1. Introduction
In the United States, statistics compiled by the National Weather Service's Office of Meteorology using Storm Data indicate that from 1990 through 1998, ice storms were responsible for 69 fatalities, nearly 3000 injuries, and roughly $2 billion in damage. (These data are available online at www.nws.noaa.gov/om/hazstats.htm under the U.S. Summary menu.) These statistics reveal the detrimental impact of these storms on the activities of a nation and, thus, the significant responsibility assigned to forecasters to anticipate them with sufficient lead time. Despite the devastating nature of freezing rain on life and property, relatively few studies have analyzed the environmental conditions associated with this phenomenon.
The basic environmental conditions necessary for freezing rain to occur were first determined when W. Meinardus, a German scientist, investigated an ice storm that occurred over middle and eastern Germany on 20 October 1898 (Okada 1914). Using mountaintop observations to ascertain a vertical thermodynamic profile, he found that there was a moist (nearly saturated) layer of air aloft that was warmer than 0°C. He inferred that this layer must have had sufficient upward vertical velocity to cause condensation of water vapor in that air, and that subfreezing air existed below the warm layer through which the raindrops fell and became supercooled. These supercooled raindrops froze on contact with a subfreezing surface. Later studies by Frankenfield (1915) and Meisinger (1920) verified the early findings of Meinardus, although their studies were also limited to the investigation of one storm or a series of storms that affected one location. Beyond these initial works and sporadic descriptions of significant ice storms, few additional efforts were made to better understand the atmospheric processes associated with freezing rain until the 1950s.
Riehl et al. (1952, 48–49) provided a general treatment of the synoptic features that accompany freezing-rain events. While their study focused primarily on ice storms in and around Chicago, Illinois, Riehl et al.'s results can be applied to many parts of the United States. They identified two scenarios that seem to be the most common for producing freezing rain. In the classic cyclone scenario, the southern edge of the icing area averages 50 mi (92.5 km) or more to the north of the warm front, snow falls to the north of the freezing-rain area, and rain falls to the south. Freezing rain ends or quickly changes to snow after the passage of the north–south axis of the surface cyclone. During the second scenario, a stationary front exists at the surface and a broad southwesterly current ascends over the cold dome. Precipitation is usually more widespread, lighter, and lasts longer in this scenario.
Bennett (1959) also discussed the synoptic conditions of freezing rain and ice storms for various regions of the United States using weather observations obtained from the U.S. Weather Bureau. While he agreed with the findings of Riehl et al. (1952, 48–49) that freezing-rain events occur most frequently in association with warm frontal boundaries, Bennett suggested that, in some parts of the United States, freezing rain may occur frequently with stationary or occluded polar fronts. Bennett also found that the cold surface air mass involved in the formation of freezing rain almost invariably is continental polar (cP) in nature with the exception of certain areas such as near the New England coast where cP air masses have usually been modified to maritime polar. The warm air that ascends over the subfreezing surface air is usually maritime tropical except in the Pacific Northwest where it is more likely to be maritime polar.
In a recent study, Bernstein et al. (1998) studied the location of freezing-rain observations relative to surface weather features, such as cyclone centers, fronts, troughs, and the characteristics of the freezing-rain air mass during 35 cases of winter weather between November and March of 1993–95. Like Bennett (1959) and Riehl et al. (1952, 48–49), they found that most freezing rain occurs on the cold side of warm fronts in arctic air masses or Atlantic maritime air masses that are found east of the Appalachian Mountains. Additionally, they found that the Atlantic air masses appeared to be more efficient at creating freezing rain than arctic air masses because of the relatively high moisture content and the higher frequency of a favorable thermal stratification for freezing rain (a layer of warm air over a layer of subfreezing surface air).
More recently, case studies have utilized rawinsonde observations (Young 1978; Bocchieri 1980; Zerr 1997; Bernstein 2000) and remote sensing techniques, such as wind profilers and Doppler radar (Martner et al. 1993; Ramsey 1995; Thoreson 1995; Zerr 1997), to view the atmosphere's vertical structure during freezing-rain events. Bernstein (2000) used data at six rawinsonde sites to identify important regional and local influences on freezing precipitation in the United States. He found that warm advection was the primary forcing associated with freezing-rain events. In addition, he determined that the thermodynamic profile during freezing rain usually had an elevated warm layer with a depth of up to 2800 m, and a maximum temperature within the layer that ranged from 1° to 10°C. Below the warm layer, was a subfreezing layer, which had a depth of up to 1400 m, and a minimum temperature within the layer that ranged from −1° to −7°C. Cortinas (2000) found similar results for freezing-rain events in the Great Lakes region of North America.
An understanding of the regional differences associated with freezing rain is important for forecasters, particularly those at national centers, in order to recognize the degree to which local and synoptic-scale processes may affect the formation and evolution of freezing rain within different regions of the United States. Thus, the objectives of this study are 1) to identify and document the local and synoptic conditions associated with freezing-rain events in the continental United States from 1976 to 1990, and 2) to evaluate the temporal and horizontal variability of these conditions among areas of high freezing-rain frequency. In this paper, section 2 describes the data and methodology used in this study, section 3 examines the local surface and upper-level conditions during freezing rain, section 4 identifies synoptic-scale processes observed during freezing rain, section 5 is a discussion of the results, and section 6 contains some conclusions from the study.
2. Data and methodology
We obtained hourly surface observations taken at stations south of 49°N from 1976 to 1990 from the National Oceanic and Atmospheric Administration/National Climate Data Center (NOAA/NCDC) to determine the horizontal distribution of freezing rain. We selected this time period because of the data availability at NCDC; data prior to 1976 were not used because some station data were only available for one out of every three hours, and data after 1990 were not used to ensure that present weather observations were taken by human observers and not by the Automated Surface Observing System in the United States. The surface dataset was created by merging DATSAV2 and TD-3280 formats. Quality control procedures for each format are documented in reports published by the U.S. Air Force (1986) and the U.S. Department of Commerce (1994). To ensure reliable statistical results, each station had to have at least 80% of the total possible observations for the months of September through April, 1976–90. After applying these criteria, a dataset of 489 stations was available for analysis.
Using the surface observations dataset described above, a count of freezing-rain observations (a single observation of freezing rain occurring exclusively, irrespective of intensity, and excluding freezing drizzle) was made for each of the 489 stations for the period September through April, 1976–90. These values were divided by 15 to obtain the annual freezing-rain frequency and then objectively analyzed onto a grid using a procedure described by Koch et al. (1983; Fig. 1). We then extracted rawinsonde data from the NOAA/Forecast Systems Laboratory rawinsonde database (Schwartz and Govett 1992) within areas that have a high frequency of freezing rain. The local environments of freezing rain at these rawinsonde locations were studied by assessing the thermodynamic stratification at the time of freezing rain, and by examining trends in the evolution of temperature and moisture variables, as well as precipitation type 6 h before and 6 h after freezing rain was observed at the rawinsonde launch time (1100 or 2300 UTC). These 1100 or 2300 UTC freezing-rain observations are hereafter referred to as ST observations.
Rawinsonde data were analyzed only if freezing rain and no other precipitation types occurred at rawinsonde launch time, irrespective of duration or intensity. The 1-h offset takes into account the actual time that the rawinsonde was released, typically 45–60 min before 0000 or 1200 UTC. Dry-bulb and dewpoint temperatures were extracted and interpolated (density weighted) to 25-hPa levels for analysis. This interval was chosen to retain sufficient resolution in order to identify shallow layers of warm and cold air that may exist.
We analyzed the synoptic environment during freezing rain by creating mean composite maps of meteorological conditions surrounding various stations located near a local maximum in freezing-rain frequency. As in Cortinas (2000), composite maps of geopotential height and temperature at 850 and 500 hPa using data from the National Centers for Environmental Prediction (formerly known as the National Meteorological Center) Grid Point Dataset (Mass et al. 1987) were created. Composite maps for surface conditions were also created using available surface observations.
