Motivated by the intensity and severity of winds and temperature falls that frequently accompany rapidly developing cold fronts in northern Utah, this paper presents a 25-yr climatology of strong cold frontal passages over the Intermountain West and adjoining western United States. Using conventional surface observations and the North American Regional Reanalysis, strong cold frontal passages are identified based on a temperature fall of 7°C or greater in a 2–3-h period, a concurrent pressure rise of 3 hPa or greater, and the presence of a large-scale 700-hPa temperature gradient of at least 6°C (500 km)−1. The number of strong cold frontal passages exhibits a strong continental signature with very few events (<10) along the Pacific coast and more than 200 events east of the Continental Divide. The number of events increases dramatically from the Cascade Mountains and Sierra Nevada to northern Utah, indicating that the Intermountain West is a frequent cold front breeding ground.
A composite of the 25 strongest events at Salt Lake City (based on the magnitude of the temperature fall) reveals that confluent deformation acting on a broad baroclinic zone over central Nevada commonly initiates Intermountain frontogenesis. The confluent deformation develops in southwesterly large-scale flow and appears to be enhanced by flow deflection around the Sierra Nevada. Quasi-stationary development and intensification of the southwest–northeast-oriented cold front then occurs as a mobile upper-level trough approaches from the west. The front becomes mobile as cold advection and ascent associated with the upper-level trough overtake the low-level front. Cloud and precipitation observations suggest that differential diabatic heating contributes to the rapid frontal intensification in many events.
Intense, rapidly developing cold fronts threaten lives and property over the Intermountain West several times a year. High winds, which may occur in the pre- or postfrontal environment, can cause major wildfire runs, road closures because of blowing dust, power outages, and property damage. For example, 40 m s−1 wind gusts produced more than $15 million of damage as a cold front moved across northern Utah on 5 June 1995 (NCDC 1995, p. 356), while the cold front accompanying the 2002 Tax Day Storm (15 April) produced dramatic temperature falls of 7°C in 10 s and 19°C in 2 h, wind gusts of more than 30 m s−1, and $2 million in property damage (NCDC 2002, p. 243).
Previous studies of cold fronts over western North America are limited primarily to landfalling Pacific fronts (e.g., Elliott 1958; Hobbs et al. 1980; Hobbs and Persson 1982; Reynolds and Kuciauskas 1988; Braun et al. 1997; Colle et al. 1999; Yu and Smull 2000; Chien et al. 2001) and cold fronts along the eastern slopes of the Rocky Mountains (e.g., Sanders 1955; Shapiro et al. 1985; Colle and Mass 1995; Marwitz and Toth 1993; Schultz et al. 1997, 1998; Neiman and Wakimoto 1999; Darby et al. 1999; Neiman et al. 2001; Schultz 2004; Bosart et al. 2008). Although landfalling Pacific cold fronts may feature a narrow cold frontal rainband and a gravity current–like structure (e.g., Carbone 1982), they are often accompanied by modest temperature changes of 2°–4°C or less (e.g., Braun et al. 1997; Colle et al. 1999; Chien et al. 2001). In contrast, cold fronts east of the Rockies can feature intense horizontal temperature gradients. For example, Shapiro (1984) describes a cold front that produced temperature falls of 6°C in 10 s and 13°C in 2.5 h, and Schultz (2004) documents a cold front that produced temperature falls of 10°C in 2 min and nearly 28°C in 10 h. Nonclassical frontal structures, such as the cold front aloft (e.g., Hobbs et al. 1996), and complex mergers between elevated and surface-based frontal zones (e.g., Neiman et al. 1998; Neiman and Wakimoto 1999) are also common east of the Rockies.
Knowledge of frontal evolution over the Intermountain West is based on a handful of studies that describe topographic frontal distortions (e.g., Steenburgh and Blazek 2001), deep warm-core seclusions (e.g., Horel and Gibson 1994), and the modulation of orographic precipitation processes by surges of low-θe air aloft and ahead of the surface cold front (e.g., Long et al. 1990; Steenburgh 2003; Shafer et al. 2006). Although many of these studies emphasize the lack or loss of low-level baroclinity due to upstream orographic effects, intense cold fronts may develop over the Intermountain West and can be accompanied by damaging winds (e.g., Schultz and Trapp 2003).
Despite being known for complex topography, “aridity, more than anything else, gives the western landscape its character” (Stegner 1992). This aridity leads to high Bowen ratios [i.e., the ratio of surface sensible to latent heat flux; Bowen (1926); Lewis (1995)], large diurnal variations in sensible heating, deep daytime convective boundary layers, and strong nocturnal surface inversions. Studies of subtropical Australian cold fronts, which also form in an arid region with strong daytime sensible heating, suggest that if the surface heating is homogeneous, it has an overall frontolytic effect. This is primarily due to differential heating of the mixed layer, which is shallower in the postfrontal environment (Reeder 1986; Smith et al. 1995; Reeder and Tory 2005; Tory and Reeder 2005). In contrast, inhomogeneous surface heating, which may arise from differential cloud cover and/or precipitation, can produce intense frontogenesis during the day (e.g., Koch et al. 1995, 1997; Gallus and Segal 1999; Segal et al. 2004; Bosart et al. 2007). These effects may be magnified by low-level evaporative cooling (e.g., Huang and Emanuel 1991; Katzfey and Ryan 1997; Schultz and Trapp 2003), which can be pronounced over the climatologically dry lower troposphere of the Intermountain West.
