A Diagnostic Examination of Consecutive Extreme Cool-Season Precipitation Events at St. John’s, Newfoundland, in December 2008

a. Data

This study utilizes 6-hourly precipitation data for CYYT, obtained from the Environment Canada 6-h corrected precipitation database. The corrected precipitation data are based on work done by Mekis and Hogg (1999), whereby the data were adjusted to accurately reflect precipitation gauge changes, wind conditions, and changes in station location. All precipitation data in this study were observed in liquid equivalent form (i.e., no data were acquired using ruler methods), limiting errors associated with frozen precipitation.

The National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) analysis, with a horizontal resolution of ½°, is used as the dataset for most synoptic analyses, which differs from the choice of analyses in Milrad et al. (2009, 2010), where the NCEP–NCAR Global Reanalysis with a horizontal resolution of 2.5° (Kalnay et al. 1996), and NCEP North American Regional Reanalysis (NARR; Mesinger et al. 2006), with a horizontal resolution of 32 km, were used. The authors believe that while the dataset does not extend far enough back historically to be used for the composites in Milrad et al. (2009, 2010), the GFS ½° analysis is a superior tool for MAG 1 and 2, primarily in terms of resolution compared to the NCEP–NCAR Global Reanalysis, and accuracy compared to the NARR; the latter of these concerns revolves around the precipitation assimilation scheme over Canada in the NARR (M. Carrera 2008, personal communication), which will be further detailed in the future. However, the authors note that the NCEP–NCAR Global Reanalysis is used here for diagnostics of Q-vector divergence and for gridpoint diagnostics of the QG-forcing parameters defined in Milrad et al. (2010). This is done for two reasons: 1) to be consistent with the QG-forcing synoptic-typing methodology of Milrad et al. 2010, which used the NCEP–NCAR Global Reanalysis, and 2) while the authors believe the GFS ½° analysis is generally a superior tool to the NCEP–NCAR Global Reanalysis for dynamic and thermodynamic diagnostics, errors were found in relatively noisy fields (e.g., Q-vector divergence) and point calculations using the higher-resolution analysis. For the sake of completeness, it has been found that the large-scale dynamics and thermodynamics shown in this paper display qualitatively similar results in both analysis tools (not shown). These points are discussed further in sections 3a and 4a.

The backward-trajectory analysis is performed using the National Oceanic and Atmospheric Administration/Air Resources Laboratory’s (NOAA/ARL) Hybrid Single Particle Lagrangian Integrated Trajectory Model (HYSPLIT; information available online at http://www.arl.noaa.gov/HYSPLIT.php). The radar data were provided by Environment Canada’s Atlantic Climate Centre and the satellite imagery was obtained from NOAA’s Comprehensive Large Array Data Stewardship System (CLASS; available online at http://www.class.ncdc.noaa.gov/saa/products/welcome). The vast majority of the calculations and analyses in this study are displayed using the General Meteorological Package (GEMPAK), version 5.11.1 [updated from the original package devised by Koch et al. (1983)], a data manipulation and visualization software package.

b. Methods used

The methods used in this study revolve around two main points:

• The application of both methods of manual synoptic typing outlined in Milrad et al. (2010) to MAG 1 and MAG 2, as well as a validation of the composite synoptic structures observed for the median extreme cases (Milrad et al. 2009) and the various synoptic types (Milrad et al. 2010).

• The dynamic and thermodynamic analyses in sections 3b and 4b for MAG 1 and MAG 2, respectively, strive to illustrate the relative contributions of dynamic forcing and the thermodynamic properties of the air mass involved, to the heavy precipitation recorded in both events. Equation (1) relates the precipitation rate to the integrated ascent (assumed for the purpose of this paper to be quasigeostrophic) and incipient air mass, where precipitation (P) is assumed equal to condensation, g is the acceleration due to gravity, ω is the vertical motion in pressure coordinates, and (dr_s/dp)_ma is the saturation mixing ratio lapse rate along the moist adiabat. This equation was first published by Gyakum (2008), who states that “the transports of tropical air mass characteristics are crucial to the production of heavy precipitation in the extratropical latitudes.” In addition, using the lapse rate along the moist adiabat as a measure of effective static stability, as presented in Eq. (1), suggests that weaker static stability, often associated with moist subtropical air masses, implies higher values of precipitation rate for a given value of ascent (Gyakum 2008):

  

