Abstract

Strong winds equatorward and rearward of a cyclone core have often been associated with two phenomena: the cold conveyor belt (CCB) jet and sting jets. Here, detailed observations of the mesoscale structure in this region of an intense cyclone are analyzed. The in situ and dropsonde observations were obtained during two research flights through the cyclone during the Diabatic Influences on Mesoscale Structures in Extratropical Storms (DIAMET) field campaign. A numerical weather prediction model is used to link the strong wind regions with three types of “airstreams” or coherent ensembles of trajectories: two types are identified with the CCB, hooking around the cyclone center, while the third is identified with a sting jet, descending from the cloud head to the west of the cyclone. Chemical tracer observations show for the first time that the CCB and sting jet airstreams are distinct air masses even when the associated low-level wind maxima are not spatially distinct. In the model, the CCB experiences slow latent heating through weak-resolved ascent and convection, while the sting jet experiences weak cooling associated with microphysics during its subsaturated descent. Diagnosis of mesoscale instabilities in the model shows that the CCB passes through largely stable regions, while the sting jet spends relatively long periods in locations characterized by conditional symmetric instability (CSI). The relation of CSI to the observed mesoscale structure of the bent-back front and its possible role in the cloud banding is discussed.

1. Introduction

The potential to generate strong surface winds and gusts as they pass is one of the most important aspects of extratropical cyclones due to the direct impact on society. The aim of this article is to analyze the three-dimensional structure of the region of strong winds near the center of an intense extratropical cyclone and determine the origin of the airstreams within that region. The study is focused on a cyclone that developed according to the Shapiro–Keyser conceptual model (Shapiro and Keyser 1990). This model is characterized by four stages of development: 1) incipient frontal cyclone, 2) frontal fracture, 3) frontal T bone and bent-back front, and 4) warm-core seclusion. Frontal fracture describes the break of a continuous thermal front as the cyclone intensifies so that the cold front is dislocated eastward from the warm front with a weaker gradient in between. This region is termed the “frontal fracture zone” and is associated with air descending cyclonically from the northwest to the south of the frontal cyclone. The descending air gives rise to a pronounced “dry slot” in satellite imagery. The extensive cloud wrapping around the poleward side of the cyclone core is described as the “cloud head” (Böttger et al. 1975), and its leading extremity is described as the “cloud head tip” (Browning and Roberts 1994). Figure 1 shows a schematic diagram of the structure of a Shapiro–Keyser cyclone during development stage 3.

There are two separate regions usually associated with strong winds in Shapiro–Keyser cyclones. The first region is the low-level jet ahead of the cold front in the warm sector of the cyclone. This low-level jet is part of the broader airstream known as the warm conveyor belt, which transports heat and moisture northward and eastward while ascending from the boundary layer to the upper troposphere (Browning 1971; Harrold 1973). The second region of strong winds develops to the southwest and south of the cyclone center as a bent-back front wraps around the cyclone. The strong winds in this region are the focus of this contribution.

Two different airstreams have been associated with strong winds in this region: the cold conveyor belt (Carlson 1980; Schultz 2001) and sting jets (Browning 2004; Clark et al. 2005). The cold conveyor belt (CCB) is a long-lived synoptic-scale airstream on the poleward (cold) side of the warm front that flows rearward relative to the cyclone motion in the lower troposphere. It extends round the poleward flank of the cyclone and in some mature cyclones it wraps around the west and then equatorward flank where it provides a wind component aligned with the system motion and therefore strong ground-relative winds. A key aspect of the CCB is that the wind maximum is near the top of the boundary layer and slopes radially outwards with height on the cold side of the bent-back front, as would be expected from gradient thermal wind balance.

The term “sting jet” was introduced by Browning (2004) (see also Clark et al. 2005) to describe strong low-level winds in the cold air between the bent-back front and the cold front on the basis of observations of the Great October storm of 1987 from satellite, precipitation radar, and the surface wind network (Browning 2004). The air associated with the sting jet descends from the cloud head tip, moving ahead of it around the cyclone into the dry slot behind the cold front. As the cyclone develops into phase 3, the region of weak gradients between the bent-back front and cold front expands and the sting jet airstream descends into this region. Here, the boundary layer has near-neutral stability or potential instability (Browning 2004; Sinclair et al. 2010); these characteristics have been hypothesized to enhance turbulent mixing of high-momentum air down to the surface.

Clark et al. (2005) analyzed simulations of the same case using the Met Office Unified Model (MetUM) and identified distinct clusters of trajectories calculated using model winds with sting jet airstreams. A key characteristic of sting jet trajectories is that they descend as they accelerate. There are several influences on vertical motion in this sector of a cyclone. On the largest scale, the cyclone forms as part of a baroclinic wave. On isentropic surfaces cutting through a baroclinic wave in the midtroposphere the generic structure of motion gives rise to four air masses: air ascending poleward and splitting into a cyclonic and anticyclonic branch and air descending equatorward and also splitting into a cyclonic and anticyclonic branch (Thorncroft et al. 1993). The two cyclonic branches wrap around the cyclone core. On higher-isentropic surfaces they are described as the cyclonic branch of the warm conveyor belt (ascending) and the dry intrusion (descending). Both the CCB (ascending or horizontal) and sting jet airstreams (descending) also turn cyclonically and are found on lower-isentropic surfaces that can intersect the ground in the warm sector. In addition to the primary circulation of the baroclinic wave, cross-frontal circulations contribute to vertical motion. For example, frontogenesis at the cold front contributes to the ascent of the warm conveyor belt and descent of the dry intrusion behind. Semigeostrophic theory shows that the cross-frontal circulations are necessary to maintain approximate thermal wind balance in a time-dependent flow and therefore depend on its rate of change (Hoskins and Bretherton 1972). Schultz and Sienkiewicz (2013) have used model diagnostics to show that descent can be enhanced in the region beyond the cloud head tip, where the sting jet airstream descends, as a result of frontolysis. The airstream leaves the tight gradient of the bent-back front at the west of the cyclone, and therefore the gradient must decrease with time in a Lagrangian frame. Similarly, ascent is expected in the CCB where the bent-back front strengthens.

Several studies have investigated the mechanisms leading to sting jets. Browning (2004) proposed that the sting jets (local wind maxima) occur beneath the descending branches of slantwise circulations generated by the release of conditional symmetric instability (CSI) in the frontal fracture region between the cloud head tip and the cold front. Numerical simulations represented some form of slantwise motion in that region (Clark et al. 2005). Analysis of model humidity and equivalent potential temperature along trajectories indicated that the airstream originated from a saturated region within the cloud head, but became unsaturated on descent. This would be consistent with the evaporation of cloud and banding in the cloud. A necessary condition for CSI to give rise to slantwise convection is that the air is saturated (at least initially). Further case studies of storms with strong winds in the sting jet region clearly identify regions meeting the CSI criterion that also exhibit banding in the cloud head (Gray et al. 2011). Martínez-Alvarado et al. (2012) used CSI diagnostics to construct a regional sting jet climatology. They found that up to a third of a set of 100 winter North Atlantic cyclones over the past two decades (1989–2009) satisfied conditions for sting jets (Martínez-Alvarado et al. 2012). However, in other studies the importance of CSI is not as clear (Baker et al. 2014; Smart and Browning 2014). In addition, there have not been detailed in situ observations in the appropriate region of Shapiro–Keyser-type cyclones that could have established the existence of slantwise rolls or connection to instability with respect to CSI. Finally, Browning (2004) and Clark et al. (2005) proposed that evaporative cooling may also enhance the descent rate of sting jet airstreams, although Baker et al. (2014) found little impact in an idealized cyclone simulation.

The cyclone analyzed here produced very strong winds over the United Kingdom on 8 December 2011 and was the focus of the intensive observing period 8 (IOP8) during the second field campaign of the Diabatic Influences on Mesoscale Structures in Extratropical Storms (DIAMET) project. The storm has been the subject of extensive investigation involving not only the present article. Baker et al. (2013) described the flights and summarized the severe societal impacts of the storm. Vaughan et al. (2014, manuscript submitted to Bull. Amer. Meteor. Soc.) give more details of the DIAMET experiment and present the results of research on high-resolution ensemble simulations and further in situ aircraft observations, as well as observations from automatic weather stations across the north of the United Kingdom. The cyclone was named Friedhelm by the Free University of Berlin’s adopt-a-vortex scheme (http://www.met.fu-berlin.de/adopt-a-vortex/).

With its aircraft field campaigns, DIAMET joins worldwide efforts to sample weather systems through aircraft observations (e.g., Schäfler et al. 2011; Sapp et al. 2013). To the authors’ knowledge there have only been two previous research flights into an intense cyclone of this type, crossing the strong wind regions near the cyclone center. Shapiro and Keyser (1990) show dropsonde sections across a similar storm observed on 16 March 1987 during the Alaskan Storms Programme. A second cyclone that developed extremely rapidly was observed at three stages in its evolution in IOP4 of the Experiment on Rapidly Intensifying Cyclones over the Atlantic (ERICA) experiment. Neiman et al. (1993) present dropsonde sections through this storm and Wakimoto et al. (1992) present data from the aircraft radar in more detail. Some common aspects of the observed structures will be compared in this paper. Friedhelm also passed over Scotland where there is a high-density automatic weather station network and radar network estimating precipitation rate from reflectivity (discussed by Vaughan et al. 2014, manuscript submitted to Bull. Amer. Meteor. Soc.). Also, numerical models have improved considerably in the last 20 years. Here, a state-of-the-art numerical weather prediction model is evaluated against in situ and dropsonde observations and then used to analyze the history of air masses passing through the regions of strongest low-level winds. The scientific questions addressed are as follows:

  • How are the strong wind regions southwest of the cyclone core related to the characteristic airstreams that have been proposed to exist there (CCB and sting jet)?

