Abstract

A climatology of cold air outbreaks (CAOs) in the high latitudes of the South Pacific and an analysis of the dynamical mechanisms leading to their formation are presented. Two major and distinct regions with frequent CAOs from autumn to spring are identified: one in the Ross Sea and another in the Amundsen and Bellingshausen Seas. Using an objective method to attribute CAOs to extratropical cyclones, it is shown that about 80% of the CAOs occur in association with the cyclonic flow induced by the passage of extratropical cyclones. Based on kinematic backward trajectories it is quantified that more than 40% of the air masses leading to CAOs originate from Antarctica and descend substantially, with the Ross Ice Shelf corridor as the major pathway. CAO trajectories descending from Antarctica differ from those originating over sea ice by a much lower specific humidity, stronger diabatic cooling, and much more intense adiabatic warming, while potential vorticity evolves similarly in both categories. In winter, CAOs are the major contributor to the net turbulent heat flux off the sea ice edge and CAO frequency strongly determines its interannual variation. Wintertime variations of the frequency of extratropical cyclones are strongly imprinted on the frequency of CAOs and the net turbulent heat and freshwater fluxes. In particular, much of the precipitation associated with the passage of extratropical cyclones is compensated by intense evaporation in cyclone-induced CAOs. This highlights the dominant role of the extratropical storm track in determining the variability of the buoyancy flux forcing of the Southern Ocean.

1. Introduction

The Antarctic continent and the belt of sea ice surrounding it provide a gigantic pool of radiatively cooled air, in particular during the cold season from early autumn to late spring. Associated with the pool of cold air is a strong temperature inversion in the interior of the continent, which over the most elevated regions exceeds 25 K in the winter mean (Phillpot and Zillman 1970). Equatorward excursions of these cold air masses across the sea ice boundary over the relatively warm open ocean are termed marine cold air outbreaks (CAOs). They are characterized by a large air–sea temperature difference and strong baroclinicity at their margins, where the cold air masses impinge on warmer midlatitude air. The first systematic study of CAOs in the Southern Hemisphere was performed only recently by Bracegirdle and Kolstad (2010), who found strong CAOs during winter close to the sea ice boundary. They also identified the largest day-to-day variability of the air–sea temperature difference in the Southern Hemisphere to be located in the South Pacific. Kolstad (2011) identified two hot spots where CAOs are particularly intense and have a strength comparable to CAOs in the Northern Hemisphere. These CAO hot spots are located in the Ross Sea (RS) and the Amundsen and Bellingshausen Seas (ABS).

CAOs are well known as hotbeds of convective and often severe weather systems, like boundary layer fronts (Drüe and Heinemann 2001) and mesocyclones, including polar lows (Rasmussen and Turner 2003), which on account of gale-force winds and tremendous amounts of snowfall pose a serious threat to human activities in the polar regions. In the RS and the ABS a high density of mesocyclones off the sea ice boundary has been found (Irving et al. 2010). Mesocyclones can develop spontaneously along the baroclinic outer edge of the flow that descends from the Antarctic continent and in particular when the air mass reaches the open ocean (i.e., when it becomes a CAO). Surface sensible heat fluxes then favor convection by destabilizing the air mass and latent heat fluxes provide additional energy for the intensification of the vortices (Gallée 1995; Heinemann and Klein 2003). Furthermore, the magnitude of potential vorticity (PV) in the downslope flows from Antarctica is often high due to the strong stratification of the involved air mass. Klein and Heinemann (2002) proposed that vertical stretching of this high-PV air imposed by the large-scale flow or by topography can cause the spinup of cyclonic vorticity. Carleton and Song (1997) found that during winter, mesocyclone outbreaks, organized as a series of consecutive vortices, occur in the ice-free part of the RS associated with the cold, southerly flow to the west of large-scale extratropical cyclones. Furthermore, they found a strong linkage between the occurrence of mesocyclone outbreaks and large-scale baroclinic wave patterns, a relationship that also exists for CAOs (Bracegirdle and Kolstad 2010).

The steep Antarctic coastline shields the interior of the Antarctic continent from the storm-battered high-latitude Southern Ocean (e.g., Stohl and Sodemann 2010, their Fig. 3). At the same time, the topography of the Antarctic continent exerts a strong control on the low-level flow in and around Antarctica. Idealized rotating tank experiments by Baines and Fraedrich (1989) revealed that in a mean easterly flow the asymmetry of the Antarctic topography enforces the development of flow patterns with a barotropic wavenumber-3 structure corresponding closely to mean 700-hPa geopotential height maps, whereby the cyclonic eddies associated with this pattern are approximately collocated with maxima of cyclone frequency (Simmonds et al. 2003; Wernli and Schwierz 2006). As noted by Fogt et al. (2012), there is a close correspondence between variations in the position of the climatological low seen in sea level pressure (SLP) in the RS and ABS (also called the Amundsen Sea low; e.g., Turner et al. 2013) and the frequency of synoptic-scale cyclones. These findings indicate a complex interplay between the flow induced by extratropical cyclones impinging on the Antarctic continent, the modification of the flow by the orography, and its subsequent effect on the strength and lifetime of the cyclones. A region where this interaction has been studied in some detail is the steep coast of Adélie Land west of the RS. There, it was found by Bromwich et al. (2011b) that topographically induced barrier winds, occurring when an extratropical cyclone approaches the coast, often act in concert with the drainage flow from the Antarctic continent to enhance low-level baroclinicity such that the cyclone undergoes a secondary intensification, allowing it to subsequently propagate into the RS.

Over the continent the Antarctic topography influences the low-level flow by two mechanisms. First, longwave radiative cooling of the snow surface during the nonsummer months causes a tremendous potential temperature deficit over the Antarctic continent (Parish and Cassano 2003a). In the case of sloping topography, this results in a downslope directed pressure gradient force, leading to katabatic drainage flows prevalent over almost all of Antarctica (e.g., Parish and Bromwich 1991). These airstreams forced by the katabatic mechanism alone are shallow and typically restricted to the lowermost 100 m of the boundary layer, as indicated by vertical profiles of purely katabatically forced flows measured in Coats Land adjacent to the Weddell Sea (Renfrew and Anderson 2006). Second, when rather stably stratified air impinges on the orography, it is blocked and a downslope pressure gradient force is established. As shown by Parish and Cassano (2003a,b) subsequent adjustment leads to low-level winds closely resembling those of katabatic flows in their direction and strength.

A major gap in the Antarctic coastline adjacent to the RS and the ABS (Fig. 1) is the Ross Ice Shelf, a flat floating ice sheet that connects to the interior of the Antarctic continent. It provides a relatively narrow corridor through which cold air masses can leave the Antarctic continent. The airstream through the Ross Ice Shelf corridor, also known as the Ross Ice Shelf airstream (RAS; Parish et al. 2006; Seefeldt and Cassano 2012), presents a major contribution to the drainage of atmospheric mass from the Antarctic continent (Parish and Bromwich 1998, 2007). Nigro and Cassano (2014) found that the RAS is present over more than 34% of the time. As indicated by satellite observations, the air masses on the ice shelf are often advected into the interior of the ice-covered RS basin and farther toward the sea ice edge when synoptic forcing from an extratropical cyclone is present (Bromwich et al. 1992). Thereby, this cold flow assists in the generation of open polynyas (Bromwich et al. 1993) and affects mechanical sea ice export, causing a strong correlation between the amount of cold air exported from Antarctica and sea ice extent on seasonal time scales (Haumann 2011; Pezza et al. 2012).