3. Freezing-rain environment
The horizontal distribution of freezing rain compiled using the surface observations database reveals that there are four regions in the continental United States with a relatively high frequency of freezing rain: the Catskill and Allegheny Mountain region of the northeast, the Piedmont of North Carolina and Virginia, the central United States from southwest Missouri to Pennsylvania, and portions of the Pacific Northwest (Fig. 1). These results are similar to those obtained by Baldwin (1973), Robbins and Cortinas (1996), and Bernstein and Brown (1997). Using rawinsonde data at sounding sites closest to these areas, the local environments were investigated at Albany, New York; Buffalo, New York; Greensboro, North Carolina; Peoria, Illinois; and Spokane, Washington. For each of the five sites, median freezing-rain soundings were created and used to calculate various parameters (described later) that we believe characterize the thermodynamic stratification and can be used by forecasters to assess the potential for freezing rain. The evolution of surface variables during a 12-h period beginning 6 h before the synoptic time observation of freezing rain was also examined to identify similarities and differences among freezing-rain events at different geographical locations. This analysis also enabled us to compare the local environments of many freezing-rain events with individual events documented in the literature.
a. Surface conditions
Since surface [2 m above ground level (AGL)] temperature and moisture are critical to determining precipitation type near the ground, we evaluated the evolution of these variables near the surface by computing 6-h trends in temperature and dewpoint depression (dry-bulb temperature minus dewpoint temperature) before and after the ST observation. An analysis of these weather conditions shows that at most locations the evolution of weather conditions at the surface is similar to that associated with the passage of a warm front. The surface temperature trends usually show an increase in the 6 h prior to the ST observation for more than 80% of the events at all locations, except Greensboro and Albany (Table 1).
The trends at Greensboro and Albany differed somewhat from those described in previous freezing-rain studies (e.g., Stewart and King 1987; Stewart et al. 1990; Martner et al. 1993; Rauber et al. 1994) during which the surface temperature usually increased throughout a significant portion of most freezing-rain events. At Greensboro the largest percentage of cases was associated with surface cooling prior to the ST observation. At Albany only slightly more than half of the events were accompanied by an increasing surface temperature prior to the ST observation, whereas the temperature decreased prior to one-third of the ST observations. At all locations these temperature changes were accompanied by nearly saturated conditions throughout the 12-h period surrounding the ST observation, with a trend toward saturation up to the time of the ST observation.
During the ST freezing-rain observation, the median temperature at each site (Table 1) was near −1°C and the air was nearly saturated, similar to previous studies of freezing rain. After the ST observation, the temperature increased in most cases and a constant dewpoint depression trend was observed most often at all sites. At Buffalo and Peoria, all dewpoint depression trends were observed nearly equally.
Accompanying the changes in temperature and moisture, the evolution of precipitation type during the 12-h period was quite complex in each case and suggests that simple conceptual models to describe the evolution of these features probably do not accurately describe their evolution and should be used with caution. In only 25% of the cases, freezing precipitation (freezing rain and freezing drizzle) was the only precipitation type during the 12-h period. This value is representative of most locations, except for Peoria, where a smaller percentage of freezing-rain-only observations were reported during the 12-h period.
The remaining cases all involved two or three different precipitation types that followed a complex evolution (Table 2). At most locations, freezing precipitation and freezing precipitation mixed with snow were the most common types observed prior to the freezing-rain ST observation. After the ST observation, freezing precipitation, liquid precipitation, and mixtures thereof were most frequently observed, although they did not constitute the majority of the observations. In only 13% of all cases did the precipitation type begin as snow and/or ice pellets, change to freezing precipitation, then to liquid precipitation as the surface air warmed above the freezing point, an evolution predicted by the conceptual models of Stewart and King (1987) and Martner et al. (1993). Instead, the evolution of precipitation type during most cases was variable and complex, sometimes changing from one type to another and back to the same type several hours later. These observations indicate that the evolution of surface conditions that accompany freezing rain is not always the same among events or locations and that forecasters should be cautious when applying oversimplified conceptual models to all freezing-rain events.
b. Rawinsonde analysis
In addition to surface conditions, the thermodynamic stratification up to 500 hPa is critically important in the determination of precipitation type at the ground since diabatic processes such as melting, supercooling, and freezing are dependent on the temperature and relative humidity of the environment. An examination of the upper-level conditions within the local environment began with our hypothesis that the characteristics of the thermodynamic stratification observed during freezing rain at any one of the selected locations are similar in magnitude to those observed at the remaining locations. In other words, the climatological thermodynamic profile observed during freezing rain at one location is characteristic of the profile at any other. If true, this hypothesis implies that precipitation-type forecast techniques that utilize empirically derived parameters developed using thermodynamic data from single or multiple locations can be applied accurately at other locations without the need to recompute these parameters for each region. We recognize that although a robust, nonparametric hypothesis test was used, the results presented in this study may not be representative of all freezing-rain events due to the small sample size of rawinsonde observations. We suggest that additional studies with larger datasets should be pursued to verify these results.
To test our hypothesis, we characterized the thermodynamic profile at each rawinsonde site using various parameters and the Wilcoxon–Mann–Whitney rank-sum test with a rejection level of 5% [see Wilks (1995) for a complete explanation of this rank-sum test]. We determined statistically significant differences between each site by calculating sounding parameters (described in the next paragraph) at each site. Then, we compared the site sample for each station to the composites generated from the remaining sites. If the parameter from the site sample passes the significance test, then the sample was not statistically different from the sample parameter created using the rest of the sites. This means that there is a 95% probability that the site sample is similar to the remaining-sites sample. A hypothesis test failure indicates that the site sample is statistically different from the remaining-sites sample.
Analyzing these sounding parameters for all of the freezing-rain events revealed that, even though local freezing-rain environments possess basic common characteristics (i.e., a warm layer over a subfreezing surface-based layer), a notable amount of variability exists as well (Table 3), which should be considered by forecasters when anticipating freezing-rain conditions. Our analysis of the rawinsonde data for all sites in this study showed that general characteristics of the median profile during freezing rain include a midlevel warm layer with a median depth (MAD) of 1324 (380) m above a subfreezing layer having a median depth (MAD) of 613 (300) m. Interestingly, the lowest temperature below the warm layer did not always occur at the surface. In fact, the minimum temperature in the subfreezing layer occurred at a median height of 217 m AGL for all soundings and ranged from 42 m AGL at Spokane to 322 m AGL at Peoria (Table 3).
Combining the depth data with the thermodynamic profile yields a median (MAD) EMP of 2411 (1519) °C m and a median (MAD) EFP of 1247 (761) °C m (Fig. 3). The EMP–EFP ratio also showed significant variability, with a median (MAD) value of 1.7 (1.6) °C m. These values indicate that the magnitude of the melting process usually dominates over that of the freezing process during freezing rain, although the magnitude of these processes is quite variable. Zerr (1997) obtained similar results in his analysis of 34 freezing-rain soundings gathered from 19 different sounding locations. Unfortunately, he did not stratify his sounding analysis by location; instead, he combined all 34 soundings and subsequently discovered the significant variability that exists in the magnitudes of both the EMP and the EFP. Consequently, none of this variability was attributed to regional differences in local and synoptic processes associated with freezing rain.
In our study, only Greensboro failed the hypothesis test for the EMP while Albany and Spokane failed the test for the EFP. Peoria and Buffalo both passed the EMP and EFP hypotheses testing. This implies that the parameters computed for these sites are statistically similar to those contained in the collection of freezing rain soundings for the remaining sites combined and could be considered representative parameters for characterizing freezing-rain environments within any of the regions examined.