  

  View Expanded

  

The definition of time t = 0 h is consistent with that established in Milrad et al. (2009, 2010) and refers to the onset of the 6-h period of maximum recorded precipitation at CYYT. Comparisons of backward trajectories and synoptic structures to the NCEP–NCAR Global Reanalysis and NARR have been made (not shown) and were confirmed to be similar.

a. Overview and synoptic typing

MAG 1 affected CYYT from 1200 UTC 7 December 2008 until 1200 UTC 9 December 2008, where t = 0 h is 0600 UTC 8 December 2008. Overall, 73.6 mm of precipitation fell at CYYT during MAG 1 (Fig. 2), which ranks as the fifth most extreme cool-season precipitation event since 1979 (Milrad et al. 2009). MAG 1 exceeded the median extreme event defined in Milrad et al. (2009) by approximately 30 mm. As shown in Fig. 2, most of the precipitation (63 mm) fell in a 24-h period from 1200 UTC 7 December to 1200 UTC 8 December. During this time period, the boundary layer at CYYT was almost completely saturated (Figs. 2a and 2b) and the temperature slowly rose from around 0°C at 1200 UTC 7 December to just above 10°C shortly after 0600 UTC 8 December. Subsequently, the surface proceeded to slowly cool, reaching a temperature of around −3°C by 1200 UTC 9 December, at the end of the event. Observed surface winds during MAG 1 shifted from easterly at 1200 UTC 7 December to southerly during the period of heaviest precipitation, from 0000 UTC 8 December to 1400 UTC 8 December and, finally, to westerly from 1400 UTC 8 December until the end of the event (Fig. 2). Light to moderate snow fell for the first few hours of the event (Fig. 2a) but quickly changed to light to moderate rain by 1700 UTC 7 December (Fig. 2a) and tapered to rain showers by 0000 UTC 9 December (Fig. 2c) and, finally, light snow by 1000 UTC 9 December (Fig. 2c). The observations of temperature trends, wind shifts, and precipitation types are consistent with the passage of a typical midlatitude cyclone at CYYT. Moreover, the wind shift to the south just before 0000 UTC 8 December (Fig. 2a) is consistent with a spike in surface temperature associated with a warm front that moved through CYYT in advance of the sea level cyclone. The subsequent wind shift to the west after 1800 UTC 8 December (Fig. 2b) and the accompanying surface temperature drop were associated with the passage of a trailing cold front. The heaviest precipitation at CYYT during MAG 1 occurred when the station was located in the warm sector of the cyclone, while the surface winds were onshore (southerly). Finally, the observation of easterly surface winds during part of MAG 1 is consistent with the extreme composite in Milrad et al. (2009).

The evolution of MAG 1 is displayed in infrared satellite imagery in Fig. 3. A coastal cyclone was not evident until t = −24 h (0600 UTC 7 December), when an area of clouds was present off the coast of Virginia and North Carolina (Fig. 3d). By t = −12 h, the cyclone had acquired the typical comma-cloud structure often seen in intensifying extratropical cyclones and was located just south of the Nova Scotia coast (Fig. 3e). At t = 0 and +6 h, the trailing cold front remained just west of CYYT as the sea level cyclone moved northward into Labrador (Figs. 3f–h). Between t = +6 and +12 h, the trailing cold front (Figs. 3h–i) moved through CYYT, coinciding with the surface wind shift to the west and a surface temperature drop.

In Fig. 4, a time series of radar imagery from the Holyrood, Newfoundland, radar is presented, with images every 6 h from t = −18 to +12 h. Figure 4 shows that light precipitation was present at CYYT at t = −18 h (Fig. 4a) and bands of heavy precipitation were located just west of CYYT at t = −12 h (Fig. 4b) and t = −6 h (Fig. 4c). At t = 0 h (Fig. 4d), heavy precipitation was located directly over the station, and by t = +6 h (Fig. 4e) the precipitation has moved east of the station, following the passage of the trailing cold front.