  • Where trajectory analysis identifies different airstreams, are they observed to have distinct airmass properties?

  • What dynamical mechanism is responsible for the observed cloud banding in the cloud head and to the south of the cyclone?

Dropsonde and in situ measurements are used to link the observed system to the structure simulated in the MetUM. The model is then used to calculate the airstreams and the evolution of their properties as they move into regions of strongest winds. Throughout the paper, the term “airstream” is identified with a coherent ensemble of trajectories that describes the path of a particular air mass arriving in a region of strong winds. Wind speed is not a Lagrangian tracer and typically regions of strong winds move with the cyclone and change structure as it develops. Therefore, air flows through the strong wind regions (local wind maxima or jets) and each airstream must be identified with the time when it is in the associated strong wind region. Trajectory analysis is combined with potential temperature θ tracers to investigate the processes responsible for the evolution of each identified air mass. Tracer observations from the aircraft are used to investigate whether the airstreams identified are distinct in composition or not.

The article is organized as follows: The aircraft observations, numerical model, and trajectory and tracer tools are described in section 2. A synoptic overview of the case study and a detailed account of the evolution of strong wind regions near the cyclone center are given in section 3. In section 4, the air masses constituting strong wind regions are identified and classified as CCBs or sting jets, according to their evolution and properties. The conditions for mesoscale atmospheric instabilities in the vicinity of the identified airstreams are investigated in section 5. Finally, discussion and conclusions are given in section 6.

2. Methodology

a. Available aircraft observations

Cyclone Friedhelm was observed with the instruments on board the Facility for Airborne Atmospheric Measurements (FAAM) BAe146 research aircraft. The instruments allowed in situ measurements of pressure, wind components, temperature, specific humidity, and total water (all phases) as well as chemical constituents such as carbon monoxide (CO) and ozone. The aircraft was equipped with comprehensive cloud physics instrumentation characterizing liquid droplet and ice particle size and number distributions. A summary of the instruments, their sampling frequency, and uncertainty on output parameters is given in Vaughan et al. (2014, manuscript submitted to Bull. Amer. Meteor. Soc.). The observations are shown here at 1 Hz. In addition 21 dropsondes (Vaisala AVAPS RD94) were launched from approximately 7 km. The dropsondes contributed measurements of temperature, pressure, and specific humidity as a function of latitude, longitude, and time (at a sampling frequency of 2 Hz). Horizontal wind profiles were obtained by GPS tracking of the dropsondes (logged at 4 Hz). Two sondes could be logged on the aircraft at any one time, and the average time for sonde descent was 10 min, limiting the average sonde spacing to 5 min along the flight track or 30 km at the aircraft science speed of 100 m s−1. The vertical resolution is about 10 m. Table 1 lists the sonde release times along the three dropsonde curtains across the cyclone.

b. Numerical model

The case study has been simulated using the MetUM version 7.3. The MetUM is a finite-difference model that solves the nonhydrostatic deep atmosphere dynamical equations with a semi-implicit, semi-Lagrangian integration scheme (Davies et al. 2005). It uses Arakawa C staggering in the horizontal (Arakawa and Lamb 1977) and is terrain following with a hybrid height Charney–Phillips (Charney and Phillips 1953) vertical coordinate. Parameterization of physical processes includes longwave and shortwave radiation (Edwards and Slingo 1996), boundary layer mixing (Lock et al. 2000), cloud microphysics, and large-scale precipitation (Wilson and Ballard 1999) and convection (Gregory and Rowntree 1990).

The simulation has been performed on a limited-area domain corresponding to the Met Office’s recently operational North Atlantic–Europe (NAE) domain with 600 × 300 grid points. The horizontal grid spacing was 0.11° (~12 km) in both longitude and latitude on a rotated grid centered around 52.5°N, 2.5°W. The NAE domain extends approximately from 30° to 70°N in latitude and from 60°W to 40°E in longitude. The vertical coordinate is discretized in 70 vertical levels with lid around 80 km. The initial and lateral boundary conditions were given by the Met Office operational analysis valid at 0000 UTC 8 December 2011 and 3-hourly lateral boundary conditions (LBCs) valid from 2100 UTC 7 December 2011 for 72 h.

Several previous studies have used resolutions of this order to study this type of storm (e.g., Clark et al. 2005; Parton et al. 2009; Martínez-Alvarado et al. 2010), motivated on the basis that the fastest growing mode of slantwise instability should be resolvable at these horizontal and vertical resolutions (Persson and Warner 1993; Clark et al. 2005). Vaughan et al. (2014, manuscript submitted to Bull. Amer. Meteor. Soc.) provide an analysis of an ensemble at 2.2-km grid spacing, including the low-level wind structure, but the domain in that case is restricted to the United Kingdom; the use of the 12-km grid spacing allows the simulation of a larger domain that includes the full cyclone without dominant effects from the LBCs. Moreover, the trajectory analysis (see sections 2c and 2d) requires a large domain to allow long trajectories to be calculated without the majority of them leaving the domain.

c. Trajectory analysis

Two trajectory models are used in the paper. The first model is the Reading Offline Trajectory Model (ROTRAJ) as developed by Methven (1997). Its application to aircraft flights is detailed in Methven et al. (2003). It calculates trajectories using European Centre for Medium-Range Weather Forecasts (ECMWF) analysis data. In this paper, the ECMWF Interim Re-Analysis (ERA-Interim) dataset has been used in its native configuration (T255L60 in hybrid sigma-pressure vertical coordinates every 6 h). A fourth-order Runge–Kutta scheme is used for the trajectory integration (with a time step of 15 min). The boundary condition on vertical velocity is used during the interpolation to ensure that trajectories cannot intercept the ground.

The second model, based on the Lagrangian Analysis Tool (LAGRANTO) model of Wernli and Davies (1997), calculated trajectories using hourly output from the MetUM (in the model’s native vertical coordinate). The time-stepping scheme is also fourth-order Runge–Kutta. Previous comparison has shown the LAGRANTO model and the trajectory model used here perform similarly even though there are differences in interpolation (Martínez-Alvarado et al. 2014). Atmospheric fields, such as θ and specific humidity, were interpolated onto the parcel positions to obtain the evolution of those fields along trajectories. The material rate of change of the fields along trajectories was computed using a centered difference formula along the temporal axis. Thus, rather than being interpreted as instantaneous values, rates of change along trajectories should be interpreted as an estimate of hourly mean values.

d. Potential temperature tracers

The θ tracers used in this work have been previously described elsewhere (Martínez-Alvarado and Plant 2014). They are based on tracer methods developed to study the creation and destruction of potential vorticity (Stoelinga 1996; Gray 2006). Potential temperature is decomposed in a series of tracers so that . Each tracer ΔθP accumulates the changes in θ that can be attributed to the parameterized process P. The parameterized processes considered in this work are (i) surface fluxes and turbulent mixing in the boundary layer, (ii) convection, (iii) radiation, and (iv) large-scale cloud and precipitation. The tracer θ0 matches θ at the initial time. By definition, this tracer is not modified by any parameterization but it is, nevertheless, subject to advection.

The θ tracers and trajectory analysis provide different approximations to the Lagrangian description of the flow field. The θ tracers are computed online whereas trajectories are computed offline from hourly velocity data on the model grid. Tracer θ0 experiences transport only, without diabatic modification. Therefore, in the absence of subgrid mixing or numerical advection errors (in the tracer or trajectory schemes), it is expected that θ0 conserves the same value when sampled along a trajectory. To focus on results where the θ tracers and trajectories are consistent, the criterion

 
formula

is applied where the tolerance on nonconservation is Δθ0 = 3.5 K. Here, xi refers to a point along trajectory i, tarr is the arrival time of the trajectory in the strong wind region (the release time of the back trajectories), and torigin is a common reference time (0100 UTC 8 December 2011) described as the trajectory origin. Approximately 20% of trajectories are rejected by this criterion, although the identification of airstreams is insensitive to this filter.

e. Diagnostics to identify regions of atmospheric instability

Previous studies on sting jets have shown that the necessary conditions for CSI are satisfied in the regions that sting jet airstreams pass through (Gray et al. 2011; Martínez-Alvarado et al. 2013; Baker et al. 2014). Here, we identify regions that satisfy necessary conditions for instability in the analyzed case and their locations relative to the airstreams.

Conditional instability (CI) with respect to upright convection is identified in regions where the moist static stability [, defined as in Durran and Klemp (1982)] is negative. A necessary condition for inertial instability (II) is that the vertical component of absolute vorticity ζz is negative. Inertial instability can be regarded as a special case of (dry) symmetric instability (SI); in the limit that θ surfaces are horizontal, SI reduces to II. A necessary condition for CSI is that the saturation moist potential vorticity (MPV*) is negative (Bennetts and Hoskins 1979). MPV* is given by

 
formula

where ρ is density, ζ is the absolute vorticity, and is the saturated equivalent potential temperature. Note that is a function of temperature and pressure, but not humidity (since saturation is assumed in its definition). Following Schultz and Schumacher (1999), a point is only defined as having CSI if inertial and conditional instabilities are absent. If the necessary conditions for CI or CSI are met then they can only be released if the air is saturated, so we apply an additional criterion on relative humidity with respect to ice: RHice > 90%. As in Baker et al. (2014), we use the full winds rather than geostrophic winds in these CSI and II diagnostics. The diagnostics for the conditions for instability are applied at each grid point; a grid point is labeled as stable (S) if none of the three instabilities are identified.