The sounding statistics computed for Greensboro revealed that the warm-layer characteristics were more pronounced there than for any other region during freezing rain. Specifically, Greensboro freezing-rain soundings exhibited the largest median EMP (Fig. 3a), the greatest median warm-layer depth (Fig. 4a), the highest median maximum wet-bulb temperature in the inversion (Fig. 5a), and one of the lowest median levels of the maximum wet-bulb temperature (Fig. 5b). The median EMP was 64% higher than the median value for the aggregate dataset; the median warm-layer depth was 26% higher. All of these warm-layer parameters, except the warm-layer depth, failed the hypothesis test. These test results suggest that the close proximity to the Gulf of Mexico, the Atlantic Ocean, and the Appalachians may play an important role in the generation of freezing-rain conditions in Greensboro, and possibly the entire Southeast, as suggested by Robbins (1998) and Bernstein (2000). This will be discussed in the following section.
In the Northeast, Albany has the most pronounced cold-layer characteristics of all the rawinsonde sites during freezing rain. Specifically, Albany freezing-rain soundings have the largest median EFP (Fig. 3b), roughly 33% higher than the median value for all soundings considered collectively. Further, the median cold layer is deeper than any other sounding site examined and approximately 63% deeper than the median value for the collective group of soundings (Fig. 4b). Albany's median lowest wet-bulb temperature below the melting layer and the median height of this temperature above the ground are similar to the other sites, except Spokane, which failed the hypothesis test for these two variables (Figs. 5c and 5d). The EFP and the median cold-layer depth at Albany were rejected by the hypothesis test, whereas at Buffalo, all the sounding parameters passed the hypothesis test.
The results from Buffalo and Albany suggest that intraregional differences in freezing-rain sounding characteristics also exist, particularly at low levels. The greatest differences between the Buffalo and Albany median soundings are related to the characteristics of the cold layer that we speculate is largely determined by the local terrain. At Albany, the Hudson Valley restricts the movement of low-level cold air out of the area, causing it to be deeper than at Buffalo, where no similar valley structure surrounds the city. It also appears that Lakes Ontario and Erie have a minimal effect on the freezing-rain frequency observed in the Northeast. Since freezing rain has a climatological tendency to occur north of a stationary front or a warm front associated with an advancing surface cyclone as suggested by Riehl et al. (1952, 48–49), then it would be expected that surface winds in these areas would be from an easterly or northeasterly direction. This would largely eliminate any thermal advection from the lakes and is consistent with Cortinas (2000) who found that the primary affect of the Great Lakes on freezing-rain frequency was most noticeable on their western shores.
In contrast to both Albany and Greensboro, Spokane experiences the least pronounced warm and cold layers of all the sounding sites examined (Table 2). Specifically, Spokane has the lowest median EMP and EFP (Fig. 3), the smallest median maximum wet-bulb temperature in the inversion (Fig. 5a), the greatest median wet-bulb temperature below the melting zone (Fig. 5c), as well as the shallowest median cold- and warm-layer depths (Fig. 4). The EMP is about 42% smaller and the median warm-layer depth is approximately 24% shallower than the median values for the aggregate dataset. The depths of the warm and cold air layers and the maximum upper-level temperature all failed the hypothesis test at Spokane, indicating that the conditions at Spokane are significantly different than those at the other sites. Of particular interest is the relative shallowness of the surface cold layer during freezing rain at Spokane and the closeness of the coldest temperature to the surface (Table 3). At Spokane, the lowest low-level temperature was observed at the surface during 33% of the freezing-rain reports, compared to 28% at Albany, 19% at Buffalo, 6% at Greensboro, and 10% at Peoria.
The characteristics of the cold dome during freezing rain at Spokane appear related to the “trapping” effect of the terrain surrounding the Columbia River basin as suggested by Bernstein (2000). Zishka and Smith (1980) showed that the climatological track of surface anticyclones from the Northwest Territories of Canada lies east of the basin during the winter. Although cold surface air associated with these systems is advected into the basin through numerous northern valleys, the track suggests that the deepest cold air remains east of the basin. After the passage of the anticyclone, the cold air becomes trapped since the mountainous terrain surrounding the basin restricts its movement. The high static stability at the top of these pools limits vertical mixing, which sometimes creates persistent low-level cloudiness that reduces insolation at the ground, strengthening the cold pool further (Whiteman et al. 2000). Storm systems that subsequently affect the Pacific Northwest advect midlevel warm air and may generate freezing precipitation over the basin.
While the sounding parameters for Spokane differ significantly from the other sounding sites considered collectively, the differences in the warm-layer characteristics between Spokane and Greensboro, and the cold-layer characteristics between Spokane and Albany, are remarkable. Recall that the warm layer during freezing rain is most pronounced at Greensboro and least pronounced at Spokane. The median EMP at Greensboro is 1.8 times higher than at Spokane. Similarly, Albany has the most pronounced surface cold layer during freezing rain and Spokane has the least. The EFP at Albany is 8.2 times higher than at Spokane.
The results of the hypothesis test suggest that these significant differences may be caused by local geography or the strength of the atmospheric processes that create favorable thermodynamic profiles for freezing rain as suggested by Bernstein (2000) and Cortinas (2000). In the following section, we investigate the regional differences in freezing-rain environments by analyzing the mean synoptic patterns associated with freezing rain in four regions surrounding the rawinsonde sites discussed in this section. Given these results, we speculate about the reasons for the interregional variability seen in the thermodynamic parameters.
4. Synoptic environment
The extent to which the local environments during freezing rain vary from region to region suggests that the synoptic-scale processes that create those environments also show a similar amount of variability. Mean composite maps for the surface, 850-hPa, and 500-hPa pressure levels were created for freezing-rain cases at seven stations in order to examine the synoptic environment during freezing rain. We selected these cases from surface stations located in areas of relatively high freezing-rain frequency (Fig. 1). Since we used gridded upper-level data to create these composites, we were not limited to rawinsonde locations; instead, we were able to increase the sample size of freezing-rain observations by using sites, which were often nonrawinsonde sites, collocated with local maxima in relative freezing-rain frequency. The sites we chose were Albany and Binghamton, New York; Greensboro, North Carolina; Springfield, Illinois; Columbia, Missouri; Missoula, Montana; and Pendleton, Oregon (Fig. 1). As in the previous section, only those reports during which freezing rain occurred at 1100 or 2300 UTC were used. Since surface data were not available in the Grid Point Dataset, we used the surface observations dataset to create surface composite maps of dry-bulb and dewpoint temperatures, winds, and mean sea level pressure.
a. Cold-air damming region of the Piedmont
Cold-air damming events in the Piedmont occur year-round; however, they occur most frequently during the winter months [see Bell and Bosart (1988) and Stauffer and Warner (1987) for a more information on cold-air damming events]. Damming situations are often characterized by a strong anticyclone centered over the northeast United States, a northeasterly flow that sometimes includes a slight upslope component, low clouds, and cool temperatures.
The orientation of the local maximum in freezing-rain frequency centered over Greensboro, North Carolina, relative to the Appalachian Mountains suggests that it may be the footprint of cold-air damming. Indeed, our examination of the surface maps at the time of freezing rain for each of the cases reveals that 68% were associated with cold-air damming conditions. The surface composite for Greensboro strongly reflects this, with a mean northeasterly wind parallel to the Appalachian Mountains, forming a mean thermal trough and an inverted pressure ridge there (Fig. 6a). Surface cooling within the 6-h period prior to precipitation onset takes place in a significantly higher percentage of freezing rain cases at Greensboro than at other locations (Table 1). This cooling is indicative of the advective, diabatic, and adiabatic processes that take place within the damming region during the creation of the cold dome.