Fig. 1.

Location map of the study domain showing the RS (gray) and the ABS boxes (black) in which CAOs are investigated. The terrain height in ERA-Interim is indicated by the gray shading in intervals of 500 m.

Fig. 1.

Location map of the study domain showing the RS (gray) and the ABS boxes (black) in which CAOs are investigated. The terrain height in ERA-Interim is indicated by the gray shading in intervals of 500 m.

The Ross Ice Shelf and the southern part of the RS are bounded to the west by the 2000-m-high Transantarctic Mountains, which block stably stratified flows and give rise to frequent events of intense geostrophically balanced barrier winds (O’Connor et al. 1994; Parish et al. 2006; Chenoli et al. 2013). These barrier winds amplify the drainage of Antarctic air masses via the Ross Ice Shelf corridor. Furthermore, when an extratropical cyclone is located in the southern RS or off the coast of Mary Byrd Land, its cyclonic circulation can enforce intense downslope winds descending from the Transantarctic Mountains (Breckenridge et al. 1993), which frequently support the genesis of mesocyclones (Bromwich 1991).

CAOs cause intense sensible and latent heat fluxes from the ocean surface to the atmosphere, as suggested by aircraft measurements. Shapiro et al. (1987) estimated sensible and latent heat fluxes in the vicinity of a polar low, which developed in association with a CAO in the Norwegian Sea, to be as large as 500 W m−2 each. Such enormous sensible and latent heat fluxes with approximately equal contributions of both types of fluxes were also suggested by measurements in a CAO over the Gulf Stream along the eastern coast of the United States (Grossman and Betts 1990). In contrast, Renfrew and Moore (1999) obtained a sensible heat flux of about 500 W m−2 in a CAO over the Labrador Sea off the sea ice edge, whereas the latent heat flux was about five times weaker, characteristic of the very low air and sea surface temperatures. Fluxes with similar intensity and with a comparable ratio of sensible to latent heat fluxes were found over a polynya in the western RS by Knuth and Cassano (2014). From these observations and the study by Grossman and Betts (1990) it is clear that CAOs have the potential to leave a distinct imprint in the spatial and temporal distribution of the seasonal fluxes. In the high-latitude South Pacific the climatological mean fluxes during winter are an order of magnitude weaker than instantaneous fluxes observed in CAOs. Figure 2 shows the interannual seasonal mean turbulent heat flux (sensible and latent heat fluxes) in the high-latitude South Pacific in the Interim European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-Interim; Dee et al. 2011), issued by the ECMWF, and its interannual variability, as characterized by the interquartile range. Both are weakest in December–February and have a maximum in June–August. There are two regions close to the sea ice edge where the winter turbulent heat flux from the ocean into the atmosphere is particularly large, one in the western RS and a second in the ABS, whereas in between, in the eastern RS, the wintertime turbulent heat flux is less than 50% of the neighboring maxima. It is interesting to note that in between the two maxima the contribution of evaporation to total precipitation over Antarctica is at a minimum (Sodemann and Stohl 2009, their Fig. 1). The maxima are closely bounded by the winter mean sea ice extent and they persist into late spring, when the sea ice extent gradually declines. Furthermore, they are approximately collocated with the CAO hot spots found by Kolstad (2011), which suggests that CAOs are of key importance for the spatial distribution of the winter mean fluxes. The interquartile range of the fluxes locally amounts to almost 50% of the climatological mean. The variability in the ABS in winter is considerably larger than in the RS, which possibly is an imprint of the strong interannual variations of cyclone frequency in the ABS unequivocally exceeding variations in the RS (Simmonds et al. 2003). The narrow band of relatively strong turbulent heat flux immediately off the Ross Ice Shelf occurring in autumn (Fig. 2b) might be related to barrier winds along the Transantarctic Mountains, which cause large sensible and latent heat fluxes when sea ice extent is still small.

Fig. 2.

Interannual variability of seasonal turbulent heat fluxes (sensible and latent heat fluxes) shown in terms of the interquartile range (IQR; shading). Seasonal turbulent heat fluxes are shown by gray contours in intervals of 20 W m−2. The sea ice boundary (50% sea ice concentration) derived from passive microwave brightness temperature is indicated by the solid black line.

Fig. 2.

Interannual variability of seasonal turbulent heat fluxes (sensible and latent heat fluxes) shown in terms of the interquartile range (IQR; shading). Seasonal turbulent heat fluxes are shown by gray contours in intervals of 20 W m−2. The sea ice boundary (50% sea ice concentration) derived from passive microwave brightness temperature is indicated by the solid black line.

Strong latent and sensible heat fluxes, as occurring in association with CAOs, considerably alter the buoyancy of oceanic surface waters, which is a critical parameter for the formation rates of water masses (e.g., Talley 2008). The latent heat flux affects the buoyancy of surface waters by changing both temperature and salinity, and thus it represents an important term in the buoyancy budget of the ocean. Over most regions of the Southern Ocean the interannual variability of the wintertime freshwater flux (evaporation minus precipitation) is driven by variations of precipitation attributed to cyclones and fronts and therefore is closely linked to shifts and changes in the intensity of the extratropical storm track (Papritz et al. 2014). In particular, more frequent storms lead to negative anomalies of the freshwater flux. However, Papritz et al. (2014) also showed that close to the sea ice edge in the South Pacific this signal is complicated by enhanced evaporation in the presence of extratropical cyclones, which partly compensates for the freshwater input due to cyclone and front precipitation. Cyclone-induced CAOs, which were not explicitly accounted for in the aforementioned study, are likely to further enhance this compensation and locally they might even balance cyclone and front precipitation.

Climatological studies of CAOs that included the Southern Hemisphere (Bracegirdle and Kolstad 2010; Kolstad 2011) focused on the identification of favorable conditions for polar low development. The dynamical mechanisms behind the formation of CAOs and their linkage to the extratropical storm track, as well as the contribution of the downslope flows from Antarctica to and the origin and thermodynamic history of CAO air masses in the RS and the ABS, are so far uninvestigated. Also, the climatological contribution of CAOs to the turbulent heat flux and to extreme events of such fluxes has not yet been systematically quantified. It is the purpose of this study to close this gap.

We present an objective climatology of CAOs in the high-latitude South Pacific based on an Eulerian identification criterion, complemented by trajectory calculations. In section 2 the underlying dataset and the methodology employed are presented, followed by an overview on the climatological frequency of CAOs and an analysis of the origin of the air masses leading to CAOs in section 3. Thereby, CAOs are separately investigated in the subregions of the RS and the ABS as outlined in Fig. 1 by the gray and black boxes, respectively. Here, also the characteristics of air masses descending from Antarctica and those originating from the sea ice covered ocean leading to CAOs are investigated. It follows a quantification of the net turbulent heat flux associated with CAOs and an investigation of the relevance of CAOs for events of extreme fluxes in section 4. CAOs are then attributed to extratropical cyclones using an objective method in section 5. Finally, interannual variations of the frequency of CAOs and of the net turbulent heat and freshwater flux in winter are linked to the variability of cyclone frequency. The major results are summarized in section 6 and complemented by an outlook on directions for future research.