All these diagnostics indicate necessary, but not sufficient, conditions for instability. The most basic theories for each of these instabilities rely on different assumptions regarding the background state upon which perturbations grow, namely, uniform flow for CI, uniform PV for CSI, and uniform pressure in the horizontal for inertial instability. These conditions are far from being met in an intense cyclone where there are strong pressure gradients, wind shears, and PV gradients. Shear instability is also present on all scales and grows as a result of opposing PV gradients in shear flows.

3. Synoptic overview and identification of regions of strong winds

a. Synoptic overview

On 6 December 2011, extratropical cyclone Friedhelm started developing over Newfoundland (50°N, 56°W). Its development was part of a baroclinic wave, in tandem with another strong cyclone to the west (named Günther) that, as Friedhelm, reached maturity on 8 December 2011, but near Newfoundland. Traveling to the northeast, Friedhelm continued its development according to the Shapiro–Keyser cyclogenesis model (Shapiro and Keyser 1990), as shown in Table 2. The cyclone satisfied the criterion to be classified as an atmospheric “bomb” by consistently deepening by more than 1 Bergeron (Sanders and Gyakum 1980). At 1200 UTC 8 December 2011, the cyclone center was located around 59°N, 7°W, just northwest of Scotland. The FAAM aircraft reached the cyclone center at 1234 UTC when satellite imagery (Fig. 2a) shows a very well-defined cloud head hooking around the cyclone center (early stage 4). This image also shows prominent cloud banding, especially southeast of the cloud head tip, to the southwest and south of the cyclone center.

The frontal system and the intensity of the cyclone are depicted in the Met Office analysis valid at 1200 UTC 8 December 2011 (Fig. 2b). Figure 2c shows the synoptic situation in the 12 h forecast using the MetUM. The similarity with the Met Office analysis chart at this time is remarkably good in terms of the depth of the cyclone (957 hPa in both charts) and the location of the surface fronts. The position error of the low pressure center in the simulation is less than 50 km.

b. Development of regions of strong winds

The structure of regions of strong winds to the south of the cyclone center varied throughout the interval under study. Before 0500 UTC, the only airstream associated with strong winds was the warm conveyor belt ahead of the surface cold front (not shown). Although this region of strong winds continued to exist throughout the interval under study, it was excluded from the airstream analysis to focus on the strong low-level winds behind the surface cold front to the south of the cyclone center. These winds first exceeded 40 m s−1 at 0500 UTC when a distinct jet developed at 600 hPa. By 0600 UTC, the maximum winds (47 m s−1) had descended to 700 hPa. Figures 3a–d show the development of the ground-relative wind field on the 850-hPa isobaric level every 3 h from 0900 to 1800 UTC. At 0900 UTC (Fig. 3a), the region of maximum winds was about 50 km wide with winds up to 49 m s−1 spanning 600–800 hPa. By 1200 UTC (Fig. 3b), the region of maximum winds had moved over Scotland and orographic effects might have influenced its structure. The first dropsonde curtain (D–C) crosses just to the west of the low-level wind maximum.

By the time (1500 UTC) of the in situ aircraft legs west of Scotland (F–G in Fig. 3c), the wind maximum had reached the eastern side of Scotland. However, it was important for the aircraft to remain upstream of the mountains to reduce the orographic influence on the observed winds, cloud, and precipitation. The second flight dropped sondes across the low-level wind maximum at 1800 UTC (J–H in Fig. 3d), and the subsequent in situ legs (continuing until 2000 UTC) were at the longitude of the jet maximum but on its northern flank.

c. Identification of regions of strong winds

The structure of strong wind regions, and associated temperature and humidity fields near the cyclone center, were measured using dropsonde observations for three sections across the storm during the two FAAM research flights (see Table 1). The dropsonde data were relayed to the Global Transmission System (GTS) from the aircraft and was assimilated by the global forecasting centers. All 17 sondes from the first two legs made it into the assimilation window for the 1200 UTC global analysis of both the Met Office and ECMWF and would have influenced subsequent operational forecasts. However, the simulation shown here starts from the global Met Office analysis for 0000 UTC 8 December 2011 and therefore is independent of the dropsonde data.

The first dropsonde leg (1130–1234 UTC) was from south to north toward the low pressure center (D–C in Fig. 3b). During this leg the aircraft flew from just north of the surface cold front, crossing above the cloud bands into the cyclone center. Surface pressure measured by the tenth sonde was 959 hPa, just above the minimum in the analysis at 1200 UTC. Figure 4a shows the structure of wind speed, θe, and RHice obtained from the sondes. The southern arm of the bent-back front was crossed between 57° and 57.3°N and divides two distinct air masses: the cyclone’s warm seclusion to the north and the frontal fracture zone to the south. The strongest winds are confined below 720 hPa near the bent-back front with a maximum at 866 hPa, just above the boundary layer (51 m s−1). At this level, the strong winds extend southward to about 55.5°N into a region of near saturation and moist neutrality (∂θe/∂z ≈ 0). At about 56.5°N, the strong winds extend upward to meet the upper-level jet. Between 800 and 600 hPa, there is subsaturated air on the southern flank of this wind maximum and saturated air to the north of it. In section 4b, it will be shown that this humidity structure indicates an airmass boundary. At 600 hPa, there is a second wind speed maximum to the south (55°N) associated with the dry intrusion descending beneath the poleward flank of the upper-level jet. It is well separated from the lower-level wind maximum discussed above and also the low-level subsaturated air at about 56°N. The average sonde spacing was 30 km, but the low-level cloud and precipitation banding (oriented perpendicular to the section) has a spacing of 25–50 km and is therefore underresolved by the dropsonde data. Therefore, a more finescale structure in humidity cannot be ruled out.

Figure 4b shows an approximately corresponding straight section derived from model output at 1200 UTC. In the model, the bent-back front is displaced southward by approximately 0.2°–0.3° latitude (in terms of both temperature and wind). The strongest low-level winds are also confined to a latitudinal band between 55.5° and 57°N with nearly neutral moist stability. However, in the model the region of strong winds extends upward as an unbroken region between 950 and 600 hPa, and the distinctive low-level maximum adjacent to the bent-back front is missing. Moreover, the dropsonde observations reveal stronger winds near the surface than those produced by the model between 56° and 57°N. The moisture distribution shows the greatest differences between observations and the model simulation. This may be associated with the cloud bands that are too narrow to be resolved in the 12-km grid spacing model. Furthermore, the model has cloud spanning the wind gradient at the bent-back front into the warm seclusion, while the observations show saturation only to the south of the gradient.

The second dropsonde leg (1243–1318 UTC) was in a southwest direction radially away from the cyclone center (C–E in Fig. 3b), across the cloud head tip. Figure 4c shows the structure of wind speed, θe, and RHice obtained from the dropsondes during the second dropsonde leg. The two distinct air masses are again evident, divided by the bent-back front around 8.5°W in this section. Warm seclusion air is located to the northeast, characterized by weak winds (|V| < 20 m s−1) and low-level CI (below 700 hPa). The strong winds are again confined to a band on the thermal gradient and at greater radius, in this section between 8° and 9.5°W, with the maximum at 859 hPa (48 m s−1). Note that θe and wind speed contours are aligned and slope radially outwards with altitude (above 850 hPa). This structure was observed on several sections across the ERICA IOP4 case (Neiman et al. 1993). Thorpe and Clough (1991) pointed out that where the absolute momentum and saturated θe surfaces are almost parallel the MPV* must be near zero, consistent with conditions for CSI.

Figure 4d shows an approximately corresponding straight section derived from model output at 1300 UTC. This model section shows good agreement in terms of wind and thermal structure. However, the agreement is not so good in moisture. The aircraft crossed several cloud bands that were too narrow to be adequately resolved by sondes or the model. For example, the second sonde (7.5°W) fell through much higher humidity than the first and third. It was released approximately when the aircraft crossed the closest cloud band to the cyclone center. However, it must have fallen just outside the cloud, and the 80% RHice contour indicates the higher humidity. The fourth sonde was released into the second cloud band and clearly measured saturation. This band was collocated with the thermal gradient of the front. The sea surface could often be seen from the aircraft (at 400 hPa) when flying between these cloud bands. The wind speed and θe surfaces are almost vertical, so if slantwise convective circulations did emerge as a result of CSI release, the motions would also be nearly vertical along these surfaces; however, CSI release still is a plausible candidate for the origin of the banding. In contrast with the observed banding, the model has saturated air spanning the front, as it did on the first dropsonde curtain. Although the model has some subsaturated air within the warm seclusion (7.5°–8°W), it has too much moisture near the cyclone core. The flight leg returning along this section at 643 hPa (not shown) encountered high relative humidity only within 0.5° of the center with much drier air surrounding. Model humidity on 640 hPa (not shown) indicates subsaturated air within the seclusion, wrapping around the cloudy cyclone core. This feature can be identified in the satellite image (Fig. 1a); however, humidity in the model extends over larger areas.