Although stronger than most of the other regions examined, surface thermal advection (not shown) at Greensboro during freezing rain is extremely weak, with an average value of approximately −0.27°C (3 h)−1. Surface cold advection during freezing rain is stronger to the southwest over eastern Georgia implying that the cold dome is well established by the time precipitation begins at Greensboro and that cold advection there has begun to abate (Fig. 6b). These results are consistent with those of Forbes et al. (1987), Gay and Davis (1993), Rauber et al. (2000), and Bernstein (2000), who also found a significant number of freezing-rain events in the Southeast associated with cold-air damming.
To the west, a weak, inverted trough at the surface extends from the Gulf coast to Wisconsin (Fig. 6a). This trough seems to be associated with a deep short-wave trough at 850 hPa (Fig. 6c). While a closed surface cyclone does not show up in the mean surface composite map, it was found through examination of the individual cases for Greensboro that 75% were associated with an identifiable, but relatively weak, surface cyclone that usually moved northeast from the Gulf of Mexico across northern Florida and into the Atlantic Ocean.
The southerly flow east of the 850-hPa trough axis results in midlevel warm advection directly over the damming region (surface cold dome) and is consistent with Bell and Bosart (1988), who state that the cold dome during damming events is almost always located below 850 hPa, with strong warm advection occurring at 850 hPa. Indeed, a comparison of the 850-hPa thermal advection composite map with a similar map at 500 and 700 hPa (not shown) indicates that the strongest warm advection occurs near 850 hPa. The result of this warm advection is an elevated warm layer, with a maximum wet-bulb temperature that occurs at an average height of 853 m (Table 2). In addition to the importance of this advection to the creation of an elevated warm layer, the quasigeostrophic analysis indicates that it is also important, along with differential vorticity advection, in creating upward motion over the region (Fig. 6d).
b. Central United States
Surface and upper-level composite maps were created for Columbia, Missouri, and Springfield, Illinois, which were both located within the freezing-rain maximum in the central United States. These sites were chosen because they were located near local maxima in relative frequency of freezing rain. Unfortunately, in the central United States, these surface sites were not collocated with rawinsonde sites as was the case at Greensboro and Albany. The nearest rawinsonde site to Springfield and Columbia was Peoria, Illinois, which was analyzed in section 3. However, because of the proximity of these locations to Peoria and the homogeneity of the terrain in this region, we believe that the thermodynamic stratification during freezing rain at Peoria would be similar to that observed at Columbia and Springfield as well.
Generally, freezing rain appears to form over this region as a result of two possible scenarios: 1) isentropic lifting over a stationary arctic air mass with a surface baroclinic zone that usually lies 300–500 km south of the freezing-rain area, and 2) closed cyclones, associated with stronger vorticity dynamics than in the first scenario, passing within 650 km to the south of the composite sites. These results are similar to the findings of Riehl et al. (1952, 48–49) for freezing rain in Chicago and those of Bernstein (2000) for other locations in the United States. Because of the similarity in the composite maps at Columbia and Springfield, the close proximity of the stations, the homogeneity of the terrain in this area, and the desire to increase the sample size, we combined the events at Columbia and Springfield. We then separated the combined dataset by scenario to determine if there were any differences in the synoptic-scale processes associated with each one.
Examination of the individual freezing-rain events revealed that 57% were associated with a stationary or a very slow moving arctic front 300–500 km south of the station with the center of a strong high pressure system located near the central U.S.–Canadian border (Fig. 7a). The mean surface pressure composite for the stationary-front cases shows an inverted surface pressure trough extending from southern Arizona and New Mexico to near the Gulf coast of Texas and then northeast to Kentucky. At the time of freezing rain, the cold temperatures and light winds (Fig. 7b) produced thermal advection at the surface that was extremely weak and ranged from −0.1°C (3 h)−1 to −0.2°C (3 h)−1. The minimal advection indicates that the subfreezing surface air was already established before the freezing rain began, similar to that observed in the Southeast. The most significant cold advection at the surface was 800–900 km southwest of Columbia and was associated with the surface baroclinic zone.
Above the surface, a deep, positively tilted, 500-hPa long-wave trough (not shown) over the Rocky Mountain region always was associated with producing southwesterly flow over the freezing-rain area. The low- to midlevel warm advection decreased from roughly 1°C (3 h)−1 at 850 hPa (Fig. 7c) to around 0.5°C (3 h)−1 at 500 hPa (not shown) and was comparable to that of the other regions with the exception of the Northeast.
While the results presented thus far suggest that freezing rain in this region occurred most often as a result of warm advection over a stationary arctic air mass, freezing rain also occurred with closed surface cyclones that formed over the Southwest in New Mexico, southern Colorado, or western Texas. These cyclones usually tracked into southern Texas then northeast through southern Arkansas, across western Tennessee and Kentucky, and into the Northeast (Fig. 7d). Inspection of the individual events revealed that 43% were associated with a closed surface cyclone passing within 650 km to the south of the freezing-rain location.
Freezing rain associated with these closed surface cyclones did not always occur directly north of the attendant warm front, as suggested by Riehl et al. (1952, 48–49). In fact, only 33% of the freezing-rain observations at these sites occurred directly north of the warm front. During the other freezing-rain observations, the surface cyclone was located east, southeast, or south of the site. This tendency is evident in the mean surface wind and temperature composite map (Fig. 7e), which shows a surface cyclone south of the area and centered over northeast Arkansas associated with a closed circulation at 850 hPa (Fig. 7f). The mean surface cold advection is slightly stronger over this region during this scenario than any other region, and ranges between −0.2°C (3 h)−1 and −0.5°C (3 h)−1. There is evidence of a warm front east of the surface low with the warm advection maximized over Tennessee. Thus, freezing rain northwest of the surface low axis may be more common than Riehl et al. implied.
Unlike the cases associated with a stationary front, differential vorticity advection in the cases involving a surface low was typically much stronger. Comparing the mean differential vorticity advection field between the stationary front and closed cyclone (Figs. 7g and 7h) cases shows that the differential vorticity advection over the sites is four to six times greater during the closed cyclone cases than during the stationary front cases. In both scenarios, the contribution of the thermal advection to the vertical motion field is similar. With weaker vorticity dynamics in the stationary front cases, the primary forcing for the upward motion needed to induce precipitation and provide the warming aloft favorable for freezing rain appears to be warm advection near 850 hPa.
c. Allegheny and Catskill Mountains region
We analyzed freezing rain within the Allegheny and Catskill Mountains region by examining events at Binghamton and Albany. For these events, the primary track of the surface low was from the southern Great Plains to the Northeast, passing through the Midwest, and to the west of the location of freezing rain, with 76% of the observations at these locations associated with closed surface cyclones. This primary low track is consistent with Cortinas (2000), who found that surface lows associated with freezing rain in the Great Lakes region usually formed in the southern Great Plains and moved to the northeast across the lakes west of the freezing-rain region.
The distribution of precipitation following these surface cyclones as they move across the Midwest and into the Northeast suggests that the climatological distribution of freezing rain would show an area of relatively high frequency across this area as well. Indeed, the climatological distribution (Fig. 1) shows a nearly continuous area of freezing rain across this region with increasing frequency toward the Northeast. The relationship between freezing rain in the central and northeast United States was investigated by determining the number of times that an observation of freezing rain at Binghamton and/or Albany occurred within 48 h after a report of freezing rain at Columbia and/or Springfield. Likewise, the number of times that freezing rain occurred at Columbia and/or Springfield 48 h before an observation of freezing rain at Binghamton and/or Albany was also determined.
The analysis shows that 62% of the freezing-rain observations at Columbia and/or Springfield were followed by freezing rain at Binghamton and/or Albany. This connection between freezing rain at two locations in the central United States and two in the northeast provides additional support to the notion, proposed by Riehl et al. (1952, 48–49), that freezing rain in the midwest is associated with stationary fronts that extend from the midwest to the northeast. The opposite, however, is not true: only 2% of the ST freezing-rain observations at Binghamton and/or Albany were preceded by freezing rain at Columbia and/or Springfield. Thus, whereas freezing rain in the central United States usually does not precede freezing rain in the northeast, there is a moderate probability that it will occur in the northeast within 48 h of its occurrence in the central United States. Because of the many possible scenarios that could produce this relationship, the small sample size, and the limited scope of this study, the reasons for the low frequency of freezing rain in the northeast preceded by freezing rain in the central United States are unclear.