This paper describes the characteristics of strong Intermountain cold fronts and investigates the large-scale, orographic, and diabatic processes responsible for their development. Our findings show that the frontal development begins in response to large-scale confluent frontogenesis, which appears to be enhanced by flow deflection around the Sierra Nevada, and the subsequent rapid frontal intensification appears to be a result of differential diabatic processes. These results contribute to our knowledge of frontal processes over arid and mountainous regions and should be valuable for improving short-term forecasts over the Intermountain West.
2. Data and methods
a. Data sources
Strong cold fronts were identified during a 25-yr study period (1979–2003) using surface temperature and pressure observations from the National Climatic Data Center (NCDC) Global Integrated Surface Hourly database (DS-3505), which includes standard hourly and special observations, and gridded analyses from the North American Regional Reanalysis (NARR; Mesinger et al. 2006). To ensure reliable results, we considered only first-order observing stations with at least 90% temperature and pressure data availability, yielding a total of 77 sites (Fig. 1). To increase station density over the Intermountain West, one exception to this requirement was allowed at Wendover, Utah (ENV), where 21% of the hourly reports were missing prior to 1996. Other surface variables were not used for cold front identification but were used to examine the sensible weather accompanying cold fronts (e.g., 1-h accumulated precipitation and wind gust).
Changes in instrumentation and reporting standards during the study period affect our results. The most significant change occurred on 1 July 1996 with the transition from surface aviation observation (SA) to aviation routine weather report (METAR) reporting code. After this transition, the average annual number of special observations nearly doubled (Fig. 2a) and the frequency of identified strong cold frontal passages increased (Fig. 2b) since special nonhourly observations more effectively resolve rapid temperature and pressure changes. The most significant instrumentation change occurred with the installation of Automated Weather Observation Stations (AWOS), a project that began in 1991 and was completed at most locations by 1997. This instrumentation change, however, appears to have had minimal effect compared to the reporting code transition.
The NARR was also used to identify strong cold frontal passages and to generate a composite of events at Salt Lake City, Utah (SLC). The NARR is a 32-km, 45-layer regional reanalysis based on the eta model three-dimensional variational data assimilation (3DVAR) system providing analyses covering North America beginning in 1979 (Mesinger et al. 2006). NARR data were obtained from the National Oceanic and Atmospheric Administration Operational Model Archive Distribution System (NOMADS) at the National Climactic Data Center Web site (http://nomads. ncdc.noaa.gov/#narr_datasets); the data were processed and plotted using the Grid Analysis and Display System (GrADS). Composites were generated using the original NARR analyses, which for 2003 are affected by an oceanic precipitation data processing error (Mesinger et al. 2006; West et al. 2007). Only one of the composite events, however, occurred in these years and was not significantly affected.
b. Strong cold front identification
Numerous salient weather changes may accompany the passage of a cold front, including a temperature fall, a pressure rise, a wind direction and/or speed change, a cloud cover increase or decrease, and precipitation (e.g., Petterssen 1956; Sanders and Doswell 1995). In this study, we use three criteria to identify strong cold frontal passages: 1) a surface temperature fall of at least 7°C over a 2–3-h period, 2) a corresponding altimeter pressure rise of at least 3 hPa, and 3) the presence of a 700-hPa temperature gradient of at least 6°C (500 km)−1. Wind changes are not used because local and regional complex topography affect wind direction and speed (e.g., Williams 1972; Steenburgh and Blazek 2001), whereas cloud cover and precipitation are not used because of their inherent variation with cold frontal structure (e.g., Moore and Smith 1989; Sanders and Doswell 1995).
The temperature fall criterion is based in part on Sanders (1999), who defines a strong surface baroclinic zone (i.e., front) as having a gradient of 8°C (110 km)−1, or approximately 7°C (100 km)−1. Our approach assumes that most of the baroclinity (7°C or greater) passes within a 2–3-h period, which favors fast-moving cold fronts and/or cold fronts with sharp edges. We use a 2-h minimum time period rather than the 1-h standard reporting interval to help limit contamination by erroneous data reports and a 3-h maximum time period because our goal is to sample strong cold fronts with abrupt changes; a longer period may include weaker events that feature broad baroclinity. However, we recognize that one may also categorize a cold front as being “strong” if a large temperature fall occurs over a larger time interval.
We include the 3-hPa altimeter pressure change criterion because large pressure rises typically occur behind strong cold fronts over the Intermountain West (e.g., Koppel et al. 2000; Schultz and Trapp 2003), and because 3 hPa is much larger than the early morning and early evening pressure rises of 0.5–1 hPa in 2–3 h that are produced by the diurnal and semidiurnal tides (Mass et al. 1991). The pressure change criterion also helps eliminate events where large temperature changes are produced by rapid nocturnal cooling after sunset. Since changes in temperature can produce an artificial change in sea level pressure (Mass et al. 1991), we use the altimeter setting for the pressure change calculations; this was also the most reliably reported pressure variable.
The third criterion requires the temperature and pressure changes to occur in the presence of a 700-hPa baroclinic zone. The 700-hPa level lies near the top of the Intermountain boundary layer and is less influenced by diurnal surface temperature variability. Furthermore, cursory inspection of potential events at low-elevation Pacific coast stations suggests 700 hPa was as effective as 850 hPa for diagnosing the large-scale temperature gradient.