A 5-day backward-trajectory analysis was performed for MAG 1 at t = 0 h (see Milrad et al. 2010 for methodological details). Trajectories end at approximately 300, 500, and 700 hPa. In Milrad et al. (2010), most of the extreme cases fit into one of three types based on parcel origin: west, southwest, and south. This classification system was based on the majority of trajectories originating from a particular direction (south, southwest, and west). In a few cases in Milrad et al. (2010), the low- and midlevel trajectories originated from the source region in which the event was eventually classified (e.g., southwest), while some upper-level trajectories originated from the west. The latter type of evolution was evident in the case of MAG 1, where the lower-level parcels primarily originated from the Atlantic Ocean (south), the midlevel parcels from the Gulf of Mexico (southwest), and the upper-level parcels from both the west and southwest (Fig. 5). In following Milrad et al. (2010), since the majority of the trajectories originate in the Gulf of Mexico, MAG 1 is classified as a southwest event.

A time series of three dynamically relevant QG ascent-forcing parameters from the NCEP–NCAR Global Reanalysis is displayed in Fig. 6a, based on the partitioning method of Milrad et al. (2010). Both the geostrophic and real-wind 1000–700-hPa frontogenesis results are shown in Fig. 6a following Milrad et al. (2010) to show that the two parameters exhibit the same general time evolution. The evolution of MAG 1 in Fig. 6a was in general agreement with the “cyclone” cases in Milrad et al. (2010), such that the values of the 1000–700-hPa horizontal temperature advection and 1000–700-hPa frontogenesis were relatively high at CYYT at the start of the event but rapidly became less positive around t = 0 h, whereas the midlevel vorticity advection became positive after t = 0 h (Fig. 6). This suggests that a midlatitude cyclone passed just to the west of CYYT, such that a warm front moved through CYYT ahead of the sea level cyclone and was followed 12–36 h later by the passage of a trailing cold front and the approach of the associated upper-level trough. It should be noted that the largest magnitudes of QG forcing are at t = −6 h (Fig. 6a), not t = 0 h; the classification system in Milrad et al. (2010) was primarily designed to describe the large-scale flow pattern for use by an operational forecaster and is not necessarily designed to unequivocally assess the time of heaviest precipitation, although in most cases, the parameters do line up accordingly.

Figures 6b shows a time series of precipitable water over the course of MAG 1. It is evident that remarkably large values of precipitable water were present at CYYT from t = −12 to +6 h. In particular, at t = 0 h, in the middle of December at 47°N, the precipitable water value at CYYT was over 32 mm (Fig. 6b). This value of precipitable water is over 20 mm greater than the climatological (1971–2000) average at CYYT for the month of December (9.47 mm). In addition, the largest precipitable water anomaly observed in the extreme composite of Milrad et al. (2009) was +14 mm, which is still approximately 9 mm smaller than the maximum value observed during MAG 1 (Fig. 6b). However, while it is greater than the average extreme event, the precipitable water value observed at CYYT is not climatologically unheard of. In following with the method of Hart and Grumm (2001), it is found that the precipitable water value observed during MAG 1 is approximately 1.5 standard deviations above the climatological mean.

Figure 6c shows the NCEP–NCAR Global Reanalysis 850–500-hPa layer-averaged vertical motion (ω), while Fig. 7 shows a time series of NCEP–NCAR Global Reanalysis Q_s (Figs. 7a, 7c, and 7e) and Q_n (Figs. 7b, 7d, and 7f) divergence. As previously discussed by Keyser et al. (1988) and Milrad et al. (2010), Q_s and Q_n are the along- and across-isentrope components of the Q vector, are proportional to the corresponding components of frontogenesis, and are defined by Eqs. (5) and (6), respectively, in Milrad et al. (2010). They are related to quasigeostrophic ascent via the Q-vector form of the quasigeostrophic omega equation [Eq. (4) in Milrad et al. (2010)]. Keyser et al. (1988), albeit in a study of idealized atmospheric flows, suggested that Q_s is related to forcing from temperature advection as a result of a synoptic-scale wave, while Q_n relates to mesoscale forcing.

In the case of MAG 1, Figs. 7a, 7c, and 7e show that there was substantial Q_s convergence in association with the approaching sea level cyclone during the period in which 52 mm of the 73.6 mm of precipitation was recorded at CYYT (t = −12 to +6 h). At t = −12 and −6 h, Q_n convergence was also present (Figs. 7b and 7d), albeit smaller in magnitude than the Q_s convergence. This implies that the majority of the forcing for ascent was in conjunction with the synoptic-scale warm-air advection associated with the sea level cyclone discussed in section 3b. However, the fact that Q_n convergence was observed at both t = −12 and −6 h suggests that mesoscale frontogenetical forcing for ascent plays a more substantial role during part of MAG 1 than in the cyclone group composite members of Milrad et al. (2010), although certainly less so than in MAG 2 (section 4a). This is not the case at t = 0 h, where substantial Q_s convergence is observed (Fig. 7e) alongside Q_n divergence.

b. Synoptic–dynamic analysis

It was shown in section 3a that MAG 1 produced 73.6 mm of precipitation. Satellite imagery demonstrated that this extreme event was associated with an intensifying sea level cyclone approaching St. John’s from the south and west. In this section, the dynamics of the cyclone as well as the thermodynamics of the air mass at CYYT are analyzed.