The third dropsonde leg (1754–1806 UTC) was on the second flight to the east of Scotland when the storm had wrapped up further into the seclusion stage 4. The northward leg crossed the low-level jet spanning only 1° latitude (J–H in Fig. 3d). Strong winds (|V| > 40 m s−1) are located below 700 hPa and span the whole section horizontally, although the maximum (48 m s−1) is located at 816 hPa on the first (southern) sonde profile (Fig. 4e). The wind speed (momentum) surfaces again slope radially outwards from the cyclone with height. The term θe is well-mixed throughout the region of strongest winds, and the gradient aloft is weak. The turbulence was observed to be strong along this section on the later in situ legs. The turbulent kinetic energy, calculated from 32-Hz turbulence probe data on 2-min segments, was 7–10 m2 s−2 at 500 m above the sea. The maximum wind speed observed at this level was 47 m s−1 at the southern end (point J). Observations of turbulence throughout the DIAMET experiment are reported in Cook and Renfrew (2014).

The corresponding model section (Fig. 4f) reproduces the location and strength of low-level winds even at this long lead time (T + 18). However, the θe gradient across the frontal surface appears too strong and wind speed decreases too rapidly in the boundary layer approaching the sea surface. These deficiencies are both consistent with turbulent mixing being too weak in the model.

The few dropsonde sections that have previously been reported through intense extratropical cyclones did not capture the mesoscale detail observed in the DIAMET IOP8 case. Dropsonde sections along a similar radial to curtain 1 were flown through the Alaska storm and ERICA IOP4 case and are presented using manual analysis in Shapiro and Keyser (1990). The Alaska storm section (their Fig. 10.19) is most similar, although the region between the cyclone center and the south of the low-level wind maximum is sampled by only 5 sondes rather than 8. The low-level wind maximum in that case also just exceeds 45 m s−1 and is confined below 750 hPa. The θe surfaces are almost vertical at this location along the bent-back front, while they slope radially outwards with height where the bent-back front was crossed north of the cyclone center. In the ERICA IOP4 case, only two dropsondes were used in this cyclone sector, and therefore the mesoscale wind structure is not well resolved (their Fig. 10.26). However, Neiman et al. (1993) show a cross section similar to curtain 2 in stage 4 (seclusion). They estimated that the radius of maximum wind increased from 75 to 200 km with altitude, and they describe it as an outward sloping bent-back baroclinic ring (Neiman and Shapiro 1993, p. 2162; Neiman et al. 1993, p. 2194). At each radius, the decrease in azimuthal wind with height above the boundary layer is required for thermal wind balance with the temperature gradient across the bent-back front with warm air in the center. The more general form of thermal wind balance arises from a combination of gradient wind balance in the horizontal with hydrostatic balance. Thorpe and Clough (1991) estimated thermal wind imbalance from dropsonde curtains across cold fronts and showed that it could be substantial. Thermal wind imbalance implies transient behavior in the flow, either associated with a cross-frontal circulation or perhaps CSI release.

Although there are systematic model deficiencies identified from the three dropsonde curtains, the wind and potential temperature are in reasonable agreement, both in terms of structure and values either side of the bent-back front. The humidity field is less well represented (which also affects θe). The model is now used to reconstruct the development of regions of strong winds in the immediate vicinity of the cyclone center. The trajectory and tracer analysis depend only upon the wind and potential temperature evolution.

4. Air masses arriving at regions of strong winds

a. Identifying airstreams associated with strong low-level winds in the model simulation

The aim of this section is to relate the mesoscale structure of strong winds in the lower troposphere with airstreams. It is determined whether each strong wind structure is associated with a single coherent airstream, multiple airstreams that are distinct from one another, or a less coherent range of trajectory behaviors. The airstreams are then used to examine the evolution of air coming into strong wind regions, its origins, and diabatic influences on it.

Boxes surrounding regions of strong winds were defined at 0900, 1300, 1600, and 1800 UTC. Back trajectories from these boxes were computed using the winds of the forecast model. A selection criteria based on a wind speed threshold (|V| > 45 m s−1) was applied to retain only those trajectories arriving with strong wind speeds. Note that this threshold is almost as high as the maximum wind speeds observed by the aircraft on its low-level runs just after 1500 and 1900 UTC. However, the aircraft did not sample the associated air masses at their time and location of greatest wind speed; for example, the first dropsonde curtain (Fig. 3a) has a substantial region with observed winds exceeding 45 m s−1. Visual inspection of the trajectories revealed distinct clusters with distinct origin and properties. The trajectories were subdivided by choosing thresholds on θe, pressure, and location that most cleanly separated the clusters. The thresholds differ for each arrival time such as to get the cleanest separation into one, two, or three clusters. The “release time” of the back trajectories will also be referred to as the “arrival time” of the airstreams (considering their evolution forwards in time).

S1 airstreams all follow a highly curved path around the cyclone core, arriving at pressure levels around 800 hPa (Figs. 5a,c,f,i). S3 airstreams follow a similar path but in general arriving at pressure levels below S1 airstreams. S3 was only identified as a cluster distinct from S1 for the arrival times 1300 and 1600 UTC. As well as lower arrival positions, they follow a path at a slightly greater radius from the cyclone center; it will be shown later that they also have a distinct history of vertical motion. S2 airstreams follow a more zonal path, descending in from greater radius on the west flank of the cyclone (Figs. 5b,d,g). The S2 cluster was not found in back trajectories from 1800 UTC.

In Fig. 5, the locations of the back trajectories from 1300 and 1600 UTC are shown as black dots at the times of 1200 and 1500 UTC, respectively (with the corresponding pressure map). This is to tie in with the first dropsonde curtain centered on 1200 UTC and the in situ flight legs near 1500 UTC—the back trajectories from the strong wind regions (farther east) span the line of the observations at these two times.

Figure 6 shows vertical sections of horizontal wind speed, θe, and RHice at 0900 and 1200 UTC along sections marked in Figs. 3a and 3b. The positions of the airstream trajectories crossing the vertical sections at the two times are overlain. The section at 0900 UTC (Fig. 6a) shows trajectories whose arrival time is also 0900 UTC, which explains their orderly distribution. By definition the two airstreams (S1 and S2) are located in the region of strong winds. However, there is a clear separation between them, with S1 trajectories (white circles) located beneath S2 trajectories (gray circles). S1 trajectories are near saturation with respect to ice, while S2 trajectory locations are subsaturated. However, they are characterized by similar θe values [293 < θe < 296 (K)].

The section at 1200 UTC (Fig. 6b) shows back trajectories released from the strong wind regions at 1300 UTC. Even though these trajectories are 1 h away from their arrival time, they have already reached the strong wind region to the south of the bent-back front. At 1200 UTC, the trajectories classified as S1 (white circles) span a deeper layer from 800 hPa to about 600 hPa. S2 trajectories are located to the south of S1. As a result, the two trajectory sets are now characterized by slightly different θe values. The S3 air mass is located beneath S2 and parts of S1. Referring back to the dropsonde observations in Fig. 4a, it can be seen that the S1 and S3 airstreams coincide with cloudy air, while the S2 airstream (56°N; 600–800 hPa) is characterized by lower RHice (50%–80%).

b. Identifying airstreams with distinct composition using the aircraft data

The FAAM aircraft conducted three level runs on a descending stack through the strong wind region south of the cyclone, just to the west of Scotland. The legs were over the sea between the islands of Islay and Tiree (Vaughan et al. 2014, manuscript submitted to Bull. Amer. Meteor. Soc.) perpendicular to the mesoscale cloud banding. Figure 7a presents measurements of wind speed (black), CO (blue), θ (red), θe (orange), and pressure altitude (dark red). At the beginning of the time series the aircraft was within the warm seclusion heading south from the cyclone center at 643 hPa (≈3.7 km). There is a marked change in airmass composition (CO increase) at point R nearing the radius of maximum winds at this level. The composition was fairly uniform (labeled O1) until an abrupt change moving into air mass O2 at 1454 UTC (14.9 h); this was also seen in other tracers such as ozone. Across O1, wind speed dropped slowly with distance and several narrow cloud bands were crossed (seen as spikes in θe, marked C).

Air mass O2 was characterized by higher CO than the rest of the time series shown. Two lower dips coincide with peaks in θe indicating changes in composition associated with banding. The aircraft began to descend on the same heading leaving air mass O2. At point T1 (15.1 h), it performed a platform turn onto a northward heading and then continued descent to a level northward run at 840 hPa in air mass O3 (radar height above the sea surface of 1400 m). This run crossed two precipitating cloud bands (see Vaughan et al. 2014, manuscript submitted to Bull. Amer. Meteor. Soc.). The CO is variable along this run but drops toward the end entering air mass O4). The wind speed increased generally along this northward run, interrupted by marked drops within the cloud bands. The aircraft descended again into cloud to the 180° turn T2 and descent to a third-level run heading southward at 930 hPa (height 500 m). The maximum wind speed observed was 49 m s−1 after turning at 500 m. The same two cloud bands were crossed at this lower level.

The lower panels in Fig. 7 show back trajectories calculated from points spaced at 60-s intervals along the flight track. The calculation uses ERA-Interim winds, interpolated in space and time to current trajectory locations, as described in section 2c. This technique has been shown to reproduce observed tracer structure in the atmosphere with a displacement error of filamentary features of less than 30 km (Methven et al. 2003). The back trajectories are colored using the observed CO mixing ratio at each release point. The color scale runs from blue to red (low to high), and the corresponding mixing ratios can be read from the time series.