For each potential event (identified with surface temperature and pressure changes), we used a hub-and-spoke method to determine the difference in 700-mb temperature between the station location and a point 500 km away along 8 radials (every 45°). The magnitude of the temperature difference along the radial with the most negative value, which points toward the coldest air, must equal or exceed 6°C.
The 6°C (500 km)−1 temperature gradient value represents the mode for all potential Intermountain cold front events (i.e., those meeting the temperature and pressure criteria) and is about 2°C greater than the NARR climatological 700-hPa temperature gradient at the 18 Intermountain stations (Fig. 3; Intermountain stations are identified by asterisks in Fig. 1). This value attempts to discriminate between events that were associated with weak large-scale baroclinity and those that were likely produced by a synoptic-scale cold front, and it also eliminates 40% of the potential events identified with only surface temperature and pressure changes (Fig. 3).
a. Geographic and temporal characteristics
The climatology identified a total of 4580 events over the western United States, with 948 events at Intermountain locations. The number of strong cold frontal passages exhibits a strong continental signature with very few events (<10) along the Pacific coast and more than 200 events at stations east of the Continental Divide (Fig. 4). Strong cold frontal passages are infrequent at coastal locations primarily because landfalling cold fronts rarely meet the 7°C surface temperature fall criterion. The highest number of events occurs near the eastern slopes of the Rocky Mountains from north-central Montana equatorward to southeastern Colorado, with the greatest number of events (295) at Pueblo, Colorado (PUB; see Fig. 1 for location).
The number of events increases from the Sierra Nevada–Oregon Cascades (e.g., near RNO, Fig. 1; 17 events) to a local maximum over northern Utah (SLC, 100 events), indicating that strong cold fronts frequently develop over the intermediate Intermountain West. One Intermountain station, however, is not consistent with this pattern. The total of 31 events at ENV is fewer than nearby locations because of missing hourly observations prior to 1996 (see section 2) and the fact that northwesterly postfrontal winds experience strong, localized downslope warming at this location, which decreases the temperature change accompanying cold fronts.
For all the western stations, there is a modest seasonal cycle in the frequency of strong cold frontal passages, with a maximum during late spring (April–June) and a minimum in the late fall and early winter (November–February; Fig. 5). This pattern is more pronounced for Intermountain stations where the event frequency increases in the spring and reaches a maximum in June (Fig. 6). Geographically, the month of maximum event frequency generally shifts from spring to summer as one moves poleward (Fig. 6, events clustered for clarity). The only locations featuring a pronounced cold season maximum are coastal or in the extreme desert Southwest. These results suggest that the frequency of strong cold front passages is only partly related to synoptic activity. For example, although Nevada cyclogenesis is most common in March (Lee 1995) and cold frontal passages at SLC peak in April (Astling 1984), the frequency of strong cold frontal passages at SLC peaks in late May. Thus, at SLC, cold fronts tend to be less common but more intense as the warm season approaches. Further, although there is a secondary maximum in Nevada cyclogenesis in November (Lee 1995), no such maximum exists in strong cold front frequency, suggesting that only a small percentage of frontal passages are strong during the fall Nevada cyclone season.
Strong Intermountain cold frontal passages occur most frequently in the late afternoon and evening, with 76% of all events occurring between 1400 and 2100 LST (Fig. 7a). The peak frequency (14%) occurs at 1800 LST, whereas the minimum of 0.7% is at 0700 and 0800 LST. This pronounced diurnal cycle is found at all locations in the western United States and is similar to the evening–early nighttime maximum for cold fronts at Denver, Colorado (Wiesmueller 1982). Although the number of late afternoon–early evening events might be enhanced by diurnal cooling, which increases the likelihood of meeting our temperature criterion, the large number of late-day events suggests that many fronts strengthen in response to daytime diabatic heating (e.g., Koch et al. 1995). The time of maximum frequency also varies by time of year, tending to peak later in the day during the warm season and earlier in the day during the cool season (Fig. 7b), which further supports the importance of daytime heating for frontal intensification.
The origin of the postfrontal air mass was estimated using the direction of the 700-hPa temperature ascendant (−∇T), which is oriented normal to the isotherms and directed toward lower temperatures. At most locations, the coldest air varies from the west to the north of the stations (Fig. 8a), implying that the postfrontal air mass usually comes from these directions. However, this approach cannot be used to approximate the surface motion of the cold front since the complex terrain distorts the orientation of the surface front (e.g., Steenburgh and Blazek 2001).