Two days prior to the heaviest precipitation at CYYT (t = −48 h), and in association with an upper-level trough, a weak sea level cyclone is located over the northern Great Lakes (Fig. 8a), which proceeds to propagate eastward and be situated near Lake Ontario at t = −24 h (Fig. 8b). By t = −24 h (Fig. 8b), the trough on the DT, defined here as the 2-PVU surface, where 1 potential vorticity unit (PVU) = 10⁻⁶ m² s⁻¹ K kg⁻¹, approaches the eastern seaboard of the United States. Concurrently, a secondary area of low pressure develops off the North Carolina coast (Fig. 8b). At t = −12 h (Fig. 8c), the initial sea level cyclone over the Great Lakes disappears and is replaced by a 992-hPa sea level cyclone off the eastern seaboard. The trough on the DT is now located southwest of the developing sea level cyclone, as the system establishes a westward tilt with height. From the initial sea level cyclone development at t = −24 h (1008 hPa; Fig. 8b) to its position just west of CYYT at t = 0 h (980 hPa; Fig. 8e), the cyclone intensifies by 28 hPa, surpassing the “bomb” criterion established by Sanders and Gyakum (1980). By t = +12 h, after the trailing cold front has passed through CYYT, the cyclone center is located northeast of Labrador, with a minimum sea level pressure of 968 hPa, completing a 50-hPa intensification in 36 h.

At t = 0 h, as the sea level cyclone passed just west of the station, high values of θ on the DT are located over CYYT (Fig. 8e). In fact, the values of θ near CYYT steadily increase from t = −36 h (not shown) until the passage of the trailing cold front between t = +6 and +12 h (Fig. 8f), and are quantitatively similar to values observed over the southern United States and Gulf of Mexico. The extremely warm DT at CYYT, while the station is in the warm sector of the cyclone, suggests the presence of a subtropical air mass. In addition, it is implied that the downstream ridge on the DT intensifies from t = −24 h (Fig. 8b) to t = +6 h (Fig. 8f) due to (a) strong low-level warm-air advection ahead of the sea level cyclone and (b) latent heat release as a result of the heavy precipitation seen in Figs. 4b–d. This is consistent with the findings of Milrad et al. (2009, 2010) for the extreme and cyclone group composites, respectively.

MAG 1 is a classic case of rapid East Coast cyclogenesis; however, the climatology of Milrad et al. (2009) shows that 70+ mm of precipitation at CYYT during one precipitation event has occurred only 5 times in the past 30 yr. Therefore, it is likely that the extreme precipitation as a result of MAG 1 is due to a potent combination of a myriad of factors. While it is established that large amounts of QG forcing for ascent (low-level warm-air advection and low-level frontogenesis) are present at CYYT from t = −18 to −6 h (Fig. 6a), both moisture availability and the thermodynamic properties of the air mass at CYYT during this time period are also crucial in explaining the amount of precipitation.

Warm moist air at CYYT, on the east side (warm sector) of the rapidly intensifying cyclone, is shown by the presence of large values of equivalent potential temperature (θ_e) at 850 hPa (Fig. 9). These θ_e values are greater than those present in the Gulf of Mexico, as a plume of moist subtropical air extends from near Bermuda northward to Atlantic Canada from t = −12 to +6 h (Figs. 9c–f), before finally moving east of Newfoundland after t = +6 h (Fig. 9f), as the sea level cyclone propagates away.