2. Methodology

We use the ERA-Interim reanalysis dataset in the period from 1980 to 2010 (Dee et al. 2011). The data is interpolated to a 1° × 1° horizontal grid and is available in 6-hourly intervals. Caution must be exercised when dealing with reanalyses in the data-sparse southern high latitudes as it may contain biases. Here we refrain from an extensive discussion of the quality and reliability of ERA-Interim in these regions and refer to the literature (e.g., Dee et al. 2011; Bromwich et al. 2011a; Nicolas and Bromwich 2011; Bracegirdle and Marshall 2012). However, we note that sensible and latent heat fluxes and precipitation in ERA-Interim are not directly constrained by observations, but are obtained from short-range model forecasts. They depend on the details of the parameterization as much as on the accuracy of the representation of the atmospheric state. Accordingly, there are differences in turbulent heat and freshwater fluxes between various current reanalyses (Trenberth et al. 2011) as well as estimates obtained from the ocean state (Cerovečki et al. 2011). Based on a comparison of aircraft measurements conducted during the Greenland Flow Distortion Experiment in cold air outbreaks over Denmark Strait and Irminger Sea and operational analyses from the ECMWF, Renfrew et al. (2009) concluded that sensible and latent heat fluxes in the ECMWF model1 are within the range of observational uncertainty. Furthermore, Lindsay et al. (2014) found that in the Arctic, ERA-Interim has high correlations and small biases with respect to observations for 2-m air temperature, wind speed, and surface fluxes and stands out in terms of precipitation. While, according to their study, other reanalyses achieve better scores for some specific variables, they are not consistently good across all variables. Although observational coverage is better in the Arctic, we anticipate that the good performance of ERA-Interim translates also to the more remote South Pacific, an assumption supported by the high accuracy of ERA-Interim in reproducing independent measurements of SLP taken by buoys in the Bellingshausen Sea (Bracegirdle 2012). This makes ERA-Interim the best-suited foundation for the purpose of this climatological study. In addition to the ERA-Interim dataset, climatological monthly sea ice concentrations derived from passive microwave brightness temperature at 25-km resolution (Comiso 2012)2 are used to delineate the sea ice boundary in the figures.

Extratropical cyclones are identified from minima in SLP with a slightly updated version of the algorithm of Wernli and Schwierz (2006). For every cyclone a mask representing its area is determined by a contour search algorithm that identifies the outermost closed contour enclosing one or several SLP minima. Temporal averaging of these binary cyclone fields, which have a value of 1 inside and 0 outside of the mask, yields cyclone frequencies.

a. Identification of cold air outbreaks

CAOs are identified from the air–sea potential temperature difference θSKTθ850, where θ850 denotes potential temperature at 850 hPa and θSKT is potential skin temperature. For the calculation of potential temperature a reference pressure of 1000 hPa is used. The use of potential skin temperature instead of skin temperature accounts for the fact that in the area under consideration the frequent passage of extratropical cyclones leads to large variations of SLP. All grid points where the air–sea potential temperature difference exceeds a threshold θthresh, namely θSKTθ850θthresh, receive a flag (with a value of 1) indicating a CAO at this grid point. Grid points over land are excluded from the analysis. As the skin temperature over sea ice is very low and therefore the potential temperature difference is always well below the threshold, no special treatment of grid cells covered by sea ice is necessary. For each 6-hourly time step the grid points flagged as CAO are then clustered by spatial connectivity to obtain individual masks for each CAO. All CAO masks are kept for further analysis, irrespective of their size. The thresholds θthresh ∈ {0, 2, 4, 6, 8} K are used, leading to five intensity classes of CAOs. Throughout the paper we use the convention that results refer to CAOs using a threshold of 4 K unless stated differently. One of the key quantities considered in this study is the frequency of CAOs. As for the cyclone frequency, it denotes the fraction of time at which the CAO criterion is satisfied at a certain grid point and it is obtained from temporally averaging the CAO masks.

Finally, from each CAO grid point, 4-day kinematic backward trajectories starting at a height of 850 hPa are calculated using the Lagrangian Analysis Tool (LAGRANTO; Wernli and Davies 1997). The trajectories are calculated using the three-dimensional winds on model levels and the position of the trajectories is written out every 6 h. Meteorological parameters such as pressure, potential temperature, and specific humidity are traced along the trajectories by interpolation of these fields to the respective position of the trajectories. A gridding approach with linearly decreasing weights within a radius of 111 km is used to derive trajectory density maps.

Similar indices based on the air–sea potential temperature difference were used to identify CAOs in previous studies focusing on polar low development (e.g., Bracegirdle and Gray 2008; Kolstad 2011) and the difference of temperature at 500 hPa and skin temperature is operationally used by the Norwegian Meteorological Institute to forecast regions where polar lows are likely to occur. Although similar, there are differences in the precise definition of the indices so far used in the literature and our index. The index used by Kolstad (2011) is a measure for vertical static stability, whereas in contrast θSKTθ850 is intimately related to turbulent heat fluxes and thus better suited for the purpose of this study. In particular large θSKTθ850 will be associated with intense fluxes. According to Bracegirdle and Gray (2008), it can be advantageous to choose wet-bulb potential temperature instead of dry potential temperature, since this also accounts for the dryness of the air in CAOs. However, we did not find significant differences between CAOs identified with dry, equivalent, or wet-bulb potential temperature in the region considered in this study. The results are more sensitive to the choice of the level at which the air potential temperature is determined. For studying the development of polar lows typically the 700-hPa level is chosen (e.g., Bracegirdle and Kolstad 2010) because it is well above the boundary layer. However, the cold air masses advected across the sea ice boundary are often shallow and restricted to the boundary layer and as such they rarely extend to levels above 800 hPa. As shown below, they lead to intense turbulent heat fluxes when advected over open ocean despite their shallowness, but remained undetected if θ700 was used. As one of the purposes of this study is to quantify the turbulent heat flux associated with CAOs we use θ850.

b. Attribution of cold air outbreaks to extratropical cyclones

The CAOs are attributed to extratropical cyclones using an overlap criterion similar to Papritz et al. (2014). To this end the masks obtained from the clustering by connectivity of the CAO grid points and the cyclone masks, defined as the grid points within the outermost closed SLP contour, are used. A particular CAO is attributed to an extratropical cyclone if the two masks overlap by at least one grid point. To account for cases when the outermost closed SLP contour does not cover the entire area affected by the cyclonic circulation and consequently northward advection of cold air occurs to the west of the cyclone mask, the attribution to cyclones is also done with CAO masks that are expanded to the east by 2° before checking for an overlap. The convergence of the meridians toward the pole potentially introduces a latitude-dependent bias in the attributed fraction using the expanded masks, which by manual inspection of a sequence of cases was found not to significantly affect our findings.

c. An example episode

Figure 3 shows an episode of cyclone-induced CAOs in the period from 24 to 27 June 2010. At 0000 UTC 24 June, a long-lived, deep multiple cyclone system (SLP below 960 hPa) is located over the Bellingshausen Sea close to the Drake passage (Fig. 3a), which gradually fills up and subsequently decays within a day. A second cyclone (cyclone C; SLP below 950 hPa) had genesis two days before in the right jet exit off the coast of Oates Land. Ahead of an upper-level trough (not shown) it quickly deepens and propagates into the eastern RS with its center located over the sea ice at 0000 UTC 24 June. Southerly winds on the western side of the cyclones advect air from the sea ice over the open ocean, a flow that in the case of the decaying system in the Bellingshausen Sea extends into the midlatitudes. The air advected by this southerly flow is significantly colder than the sea, as is evident from θSKTθ850 (shading in Figs. 3a–d). Collocated with these CAOs is a region of strongly positive (upward directed) turbulent heat flux with values above 200 W m−2 (Fig. 3e), indicating a strong heating of the atmosphere from below. Within two days cyclone C intensifies and propagates into the Amundsen and subsequently into the Bellingshausen Sea (Figs. 3b,c). The persistence and increasing strength of the cyclonic circulation contributes to the northward advection of more and more air from the large, rebuilding cold air pool residing over the sea ice in the Ross and Amundsen Seas. On 26 June close to the edge of the sea ice (Fig. 3g), and one day later in the center of the CAO (Fig. 3h), the turbulent heat flux reaches peak values of 500 W m−2. The CAO attains its maximum extent at 0000 UTC 27 June, when cyclone C becomes stationary in the ABS and continuously exports cold air over the ocean (Figs. 3d,h).