Figure 7b shows back trajectories from the southward leg to turn T1 (1437–1503 UTC). Three coherent levels of CO are associated with distinct trajectory behaviors. Trajectories for the lowest CO (blue) wrap tightly into the cyclone center and are identified with air entering the warm seclusion. They originate from the boundary layer (almost 1 day beforehand) and ascend most strongly around the northern side of the cyclone center. The intermediate CO values (green) are labeled O1 and also wrap around the cyclone center, ascending most strongly as they move around the western flank of the cyclone. The aircraft intercepted them on the higher leg (640 hPa) where the trajectories are almost level. The marked jump to the higher CO (red) in air mass O2 is linked to a change in the analyzed trajectory behavior. All the O2 back trajectories reach a cusp (i.e., a change in direction at a stagnation point in the system-relative flow) on the northwest side of the cyclone (at around 1800 UTC 7 December 2011); before the cusp, the trajectories were ascending from the southwest. The correspondence of airstream O1 defined using aircraft composition data with airstream S1, identified using the forecast region of strong low-level winds (Fig. 5f), is striking. The O2 and S2 airstreams (Fig. 5g) are also very similar, although some are included in the S2 cluster that loop around the cyclone, rather than changing direction at a cusp as in the O2 cluster. The implication is that the two abrupt changes observed in composition are associated with different airstreams identified by their coherent trajectory behavior in both the ERA-Interim analyses and MetUM model forecasts.

Figure 7c shows back trajectories from the northward leg at lower levels from T1 to T2. CO increases from turn T1 on moving into the air mass labeled O3, but reduces slightly again on entering air mass O4. Again, observed changes in CO are linked to marked changes in trajectory behavior. All the O4 trajectories (including those from the lowest leg, not shown) wrap around the cyclone and are similar to the model airstreams S3 or S1. In contrast, most of the O3 trajectories approached a cusp from the southwest, while only a few wrap around the cyclone, traveling at 850 hPa. The O3 trajectories that approach from the southwest are similar to S2 trajectories (Fig. 5g); those that wrap around the cyclone center are similar to S3 trajectories (Fig. 5h).

c. Location of the airstreams relative to the frontal structure

The flight track is overlain on a vertical section through the MetUM simulation in Fig. 8 for the time interval shown in Fig. 7. The colors in the pipe along the flight track in Fig. 8a show observed wind speed on the same color scale as the model wind field. The wind structure in the model appears to be displaced southward of the observed wind structure. However, the flight track crossed the radius of maximum wind (point R in Figs. 7a, 8) at 1443 UTC and the front was shifting southward with time, so the mismatch in part reflects the asynchronous observations. However, the turn T1 was at 1503 UTC, and the winds still indicate a forecast displacement of 0.2°–0.3° southward. This is consistent with the observed displacement of the cold front and cyclone center in the forecast.

The gray shading inside the flight track pipe in Fig. 8a represents RHice. For the southward run at 640 hPa, the position of the cloud in the model appears to correspond with the observed cloud (black). However, the model has the cloud within the seclusion on the north side of the wind gradient, while the observations show the cloud farther south spanning the maximum winds. This humidity error is consistent with that seen already on the first and second dropsonde sections. In the southern section of the flight track, the observations suggest that the aircraft was flying through relatively dry air that is only saturated on crossing the cloud bands at low levels in air masses O3 and O4. In contrast, the model forecast shows a deep layer of RHice > 80% extending from around 950 hPa up to around 750 hPa. There is no indication of cloud banding along this section in the model. However, since the observed spacing was 20–25 km, a model with grid spacing of 12 km could not resolve these bands.

The model section at 1500 UTC is shown again in Fig. 8b, but the flight track is shaded with observed CO mixing ratio. The location of back trajectories at 1500 UTC, extending from the strong wind regions in the model at 1600 UTC, is also plotted on the section (showing parcels lying within 25 km of the section). The three airstreams S1, S2, and S3 are shown in different gray shades. S1 parcels (white circles) span a deep layer between 900 and 550 hPa on or north of the wind maximum. S2 parcels (light gray circles) are contained in a shallower layer between 650 and 550 hPa and are located to the south of S1 parcels. S3 parcels (dark gray circles) are restricted to a lower layer between 850 and 750 hPa and are also located to the south of S1 parcels. Following the flight track southward from R, the stretch of lower CO (black) identified as O1 coincides with the model airstream S1. The sharp CO increase moving from O1 to O2 coincides with a transition to the S2 airstream. At the lower levels the model airstream S3 lies within the stretch of higher CO (white) identified as air mass O3, and the transition to the S1 airstream occurs just south of the drop in CO associated with entering the O4 air mass (consistent with the 0.2° southward displacement error of the model). Therefore, the airstreams are bounded by abrupt changes in chemical composition, which lends credence to the identification of three clusters at this time and their different pathways.

d. Evolution of airstream properties

In the previous section it was shown that distinct air masses exist in the regions of strong winds near the cyclone center, and they are associated with three types of airstream labeled S1, S2, and S3. The evolution of these airstreams is now investigated.

Figure 9a shows the ensemble median evolution of pressure for each of the identified airstreams with arrival times of 0900, 1300, 1600, and 1800 UTC. The consistency between the airstream types at different arrival times is immediately apparent. The median pressure in S1 trajectories (green lines) remains at low levels (below 700 hPa) at all times. However, they experience slow average ascent (approximately 150 hPa in 4–7 h). As seen with the ERA-Interim trajectories, ascent occurs on the cold side of the bent-back front on the northern and western flank of the cyclone. The two S3 airstreams experience less ascent than S1 and arrive below 800 hPa. However, they start at a similar pressure level (900 hPa). The S2 airstreams exhibit very different vertical motion. They ascend to 550 hPa on average and then descend slowly to an average of 700 hPa (however some descend considerably further). The peak altitude of S2 trajectories occurs directly west of the cyclone center, where they exhibit a cusp between the westward-moving air near the bent-back front and the eastward-moving air approaching from the west.

Figures 9b–e show the ensemble median evolution of θe, RHice, and ground- and system-relative horizontal wind speed along trajectories with arrival time 1600 UTC. Trajectories corresponding to other arrival times exhibit similar behavior to those arriving at 1600 UTC. The changes in θe are small (less than 4 K), as would be expected since θe is materially conserved, in the absence of mixing, for saturated or unsaturated air masses. In S1 and S3, the median RHice is above 80% throughout the analyzed interval with an increase between 0600 and 0700 UTC from 80% to saturation (Fig. 9c) associated with the weak ascent. These airstreams exhibit an increase from θe < 290 K up to θe > 293 K during the 15 h of development (Fig. 9b). This may be a result of surface fluxes from the ocean into the turbulent boundary layer. The median trajectories are below 850 hPa until 1000 UTC and therefore likely to be influenced by boundary layer mixing. In contrast, after 1000 UTC the RHice of airstream S2 decreases rapidly associated with descent, arriving with an average of 30%. During these 7 h, θe decreases by less than 1 K, which is slow enough to be explained by radiative cooling. It implies that the effects of mixing do not alter θe. The next section will investigate diabatic processes in more detail.

The ground-relative horizontal wind speed of the three streams S1, S2, and S3 start and end at similar values (Fig. 9d). They exhibit slight deceleration down to |V| < 10 m s−1 during the first 3 h after 0100 UTC and then steady acceleration to reach wind speeds |V| ≃ 45 m s−1 at 1600 UTC. The kinematic differences between the two types of trajectories can be fully appreciated by considering system-relative horizontal wind speeds (Fig. 9e). The system velocity was calculated at every time step as the domain-average velocity at the steering level, assumed to be 700 hPa. The eastward component is dominant and decreases steadily from 14.5 to 11.5 m s−1 over 24 h from 0000 UTC 8 December 2011. In S1 and S3, system-relative acceleration takes place at early times, as they wrap around the eastern and northern flank of the cyclone center. In contrast, acceleration in the S2 airstream takes place during the final few hours (between 1000 and 1600 UTC), while trajectories descend from a cusp to the west of the cyclone toward the east-southeast.

All the characteristics of the S1 airstream described above are consistent with the definition of a CCB (Schultz 2001) wrapping around three-quarters of the cyclone to reach the strong wind region south of the cyclone center. The air accelerates in a system-relative frame ahead of the cyclone along the warm front and on its northern flank on the cold side of the bent-back front. It ascends to the northeast and north of the cyclone, giving rise to cloud there, and is then advected almost horizontally on a cyclonic trajectory with the bent-back front. The behavior of S3 is also consistent with a CCB, but traversing the cyclone at a slightly greater radius than the S1 and with less ascent.

The descent of the S2 airstream from within the cloud head to the west of the cyclone center, with a corresponding rapid decrease in RHice, is consistent with the behavior of a sting jet. The descent rate is comparable to that found in sting jets (e.g., Gray et al. 2011). Further evidence to characterize S2 trajectories as part of a sting jet is the small variation in θe in comparison with the CCBs and the rapid acceleration in both ground- and system-relative winds during the descent toward the south side of the cyclone. The ERA-Interim and MetUM trajectories both show that air that becomes the sting jet airstream enters the cloud head over a range of locations spanning the northwest side of the cyclone center.

e. Partition of diabatic processes following airstreams

Having shown that the S1 and S3 airstreams have a very different history of relative humidity compared to S2, linked to vertical motion, the Lagrangian rates of change associated with diabatic processes are now investigated in more detail using tracers within the MetUM simulation. Figure 10 shows the median of the heating rate /Dt and the rate of change of specific humidity Dq/Dt within the airstreams.