The coupling index (CI), a measure of the bulk atmospheric stability, was used by Bosart and Lackmann (1995), Roebber and Gyakum (2003), and Galarneau and Bosart (2006) as θ on the DT minus the low-level θ_e. Here, the DT is the 2-PVU surface and the low-level θ_e is taken at 850 hPa (Galarneau and Bosart 2006). Figures 9d–f shows that an area of relatively low CI (<8 K) approaches CYYT from the southwest, concomitant with the highest values of 850-hPa θ_e. The authors believe the gradient in CI evident near and just east of CYYT at t = 0 h (Fig. 9e) supports the assertion that the warm, moist sector (high θ_e) of the cyclone moves east of St. John’s shortly after t = 0 h and before t = +6 h (the period of heaviest precipitation). After t = +6 h, the area of low CI moves well east of CYYT as the trailing cold front ushers in more stable air from the west, concomitant with lesser amounts of precipitation at the end of the event.

Figure 10 displays the 1000–700-hPa moisture transport vectors and associated moisture flux divergence. Even before the development of the sea level cyclone off the North Carolina coast at t = −24 h (Fig. 10b), substantial moisture transport from the Atlantic Ocean and Gulf of Mexico is present on the western side of the downstream anticyclone. As the sea level cyclone develops and rapidly intensifies, the moisture transport on the eastern flank of the cyclone increases, resulting in a broad area of moisture convergence at CYYT from t = −12 to 0 h (Figs. 10c–e).

Figure 11 displays a time series of observed soundings taken every 12 h at CYYT. As the rapidly intensifying cyclone moves toward CYYT and subtropical moisture is advected northward, the atmosphere quickly saturates by t = −18 h (Fig. 11b). From t = −18 h (Fig. 11b) to t = +6 h (Fig. 11d), the tropopause remains extremely elevated (near 200 hPa), and nearly saturated, supporting the earlier observation that a warm moist air mass exists at CYYT ahead of the sea level cyclone. Following the passage of the trailing cold front and the approach of the main upper-level disturbance between t = +6 and +12 h, the thermodynamic structure of the atmosphere shifts quickly to one with a depressed tropopause of around 400 hPa at t = +18 h (Fig. 11e) and 500 hPa at t = +30 h, and much lower temperature and mixing ratio values. This illustrates the stark difference in the air mass on either side of the trailing cold front and suggests that the bulk of the precipitation during MAG 1 fell during a period marked by strong QG forcing for ascent in the presence of a warm moist air mass.

Emanuel (1983) stated that conditional symmetric (slantwise) instability (CSI) can be largely responsible for some mesoscale precipitation bands within larger-scale storms. Schultz and Schumacher (1999) presented an ingredients-based methodology for forecasting slantwise convection, the components of which are moisture, lift, and instability. Since it has already been established in this section that there was substantial moisture and forcing for ascent, a check for both slantwise and gravitational instability is presented is Fig. 12. Figure 12 displays negative (shaded) values of saturated equivalent geostrophic potential vorticity (MPV*_g), defined as the criterion for CSI by Schultz and Schumacher (1999) along with saturated equivalent potential temperature. The cross section is roughly perpendicular to the thermal wind, although Schultz and Schumacher (1999) point out that MPV*_g is much less sensitive to the orientation of the cross sections than momentum (M) surfaces. Finally, regions of vertical convective instability (dθ_e/dz < 0; Bluestein 1992) can often be collocated with CSI (Schultz and Schumacher 1999). However, the convective instability mode may dominate a situation where both types of instability are present (Emanuel 1983).

Figure 12 shows that the criterion for CSI (MPV*_g < 0) in the vicinity of CYYT is satisfied from t = −6 to +12 h. Since the majority of the event precipitation fell before t = +6, the focus is on Figs. 12c–e. In Fig. 12c, an area of weak CSI is observed in the 800–500-hPa layer, in the vicinity of CYYT. This area has expanded eastward and upward from its location at t = −12 h (Fig. 12b). By t = 0 h (Fig. 12d), the region of CSI is located in the lowest 300 hPa of the troposphere and the value of MPV*_g is more negative than at previous times. Additionally, there is a region of convective instability present near 850 hPa (Fig. 12d). This suggests that at t = 0 h, both gravitational and symmetric instabilities are present, supporting the assertions of Schultz and Schumacher (1999) that regions of CSI are often observed prior to the development of convective instability, and that it is not uncommon to observe both instabilities simultaneously. The Holyrood radar imagery (Figs. 4c–e) lends credence to this idea in showing banded precipitation structures over Newfoundland at t = −6 h (Fig. 4c), more cellular structures at t = 0 h (Fig. 4d), and banded precipitation at t = +6 h (Fig. 4e). Accordingly, at t = +6 h, just prior to the passage of the surface cold front, the region of CSI is still present, located in the lowest 150 hPa, while the area of convective instability has dissipated.