To approximate the surface motion of cold fronts, we examined the surface wind direction at 1 h after frontal passage, focusing near the Utah–Idaho–Wyoming border. The postfrontal wind direction frequency shows a strong contrast across the higher terrain that runs from western Wyoming to northeastern Utah. Northerly to northeasterly postfrontal winds are common at Lander, Wyoming (LND), whereas northwesterly and westerly postfrontal winds are common at SLC and Burley, Idaho (BYI), respectively (Fig. 8b). Idaho Springs, Colorado (IDA), and Rock Springs, Wyoming (RKS), feature bimodal distributions, with easterly postfrontal winds at the latter reflecting the westward penetration of continental cold fronts. At IDA, southwesterly postfrontal winds result from the topographic channeling of cold fronts up the Snake River Plain (Steenburgh and Blazek 2001), whereas northerly postfrontal winds accompany cold fronts that have traversed the Continental Divide. These results, combined with the rarity of strong cold fronts along the Pacific coast (Fig. 4), support the hypothesis that, except near the Continental Divide, most strong Intermountain cold fronts are of Pacific origin and develop or intensify over the Intermountain West.
b. Surface weather accompanying strong Intermountain cold fronts
High winds frequently accompany strong Intermountain cold fronts. For Intermountain stations, 45% of the strong cold frontal passages feature a maximum wind gust equal to or greater than the National Weather Service high-wind advisory criterion (≥20 m s−1) within ±12 h of frontal passage (Fig. 9a). The peak wind gust occurs most frequently (66% of the time) at or within an hour of frontal passage (Fig. 9b). Outside this time window, peak gusts are more commonly prefrontal than postfrontal. A similar frequency and magnitude of wind relative to frontal passage is found for all western U.S. cold frontal passages (not shown).
Precipitation data show that most strong Intermountain cold frontal passages are dry or produce light precipitation, at least at the valley locations examined in this study. Although precipitation (trace or greater) within a 30-h period1 centered on frontal passage occurs during approximately 57% of Intermountain events, total precipitation is ≤5.1 mm 76% of the time. Eleven percent of the Intermountain events feature only a trace of precipitation (Fig. 10a). In contrast, the average precipitation for the 25 strongest cold frontal passages (based on the magnitude of temperature fall; Table 1) at SLC (6.7 mm) was well above the Intermountain average (2.0 mm).
There is a distinct precipitation pattern relative to cold frontal passages. Measurable precipitation is most common during the 6-h period of frontal passage and, to a lesser degree, the subsequent postfrontal periods (Fig. 10b). Prefrontal precipitation is less common. There was little geographic variability to this distribution (not shown), suggesting a similar precipitation distribution for strong cold frontal passages over much of the western United States; however, these results are biased toward continental regions because of the limited number of strong coastal fronts.
c. Composite evolution of strong cold fronts at SLC
As shown in the previous section, the frequency of strong cold frontal passages increases eastward across the Intermountain West, reaching a maximum at SLC. This section uses a composite of the 25 strongest events at SLC (Table 1) to investigate how the interaction between the evolving large-scale flow, regional topography, and diabatic processes leads to the development of the strongest Intermountain cold fronts. The 25 events occur from March to September and primarily in the late afternoon and early evening, which is similar to the most frequent months and times of all strong Intermountain cold frontal passages (cf. Table 1; Figs. 5, 7).
An important question concerns the representativeness of the composite evolution relative to the individual cases. Surface potential temperature analyses for individual cases at 6 h prior to frontal passage (−6 h) show that 17 of the 25 events closely resemble that of the composite (see Fig. 3.20 of Shafer 2005). The remaining events feature cold fronts that are more meridionally (three events) or zonally oriented (five events), with the latter cases featuring the development of a preexisting surface front over northern Utah. Thus, the composite is representative of the majority of strong cold fronts.
At 24 h prior to frontal passage at SLC (−24 h), a broad low-level baroclinic zone extends across the northwestern United States with the leading edge of baroclinity at the surface and 700 hPa bisecting Nevada from southwest to northeast (Figs. 11a,b). A broad area of low pressure encompasses much of the interior western United States, with weakly confluent southwesterly surface flow denoting the location of incipient frontogenesis. At 700 hPa, the winds are mainly parallel to the isotherms and are therefore minimally frontogenetic (Fig. 11b). In the upper troposphere (250 hPa), a mobile upper-level trough is approaching the Pacific coast with stronger winds on the western side of the trough axis (Fig. 11d), a favorable pattern for trough digging (Bluestein 1993, p. 71). At 500 hPa, a broad area of weak ascent is present ahead of the trough axis and within the western or cold side of the baroclinic zone, a pattern favorable for the formation of clouds behind the incipient cold front (Fig. 11c).
Over the next 12 h, the low-level baroclinic zone and pressure trough intensify over central Nevada as confluence and inferred kinematic frontogenesis strengthen at the surface and 700 hPa (Figs. 12a,b). Aloft, the 250-hPa trough digs southeastward (Fig. 12d) and ascent intensifies downstream of the 500-hPa trough axis (Fig. 12c). Comparison of the analyses at −24 h and −12 h clearly illustrates that the upper-level trough, 500-hPa ascent, and the region of 700-hPa cold advection are advancing toward the quasi-stationary low-level confluence zone and incipient cold front (cf. Figs. 11, 12).
Confluent flow over Nevada establishes a boundary between warm Intermountain air and colder Pacific air. By −9 h, the confluent flow is convergent (Fig. 13) and, as will be shown later, strongly frontogenetic. Orographic flow deflection around the Sierra Nevada appears to enhance the confluence and is possibly responsible for its quasi-stationary development as the upper-level trough approaches, although the extent to which the Sierra Nevada contribute to the location and intensity of the frontal development remains for future study.