Fig. 3.

Episode of CAOs in the period 24–27 Jun 2010. Colors show (a)–(d) θSKTθ850 in units of K and (e)–(h) 6-hourly averaged (±3 h) turbulent heat flux (sensible and latent, upward positive) in units of W m−2. Contours of θSKTθ850 at 0, 2, 4, 6 and 8 K are shown in gray, with the 4-K contour highlighted in bold. In (a)–(d) SLP in intervals of 10 hPa is also shown (thin black). The 50% isoline of ERA-Interim fractional sea ice cover is indicated by the solid black line.

Fig. 3.

Episode of CAOs in the period 24–27 Jun 2010. Colors show (a)–(d) θSKTθ850 in units of K and (e)–(h) 6-hourly averaged (±3 h) turbulent heat flux (sensible and latent, upward positive) in units of W m−2. Contours of θSKTθ850 at 0, 2, 4, 6 and 8 K are shown in gray, with the 4-K contour highlighted in bold. In (a)–(d) SLP in intervals of 10 hPa is also shown (thin black). The 50% isoline of ERA-Interim fractional sea ice cover is indicated by the solid black line.

An interesting feature is the development of a sequence of three mesocyclones (labeled M1–M3 in Figs. 3b–d) associated with the CAO of cyclone C in the region where the cold air impinges on warmer air masses and forms a strong baroclinic zone. The interplay of the strong turbulent heat flux from the ocean in CAOs and the strong baroclinicity at the boundary of the outbreaks provides an environment in which mesoscale cyclones and polar lows form preferentially (e.g., Bracegirdle and Gray 2008; Linders and Saetra 2010). These mesoscale cyclones show both convective and baroclinic cloud structures, as can be seen from the Advanced Very High Resolution Radiometer (AVHRR) infrared satellite images (Fig. 4), which are typical for mesoscale systems developing in association with CAOs (e.g., Rasmussen and Turner 2003). The mesoscale systems themselves can act to amplify the CAO by creating secondary tongues of cold air (e.g., M1 on 25 June; Fig. 3b) or intensifying the northward advection (e.g., M2 on 26 June and M3 on 27 June; Figs. 3c,d). These tongues of cold air appear as streets of cumuliform clouds in the satellite images (Fig. 4).

Fig. 4.

Composition of AVHRR infrared (channel 4) satellite images from the National Oceanic and Atmospheric Administration (NOAA) at (left) 0009 and (right) 0137 UTC 26 Jun 2010. Where the images overlap only the earlier image is plotted. Overlaid in red is ERA-Interim SLP at 0000 UTC 26 Jun 2010 in intervals of 10 hPa.

Fig. 4.

Composition of AVHRR infrared (channel 4) satellite images from the National Oceanic and Atmospheric Administration (NOAA) at (left) 0009 and (right) 0137 UTC 26 Jun 2010. Where the images overlap only the earlier image is plotted. Overlaid in red is ERA-Interim SLP at 0000 UTC 26 Jun 2010 in intervals of 10 hPa.

The SLP minima in ERA-Interim are in close proximity to the cloud structures in the satellite images. This impressively demonstrates that despite the sparse observations in the study region and the limited resolution of the model, even many of these mesoscale systems are represented in ERA-Interim.

3. Cold air outbreaks

a. Climatology

The 4-K isoline of θSKTθ850 gives a meaningful estimate of the shape of CAOs and it encloses the areas of most intense fluxes, as evident from the example episode considered above, where in Fig. 3 the 4-K isoline is highlighted. The threshold θthresh = 4 K will be used in the following as a reference for the spatial distribution and seasonal cycle of the frequency of CAOs. In Figs. 5a–d the spatial distributions of CAO frequency in the four seasons December–February (DJF), March–May (MAM), June–August (JJA), and September–November (SON) are shown. Two distinct maxima can be observed from autumn to spring: one in the western RS and a second, slightly weaker maximum in the ABS. Furthermore, Fig. 5e shows monthly CAO frequencies averaged over the RS and the ABS boxes as indicated in Fig. 5c.

Fig. 5.

(a)–(d) Seasonal frequency of CAOs for θthresh = 4 K. The sea ice boundary (50% sea ice concentration) derived from passive microwave brightness temperature is indicated by the solid black line. Also shown are (e) monthly area averaged CAO frequencies for θthresh = 4 K and (f) winter CAO frequencies for thresholds θthresh ∈ {0, 2, 4, 6, 8} K. The RS (gray) and the ABS (black) target boxes, as well as the box used for the selection of CAO trajectories passing the Ross Ice Shelf corridor (orange), are shown in (c).

Fig. 5.

(a)–(d) Seasonal frequency of CAOs for θthresh = 4 K. The sea ice boundary (50% sea ice concentration) derived from passive microwave brightness temperature is indicated by the solid black line. Also shown are (e) monthly area averaged CAO frequencies for θthresh = 4 K and (f) winter CAO frequencies for thresholds θthresh ∈ {0, 2, 4, 6, 8} K. The RS (gray) and the ABS (black) target boxes, as well as the box used for the selection of CAO trajectories passing the Ross Ice Shelf corridor (orange), are shown in (c).

In summer the frequency is close to zero almost everywhere except for the waters off the Ross Ice Shelf. In March the CAOs off the Ross Ice Shelf benefit from the considerably more frequent strong wind events occurring along the Transantarctic Mountain range (Chenoli et al. 2013). One major factor contributing to these events is the more frequent presence of cyclones beginning in March in the southeastern RS (see Simmonds and King 2004, and section 5), which amplify the drainage of Antarctic cold air through the Ross Ice Shelf corridor (Parish and Bromwich 1998). Indeed, as quantified later in section 5, the majority of CAOs occur in association with the flow induced by cyclones. In autumn sea ice extent grows and the air over the Ross Ice Shelf and the Antarctic continent is more strongly cooled, leading to a growing pool of cold air. As a consequence the maximum in the frequency of CAOs strengthens and extends farther north. In April and May, the CAO frequency increases rapidly in the RS and the maximum of around 15% is shifted to the north between 60° and 68°S. Along with the growing sea ice extent, CAOs also start to occur in the ABS. The maximum frequency is reached in June in the RS, when the frequency of RS cyclones and the net surface radiative cooling over the ice shelf and the Antarctic continent are both maximum. In the winter mean, a CAO is present during more than 21% of the time immediately off the sea ice edge. Until July, CAOs in the ABS are less frequent and the seasonal peak is shifted to August, but is weaker than the June peak in the RS (Fig. 5e). During late winter and spring, the frequency slowly decreases, whereby clear local maxima in both regions persist. Afterward, the frequency decreases rapidly in early summer in conjunction with the retreat of sea ice. It is worth noting that many of the wintertime CAOs also affect midlatitudes as indicated by the 6% contour reaching 50°S (Fig. 5c).