A subjective comparison of all the composite maps for Binghamton and Albany reveals that the general synoptic-scale conditions associated with freezing rain are similar for both locations (not shown); therefore, only data from the Albany cases were used as the basis for our discussion. During freezing rain in the Northeast, a well-defined mean surface cyclone was located roughly 500 km to the west of Albany (Fig. 8a). The composite surface wind field during freezing rain reveals a secondary circulation near the warm front, located near the Maryland–Pennsylvania border (Fig. 8b). This circulation appears to be associated with secondary cyclogenesis, referred to as a type B cyclone in the Atlantic coastal region of the United States by Miller (1946). Average surface conditions include very weak cold advection, greater (less negative) than −0.1°C (3 h)−1, near Albany, located approximately 300 km north of the warm frontal boundary (not shown). This observation along with pronounced cold-layer characteristics from the sounding analysis suggest that the cold dome is well established at the time of freezing rain, similar to that found within other regions.
Aloft, southwesterly flow at 850 hPa is associated with a long-wave trough west of the region (Fig. 8c). The mean warm advection at 850 hPa is greater than that at the other composite sites and this warm advection exists up to 500 hPa. Specifically, the mean thermal advection at Albany decreases from 2°C (3 h)−1 at 850 hPa (Fig. 8c) to near 0.5°C (3 h)−1 at 500 hPa. The strong warm advection suggests that the warm layer, similar to other locations, develops more rapidly over this region than in other regions. An analysis of the vertical motion forcing terms shows that positive differential vorticity advection and warm advection both contribute to upward motion (Fig. 8d).
d. Columbia River basin and the Bitterroot Range
Anticipating freezing rain in the Pacific Northwest poses a significant forecasting challenge because of the complex terrain. We examined the upper-level conditions in this region for freezing-rain observations at Pendleton, Oregon, and Missoula, Montana, two locations that experience a relative maximum in freezing rain (Fig. 1). A qualitative analysis of the composite maps at these locations indicates that the predominant synoptic-scale atmospheric processes associated with freezing rain at both of these locations are similar. Because of these similarities, only the mean upper-level composite maps for Pendleton will be used to document the dominant synoptic-scale processes, although we will also discuss some of the differences that exist between these locations.
Freezing rain in the Pacific Northwest typically occurs with a short-wave trough approaching the coast. At the surface, the area of freezing rain is located between strong high pressure centered over Utah and a broad area of low pressure centered near the Gulf of Alaska (Fig. 9a), similar to the surface pressure pattern identified by Bernstein (2000) for freezing-rain events in Spokane. The low-level stratification at Pendleton is affected by its location within the Columbia basin. Although freezing rain is typically associated with a surface cyclone in the Gulf of Alaska, the Cascade Mountains block mild maritime air from reaching the basin at low levels. Instead, weak surface winds and minimal cold advection during freezing rain indicate that subfreezing or near-freezing air in the basin (Fig. 9b) was established before the freezing rain began and was associated with the antecedent anticyclone. Moreover, cooling induced by upslope flow near Pendleton (Bernstein 2000), or a reduction of insolation resulting from low clouds that may persist due to a relatively high static stability configuration at the top of the cold pool (Whiteman et al. 2000), may help sustain the subfreezing layer during freezing rain. This subfreezing surface air coupled with weak surface cold advection of approximately −0.9°C (3 h)−1 provides a favorable thermodynamic stratification for freezing rain in the basin.
At the time of freezing rain, the mean thermal advection at 850 hPa is relatively strong and comparable to that found over the Southeast (Fig. 9c). In contrast to the warm-water source regions along the Southeast coast, however, cold water in the subarctic and California currents off the coast of the Pacific Northwest keeps the near–sea surface air colder than that found over the Southeast coast. Using Carlson's (1980) conveyor belt model, we believe that moist air from near the surface of the Pacific Ocean is advected northeastward over the Columbia basin ahead of approaching short-wave troughs. This air is cooled dry adiabatically, then moist adiabatically after becoming saturated, and becomes the low- and midlevel warm air observed during freezing rain.
Given this conceptual model, we believe that the maximum wet-bulb temperature of the warm layer can be determined by calculating the wet-bulb potential temperature of the source region air. In the Northwest, the mean January water temperatures in the subarctic current range from 8° to 10°C (Shea et al. 1990). Lifting this air to the median pressure level associated with the temperature maximum at Spokane (838 hPa) yields a temperature between 0° and 2°C. This is near the average 850-hPa temperature at Pendleton and slightly less than the warmest midlevel temperature recorded at Spokane (Table 2). We believe that the sea surface temperature of the nearby water source may explain the differences in the warm-layer characteristics between freezing rain events in the Northwest and those in the Southeast, where the source region is the warm Gulf of Mexico, and those elsewhere. An analysis of the vertical motion forcing terms shows that positive differential vorticity advection and warm advection both contribute to upward motion (Fig. 9d).
At Missoula, the surface and upper-level characteristics during freezing rain appear similar to those at Pendleton. As within the basin, virtually no surface thermal advection occurs during freezing rain at Missoula (not shown). Aloft, it appears that the strong westerly and southwesterly currents associated with an approaching short-wave trough over the Bitterroot Range causes downward motion that helps to initiate and maintain a low-level thermal ridge and pressure trough over central and eastern Montana (not shown). Although warming occurs with this downslope flow, Missoula is sheltered from this warming because of its valley location in extreme western Montana. Interestingly, the average warm advection at 850 hPa during freezing rain at Missoula is very weak, less than 0.3°C (3 h)−1. Our quasigeostrophic analysis indicates only minimal forcing for upward motion over this region, primarily from differential vorticity advection.
5. Discussion
The synoptic-scale analysis in the previous section revealed that there are some similarities and differences in the atmospheric processes that create a favorable freezing-rain thermodynamic structure across the United States. From the mean composite maps for each of the regions, the surface 0°C isotherm was usually located between 15 and 100 km to the south of the freezing-rain site and the average position of the 850-hPa 0°C isotherm was between 100 and 300 km to the north of the site. Additionally, surface thermal advection was usually very weak at the time of freezing rain at most stations, suggesting that the surface cold layer was firmly established by the time precipitation ensued.
Given that midlevel warm advection was responsible for producing the elevated warm layer, and given the level at which the maximum temperature occurred within this layer on the freezing-rain soundings, we estimated that warm advection usually reached a maximum near 850 hPa. Moreover, warm advection was often the dominant quasigeostrophic process associated with upward motion. In some cases involving a strong, well-developed storm system, however, the quasigeostrophic vertical motion resulted primarily from differential vorticity advection. Despite these similarities, important regional differences included possible secondary cyclogenesis associated with northeast events, cold-air damming primarily in the southeast and occasionally in the northeast, strong low-level warm advection over stationary surface fronts in the central United States, and minimal forcing for upward vertical motion over the Pacific Northwest.
In this study, we examined freezing-rain conditions surrounding certain locations in the United States where freezing rain occurs most often: most of New York, the Piedmont of North Carolina, portions of Missouri and Illinois, western Montana, and the Columbia basin. In each of these areas freezing rain generally occurred when four conditions were present: 1) nearly saturated air located near the ground, 2) midlevel upward vertical motion, 3) a deep low-level warm layer (∼1.3 km), and 4) a shallow (∼600 m) subfreezing surface layer. Although all of these conditions have been identified in previous studies, the interregional variability of these conditions and the synoptic processes associated with creating them have not.