By −6 h the intensifying composite surface cold front extends from south-central Nevada northeastward into southeastern Idaho (Fig. 14a, where a solid line denotes the cold front). The frontal intensification occurs in concert with strengthening horizontal convergence as cold advection advances toward the quasi-stationary front (Fig. 14b). Furthermore, from −12 h to −6 h, composite surface potential temperatures ahead of the front increase from 32° to 36°C, suggesting that daytime prefrontal heating also acts to intensify the front (Fig. 14a). In the NARR composite, temperatures also warm well behind the front, although observational evidence suggests that most cases feature weaker postfrontal warming and hence stronger differential heating and diabatic frontogenesis than implied by the NARR. At upper levels, the 250- and 500-hPa trough axes are moving over the Sierra Nevada crest, and the 500-hPa upward vertical motion maximum is now approaching and phasing with the developing cold front (Figs. 14c,d).
The composite prefrontal environment also features strong southerly 700-hPa flow that exceeds 15 m s−1 (Fig. 15a) and resembles a low-level jet (LLJ; e.g., Browning and Pardoe 1973); inspection of individual events shows that nearly all events feature a pronounced prefrontal 700-hPa wind maximum. Consistent with other studies (e.g., Sortais et al. 1993; Wakimoto and Cai 2002), this LLJ has a strong ageostrophic component that is directed toward the front (Fig. 15b) and, as will be shown, this ageostrophic convergence contributes to a large part of the kinematic frontal intensification. In contrast to many moisture-laden prefrontal low-level jets (Neiman et al. 2002, 2004; Ralph et al. 2003), the Intermountain prefrontal flow is dry. In fact, just prior to frontal passage (−3 h), average surface dewpoint depressions are 23°C (Fig. 16a), resulting in low relative humidities (20%–30%) and an average prefrontal lifting condensation level of ∼3 km AGL. Consequently, none of the 25 composite members features measurable prefrontal precipitation (Fig. 16b).
After a period of quasi-stationary frontal development, the surface cold front becomes mobile with the approach of the upper-level trough shortly after −6 h and passes SLC at 0 h (Fig. 17a). At this time, the surface cold front extends from northwestern Wyoming, across northern Utah and southern Nevada, to the Sierra Nevada. A pronounced wind shift and strong postfrontal cold advection are also found at 700 hPa (Fig. 17b). Consistent with the aforementioned approach of the upper-level trough and its interaction with the surface-based front, the strongest composite upper-level upward vertical motion (exceeding 12 cm s−1) is observed at this time near and along the cold front (Fig. 17c), and the 250-hPa trough is becoming increasingly confluent (i.e., with a stronger jet on its downstream side; Fig. 17d).
Throughout the frontal intensification, the NARR-analyzed boundary layer on both sides of the front deepens with daytime heating, eventually exceeding 3 km in depth near the composite surface cold front (Fig. 18). Composite prefrontal boundary layer depths are somewhat higher than postfrontal. Previous studies suggest that frontal evolution is highly sensitive to the distribution of sensible heating and boundary layer depth across a front. Homogeneous surface heating, such as that which occurs with Australian subtropical cold fronts, is frontolytic (Smith et al. 1995; Reeder and Tory 2005). In contrast, when surface sensible heating and boundary layer growth are maximized ahead of a front, frontogenesis is enhanced (e.g., Reeder 1986; Physick 1988; Hakim 1992; Koch et al. 1995; Bosart et al. 2007). The contributions of surface sensible heating and boundary layer growth on the development of strong Intermountain cold fronts, including some important weaknesses in the NARR analyses, are discussed in more detail in the next section.
The development of strong Intermountain cold fronts appears, in many cases, to be related to high-based convection and associated low-level diabatic cooling from evaporation and sublimation (e.g., Schultz and Trapp 2003). Examination of composite convective available potential energy (CAPE) at SLC (−3 h), derived using the most unstable parcel in the lowest 180 hPa, is only around 200 J kg−1 (Fig. 19). Furthermore, inspection of 14 observed prefrontal 0000 UTC soundings from the 25 composite members2 reveals a somewhat bimodal CAPE distribution, with half of the events having less than 100 J kg−1 and 4 exceeding 500 J kg−1 (not shown). Thus, in the majority of events, a deep, dry prefrontal mixed layer with limited CAPE is found. Although CAPE values are low, downdraft convective available potential energy (DCAPE; Gilmore and Wicker 1998) is very high, reaching a NARR composite value of 800 J kg−1 at −3 h (Fig. 19). This creates high potential for low-level diabatic cooling, which has been shown by Schultz and Trapp (2003) to contribute to Intermountain cold front intensification.
As the upper-level jet approaches the incipient cold front, a broad area of ascent is located on the cold side of the developing front (Fig. 20a), a configuration that favors postfrontal clouds and precipitation. The vertical motion intensifies and becomes strongest over the surface cold front as it is overtaken by the upper-level jet at −6 h (Fig. 20b). At this time, prefrontal ageostrophic flow approaches the front at low levels and feeds a zone of sloping rearward ascent beneath the upper-level jet. This configuration of upper-level and low-level jets resembles that of a vertically uncoupled jet front system with the upper jet axis parallel to and on the cold side of the developing cold front (Shapiro 1983, his Fig. 3.1.22). With such an orientation, the upward motion in the left-exit region overlies the postfrontal air mass, a pattern favorable for tilting frontogenesis and postfrontal clouds and precipitation. It will be shown, however, that the composite vertical motion pattern cannot be fully explained by classical 2D ageostrophic jet-streak circulations and that curvature effects must also be considered. By 0 h, the upper-level jet axis and low-level vertical motion maximum are located immediately over the surface-based cold front (Fig. 20c).