The dependence of the average wintertime frequency of CAOs on the choice of θthresh is shown in Fig. 5f. Interestingly very weak CAOs (θthresh = 0 K) are slightly more frequent in the ABS than in the RS, while more intense CAOs (θthresh > 0 K) occur more often in the RS sector.

b. Origin of air masses

From the 4-day backward trajectories calculated from all CAO grid points, probability maps of the trajectory positions at times t ∈ {−72, −48, −24, −12} h before the CAO are derived (Fig. 6). More precisely, the value at a certain location on these CAO origin maps represents the likelihood for a trajectory to be at this location at time t, given that at time t = 0 h it satisfies the CAO identification criterion. The likelihood is weighted by the area, such that the integral over the globe amounts to 100%. In the following, only trajectories are considered that at time t = 0 h are located in the RS or the ABS box, respectively, and the trajectory analysis is restricted to the winter months (JJA) when CAOs are most frequent. Unless stated otherwise, the θSKTθ850 ≥ 4 K criterion is used.

Fig. 6.

CAO origin maps showing probabilities for trajectories producing a CAO (θthresh = 4 K) at time t = 0 h to be at a certain location at t ∈ {−72, −48, −24, −12} h in units of ‰ (105 km2)−1. Probabilities are shown for CAO trajectories that at t = 0 h are in the (a)–(d) RS (gray target box) or (e)–(h) ABS (black target box) and for winter (JJA) only. Gray contours show wintertime CAO frequency from 5% in intervals of 5% and the black contour indicates the sea ice boundary (50% sea ice concentration).

Fig. 6.

CAO origin maps showing probabilities for trajectories producing a CAO (θthresh = 4 K) at time t = 0 h to be at a certain location at t ∈ {−72, −48, −24, −12} h in units of ‰ (105 km2)−1. Probabilities are shown for CAO trajectories that at t = 0 h are in the (a)–(d) RS (gray target box) or (e)–(h) ABS (black target box) and for winter (JJA) only. Gray contours show wintertime CAO frequency from 5% in intervals of 5% and the black contour indicates the sea ice boundary (50% sea ice concentration).

The CAO origin maps for the RS (Figs. 6a–d) show that three days before the CAO (Fig. 6a) the majority of the trajectories are located above the sea ice and over the Antarctic continent, most pronouncedly over the Ross Ice Shelf associated with the RAS from where they subsequently flow into the RS basin. This is in agreement with previous studies highlighting the Ross Ice Shelf corridor as a major pathway for air masses to leave Antarctica (e.g., Parish and Bromwich 1998; Parish et al. 2006; Parish and Bromwich 2007). Some trajectories are located over West Antarctica, from where they flow into the Ross Ice Shelf corridor too. Subsequently, these trajectories move east of the Transantarctic Mountains, and their spatial distribution narrows considerably, before fanning out when they reach the sea ice boundary. It must be noted that the resolution of ERA-Interim is not sufficiently high to properly resolve individual contributions from the different drainage pathways in the Transantarctic Mountain Range such as Byrd Glacier. This might introduce an underestimation of the contribution of descending trajectories from Antarctica to CAOs, as the channeling of the flow by these valleys contributes substantially to the drainage from Antarctica (e.g., Parish and Bromwich 2007). A number of trajectories also originate from off the coast of Oates Land and Wilkes Land and to a lesser extent from the coastal areas in the Amundsen Sea.

Surprisingly, a large fraction of the trajectories leading to a CAO in the ABS (Figs. 6e–h) originate also from the southern RS or descends via the Ross Ice Shelf corridor, while smaller fractions descend from Ellsworth Land in West Antarctica or originate from the sea ice region in the ABS. Two factors are responsible for the predominance of the RS as the origin of ABS CAO trajectories. First, the absence of a large gap in the topography such as the Ross Ice Shelf corridor prevents the efficient drainage of cold Antarctic air masses into the ABS. Second, the area covered by sea ice between the open ocean and the coast is narrower than in the RS, which leads to a shorter residence time of trajectories above the sea ice and accordingly a weaker potential of the air mass to cool sufficiently in order to produce a CAO. This results in a smaller cold air pool and a relatively larger number of weak CAOs (θthresh = 0 K) in the ABS as compared to the RS (Fig. 5f).

As quantified below, most of the CAOs occur in association with extratropical cyclones. Thus, we hypothesize that trajectories descending via the Ross Ice Shelf corridor are deflected toward the ABS when a cyclone becomes stationary in the eastern RS very close to the Antarctic coast. To support this, we consider those CAO trajectories in the RS and the ABS that are located in the Ross Ice Shelf corridor during at least one 6-hourly time step in the four days before the CAO (orange box in Fig. 5c). These subsets comprise 38% of the RS and 35% of the ABS CAO trajectories. Already four days before the CAO the median relative vorticity is by a factor of 1.5 more negative (cyclonic) for the ABS compared to the RS trajectories (Fig. 7a). The median relative vorticity peaks two days later at a value of −1.9 × 10−5 s−1 when the ABS trajectories are over the sea ice–covered part of the RS, while it remains fairly constant for RS trajectories. Also acceleration of ABS trajectories sets in almost one day earlier and more vigorously, such that they are advected more rapidly than RS trajectories (Fig. 7b), leading to a 10% longer transport distance of the ABS trajectories (4340 km for ABS and 3960 km for RS trajectories during the considered four days). Because of the fanning out of the trajectories during the last day before the CAO (Figs. 6d,h), relative vorticity increases (becomes less cyclonic) in both subsets.

Fig. 7.

Temporal evolution of (a) the median relative vorticity (10−5 s−1) and (b) wind speed (m s−1) along CAO trajectories (θthresh = 4 K) in the RS (red) and the ABS (blue) descending from Antarctica via the Ross Ice Shelf corridor. The trajectories are required to be in the Ross Ice Shelf corridor (orange box in Fig. 5c) at least at one 6-hourly time step before the CAO.

Fig. 7.

Temporal evolution of (a) the median relative vorticity (10−5 s−1) and (b) wind speed (m s−1) along CAO trajectories (θthresh = 4 K) in the RS (red) and the ABS (blue) descending from Antarctica via the Ross Ice Shelf corridor. The trajectories are required to be in the Ross Ice Shelf corridor (orange box in Fig. 5c) at least at one 6-hourly time step before the CAO.

The evolution of relative vorticity along CAO trajectories in the ABS passing the Ross Ice Shelf corridor indicates that the large-scale flow during such events is predominantly cyclonic. These conditions typically occur in the presence of a stationary (and ultimately decaying) cyclone in the eastern RS. The eastern RS is well known for its high frequency of stationary cyclones undergoing cyclolysis (Simmonds et al. 2003; Wernli and Schwierz 2006).

c. Lagrangian characteristics of CAO air masses

To identify trajectories descending substantially with the downslope flow from the Antarctic continent, trajectories are selected that descend by at least 150 hPa within the four days before creating the CAO. This corresponds to a descent of about 1500 m. The so selected trajectories are called descending trajectories, while all others, which predominantly originate from the area covered by sea ice and to a small extent also from lower elevations on the Antarctic continent, are called nondescending. Note that the descent of trajectories in the first category is not required to be density driven, but instead can also be forced by pressure gradients imposed by the large-scale flow. In fact, as indicated by idealized simulations, the southward pressure gradient off the Antarctic coast may be the dominant mechanism responsible for downslope winds on the Antarctic continent (Parish and Cassano 2003a,b). Furthermore, trajectories in this analysis are not required to pass the Ross Ice Shelf corridor.