a. Overview and synoptic typing

MAG 2 affected CYYT from 0000 UTC 11 December to 0000 UTC 13 December, with t = 0 h at 0600 UTC 11 December. The total amount of precipitation observed at CYYT was 54 mm (Fig. 13), which is approximately 10 mm above the median extreme event defined in Milrad et al. (2009). MAG 2 occurred in three stages: an initial warm stage, a subsequent cold stage following a cold-frontal passage, and a second warm stage, as a quasi-stationary northeast–southwest-oriented baroclinic zone retreated back to the north of CYYT ahead of an approaching cyclone. Over half of the total event precipitation fell during the initial warm stage and subsequent cold frontal passage (Fig. 13), between 0000 and 1200 UTC 11 December. During this time period, unseasonably warm surface temperatures close to 15°C were replaced by a much cooler yet still saturated atmosphere (Fig. 13a). Accompanying the temperature drop between 0600 and 1200 UTC (t = 0 and +6 h) 11 December was a wind shift from 20-kt (10.29 m s⁻¹) westerly surface flow to 10-kt (5.14 m s⁻¹) northerly flow (Fig. 13a). While moderate rain was reported at several time periods during the initial warm period and subsequent frontal passage, the precipitation slowly transitioned to light frozen precipitation by t = +12 h and remained frozen until around t = +27 h when slowly rising surface temperatures caused a changeover to light rain. Toward the end of the event (Fig. 13b), cold boundary layer air remained in place until around 2100 UTC (t = +39 h) 11 December when the aforementioned boundary returned as a northward-moving warm front, initiating the second warm stage, ahead of a developing cyclone off the southeastern U.S. coast.

Figure 14 shows clouds associated with the baroclinic zone, stretching from northeastern Quebec to Louisiana at t = −36 h (Fig. 14c). The baroclinic zone slowly propagated eastward toward CYYT and, by t = −6 h (Fig. 14f), stretched from an area northeast of Newfoundland southwestward into the Gulf of Mexico. Slightly upstream of the frontal zone was a closed upper-level low over the western Gulf of Mexico (Figs. 14f–h and 18c–f) that impacted the later stages of MAG 2, as well as the subsequent few days along the Atlantic seaboard. Unfortunately, the Holyrood radar was not operating from t = −10 to +12 h, the time in which most of the precipitation fell during MAG 2.

As in MAG 1 (section 3a), the evolution of the backward trajectories was more complicated than most of the cases in Milrad et al. (2010). Figure 15 shows that some of the upper-level trajectories (300 hPa) originated southwest of CYYT to the Gulf of Mexico, while others traced backward initially to the Gulf of Mexico and then into the eastern Pacific. However, while some of the midlevel (500 hPa) trajectories also trace backward to the Gulf of Mexico, most of the mid- and low-level parcels spent a large amount of time over the Gulf Stream. In sum, MAG 2 is classified as a southwest event, with the caveat that the low-level trajectories originated in the upper troposphere early in the time evolution, over the American and Canadian prairies.

Figure 16a displays time series of QG ascent-forcing parameters. As in the 11 frontal cases in Milrad et al. (2010), the magnitude of low-level frontogenesis is large, while the magnitudes of the other two QG variables in the time series (Fig. 16a) are small, especially compared to cyclone events. In addition, the frontogenesis values observed for MAG 2 exceed those observed for MAG 1, which supports the findings of Milrad et al. (2010) in differentiating cyclone and frontal events. Finally, midlevel vorticity advection values near zero for the entire period of precipitation suggest that a strong upper-level trough did not come into play during the evolution of MAG 2; MAG 2 is categorized as a frontal case.

Figure 16b shows that during the entire period of precipitation (42 h) at CYYT, the precipitable water remains at 20 mm or above. The maximum precipitable water value (∼28 mm) is at t = 0 h, coincident with the heaviest precipitation (21 mm). In following with Hart and Grumm (2001), values of precipitable water are approximately one standard deviation above the climatological mean, and slightly above the composite extreme event in Milrad et al. (2009). The observed long duration of high precipitable water (PW) values likely plays a role in the extreme amount of precipitation observed during MAG 2. Finally, Fig. 16c shows that at t = 0 h, there was ascent at St. John’s (and along the length of the baroclinic zone), but it was relatively weak. This finding is consistent throughout MAG 2 (not shown), which is marked by relatively weak ascent in a favorable environment for heavy precipitation. This is discussed further in section 4b.