The S1 and S3 airstreams exhibit little average heating from 0100 to 0900 UTC (Fig. 10a) coinciding with the rapid system-relative acceleration phase to the east and north of the cyclone. However, they do pick up moisture, presumably associated with boundary layer fluxes over the ocean. After 0900 UTC both air masses experience heating but S1 at twice the rate of S3 and with an associated faster decrease in specific humidity (Fig. 10b). This is consistent with the stronger ascent in S1, condensation, and associated latent heat release. However, the θe increase indicates that the heating is faster than could be obtained from a pseudoadiabatic process and, therefore, highlights the action of mixing near the frontal surface.

In contrast, S2 exhibits an initial period of heating and condensation, between 0100 and 0900 UTC, which takes place during ascent (see S2 at 1600 UTC in Fig. 9a). This is followed by a period of weak cooling during descent and almost no change in q. This would be consistent with subsaturated motion.

Eulerian tracer fields running online with the MetUM are used to partition /Dt into the contributions from cloud microphysics, convection, radiation, and the boundary layer scheme (see section 2d). In airstream S2, the contribution from the cloud microphysics (Fig. 10c) has the same history as the total heating but greater intensity. The cooling on descent is a result of microphysics and may indicate that on average the ensemble experiences cooling from evaporation of condensate (ice at this level) but other processes (such as convection; Fig. 10d) oppose the microphysical cooling.

The cloud microphysics contributes latent cooling to S1 and S3 at a similar rate during the initial period, but the convection parameterization scheme (Fig. 10d) contributes heating to the S3 flow but not to S1. Subsequently, S1 and S3 experience latent heating from microphysics, which for S1 is of higher intensity and occurs over a longer time interval than for S3. This is a result of resolved ascent and stratiform precipitation. Mixing in the boundary layer and radiation in both airstreams (not shown) makes only a very small negative contribution to the total heating.

5. Mesoscale instability in the vicinity of the airstreams

Each of the identified airstreams passes through sectors of the cyclone with different susceptibilities to mesoscale atmospheric instability. The diagnostic criteria for CI, CSI, and II are described in section 2e. The relevant MetUM fields (, ζz, MPV*, and RHice) were interpolated onto every trajectory point (see section 2c) and the instability diagnostic criteria applied in order to assign an instability type to each trajectory point. Figure 11 shows histograms of the number of trajectories classified by each instability type every hour along the trajectories arriving at 1600 UTC 8 December 2011. The histograms are compiled separately for the S1, S2, and S3 airstreams.

Airstream S1 is predominantly stable (Fig. 11a), with some trajectories associated with CI, a smaller number associated with CSI, and very few associated with II. Figure 12 presents the instability diagnostics on the pressure level associated with the ensemble median position at the time shown. Figure 11a shows that at 0700 UTC some of the S1 trajectories lie within a band of CI along the bent-back front north of the cyclone, while most lie in the stable air surrounding this band on its northern flank or to the northeast of the cyclone. The peak in the proportion of S1 trajectories associated with CI occurs at 0800 UTC, while these trajectories are ascending. By 1100 UTC (Fig. 12b), most of the S1 trajectories are in stable air in the northwest part of the cyclone. Likewise, airstream S3 is mostly stable (Fig. 11c), with small numbers of trajectories associated with CSI, CI, and II.

In contrast, airstream S2 shows a much larger degree of instability (Fig. 11b). For several hours (0900–1300 UTC) more than 50% of the trajectories are associated with CSI. The trajectories begin their descent during this period (Fig. 9a). Figures 12c–d show that the S2 trajectories follow a band of instability as they wrap cyclonically through the cloud head. At 0700 UTC, this band is mostly inertially unstable, and the trajectories lie within this region. By 1100 UTC, the trajectories lie nearer to the southwest end of this band, which is associated mostly with CSI, and they appear to be entering the dry slot to the west-southwest of the cyclone.

These results suggest that airstream S2 passes through part of the storm meeting the necessary conditions for CSI, while S1 and S3 are in much more stable regions. They experience CI near the bent-back front where the convection scheme in the model produced some latent heating, although latent heating related to the resolved flow was dominant (Figs. 10b,c).

Figures 12c–d show marked “fingering” in the RHice field to the west-southwest of the cyclone center. This was the location of observed cloud banding in the morning, which progressed into the south side of the cyclone by midday (Fig. 2). The banding may arise as a result of the model trying to represent active CSI rolls at 12-km grid spacing and 70 vertical levels. Note that this is the same horizontal resolution and a similar vertical resolution to that used by Clark et al. (2005) in their examination of the October 1987 storm. Thorpe and Clough (1991) describe how, as the rolls characteristic of SI develop nonlinearly, they result in ruckles in the absolute momentum surfaces that would imply bands of negative and positive values of ζz. The ζz < 0 strips labeled as II here may well indicate development of CSI rather than inertial instability in the traditional sense (which would require weak pressure gradients unlikely to be met in the extratropics). Furthermore, the vorticity strips are flanked by two oppositely signed vorticity gradients and therefore must be unstable with respect to shear instability that acts on much faster time scales. In recognition of this ambiguity the instability maps and histograms have also been produced (not shown) with an alternative definition of the CSI and II instabilities: CSI at moist (RHice > 90%) grid points where MPV* < 0 and and II where ζz < 0 and a grid point is not already assigned as either having CI or CSI. The consequence is that some grid points, especially in the cloud head, change from being diagnosed as having II to having CSI (and some dry conditionally unstable points change from being defined as stable to having II). This strengthens the argument that the airstream S2 passes through the part of the storm meeting the necessary conditions for CSI.

The typical separation of the cloud bands (25 km) observed in Figs. 2a and 7a is too small to be resolved in a 12-km model. Therefore, we cannot expect to faithfully represent finescale slantwise circulations that may give rise to the observed banding. However, since CSI and the moisture required for it to be released are present, the model will release this instability in the form of one or more slantwise circulations on a broader scale (as seen in Figs. 12c,d). The link between the region of CSI from which the S2 airstream descends and the observed banded structure in the region where the S2 airstream arrives suggests that CSI release is a plausible explanation for the banding. The cloud bands were intercepted by the aircraft at the level identified with the S3 airstream and we have associated S3 with the CCB. However, the sting jet airstream S2 was immediately above the CCB at the time of interception but continued to descend overrunning the S3 part of the CCB (Fig. 5).

6. Discussion and conclusions

The focus of this article is the region of strong winds in the lower troposphere to the equatorward and rearward side of the center of extratropical cyclones. Two airstreams have been described in the literature and related to strong winds in this region: CCBs and sting jet airstreams. The aim of this paper was to present airborne observations and model simulations of this strong wind region during an intense cyclone named Friedhelm [IOP8 of the DIAMET field campaign (Vaughan et al. 2014, manuscript submitted to Bull. Amer. Meteor. Soc.)] and relate them to airstreams. The observations include three dropsonde curtains and in situ measurements. To the authors’ knowledge there are only two previous aircraft experiments with good in situ observational coverage (beyond satellites and ground-based network) across the strong wind regions of Shapiro–Keyser cyclones (Neiman et al. 1993; Wakimoto et al. 1992). In comparison, the Friedhelm case has much higher density dropsonde coverage (separation ≈30 km) across the regions of interest.

The first dropsonde curtain was a northward section from the cold front, crossing the bent-back front into the cyclone center just as peak cyclone intensity was reached. The second followed immediately, running radially outwards toward the southwest across the prominent cloud banding and cloud head tip. The third was 5 h later crossing the bent-back front after the cyclone had crossed Scotland and wrapped up further into the warm seclusion stage. A common feature on all three sections was that wind speed was highest immediately above the boundary layer (with maxima in the range 48–51 m s−1). To the southwest, the bent-back front sloped radially outwards with height like the “bent-back baroclinic ring” described by Neiman and Shapiro (1993). In Friedhelm at the end of development stage 3 (0900 UTC), the diameter of the ring of maximum winds at 850 hPa was 290–360 km; this is broad compared with 150 km at this level for ERICA IOP4 (Neiman and Shapiro 1993) and 220 km for the October 1987 storm (Browning 2004). This structure could be expected from consideration of gradient thermal wind balance. The slope of the momentum surfaces was very steep and θe surfaces were almost parallel to the momentum surfaces, implying that MPV* was near zero, given that the air was saturated in a cloud band sloping up the warm side of the frontal surface.

A simulation of the cyclone with the MetUM, initialized at 0000 UTC 8 December 2011 from the Met Office global analysis, captured the cyclone’s major features well with an overall southward displacement error at 1500 UTC of approximately 0.2° latitude. The location and shape of the cold front and bent-back front was very close to the analysis. However, on the south side of the cyclone, the strong winds in the model extended too far upward without the marked step observed in the front that contained the strongest winds to lower levels. In the later stage, downwind of Scotland, the gradient in wind strength and θe across the frontal surface was too strong and the drop in wind speed toward the ocean surface was also too great. Both aspects are indicative of the turbulent mixing being too weak in the model at this later stage. The model simulation of the humidity field was not as good as for winds and temperature. Two systematic errors were identified. First, observations showed that subsaturated air was wrapped within the warm seclusion, but around a cyclone core that was nearly saturated. The model captured this structure but the central regions of high relative humidity were too extensive. Second, deep cloud was observed along the warm side of the bent-back frontal surface coincident with the strongest winds, but the model put the cloud farther toward the cyclone center across the wind speed gradient.