The determination of interregional differences is important to forecasters who must accurately predict dangerous freezing-rain conditions, especially those forecasters at national forecasting centers, who are responsible for issuing these forecasts for the entire United States. Interregional differences can also affect the accuracy of forecast techniques, such as those that use critical thickness values derived for particular locations to forecast precipitation type within other regions. An understanding of any significant differences can help forecasters determine if the critical values determined for one particular region are applicable within other regions as well. Our analysis of the freezing-rain sounding data suggests that any forecast technique or algorithm that applies one set of sounding parameters to forecast freezing rain at all locations will not be as accurate as those that use locally derived parameters.
Based on the limited number of cases we examined, our results suggest that forecasters with a regional area of responsibility can use techniques that account for the statistical variability of the important sounding parameters (e.g., warm-layer depth, cold-layer depth, etc.), such as that described by Keeter and Cline (1991) and the model output statistics originally described by Bocchieri (1980). The difficulty with using these forecast techniques for national forecasts, however, is that they require the use of a statistically representative sample of these parameters for many years at nearly every forecast location, and observed data at mandatory pressure levels that are not available everywhere, particularly in the western mountainous terrain. Moreover, the continuous changes to some operational models such as the Eta Model (Black 1994) precludes the development of accurate model output statistics when these statistics rely on historical model data from a particular model configuration.
Because of the interregional variability of sounding parameters associated with freezing rain, we suggest that techniques to forecast freezing rain should not be based strictly upon empirically derived parameters that are not based on processes associated with precipitation. Instead, given the availability of high-resolution model data and sufficient computer resources, we suggest that forecast techniques should evaluate the entire thermodynamic structure and be based upon microphysical processes that determine precipitation type. Because of the spatial variability of freezing-rain conditions, these techniques are particularly important for forecasters concerned with large synoptic areas. A preliminary study that looked at forecasting freezing rain across the United States and Canada by Cortinas and Baldwin (2000) showed that two algorithms based on microphysical processes, those of Czys et al. (1996) and Ramer (1993), indeed, were more accurate than one created by Cortinas and Baldwin that was based on partial thickness values found in Keeter and Cline (1991), Younkin (1967), and Zerr (1997).
6. Conclusions
In this paper, we analyzed the synoptic-scale and local-scale freezing-rain environments and these are the key findings.
Freezing rain in the United States occurred most frequently in the Pacific Northwest, the central United States, the southeast, and the northeast.
Examining observations 6 h before and after freezing rain indicates that surface warm advection usually preceded the observations at Buffalo, Spokane, Albany, and Peoria; cold advection preceded the largest percentage of freezing-rain observations at Greensboro. Minimal thermal advection usually occurred at the time of freezing rain. After freezing rain, warm advection occurred most frequently at all locations. Prior to freezing rain, there was no consistent dewpoint depression trend for all stations.
The evolution of precipitation type at a single location was complex. Among all study locations, more than 75% of the freezing-rain events also included other types of precipitation before or after the ST freezing-rain observation. The evolution of precipitation type associated with the passage of a warm front found in other studies—that is, frozen, then freezing, then liquid—only occurred during 13% of the events in this study.
Temporal and horizontal variability existed among freezing-rain soundings: elevated warm-layer characteristics included a median maximum temperature of 3.2°C at a median height of ∼1100 m and a median depth of ∼1300 m; near the ground, cold-layer characteristics included a median surface temperature of −1°C, and a median layer depth of ∼600 m. The coldest wet-bulb temperature below the melting zone (median value of −2.9°C) was often elevated and typically occurred ∼200 m above the surface. Sounding characteristics at Albany, Greensboro, and Spokane were significantly different that those at Buffalo and Peoria.
Geography appeared to play a role in the formation of freezing rain. In the presence of an elevated warm layer, upward motion, and sufficient moisture, the valleys within complex terrain were favored areas for freezing rain because of the presence of subfreezing air that settled there. While the characteristics of the subfreezing surface layer may be influenced by terrain, nearby water sources appeared to limit the maximum temperature of the elevated warm layer. In the Pacific Northwest, warm layers were shallower and colder than for areas near the Atlantic Ocean and the Gulf of Mexico.
During freezing-rain events, quasigeostrophic adiabatic ascent was caused by both warm advection and positive differential vorticity advection in the southeast, the northeast, the Pacific Northwest, and during events in the central United States that are associated with stationary fronts. In the Bitterroot Range and during central United States events associated with closed surface cyclones, differential vorticity advection was the dominant process that causes ascent.
Freezing rain did not always occur directly north of a surface front in the central United States. Although 57% of freezing-rain events occurred on the cold side of stationary fronts, the remaining events were associated with closed surface cyclones. During 77% of those remaining events, freezing rain occurred when the surface cyclone was east, southeast, or south of the freezing-rain observation.
In the northeast, the surface composite maps indicate that freezing-rain events there may have been associated with secondary cyclogenesis along the eastern coast of the United States.
Variability in the synoptic-scale and local environments associated with freezing rain suggests that forecast techniques that rely on critical parameters derived for specific geographical locations may not be applicable elsewhere. Therefore, forecasters should consider using computer algorithms that evaluate all sounding data and are based upon precipitation microphysics if their current precipitation-type techniques are performing inaccurately.
Acknowledgments
We would like to thank Messrs. Ben Bernstein, Jon Racy, Kermit Keeter, Dr. David Schultz, and two anonymous reviewers for providing us with very useful reviews of this manuscript, and Messrs. Steve Fletcher and Doug Kennedy, who created the initial surface observations database. We are grateful to Neal Lott at the NOAA/National Climatic Data Center who provided the surface dataset, and to Dr. Robert Maddox for providing the financial support from the NOAA/OAR/National Severe Storms Laboratory to pursue this research project while the first author completed his graduate degree at the University of Oklahoma. Funding for this research was provided through NOAA–University of Oklahoma/Cooperative Institute for Mesoscale Meteorological Studies Cooperative Agreement NA67RJ0150.
REFERENCES
Baldwin, J. L., 1973: Climates of the United States. U.S. Department of Commerce Doc., 113 pp. [NTIS COM-74-11708/6.].
Bell, G. D., and Bosart L. F. , 1988: Appalachian cold-air damming. Mon. Wea. Rev, 116 , 137–161.
Bennett, I., 1959: Glaze: Its meteorology and climatology, geographical distribution, and economic effects. U.S. Army Quartermaster Research and Engineering Command Tech. Rep. EP-105, Natick, MA, 217 pp.
Bernstein, B. C., 2000: Regional and local influences on freezing drizzle, freezing rain, and ice pellet events. Wea. Forecasting, 15 , 485–508.
Bernstein, B. C., and Brown B. G. , 1997: A climatology of supercooled large drop conditions based upon surface observations and pilot reports of icing. Preprints, Seventh Conf. on Aviation, Range, and Aerospace Meteorology, Long Beach, CA, Amer. Meteor. Soc., 82–87.
Bernstein, B. C., Omeron T. A. , Politovich M. K. , and McDonough F. , 1998: Surface weather features associated with freezing precipitation and severe in-flight aircraft icing. Atmos. Res, 46 , 57–73.
Black, T. L., 1994: The new NMC mesoscale eta model: Description and forecast examples. Wea. Forecasting, 9 , 265–284.
Bluestein, H. B., 1992: Principles of Kinematics and Dynamics. Vol. I. Synoptic–Dynamic Meteorology in Midlatitudes. Oxford University Press, 431 pp.
Bocchieri, J. R., 1980: The objective use of upper air soundings to specify precipitation type. Mon. Wea. Rev, 108 , 596–603.
Carlson, T. N., 1980: Airflow through midlatitude cyclones and the comma cloud pattern. Mon. Wea. Rev, 108 , 1498–1509.
Cortinas, J. V. Jr, 2000: A climatology of freezing rain in the Great Lakes region of North America. Mon. Wea. Rev, 128 , 3574–3588.