To better understand the composite ageostrophic circulation, we now examine the jet (250 hPa) and low-level (700 hPa, near crest level) ageostrophic flow characteristics at −6 h. The ageostrophic wind may be expressed as
the subscript p denoting differentiation along a pressure surface, V the total wind, Vag the ageostrophic wind, ϕ the geopotential, ω the vertical velocity, and f the Coriolis parameter. The first term on the right-hand side of Eq. (1) represents the isallohypsic component. The three terms within parentheses represent the local ageostrophic wind tendency, inertia–advective component, and convective component, respectively. Further, we diagnose a residual term from the difference between the NARR-analyzed ageostrophic wind and that calculated using Eq. (1). This residual arises from subgrid-scale processes, such as parameterized boundary layer turbulence, as well as from numerical errors. One-sided differencing with a 3-h time step is used to calculate the local ageostrophic wind tendency component.
At 250 hPa, the cyclonically curved geostrophic jet reaches its greatest magnitude near the trough axis (Fig. 21a). In contrast, the observed wind at this time is weakest in the trough axis with a jet streak found downstream (Fig. 14d). The ageostrophic flow is primarily along the isohypse, directed against the geostrophic flow (Fig. 21b), and dominated by the inertia–advective component (Fig. 21c; other components are small in magnitude and not presented). These results are consistent with gradient wind balance and illustrate that, similar to the sharply curved upper-level trough examined by Shapiro and Kennedy (1981), curvature effects modify if not overwhelm the ageostrophic transverse circulation expected if the jet streak were linear. Therefore, caution should be used in applying the four-cell jet-streak conceptual model to Intermountain front cases that also feature strong flow curvature (Moore and Vanknowe 1992).
At 700 hPa, a broad region of strong prefrontal geostrophic flow extends across Utah and the adjoining region (Fig. 22a). The ageostrophic flow is strongest in the prefrontal environment, where it is directed toward the front; postfrontal ageostrophic winds are weak (Fig. 15b). The weak postfrontal ageostrophic flow reflects a balance between the isallohypsic and ageostrophic tendency terms (Figs. 22b,c). In the prefrontal environment, however, these terms are weak and the ageostrophic flow is dominated by the residual term (Fig. 22d). Further, the strongest prefrontal ageostrophic winds are generally collocated with areas of large surface NARR turbulent kinetic energy (TKE; shading in Fig. 22d), suggesting that boundary layer drag is responsible for the prefrontal cross-isobar flow at this level. In contrast, the residual term is weaker in the postfrontal environment where the boundary layer is shallower and the TKE weaker. Inspection of the ageostrophic wind components at lower levels (e.g., 800 mb) reveals strong frictional drag effects on both sides of the front (not shown). Differential friction effects are most pronounced at 700 mb because of the contrast in boundary layer depth and TKE across the front.
Hourly surface observations within 9 h of the 25 composite events provide a clearer depiction of the sensible weather accompanying frontal passages (Fig. 16). Measurable precipitation is almost entirely postfrontal, with a peak frequency 3–6 h after frontal passage (Fig. 16b). In addition, the mean fractional cloud cover is also lower in the prefrontal environment and increases after frontal passage (Fig. 16b). Mean prefrontal temperatures increase ∼5°C during the 9 h prior to frontal passage and dewpoint depressions exceed 20°C (Fig. 16a). Cloud ceilings are also much higher in the prefrontal environment, averaging around 4000 m AGL prefrontal and around 1200 m AGL in the postfrontal environment (not shown). These cloud and precipitation contrasts between the pre- and postfrontal environment strongly suggest that differential daytime diabatic heating is important for the frontal intensification.
The composite evolution described above provides a qualitative look at the development of strong cold fronts over the Intermountain West. The composite cold front forms on the leading edge of preexisting baroclinicity as confluent southwesterly low-level flow develops over central Nevada. The location and strength of the confluent deformation appear to be influenced by flow deflection around the Sierra Nevada. Quasi-stationary frontal development occurs as a mobile upper-level trough approaches from the west and confluent deformation and low-level convergence intensify. Concurrently, a prefrontal low-level jet develops. This jet transports relatively dry boundary layer air poleward and features a strong ageostrophic component that is directed toward the front and appears to result from boundary layer drag. As the upper-level trough overtakes the surface-based front, it moves rapidly downstream and is accompanied by strong temperature falls and pressure rises. The front becomes mobile as it is overtaken by cold advection and ascent associated with the upper-level trough.
d. Frontogenesis diagnostics
To further diagnose the processes contributing to the development of strong Intermountain cold fronts, we calculate frontogenesis (Petterssen 1936), defined as the Lagrangian rate of change of the magnitude of the horizontal potential temperature (θ) gradient:
and the subscript p denotes differentiation along a constant pressure surface. For surface frontogenesis calculations, differentiation is done along the NARR surface layer.
