill, D.E., Pisano, P., Pugner, P., Rasmussen, R.M., Reynolds, D.W., Schlatter, T.W., Stewart, R.E., Tracton, S., Waldstricher, J.S., 2005. Improving short term (0–48 hour) cool-season quantitative precipitation forecasting: recommendations from a USWRP workshop. Bull. Am. Meteorol. Soc. 86, 1619–1632. Roebber, P.J., Gyakum, J.R., 2003. Orographic influences on the mesoscale structure of the 1998 Ice Storm. Mon. Weather Rev. 131, 27–50. Roger, N., 1999. Facing the unforeseeable: lessons from the Ice Storm of '98. Les Publications du Québec, Sainte-Foy Québec, p. 434. Stewart, R.E., King, P., 1987. Freezing precipitation in winter storms. Mon. Weather Rev. 115, 1270–1279. Stewart, R.E., Bachand, D., Dunkley, R.R., Giles, A.C., Lawson, B., Legal, L., Miller, S.T., Murphy, B.P., Parker, M.N., Paruk, B.J., Yau, M.K., 1995. Winter storms over Canada. Atmos.-Ocean 33, 223–248. Szeto, K.K., Stewart, R.E., 1997. Effects of melting on frontogenesis. J. Atmos. Sci. 54, 689–702.