Figure 17 suggests that during the period of heaviest precipitation at CYYT (t = −6 to +12 h), when 35 mm of the total event’s 54 mm was recorded, Q_n is substantially more convergent than Q_s, as Q_s divergence is observed at both t = 0 and +6 h (Figs. 7c,e). This is consistent with the results for the frontal composite in Milrad et al. (2010) and suggests that mesoscale frontogenetical forcing is a much bigger player (relative to synoptic-scale frontogenesis) in MAG 2 than in MAG 1.

b. Synoptic–dynamic analysis

Figure 18 displays a downstream ridge on the DT, similar to the frontal composite of Milrad et al. (2010). Prior to MAG 2, as the intense sea level cyclone responsible for MAG 1 moved away from Newfoundland, the associated upper-level trough moves over Newfoundland at t = −48 h (Fig. 18a). By t = −24 h, the sea level anticyclone is established downstream of CYYT and potentially cold air aloft is replaced by high θ on the DT, advected by strong southerly flow in association with the downstream anticyclone and a closed upper-level low in the western Gulf of Mexico (Fig. 18b).

The assertion that MAG 2 was a three-stage event is supported by Fig. 18, which shows warm θ on the DT prior to t = +6 h (Figs. 18a–e), replaced by cooler θ at t = +6 h (Fig. 18f). By t = +12 h, however, the high-θ air returns as the surface baroclinic zone retreats northward in association with the developing cyclone in the Gulf of Mexico (not shown). By t = +24 h (not shown), the baroclinic zone is fully to the north of CYYT and the station remains in the subtropical air mass until the passage of the Gulf of Mexico cyclone on 13 December.

Figure 19 highlights a plume of warm low-level θ_e extending parallel to the baroclinic zone from the Gulf of Mexico to northeast of Newfoundland. Figure 19 also confirms that the bulk atmospheric stability in MAG 2 is substantially higher than in MAG 1. While there are low (<8 K) values of the CI observed off the U.S. coast on the southern edge of the baroclinic zone, the lowest values do not make it into the CYYT area, even at t = 0 h (Fig. 19e).

Figure 20b highlights the fact that strong moisture transport on the western flank of the downstream high is already established by t = −24 h. This northward moisture transport continues unabated for more than 48 h, with the largest vectors and a persistent area of moisture convergence located over CYYT (Fig. 20). The relatively long duration of the moisture transport convergence and the high precipitable water values are likely important factors in the extreme amount of precipitation during MAG 2.

Figure 21 shows that the tropopause remains very high, hovering between 150 and 200 hPa during the t = −6 to +42-h time period. Unlike in MAG 1, the tropopause remains high for the duration of the precipitation period during MAG 2, suggesting that there was no drastic shift in air mass during the event. However, Figs. 21c–e show a shallow layer of cold air in the lower tropopause following the initial cold frontal passage; this cold air remains only in the boundary layer with a 200–300-hPa inversion layer situated above it. This supports the previously discussed cool surface temperatures following the initial cold front passage (Fig. 13).

Figure 22 displays cross-section plots oriented northwest–southeast, roughly perpendicular to the quasi-stationary baroclinic zone (Fig. 14). Over 60% of the precipitation during MAG 2 fell between t = −6 and +12 h. However, there was a region of CSI near CYYT, in the near-surface layer, prior to precipitation falling, at t = −24 h (Fig. 22a) and t = −12 h (Fig. 22b). This area is persistent through t = −6 h (Fig. 22c) and t = 0 h (Fig. 22d), with stronger values located to the northwest of CYYT. At t = 0 h (Fig. 22d), there is also a small region of convective instability around 850 hPa, in the vicinity of CYYT, which translates eastward by t = +6 h (Fig. 22e) and t = +12 h (Fig. 22f). Consequently, it appears that convective-symmetric instability was present near CYYT at t = 0 h, as was observed during MAG 1, while CSI was present during most of the heaviest precipitation period. The presence of CSI (and short-lived convective instability) helped to allow for the development of heavy precipitation in the presence of frontogenetical forcing for ascent, a process that is fairly common (Schultz and Schumacher 1999).