Here, FW is the frontogenesis due to horizontal confluence and convergence (kinematic frontogenesis), FT is the tilting frontogenesis, and FD is the frontogenesis due to horizontal gradients in diabatic heating and cooling (diabatic frontogenesis). Since the boundary layer is deep and well mixed, ∂θ/∂p is near zero during the period of rapid frontogenesis; hence, FT at the surface and 700 hPa are negligible and are not presented. To calculate FD, dθ/dp was estimated as a residual of the thermodynamic energy equation:
where Q is the diabatic heating, T is the temperature, and Cp is the specific heat at constant pressure. The local potential temperature change was evaluated with centered time differencing using values at t ± Δt, where Δt is the 3-h NARR analysis time step. The temperature change due to horizontal and vertical advection was computed using the average advection from t − Δt to t + Δt. Tests using one-sided differencing and instantaneous horizontal or vertical advection at time t showed limited sensitivity to the approach used for these estimates.
The most significant source of error in the calculation of diabatic frontogenesis is the poor representation of low-level temperatures in the NARR, which does not assimilate surface (2 m) temperature observations (Mesinger et al. 2006). For example, comparison of the observed and NARR-analyzed 0000–1200 UTC temperature difference at Intermountain observing sites in June reveals that the NARR underestimates the diurnal temperature range by roughly a factor of 2 (not shown). Therefore, diabatic heating rates are likely underestimated at low levels. Another concern is an unrealistic distribution of diabatic heating and cooling across the front, as discussed later. Finally, the NARR composite surface temperature fall during the 6 h following frontal passage is only 5.5°C (not shown), compared with the observed average value of 15.5°C (Fig. 16a).
Frontogenesis calculations are also influenced by the use of composite fields for calculations. Although the large-scale temperature pattern is similar among most composite members, averaging tends to weaken the intensity of the temperature gradient and sharpness of the wind shift and associated confluence and convergence. Consequently, the magnitude of composite frontogenesis is likely smaller than observed in the individual composite members.
We begin at −12 h, which represents a composite mean time of 0900 mountain standard time (MST), when the most rapid frontal development commences. At this time, the incipient cold front is located northwest of SLC (Fig. 23a). Total surface frontogenesis (negative values not shown for clarity) exceeds 2°C (100 km)−1 (6 h)−1 in two bands near the developing front (Fig. 23a). The first extends along or just ahead of the developing front over Nevada and central Utah. The second is located along the southern border of Idaho. Consistent with the inferred deformation and convergent frontogenesis in the prior section, kinematic frontogenesis is strongest along the front (Fig. 23b) as well as the southern Idaho border.
In contrast, diabatic frontogenesis (FD) is strongest ahead of the front and has a greater magnitude than Fw (Fig. 23c). Further, diabatic frontolysis is occurring along the incipient cold front (not shown for clarity). This FD distribution explains why the total frontogenesis maximum is shifted into the prefrontal environment and occurs because the NARR-analyzed diabatic heating minimum is located along or just ahead of the developing front (Fig. 23d). As discussed below, this represents an unrealistic diabatic heating distribution.
At −6 h, which represents a composite mean time of 1500 MST, total frontogenesis exceeds 2°C (100 km)−1 (6 h)−1 just ahead of the surface cold front (Fig. 24a). The kinematic frontogenesis maxima persist along the front and along the southern Idaho border (Fig. 24b). The NARR-analyzed diabatic heating minimum remains along or just ahead of the surface front, resulting in the unrealistic shift of FD into the prefrontal environment (Figs. 24c,d). This shift occurs, in part, because the NARR-analyzed precipitation extends well into the prefrontal environment over Utah, producing a surface sensible heat flux minimum well ahead of the surface cold front (Figs. 25a,b). In actuality, measurable precipitation is observed behind the front (Fig. 16b). Further, the mean NARR diurnal temperature range is about half of that observed in late spring (not shown). For these reasons, the diabatic heating gradient and frontogenesis are likely underestimated and more in phase with the front than suggested by the NARR composite.
Comparison of 700-hPa geostrophic and ageostrophic frontogenesis at −6 h (i.e., FW calculated using geostrophic and ageostrophic winds, respectively) shows that only a small portion of the FW can be explained by the geostrophic frontogenesis (cf. Figs. 26a,b). Instead, ageostrophic FW dominates because of ageostrophic convergence at the cold front. Inspection of other times and cross sections (not shown) indicates that the low-level ageostrophic flow is responsible for much of the kinematic surface and low-level frontal development.
The strongest composite frontogenesis occurs as the cold front passes SLC at 0 h, which represents a mean time of 2100 MST (Fig. 27a). A band of positive FW extends along the front from southern Nevada northeastward into northern Utah where two banded maxima are observed (Fig. 27b). A broad area of positive FD that exceeds 2°C (100 km)−1 (6 h)−1 exists ahead of the southern portion of the cold front, but over northern Utah positive FD is nearly coincident with the cold front (Fig. 27c). Inspection of individual cases shows that FD has become more important, increasing in magnitude from −6 h to 0 h in 14 of the 25 events, with FD exceeding Fw in 11 of 25 events (not shown). Unlike earlier, the FD maximum over northern Utah is closer to the cold front and is primarily the result of differential diabatic cooling immediately behind the front (Fig. 27d). This cooling may be attributed to evaporation and sublimation, as supported by the frequency and magnitude of observed precipitation (Figs. 16 and 25c); in addition, NARR composite precipitation shows a maximum that is roughly oriented along the cold front (not shown).
4. Discussion and conclusions
This climatology and composite study of strong cold frontal passages establishes that the Intermountain West is a breeding ground for strong cold fronts, particularly during the late spring. The frequency of strong cold frontal passages increases eastward (climatologically downstream) from the Sierra Nevada and Cascade ranges, reaching a local maximum at Salt Lake City. There is a strong diurnal cycle to strong cold front passages across the entire western United States, with most occurring during the late afternoon and early evening.