Conclusions are now drawn regarding the scientific questions posed in section 1:

  1. How are strong wind regions south of the cyclone related to the CCB and sting jet airstreams?

Back trajectory analysis within the MetUM simulation identified three distinct types of airstream arriving in the strong wind regions. Since the strong wind regions tend to move with the cyclone, the airstreams must flow through them, and so the airstreams were identified at four different times: 0900, 1300, 1600, and 1800 UTC. However, the ensemble-mean trajectory behavior for each type (S1, S2, and S3) was very consistent between the arrival times.

Airstreams S1 and S3 both traveled three quarters of the way around the cyclone, starting ahead of the warm front and staying on the cold flank of the bent-back front. Both ascended slowly on average from the boundary layer, S1 slightly faster than S3 and curving round at a slightly smaller radius. Acceleration in wind speed in a system-relative frame was greatest ahead of the cyclone (on the east) and around the northern flank with the extension of the bent-back front. Ascent was fastest on average along the northern and western flanks. Therefore, these were both identified with the CCB.

Airstream S2 descended from the cloud head on the west side of the cyclone center toward the east-southeast. The trajectories entered the cloud head from a spread of locations, some on the northern flank but many ascending from the southwest reaching a cusp at maximum altitude in the cloud head where they changed direction and descended to the east-southeast. In a system-relative frame, the cusp was associated with very light winds and the airstream accelerated rapidly on descent. S2 is associated with the sting jet airstream.

  1. Are the airstreams identified using the model observed to have distinct airmass properties?

Yes. It was shown that very marked changes observed in tracer composition (CO, ozone, and specific humidity) were explained by abrupt change in trajectory behavior and their origins. The trajectories were most sensitive to the west of the cyclone where CCB back trajectories continued around the cyclone while the sting jet trajectories experienced a cusp and originated from the southwest.

The intersection of the airstreams with the first dropsonde section at 1200 UTC was also examined. The locations of CCB airstreams S1 and S3 in the model tie in with the strongest winds at this time and were observed to be saturated. The sting jet airstream S2 is coincident with an observed region of subsaturated air (50 < RHice < 80%) above S3 and on the southern flank of the lower-tropospheric wind maximum at this time.

  1. What dynamical mechanism is responsible for the cloud banding in the cloud head and to the south of the cyclone?

Only one stack of flight legs crossed the strong wind region and cloud banding upstream of Scotland. Three distinct cloud bands were flown through at 840 hPa, and it was observed that wind speed was weaker within the cloud bands than in the clear air between. Vaughan et al. (2014, manuscript submitted to Bull. Amer. Meteor. Soc.) present further evidence that a relationship between cloud bands and surface winds was observed in the DIAMET IOP8 case across central Scotland using the precipitation radar network and automatic weather stations for the winds.

Steps were taken to determine the dynamical mechanism responsible for the banding by diagnosing the necessary conditions for mesoscale instability throughout the cyclone. The stability diagnostics were then sampled at trajectory points for each airstream separately. The CCB airstreams S1 and S3 were found to pass through largely stable regions. In contrast, over 50% of the sting jet airstream passed through regions satisfying conditions for conditional symmetric instability (CSI) as defined in section 2e. The dropsonde observations also indicate that MPV must be near zero to the southwest of the cyclone. In the model, cloud banding occurred in this region, indicative of active CSI, although the bandwidth was much greater than observed. Since the observed spacing was 20–50 km, a model with a 12-km grid length could not hope to resolve it faithfully; however, the model can still be unstable and develop its own CSI rolls. Strips of negative absolute vorticity also developed in the model in these regions. Thorpe and Clough (1991) suggest that this would be expected to happen where CSI perturbations grow into the nonlinear regime.

The results suggest that CSI is a plausible candidate for the origin of the banding. However, strong cloud bands also often develop in the boundary layer, particularly during cold air outbreaks. For example, Fig. 20 of Neiman and Shapiro (1993) shows very finescale bands in the boundary layer cloud in the cold sector of the ERICA IOP4 case. However, in that case, 6 h later, radar reflectivity from the National Oceanic and Atmospheric Administration (NOAA) WP-3D aircraft (Fig. 19b of Neiman et al. 1993) showed two parallel precipitation bands coincident with fingers of cloud extending from the cloud head tip that Browning (2004) related to the surface sting jet structures. Therefore, it is also possible that the bands on the south side of Friedhelm were initiated from upstream boundary layer structures, extending above the boundary layer through upright convection or from the release of CSI. Thus, further research is required to establish the dynamical origin of the observed banding. More detailed high-resolution experiments would be required to analyze the origin of the banding using a model where it was well resolved. Vaughan et al. (2014, manuscript submitted to Bull. Amer. Meteor. Soc.) present preliminary results from work into that direction.

All these results can finally be put together as follows: The evidence from trajectory analysis strongly indicates that the air in contact with the surface followed a trajectory similar to that of the CCB. However, the region of strong winds is not restricted to the surface, but extends from the ground into the midtroposphere with no obvious separation between the air constituting the CCBs and that constituting the sting jet. Nevertheless, our analysis shows that this region is composed of different air masses, following different trajectories, but ending up at the same horizontal location. Each air mass transports a certain amount of horizontal momentum that is transferred to the ground, generating surface shear stress and the potential for surface damage. The damage at the surface is determined not by what kind of air is in contact with the surface, but by how much shear stress the surface is subject to. In turn, the shear stress is determined by the momentum that is being transferred from the air to the ground and is proportional to the vertical wind shear and indirectly to wind strength either at a certain height (typically observed at 10 m) or as represented by friction velocity (Janssen 2004). Perhaps, there are intervals during a cyclone life cycle in which sting jets are the only streams constituting a low-level jet near the bent-back front (Browning 2004; Smart and Browning 2014). However, the general situation is given by a combination of airstreams constituting the low-level jet in which different air masses have different origins but all meet, by the intrinsic dynamics of the cyclone, on that same region. So, even though sting jet trajectories might always remain at levels above those associated with the CCB this fact does not automatically preclude the influence of these airstreams on the potentially damaging conditions experienced at the surface.

Acknowledgments

This work was funded by the Natural Environment Research Council (NERC) as part of the DIAMET project (NE/I005234/1). The authors thank the Met Office for making the Met Office Unified Model and associated start dump and LBCs files available and the NERC-funded National Centre for Atmospheric Sciences (NCAS) Computational Modelling Services (CMS) for providing computing and technical support. The BAe-146 aircraft is flown by Directflight Ltd and managed by the Facility for Airborne Atmospheric Measurements (FAAM) on behalf of NERC and the Met Office. We particularly thank Captain Alan Foster and Copilot Ian Ramsay-Rae for flying in such severe wind conditions and working hard to find an airport that would let us land for refueling during the storm. We also thank two anonymous reviewers whose comments helped to improve the manuscript.