Cortinas, J. V. Jr, and Baldwin M. E. , 2000: A preliminary evaluation of six precipitation-type algorithms for use in operational forecasting. Proc. Sixth Workshop on Operational Meteorology, Halifax, NS, Environment Canada, 207–211.
Czys, R., Scott R. , Tang K. C. , Przybylinski R. W. , and Sabones M. E. , 1996: A physically based, nondimensional parameter for discriminating between locations of freezing rain and ice pellets. Wea. Forecasting, 11 , 591–598.
Forbes, G. S., Anthes R. A. , and Thompson D. W. , 1987: Synoptic and mesoscale aspects of an Appalachian ice storm associated with cold-air damming. Mon. Wea. Rev, 115 , 564–591.
Frankenfield, H. C., 1915: Sleet and ice storms in the United States. Mon. Wea. Rev, 43 , 608.
Gay, D. A., and Davis R. E. , 1993: Freezing rain and sleet climatology of the southeastern USA. Climate Res, 3 , 209–220.
Keeter, K. K., and Cline J. W. , 1991: The objective use of observed and forecast thickness values to predict precipitation type in North Carolina. Wea. Forecasting, 6 , 456–469.
Kinzer, G. D., and Gunn R. , 1951: The evaporation, temperature and thermal relaxation time of freely falling waterdrops. J. Meteor, 8 , 71–83.
Koch, S. E., DesJardins M. , and Kocin P. J. , 1983: An interactive Barnes objective map analysis scheme for use with satellite and conventional data. J. Climate Appl. Meteor, 22 , 1487–1503.
Martner, B. E., Snider J. B. , Zamora R. J. , Byrd G. P. , Niziol T. A. , and Joe P. I. , 1993: A remote-sensing view of a freezing-rain storm. Mon. Wea. Rev, 121 , 2562–2577.
Mass, C. F., Edmon H. J. , Friedman H. J. , Cheny N. R. , and Recker E. R. , 1987: The use of compact discs for the storage of large meteorological and oceanographic datasets. Bull. Amer. Meteor. Soc, 68 , 1556–1558.
Meisinger, C. L., 1920: The precipitation of sleet and the formation of glaze in the eastern United States, January 20–25, 1920, with remarks on forecasting. Mon. Wea. Rev, 48 , 73–80.
Miller, J. E., 1946: Cyclogenesis in the Atlantic coastal region of the United States. J. Meteor, 3 , 31–44.
Okada, T., 1914: Notes on the formation of glazed frost. Mon. Wea. Rev, 42 , 284–286.
Pruppacher, H. R., and Klett J. D. , 1980: Microphysics of Clouds and Precipitation. D. Reidel, 714 pp.
Ramer, J., 1993: An empirical technique for diagnosing precipitation type from model output. Preprints, Fifth Int. Conf. on Aviation Weather Systems, Vienna, VA, Amer. Meteor. Soc., 227–230.
Ramsey, D. L., 1995: Case study of a freezing rain event utilizing Doppler radar. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 416–417.
Rauber, R. M., Ramamurthy M. K. , and Tokay A. , 1994: Synoptic and mesoscale structure of a severe freezing rain event: The St. Valentine's Day ice storm. Wea. Forecasting, 9 , 183–208.
Rauber, R. M., Olthoff L. S. , Ramamurthy M. K. , and Kunkel K. E. , 2000: The relative importance of warm rain and melting processes in freezing precipitation event. J. Appl. Meteor, 39 , 1185–1195.
Riehl, H., and and Coauthors, 1952: Forecasting in Middle Latitudes. Meteor. Monogr, No. 5, Amer. Meteor. Soc., 80 pp.
Robbins, C. C., 1998: An investigation of the local and synoptic environments associated with freezing rain in the contiguous United States. M.S. thesis, School of Meteorology, University of Oklahoma, 144 pp. [Available from Mr. Chris Robbins, NOAA/Tropical Prediction Center, 11691 S. W. 17th Street, Miami, FL 33165-2149.].
Robbins, C. C., and Cortinas Jr. J. V. , 1996: A climatology of freezing rain in the contiguous United States: Preliminary results. Preprints, 15th Conf. on Weather and Forecasting, Norfolk, VA, Amer. Meteor. Soc., 124–126.
Schwartz, B. E., and Govett M. , 1992: A hydrostatically consistent North American radiosonde database at the Forecast Systems Laboratory, 1946–present. NOAA Tech. Memo. ERL FSL-4, 81 pp. [Available from Forecast Systems Laboratory, Boulder, CO 80303.].
Shea, D. J., Trenberth K. E. , and Reynolds R. W. , 1990: A global monthly sea surface temperature climatology. NCAR Tech. Note NCAR/TN-345+STR, 167 pp.
Stauffer, D. R., and Warner T. T. , 1987: A numerical study of Appalachian cold-air damming and coastal frontogenesis. Mon. Wea. Rev, 115 , 799–821.
Stewart, R. E., and King P. , 1987: Freezing precipitation in winter storms. Mon. Wea. Rev, 115 , 1270–1279.
Stewart, R. E., Crawford R. W. , Donaldson N. R. , Low T. B. , and Sheppard B. E. , 1990: Precipitation and environmental conditions during accretion in Canadian east coast winter storms. J. Appl. Meteor, 29 , 525–538.
Thoreson, A. D., 1995: Doppler radar case study of a freezing rain event. Preprints, 27th Conf. on Radar Meteorology, Vail, CO, Amer. Meteor. Soc., 412–413.
U.S. Air Force, 1986: USAFETAC Climatic database user's handbook No.4, DATSAV2, Surface. Rep. USAFETAC/UH-86/004, 52 pp. [Available from National Climatic Data Center, Federal Building, 151 Patton Avenue, Asheville, NC 28801-5001.].
U.S. Department of Commerce, 1994: Hourly surface airways observations. TD-3280, 40 pp. [Available from National Climatic Data Center, Federal Building, 151 Patton Avenue, Asheville, NC 28801-5001.].
Whiteman, C. D., Zhong S. , Bian X. , Shaw W. , and Hubbe J. M. , 2000: Wintertime cold air pools in the Columbia Basin. Preprints, Ninth Conf. on Mountain Meteorology, Aspen, CO, Amer. Meteor. Soc., 298–303.
Wilks, D. S., 1995: Statistical Methods in the Atmospheric Sciences. Academic Press, 467 pp.
Young, W. R., 1978: Freezing precipitation in the southeastern United States. M.S. thesis, Dept. of Atmospheric Sciences, Texas A&M University, 123 pp.
Younkin, R. J., 1967: A snow index. ESSA TM WTBM NMC-40, U.S. Dept. of Commerce, 5 pp.
Zerr, R. J., 1997: Freezing rain: An observational and theoretical study. J. Appl. Meteor, 36 , 1647–1661.
Zishka, K. M., and Smith P. J. , 1980: The climatology of cyclones and anticyclones over North America and surrounding ocean environs for January and July, 1950–77. Mon. Wea. Rev, 108 , 387–401.