The composite analysis along with the summary in Fig. 28 shows that most strong cold fronts at SLC develop as follows: an amplifying upper-level trough approaches a preexisting southwest–northeast-oriented baroclinic zone (Fig. 28a), a pattern that is frequently associated with large-scale confluent frontogenesis (e.g., Petterssen 1956, his Fig. 11.9.3); this pattern also resembles that of a developing midlatitude frontal cyclone along a stationary front (Wallace and Hobbs 1977, p. 262). The incipient cold front develops in confluent southwesterly flow with flow deflection around the Sierra Nevada appearing to enhance the downstream confluence (Fig. 28b). Strong frontogenesis occurs as cold advection approaches the quasi-stationary front from the west while broad ageostrophic flow impinges from the southeast (Fig. 28b). The front becomes mobile as it is overtaken by cold advection and ascent associated with the upper-level trough (Fig. 28c).
Differential diabatic processes contribute to the frontal intensification in many events, although this hypothesis could not be shown conclusively with the NARR composite. Cloud and precipitation observations, however, suggest that postfrontal cloud shading, evaporation, and sublimation, combined with strong prefrontal sensible heating, produce daytime frontal intensification. For example, cross-front sensible heating contrasts that arise from the differential cloud cover and precipitation are highly frontogenetic during the day since the postfrontal cloud cover reduces incoming solar radiation and surface sensible heating, whereas the prefrontal environment remains strongly heated. Frontal sharpening may be further enhanced by a nonlinear interaction between the large-scale frontal circulation and the thermally driven circulation produced by the differential surface heating (Koch et al. 1995, 1997; Gallus and Segal 1999; Segal et al. 2004). The importance of differential sensible heating on cold frontal intensification is also supported by Bosart et al. (1972), Physick (1988), and Garratt (1988), who show that coastal fronts reach peak intensity with daytime heating. In these cases, however, the differential sensible heat flux arises mainly from contrasts in the latent heat of evaporation over land and sea.
A secondary contributor to diabatic frontogenesis is the cooling of postfrontal air by evaporation and sublimation. Although this effect could not be fully assessed, frequent postfrontal precipitation within a previously deep, dry convective boundary layer suggests that moist downdrafts produced by evaporation and sublimation are important to intensifying Intermountain cold fronts; this effect was noted by Schultz and Trapp (2003), who show that for an Intermountain cold front the postfrontal air mass was largely a result of downdraft air. Other studies also suggest that postfrontal evaporative cooling results in a strengthening and acceleration of cold fronts (Oliver and Holzworth 1953; Bannon and Mak 1984; Huang and Emanuel 1991; Lagouvardos et al. 1993; Katzfey and Ryan 1997). A mechanism as to how evaporation may create or modify cold fronts is offered by Seitter and Muench (1985), who suggest that evaporation beneath a wide frontal rainband evaporates and produces a low-level cold pool that spreads as a gravity current. This mechanism is similar to that in mesoscale convective systems where the mesoscale high pressure can be hydrostatically linked to the magnitude of subcloud evaporative cooling (Fujita 1959). Nonetheless, it remains for future study to fully investigate the role of evaporation and sublimation on the mesoscale development of Intermountain cold fronts, but its magnitude may be greater than that produced by differential diabatic heating resulting from cloud shading.
Subtropical Australian cold fronts also develop in an arid region that experiences strong surface sensible heating. Paradoxically, observational studies of subtropical Australian cold fronts have shown that they intensify during the night and weaken during the day (Smith et al. 1995). This is supported by a numerical study of homogeneous heating within a convective boundary layer by Reeder and Tory (2005), who find that homogeneous daytime heating is slightly frontolytic whereas nighttime cooling is frontogenetic. The discrepancy in the diurnal strength of Intermountain and continental Australian cold fronts may be explained by the high frequency of postfrontal clouds and precipitation with Intermountain cold fronts, which results in inhomogeneous surface heating. In contrast, Australian cold fronts feature homogenous surface heating.
Future work should attempt to quantify and better understand the influence of topographic, diabatic, and boundary layer processes on Intermountain frontogenesis. Of particular concern is determining the role of the Sierra Nevada in driving the quasi-stationary confluent frontogenesis, investigating the boundary layer processes that lead to the development of the prefrontal ageostrophic flow, and quantifying the complex interactions between boundary layer and diabatic processes in the rapid frontal intensification and acceleration.
This research was supported by National Science Foundation Grant ATM-0333525 while the first author was an NSF GK-12 fellow (ATM-0338340) at the University of Utah. Special thanks to Elfred Astling, Lance Bosart, Larry Dunn, Jan Paegle, and David Schultz for their constructive comments, contributions, and scientific support.
Corresponding author address: Dr. Jason C. Shafer, Department of Meteorology, Lyndon State College, 1001 College Rd., Lyndonville, VT 05851. Email: email@example.com
The most reliable precipitation data were available every 6 h at 0600, 1200, 1800, and 0000 UTC, so five periods were chosen for a total of 30 h.
Only 14 soundings were used since these cases featured soundings in the prefrontal environment (i.e., within 3 h of frontal passage), and two cases had no data.