About 48% of the RS trajectories for CAOs with θthresh = 4 K descend from Antarctica, while this applies to merely 40% of the ABS trajectories. For very strong CAOs with θthresh = 8 K, the fraction of descending trajectories is reduced to 42% and 32% in the RS and the ABS, respectively. In the following, results are shown only for RS trajectories and θthresh = 4 K. A very similar evolution of thermodynamic and dynamic quantities, however, is also found for CAO trajectories in the ABS.

Descending trajectories have very different temperature (T) and potential temperature (θ) characteristics than nondescending ones (Fig. 8). Four days before the CAO, descending compared to nondescending trajectories are located 8° latitude closer to the South Pole in the median (not shown). Initially surface radiative cooling compensates for the adiabatic warming causing the temperature to remain almost constant. During descent in the three days before the CAO, trajectories remain exposed to radiative cooling at a fairly constant rate, which, however, is overcompensated by adiabatic warming. This results in a temperature increase of slightly more than 12 K. A similar evolution of θ and T was also observed for radiatively cooled polar air masses responsible for cold temperature extremes over Europe (Bieli et al. 2014). During the last 12 h some of the trajectories are heated by sensible heat fluxes from below as they pass over open ocean, which, in combination with further descent, leads to an additional increase of temperature by 4 K. In contrast, adiabatic warming is almost absent along nondescending trajectories and therefore the θT characteristics are diabatic within the first three days (i.e., parallel to the gray line in Fig. 8). The heating from below by the sensible heat flux within the last 12 h is more important for these trajectories, as they are on average closer to the surface than descending trajectories.

Fig. 8.

Shown are θT diagrams for descending (blue) and nondescending (red) CAO trajectories (θthresh = 4 K) in the RS in winter (JJA). The descending trajectories originate from the interior of the Antarctic continent. Each dot represents the median of θ and T differences with respect to t = 0 h at 12-hourly time steps, beginning at t = −96 h. The time step t = 0 h is plotted in black. The gray line represents the change of T at 850 hPa associated with a certain diabatic change of θ.

Fig. 8.

Shown are θT diagrams for descending (blue) and nondescending (red) CAO trajectories (θthresh = 4 K) in the RS in winter (JJA). The descending trajectories originate from the interior of the Antarctic continent. Each dot represents the median of θ and T differences with respect to t = 0 h at 12-hourly time steps, beginning at t = −96 h. The time step t = 0 h is plotted in black. The gray line represents the change of T at 850 hPa associated with a certain diabatic change of θ.

A property strongly distinguishing descending from nondescending trajectories is their low specific humidity at the beginning (median value of 0.2 g kg−1 compared to 0.7 g kg−1) and the narrow distribution of specific humidity around the median, which in contrast is much wider for nondescending trajectories (Fig. 9a). The latter reflects the variety of possible synoptic conditions at their origin, while in contrast most descending trajectories have a similar history. In both classes specific humidity remains fairly constant within the first three days, while median relative humidity (Fig. 9b) of descending trajectories drops from initially 60% to 40% as they descend and warm. During the last 24 h descending trajectories mix with the surrounding more humid air, whereby the median relative humidity increases to the value of nondescending trajectories (70%) at t = 0 h. The dryness of descending trajectories has an amplifying effect on the surface latent heat flux, which at t = −12 h is 93 W m−2 in the median for descending and only 68 W m−2 for nondescending trajectories (not shown), leading to a larger increase of specific humidity in the former category. Thus, the difference in the latent heat flux between the two categories of trajectories decreases quickly as the difference in specific humidity reduces and at t = 0 h is less than 10 W m−2 in the median.

Fig. 9.

Temporal evolution of (a) specific (g kg−1) and (b) relative humidity (%), (c) potential vorticity (PVU), and (d) static stability (s−2) along wintertime (JJA) CAO trajectories (θthresh = 4 K) in the RS divided into descending (blue) and nondescending (red) trajectories. The descending trajectories originate from the interior of the Antarctic continent. The solid lines show the median of each category and the shading represents the interquartile range.

Fig. 9.

Temporal evolution of (a) specific (g kg−1) and (b) relative humidity (%), (c) potential vorticity (PVU), and (d) static stability (s−2) along wintertime (JJA) CAO trajectories (θthresh = 4 K) in the RS divided into descending (blue) and nondescending (red) trajectories. The descending trajectories originate from the interior of the Antarctic continent. The solid lines show the median of each category and the shading represents the interquartile range.

Descending air masses from Antarctica have been shown to be favorable for the genesis of mesocyclones because of their often strongly cyclonic PV, which can lead to the generation of cyclonic vorticity (Bromwich 1991; Klein and Heinemann 2002). These air masses are stably stratified as long as they are above an ice-covered surface, which impedes the spinup of mesocyclones, and some additional external forcing associated with vertical stretching is required for mesocyclogenesis to occur (Carrasco and Bromwich 1994; Gallée 1995). When the air masses are advected over open ocean and ultimately become a CAO, they are strongly destabilized by the sensible heat flux from below, while at the same time PV is no longer conserved and the strongly cyclonic PV values may quickly be reduced.

Figure 9c shows the temporal evolution of PV along CAO trajectories. Initial median PV values in CAO trajectories are more negative (cyclonic) compared to typical lower tropospheric values. The lower Coriolis parameter at the more southerly origin of descending trajectories is the major reason for their lower PV compared to nondescending trajectories. During the first three days averaged PV remains well conserved in both categories. However, this changes abruptly after −24 h, when static stability quickly drops (Fig. 9d), presumably due to heating by intense sensible heat fluxes, and consequently PV increases (Fig. 9c). At t = 0 h, PV values amount to approximately −0.5 PVU (1 PVU = 10−6 kg−1 K m2 s−1) in both categories, which is close to climatological PV values. Therefore, descending trajectories do not have systematically more negative PV than nondescending ones; when they become a CAO, their initial more negative PV with respect to the climatology is rapidly damped.

However, the distribution of PV around the median is wide and 0.7% of the descending and 0.9% of the nondescending trajectories have PV values of less than −2 PVU at t = 0 h. Before the CAO occurred, these trajectories were exposed to more intense cooling and associated stabilization (not shown), which lowered their PV. It is noteworthy that at t = 0 h their relative vorticity is clearly cyclonic (−5.5 × 10−5 s−1) and could be a sign of incipient mesocyclogenesis. Since no systematic differences between descending and nondescending CAO trajectories were found in terms of PV and vorticity, we conclude that both types are equally likely to initiate mesocyclogenesis.