Average number of freezing-rain hours per year using data from 1976 to 1990. Station locations referenced in the text are indicated by their three-letter identifier (ALB, Albany, NY; BGM, Binghamton, NY; BUF, Buffalo, NY; COU, Columbia, MO; GEG, Spokane, WA; GSO, Greensboro, NC; MSO, Missoula, MT; PDT, Pendleton, OR; PIA, Peoria, IL; SPI, Springfield, IL)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Average number of freezing-rain hours per year using data from 1976 to 1990. Station locations referenced in the text are indicated by their three-letter identifier (ALB, Albany, NY; BGM, Binghamton, NY; BUF, Buffalo, NY; COU, Columbia, MO; GEG, Spokane, WA; GSO, Greensboro, NC; MSO, Missoula, MT; PDT, Pendleton, OR; PIA, Peoria, IL; SPI, Springfield, IL)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Average number of freezing-rain hours per year using data from 1976 to 1990. Station locations referenced in the text are indicated by their three-letter identifier (ALB, Albany, NY; BGM, Binghamton, NY; BUF, Buffalo, NY; COU, Columbia, MO; GEG, Spokane, WA; GSO, Greensboro, NC; MSO, Missoula, MT; PDT, Pendleton, OR; PIA, Peoria, IL; SPI, Springfield, IL)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Sounding parameters analyzed (EMP, environmental melting parameter; EFP, environmental freezing parameter; T, dry-bulb temperature; Tw, wet-bulb temperature). Cold and warm layers are measured relative to the wet-bulb temperature. See text for an explanation of the EMP and the EFP
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Sounding parameters analyzed (EMP, environmental melting parameter; EFP, environmental freezing parameter; T, dry-bulb temperature; Tw, wet-bulb temperature). Cold and warm layers are measured relative to the wet-bulb temperature. See text for an explanation of the EMP and the EFP
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Sounding parameters analyzed (EMP, environmental melting parameter; EFP, environmental freezing parameter; T, dry-bulb temperature; Tw, wet-bulb temperature). Cold and warm layers are measured relative to the wet-bulb temperature. See text for an explanation of the EMP and the EFP
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Schematic plots of the environmental (a) melting and (b) freezing parameters for all rawinsonde locations and the combined dataset. 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). A P (F) indicates that the parameter at that location passed (failed) the hypothesis test
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Schematic plots of the environmental (a) melting and (b) freezing parameters for all rawinsonde locations and the combined dataset. 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). A P (F) indicates that the parameter at that location passed (failed) the hypothesis test
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Schematic plots of the environmental (a) melting and (b) freezing parameters for all rawinsonde locations and the combined dataset. 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). A P (F) indicates that the parameter at that location passed (failed) the hypothesis test
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Same as in Fig. 3 except for (a) warm- and (b) cold-layer depths (with respect to wet-bulb temperature)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Same as in Fig. 3 except for (a) warm- and (b) cold-layer depths (with respect to wet-bulb temperature)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Same as in Fig. 3 except for (a) warm- and (b) cold-layer depths (with respect to wet-bulb temperature)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Same as in Fig. 3 except for (a) maximum wet-bulb temperature, (b) height (AGL, m) of maximum wet-bulb temperature, (c) low-level minimum wet-bulb temperature, and (d) height (AGL, m) of minimum wet-bulb-temperature
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Same as in Fig. 3 except for (a) maximum wet-bulb temperature, (b) height (AGL, m) of maximum wet-bulb temperature, (c) low-level minimum wet-bulb temperature, and (d) height (AGL, m) of minimum wet-bulb-temperature
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Same as in Fig. 3 except for (a) maximum wet-bulb temperature, (b) height (AGL, m) of maximum wet-bulb temperature, (c) low-level minimum wet-bulb temperature, and (d) height (AGL, m) of minimum wet-bulb-temperature
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Greensboro, NC. The diamond indicates the station location. (a) Mean sea level pressure (hPa, solid line). (b) Temperature (°C, solid line), dewpoint (°C, dashed line), and winds (full barb and half-barb denote 5 and 2.5 m s−1). (c) Geopotential height (m, solid line), temperature (°C, dashed line), and winds at 850 hPa. (d) Contributions of 850–500-hPa vorticity advection (×10−13 Pa s−1 m−2, solid lines) and 850-hPa thermal advection (×10−13 Pa s−1 m−2, dashed lines) to vertical velocity. Positive (negative) values in (d) are associated with upward (downward) motion
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Greensboro, NC. The diamond indicates the station location. (a) Mean sea level pressure (hPa, solid line). (b) Temperature (°C, solid line), dewpoint (°C, dashed line), and winds (full barb and half-barb denote 5 and 2.5 m s−1). (c) Geopotential height (m, solid line), temperature (°C, dashed line), and winds at 850 hPa. (d) Contributions of 850–500-hPa vorticity advection (×10−13 Pa s−1 m−2, solid lines) and 850-hPa thermal advection (×10−13 Pa s−1 m−2, dashed lines) to vertical velocity. Positive (negative) values in (d) are associated with upward (downward) motion
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Greensboro, NC. The diamond indicates the station location. (a) Mean sea level pressure (hPa, solid line). (b) Temperature (°C, solid line), dewpoint (°C, dashed line), and winds (full barb and half-barb denote 5 and 2.5 m s−1). (c) Geopotential height (m, solid line), temperature (°C, dashed line), and winds at 850 hPa. (d) Contributions of 850–500-hPa vorticity advection (×10−13 Pa s−1 m−2, solid lines) and 850-hPa thermal advection (×10−13 Pa s−1 m−2, dashed lines) to vertical velocity. Positive (negative) values in (d) are associated with upward (downward) motion
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Columbia, MO, and Springfield, IL, associated with (a), (b), (c), and (g) a stationary front and (d), (e), (f), and (h) a closed surface cyclone. Diamonds indicate the station locations. Contours in (a) and (c), (b) and (d), (e) and (f), (g) and (h) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Columbia, MO, and Springfield, IL, associated with (a), (b), (c), and (g) a stationary front and (d), (e), (f), and (h) a closed surface cyclone. Diamonds indicate the station locations. Contours in (a) and (c), (b) and (d), (e) and (f), (g) and (h) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Columbia, MO, and Springfield, IL, associated with (a), (b), (c), and (g) a stationary front and (d), (e), (f), and (h) a closed surface cyclone. Diamonds indicate the station locations. Contours in (a) and (c), (b) and (d), (e) and (f), (g) and (h) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
(Continued)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
(Continued)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
(Continued)
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Albany, NY. Diamond indicates station location. Contours in (a)–(d) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Albany, NY. Diamond indicates station location. Contours in (a)–(d) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Albany, NY. Diamond indicates station location. Contours in (a)–(d) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Pendleton, OR. Diamond indicates station location. Contours in (a)–(d) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Pendleton, OR. Diamond indicates station location. Contours in (a)–(d) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Mean environmental conditions during freezing rain at Pendleton, OR. Diamond indicates station location. Contours in (a)–(d) are the same as Figs. 6a–d, respectively
Citation: Weather and Forecasting 17, 1; 10.1175/1520-0434(2002)017<0047:LASEAW>2.0.CO;2
Surface thermodynamic and present weather trends observed during freezing rain at Albany, NY (ALB); Buffalo, NY (BUF); Spokane, WA (GEG); Greensboro, NC (GSO); Peoria, IL (PIA). Trends were calculated using surface conditions observed at 6 h before the synoptic time (1100 or 2300 UTC) observation and 6 h after the observation (T is the dry-bulb temperature, Td is the dewpoint temperature, and t is time)
Types of precipitation occurring during the 6-h period before and after the synoptic-time (ST) freezing-rain observation (see text for an explanation of the ST observation). The last row is the percentage of events during which only freezing precipitation was observed during the 12-h period. Precipitation types: FZ, freezing rain or freezing drizzle; SN, snow; RA, rain; DZ, drizzle; IP, ice pellets. A solidus (/) indicates a mixture of multiple types. No precipitation indicates that no precipitation was reported during the indicated 6-h period
Variables derived from rawinsonde data obtained during freezing rain at Albany (ALB), Buffalo (BUF), Spokane (GEG), Greensboro (GSO), and Peoria (PIA), (MAD, median absolute deviation; EMP, environmental melting parameter; EFP, environmental freezing parameter)
Although the temperature at the surface of the ice/water particle is dependent on the radius of the ice core, it approaches the wet-bulb temperature (Kinzer and Gunn 1951) as the radius approaches zero. Therefore, the replacement of the particle surface temperature with the wet-bulb temperature represents an upper bound on the rate of melting or freezing.