REFERENCES

REFERENCES
Arakawa
,
A.
, and
V.
Lamb
,
1977
:
Computational design of the basic dynamical processes of the UCLA general circulation model
.
Methods Comput. Phys.
,
17
,
173
265
, doi:.
Baker
,
L.
,
O.
Martinez-Alvarado
,
J.
Methven
, and
P.
Knippertz
,
2013
:
Flying through extratropical cyclone Friedhelm
.
Weather
,
68
,
9
13
, doi:.
Baker
,
L.
,
S. L.
Gray
, and
P. A.
Clark
,
2014
:
Idealised simulations of sting-jet cyclones
.
Quart. J. Roy. Meteor. Soc.
, 140, 96–110, doi:.
Bennetts
,
D. A.
, and
B. J.
Hoskins
,
1979
:
Conditional symmetric instability—A possible explanation for frontal rainbands
.
Quart. J. Roy. Meteor. Soc.
,
105
,
945
962
, doi:.
Böttger
,
H.
,
M.
Eckardt
, and
U.
Katergiannakis
,
1975
:
Forecasting extratropical storms with hurricane intensity using satellite information
.
J. Appl. Meteor.
,
14
,
1259
1265
, doi:.
Browning
,
K. A.
,
1971
:
Radar measurements of air motion near fronts
.
Weather
,
26
,
320
340
, doi:.
Browning
,
K. A.
,
2004
:
The sting at the end of the tail: Damaging winds associated with extratropical cyclones
.
Quart. J. Roy. Meteor. Soc.
,
130
,
375
399
, doi:.
Browning
,
K. A.
, and
N. M.
Roberts
,
1994
:
Structure of a frontal cyclone
.
Quart. J. Roy. Meteor. Soc.
,
120
,
1535
1557
, doi:.
Carlson
,
T. N.
,
1980
:
Airflow through midlatitude cyclones and the comma cloud pattern
.
Mon. Wea. Rev.
,
108
,
1498
1509
, doi:.
Charney
,
J. G.
, and
N. A.
Phillips
,
1953
:
Numerical integration of the quasi-geostrophic equations for barotropic and simple baroclinic flows
.
J. Meteor.
,
10
,
71
99
, doi:.
Clark
,
P. A.
,
K. A.
Browning
, and
C.
Wang
,
2005
:
The sting at the end of the tail: Model diagnostics of fine-scale three-dimensional structure of the cloud head
.
Quart. J. Roy. Meteor. Soc.
,
131
,
2263
2292
, doi:.
Cook
,
P. A.
, and
I. A.
Renfrew
,
2014
:
Aircraft-based observations of air–sea turbulent fluxes around the British Isles
.
Quart. J. Roy. Meteor. Soc.
, doi:,
in press
.
Davies
,
T.
,
M. J. P.
Cullen
,
A. J.
Malcolm
,
M. H.
Mawson
,
A.
Staniforth
,
A. A.
White
, and
N.
Wood
,
2005
:
A new dynamical core for the Met Office’s global and regional modelling of the atmosphere
.
Quart. J. Roy. Meteor. Soc.
,
131
,
1759
1782
, doi:.
Durran
,
D. R.
, and
J. B.
Klemp
,
1982
:
On the effects of moisture on the Brunt–Väisälä frequency
.
J. Atmos. Sci.
,
39
,
2152
2158
, doi:.
Edwards
,
J.
, and
A.
Slingo
,
1996
:
Studies with a flexible new radiation code. Part I: Choosing a configuration for a large-scale model
.
Quart. J. Roy. Meteor. Soc.
,
122
,
689
719
, doi:.
Gray
,
S. L.
,
2006
: Mechanisms of midlatitude cross-tropopause transport using a potential vorticity budget approach. J. Geophys. Res.,111, D17113, doi:.
Gray
,
S. L.
,
O.
Martínez-Alvarado
,
L. H.
Baker
, and
P. A.
Clark
,
2011
:
Conditional symmetric instability in sting jet storms
.
Quart. J. Roy. Meteor. Soc.
,
137
,
1482
1500
, doi:.
Gregory
,
D.
, and
P. R.
Rowntree
,
1990
:
A mass flux convection scheme with representation of cloud ensemble characteristics and stability-dependent closure
.
Mon. Wea. Rev.
,
118
,
1483
1506
, doi:.
Harrold
,
T. W.
,
1973
:
Mechanisms influencing distribution of precipitation within baroclinic disturbances
.
Quart. J. Roy. Meteor. Soc.
,
99
,
232
251
, doi:.
Hoskins
,
B. J.
, and
F. P.
Bretherton
,
1972
:
Atmospheric frontogenesis models: Mathematical formulation and solution
.
J. Atmos. Sci.
,
29
,
11
37
, doi:.
Janssen
,
P.
,
2004
:
The Interaction of Ocean Waves and Wind
. Cambridge University Press, 300 pp.
Lock
,
A. P.
,
A. R.
Brown
,
M. R.
Bush
,
G. M.
Martin
, and
R. N. B.
Smith
,
2000
:
A new boundary layer mixing scheme. Part I: Scheme description and single-column model tests
.
Mon. Wea. Rev.
,
128
,
3187
3199
, doi:.
Martínez-Alvarado
,
O.
, and
R. S.
Plant
,
2014
:
Parameterized diabatic processes in numerical simulations of an extratropical cyclone
.
Quart. J. Roy. Meteor. Soc.
, doi:,
in press
.
Martínez-Alvarado
,
O.
,
F.
Weidle
, and
S. L.
Gray
,
2010
:
Sting jets in simulations of a real cyclone by two mesoscale models
.
Mon. Wea. Rev.
,
138
,
4054
4075
, doi:.
Martínez-Alvarado
,
O.
,
S. L.
Gray
,
J. L.
Catto
, and
P. A.
Clark
,
2012
: Sting jets in intense winter North-Atlantic windstorms. Environ. Res. Lett.,7, 024014, doi:.
Martínez-Alvarado
,
O.
,
S. L.
Gray
,
P. A.
Clark
, and
L. H.
Baker
,
2013
:
Objective detection of sting jets in low-resolution atmospheric datasets
.
Meteor. Appl.
,
20
,
41
55
, doi:.
Martínez-Alvarado
,
O.
,
H.
Joos
,
J.
Chagnon
,
M.
Boettcher
,
S. L.
Gray
,
R. S.
Plant
,
J.
Methven
, and
H.
Wernli
,
2014
:
The dichotomous structure of the warm conveyor belt
.
Quart. J. Roy. Meteor. Soc.
, doi:,
in press
.
Methven
,
J.
,
1997
: Offline trajectories: Calculation and accuracy. U.K. Universities Global Atmospheric Modelling Programme, Department of Meteorology, University of Reading, Tech. Rep. 44, 17 pp.
Methven
,
J.
,
S. R.
Arnold
,
F. M.
O’Connor
,
H.
Barjat
,
K.
Dewey
,
J.
Kent
, and
N.
Brough
,
2003
:
Estimating photochemically produced ozone throughout a domain using flight data and a Lagrangian model
.
J. Geophys. Res.
,
108
, 4271, doi:.
Neiman
,
P. J.
, and
M.
Shapiro
,
1993
:
The life cycle of an extratropical marine cyclone. Part I: Frontal-cyclone evolution and thermodynamic air–sea interaction
.
Mon. Wea. Rev.
,
121
,
2153
2176
, doi:.
Neiman
,
P. J.
,
M.
Shapiro
, and
L.
Fedor
,
1993
:
The life cycle of an extratropical marine cyclone. Part II: Mesoscale structure and diagnostics
.
Mon. Wea. Rev.
,
121
,
2177
2199
, doi:.
Ola
,
P.
,
G.
Persson
, and
T. T.
Warner
,
1993
:
Nonlinear hydrostatic conditional symmetric instability: Implications for numerical weather prediction
.
Mon. Wea. Rev.
,
121
,
1821
1833
, doi:.
Parton
,
G. A.
,
G.
Vaughan
,
E. G.
Norton
,
K. A.
Browning
, and
P. A.
Clark
,
2009
:
Wind profiler observations of a sting jet
.
Quart. J. Roy. Meteor. Soc.
,
135
,
663
680
, doi:.
Sanders
,
F.
, and
J. R.
Gyakum
,
1980
:
Synoptic-dynamic climatology of the “bomb.”
Mon. Wea. Rev.
,
108
,
1589
1606
, doi:.
Sapp
,
J. W.
,
S. J.
Frasier
,
J.
Dvorsky
,
P. S.
Chang
, and
Z.
Jelenak
,
2013
:
Airborne dual-polarization observations of the sea surface NRCS at C-band in high winds
.
IEEE Geosci. Remote Sens. Lett.
,
10
,
726
730
, doi:.
Schäfler
,
A.
,
A.
Dörnbrack
,
H.
Wernli
,
C.
Kiemle
, and
S.
Pfahl
,
2011
:
Airborne lidar observations in the inflow region of a warm conveyor belt
.
Quart. J. Roy. Meteor. Soc.
,
137
,
1257
1272
, doi:.
Schultz
,
D. M.
,
2001
:
Reexamining the cold conveyor belt
.
Mon. Wea. Rev.
,
129
,
2205
2225
, doi:.
Schultz
,
D. M.
, and
P. N.
Schumacher
,
1999
:
The use and misuse of conditional symmetric instability
.
Mon. Wea. Rev.
,
127
,
2709
2732
, doi:.
Schultz
,
D. M.
, and
J. M.
Sienkiewicz
,
2013
:
Using frontogenesis to identify sting jets in extratropical cyclones
.
Wea. Forecasting
,
28
,
603
613
, doi:.
Shapiro
,
M. A.
, and
D.
Keyser
,
1990
: Fronts, jet streams and the tropopause. Extratropical cyclones: The Erik Palmén Memorial Volume, C. W. Newton and E. O. Holopainen, Eds., Amer. Meteor. Soc., 167–191.
Sinclair
,
V. A.
,
S. E.
Belcher
, and
S. L.
Gray
,
2010
:
Synoptic controls on boundary-layer characteristics
.
Bound.-Layer Meteor.
,
134
,
387
409
, doi:.
Smart
,
D. J.
, and
K. A.
Browning
,
2014
:
Attribution of strong winds to a cold conveyor belt and sting jet
.
Quart. J. Roy. Meteor. Soc.
,
140
,
595
610
, doi:.
Stoelinga
,
M. T.
,
1996
:
A potential vorticity-based study of the role of diabatic heating and friction in a numerically simulated baroclinic cyclone
.
Mon. Wea. Rev.
,
124
,
849
874
, doi:.
Thorncroft
,
C. D.
,
B. J.
Hoskins
, and
M. E.
McIntyre
,
1993
:
Two paradigms of baroclinic-wave life-cycle behaviour
.
Quart. J. Roy. Meteor. Soc.
,
119
,
17
55
, doi:.
Thorpe
,
A. J.
, and
S. A.
Clough
,
1991
:
Mesoscale dynamics of cold fronts: Structures described by dropsoundings in FRONTS 87
.
Quart. J. Roy. Meteor. Soc.
,
117
,
903
941
, doi:.
Wakimoto
,
R. M.
,
W.
Blier
, and
C.
Liu
,
1992
:
The frontal structure of an explosive oceanic cyclone: Airborne radar observations of ERICA IOP 4
.
Mon. Wea. Rev.
,
120
,
1135
1155
, doi:.
Wernli
,
H.
, and
H. C.
Davies
,
1997
:
A Lagrangian-based analysis of extratropical cyclones. I: The method and some applications
.
Quart. J. Roy. Meteor. Soc.
,
123
,
467
489
, doi:.
Wilson
,
D. R.
, and
S. P.
Ballard
,
1999
:
A microphysically based precipitation scheme for the UK Meteorological Office Unified Model
.
Quart. J. Roy. Meteor. Soc.
,
125
,
1607
1636
, doi:.

Footnotes

This article is included in the Diabatic Influence on Mesoscale Structures in Extratropical Storms (DIAMET) special collection.