4. Relevance of cold air outbreaks for turbulent heat flux

a. CAOs and total heat flux

The purpose of this section is to quantify the contribution of CAOs to turbulent heat fluxes from the ocean to the atmosphere and to study the characteristics of these fluxes in CAOs. Sensible and latent heat fluxes in CAOs are considerably stronger compared to the climatological mean over open ocean in both the RS and the ABS (Fig. 10). The stronger the CAO (i.e., the larger the value of θSKTθ850), the more intense the turbulent heat flux. It is important to note that the 10th–90th percentile range of the fluxes is surprisingly narrow. This implies that the strength of a CAO is the principal factor influencing the intensity of the turbulent heat flux in CAOs.

Fig. 10.

Wintertime (a) sensible and (b) latent heat fluxes in CAOs as a function of θSKTθ850 for CAOs in the RS and the ABS regions. Median, interquartile range, and the 10th–90th percentile range are shown by a solid black line, dark gray shading, and light gray shading, respectively. In addition the climatological winter sensible and latent heat fluxes over open ocean in the RS and the ABS are indicated by red crosses. Heat flux distribution functions expressing the contributions of CAOs to sensible and latent fluxes as a function of θSKTθ850 are shown in blue (right axis). They are normalized such that the integral over the entire θSKTθ850 range yields the winter heat flux associated with CAOs.

Fig. 10.

Wintertime (a) sensible and (b) latent heat fluxes in CAOs as a function of θSKTθ850 for CAOs in the RS and the ABS regions. Median, interquartile range, and the 10th–90th percentile range are shown by a solid black line, dark gray shading, and light gray shading, respectively. In addition the climatological winter sensible and latent heat fluxes over open ocean in the RS and the ABS are indicated by red crosses. Heat flux distribution functions expressing the contributions of CAOs to sensible and latent fluxes as a function of θSKTθ850 are shown in blue (right axis). They are normalized such that the integral over the entire θSKTθ850 range yields the winter heat flux associated with CAOs.

Sensible and latent heat fluxes have different scaling characteristics with respect to the strength of the CAO. The sensible heat flux scales approximately linearly with θSKTθ850 (Fig. 10a), while the increase of the latent heat flux becomes weaker for larger θSKTθ850 and the latent heat flux does not exceed values of 200 W m−2 (Fig. 10b). For a given wind speed and stability, the latent heat flux is proportional to , where and are saturation specific humidity at the surface and in the atmosphere, respectively, and h is relative humidity. The latent heat flux is essentially limited by the saturation specific humidity with respect to the SST, . According to the Clausius–Clapeyron relationship saturation specific humidity decreases exponentially with temperature, and therefore is larger for lower air temperatures (i.e., for stronger CAOs), but the latent heat flux saturates toward the limit imposed by . In contrast, the sensible heat flux is proportional to the air–sea temperature difference leading to a linear scaling with CAO strength. Consequently for weak CAOs the latent heat flux exceeds the sensible heat flux, whereas with increasing strength of CAOs the relative importance of the sensible heat flux increases and ultimately for the most intense CAOs it strongly exceeds the latent heat flux. It is interesting to note, that for weak CAOs a small increase in their strength reduces stability, which intensifies turbulent mixing. This results in a weakly nonlinear change of the sensible heat flux for θSKTθ850 < 4 K.

The winter heat flux contribution of CAOs of a certain strength is expressed by the CAO heat flux distribution functions fXθ) (blue curves in Fig. 10), where X denotes sensible or latent heat fluxes, respectively, and Δθ = θSKTθ850 is the CAO strength. The distribution functions are normalized such that integration over the intensity range of CAOs yields the mean sensible or latent heat flux associated with CAOs ():

 
formula

Because of the skewness of the frequency distribution of CAOs, weak CAOs have the largest contributions to winter sensible and latent heat fluxes, despite the fact that fluxes in these CAOs are relatively weak. Because of the nonlinear scaling of the latent heat flux the contribution reduces more rapidly for the latent than for the sensible heat flux with increasing strength of the CAO. In winter in the ABS (RS) region, CAOs with θSKTθ850 ≥ 4 K contribute 28% (37%) to the total upward turbulent heat flux, that is, the total sensible and latent heat flux transferred from the ocean into the atmosphere (Fig. 11). This fraction is much higher than the area average frequency of CAOs of 6.7% (7.9%; see Fig. 5f). Furthermore, the strongest CAOs alone (θSKTθ850 ≥ 8 K) contribute 13% (18%). A comparison with their frequency (1.6% and 2.0%, respectively) shows that such strong CAOs are still highly relevant for the mean winter turbulent heat flux. Contributions in autumn and spring are slightly below those in winter and larger in spring than in autumn, which is consistent with the seasonality of CAOs.

Fig. 11.

Percentage of the total upward turbulent heat flux that occurs in CAOs in the ABS (a) and the RS stratified by CAO thresholds θthresh ∈ {0, 2, 4, 6, 8} K and seasons. Grid points where the fractional sea ice cover exceeds 50% were excluded.

Fig. 11.

Percentage of the total upward turbulent heat flux that occurs in CAOs in the ABS (a) and the RS stratified by CAO thresholds θthresh ∈ {0, 2, 4, 6, 8} K and seasons. Grid points where the fractional sea ice cover exceeds 50% were excluded.

b. CAOs and heat flux extremes

The intense turbulent heat flux occurring in strong CAOs and their significant contribution to the climatological turbulent heat flux indicates the relevance of CAOs for extreme heat fluxes. Here we define extremes of sensible or latent heat fluxes as events for which the respective heat flux is above its yearly 95th percentile at this particular location. Note that parts of the regions considered here are covered by sea ice during many months throughout the year and that these time steps are included when computing the 95th percentile.

The 95th percentile of the sensible heat flux shows two peaks collocated with the maxima of winter CAO frequency (Fig. 12a). Within both the RS and the ABS, between 70% and 90% of the extreme events are associated with CAOs. Thus, CAOs are the major cause of events of extreme sensible heat flux in regions where they occur sufficiently frequently and thereby they also shape the geographical distribution of the intensity of the extremes. CAOs have a smaller, but still substantial, influence on the geographical distribution of the intensity of latent heat flux extremes, which is superposed by the equatorward increase of sea surface temperature (Fig. 12b). Nevertheless, at the location of the maxima of CAO frequency, a lower when compared to the sensible heat flux but still large fraction of latent heat flux extremes occurs in association with CAOs (up to 80% in the RS).

Fig. 12.

Fraction of yearly events of extreme (above yearly 95th percentile) (a) sensible and (b) latent heat fluxes associated with CAOs for θthresh = 4 K (color). The 95th percentile of the respective climatological heat flux is shown by black contours in intervals of 25 W m−2.

Fig. 12.

Fraction of yearly events of extreme (above yearly 95th percentile) (a) sensible and (b) latent heat fluxes associated with CAOs for θthresh = 4 K (color). The 95th percentile of the respective climatological heat flux is shown by black contours in intervals of 25 W m−2.

c. Interannual variability

Despite the relatively large contribution of CAOs to the upward turbulent heat flux, the question remains how strongly the interannual variability of seasonal CAO frequency is imprinted on the net seasonal turbulent heat flux, which also includes fluxes from the atmosphere into the ocean. How well can the net turbulent heat flux be estimated from the frequency of CAOs? Given the low CAO frequency in summer this analysis is restricted to autumn to spring. In Fig. 13, the net turbulent heat flux for all autumn, winter, and spring seasons in the study period (1980 to 2010) is plotted against the frequency of CAOs separately for both basins and for CAOs with thresholds θthresh ∈ {2, 4, 6} K. It is apparent that the seasonal mean net turbulent heat flux in both regions scales almost linearly with the seasonal CAO frequency, not only in winter, but also in autumn and spring. Also the variability of the CAO frequency is large and wintertime variations often exceed 50% of the climatological mean, similar to the net turbulent heat flux (Fig. 2).

Fig. 13.

Seasonal turbulent heat flux plotted against the frequency of CAOs for thresholds θthresh ∈ {2, 4, 6} K for the (a) ABS and (b) RS. Each symbol corresponds to one specific season in the period 1980–2010 and winter values are highlighted in blue. Grid points where the fractional sea ice cover exceeds 50% were excluded. Black lines denote linear regressions for each CAO class and include autumn, winter and spring. The R2 values for separate regressions for each season are listed in Table 1.

Fig. 13.

Seasonal turbulent heat flux plotted against the frequency of CAOs for thresholds θthresh ∈ {2, 4, 6} K for the (a) ABS and (b) RS. Each symbol corresponds to one specific season in the period 1980–2010 and winter values are highlighted in blue. Grid points where the fractional sea ice cover exceeds 50% were excluded. Black lines denote linear regressions for each CAO class and include autumn, winter and spring. The R2 values for separate regressions for each season are listed in Table 1.

Linear regressions of the seasonal turbulent heat flux on seasonal CAO frequency show the highest R2 values for CAOs with θthresh = 2 K in winter and spring (Table 1) and the values are higher in the ABS than in the RS. Even though all strong CAOs are also included in the weakest intensity class, the correlation decreases when CAOs with θSKTθ850 < 2 K are included, since they contribute strongly to the frequency but their contribution to the mean flux is below that of CAOs with θSKTθ850 ≈ 2K (Fig. 10). However, in autumn when strong CAOs are still rare, mostly due to the limited sea ice extent and less efficient radiative cooling than in winter, the limiting case of θthresh = 0 K is the best predictor of the seasonal turbulent heat flux.

Table 1.

R2 values for linear regressions of autumn to spring seasonal mean turbulent heat flux on CAO frequency in the RS and the ABS sectors. Maxima in each column are highlighted in bold.

R2 values for linear regressions of autumn to spring seasonal mean turbulent heat flux on CAO frequency in the RS and the ABS sectors. Maxima in each column are highlighted in bold.
R2 values for linear regressions of autumn to spring seasonal mean turbulent heat flux on CAO frequency in the RS and the ABS sectors. Maxima in each column are highlighted in bold.

5. Cyclone-induced cold air outbreaks

Because CAOs contribute significantly to the turbulent heat flux and its variability from autumn to spring, the question arises as to how their variability is linked to variations in the large-scale circulation. These can be characterized by changes in the frequency of extratropical cyclones. In addition, the differences in the seasonal cycle of CAOs in the two regions considered here calls for an explanation. Therefore, in this section the linkage between the frequency of extratropical cyclones and CAOs is investigated.

a. Annual cycle of cyclone frequencies

A map of the winter frequency of extratropical cyclones is presented in Fig. 14a. Wintertime cyclone frequency has a spatial distribution very similar to that of mean SLP (Turner et al. 2013), with a distinct maximum in the eastern RS and a band with relatively high frequencies extending into the ABS. The median monthly cyclone frequency shows different seasonal cycles in the RS and the ABS (Fig. 14b). Note that for the calculation of area average cyclone frequency, different boxes are used than for CAOs. The boxes are chosen such that they cover the area of highest cyclone frequencies within the RS and the ABS, respectively. In the RS, the cycle is semiannual with peaks in June and October, whereby the peak in June coincides with the highest frequency of CAOs in the western RS. In contrast, the cyclone frequency in the ABS is largest in January and is at a minimum in June, with the total wintertime frequency being below 25%. This reflects the shift of the Amundsen Sea low from the eastern ABS in summer to the west in winter (Turner et al. 2013). The relative interannual variability in winter is considerably larger in the ABS, where the interquartile range is 10%, which is 40% of the wintertime cyclone frequency. Not surprisingly this region was previously termed a “pole of variability” of the southern high latitudes (Connolley 1997). In comparison, cyclone frequency in the RS is more stable throughout the study period. The wintertime interquartile range is 8%, corresponding to 27% of the median cyclone frequency.

Fig. 14.

(a) Wintertime frequency of extratropical cyclones in %. (b) Median monthly cyclone frequency for the RS (gray) and the ABS region (black). Red and blue whiskers represent the interquartile range. Averages are taken over the RS (gray) and the ABS (black) boxes shown in (a). Note that these boxes differ from those used in the analysis of CAOs.

Fig. 14.

(a) Wintertime frequency of extratropical cyclones in %. (b) Median monthly cyclone frequency for the RS (gray) and the ABS region (black). Red and blue whiskers represent the interquartile range. Averages are taken over the RS (gray) and the ABS (black) boxes shown in (a). Note that these boxes differ from those used in the analysis of CAOs.

b. Cold air outbreaks attributed to extratropical cyclones

For each of the identified CAOs, it is determined if it is related to an extratropical cyclone or not. This is done according to the overlap criterion described in section 2. The frequency of CAOs attributed to extratropical cyclones is then compared to the total frequency of CAOs. In the ABS (RS) the attributed fraction is 77.8% (76.9%) and thus the majority of CAOs are related to extratropical cyclones. Because the circulation associated with extratropical cyclones typically extends outside of the outermost closed SLP contour used to delimit the cyclone area, it can happen that a CAO occurs outside the cyclone mask, but nevertheless is caused by the cyclone’s circulation. Even though these cases are rather rare, our method of attribution slightly underestimates the number of CAOs induced by extratropical cyclones. Consistent with this argument, the fraction of attributed CAOs increases by about 5% if the CAOs are enlarged by 2° to the east before attribution. The attributed fraction is very stable throughout the years.

Given that sea ice grows fastest in autumn and its extent increases at a smaller rate in the subsequent winter months, it is not surprising that the maximum of RS CAO frequency coincides with the cyclone frequency maximum in June and then reduces in July and August (Figs. 5e and 14b). In addition to the lower cyclone frequency, this reduction of CAO frequency may be linked to the fact that the cyclone frequency maximum is relatively far south, while the sea ice edge is located farther north, such that the advection of air masses over open ocean by the cyclone circulation might be less efficient. Cyclone frequency in the ABS is at a minimum in June and consequently the peak in the CAO frequency occurs in August, when cyclone frequency has sufficiently recovered and sea ice is most extended.

From the above considerations it cannot be concluded that interannual variations of the wintertime frequency of CAOs are driven by the frequency of extratropical cyclones alone. A critical factor for the formation of CAOs is the extent of sea ice, which is affected by many different processes. A strong coupling exists between low-level winds and sea ice drift (e.g., Holland and Kwok 2012), such that a larger cyclone frequency typically leads to an increased sea ice extent in the ABS or RS. Throughout the period from 1980 to 2010, sea ice extent in the RS has shown a significantly positive trend, while it decreased mostly in the eastern part of the ABS, consistent with a deepening of the climatological low in the ABS and an associated amplification of the mean cyclonic circulation (Turner et al. 2009; Simmonds 2014).

c. Turbulent heat flux anomalies associated with cyclone-induced cold air outbreaks

To investigate the influence of interannual anomalies of wintertime cyclone frequency on the frequency of CAOs and the turbulent heat flux, wintertime composites based on cyclone frequency in the RS and the ABS are calculated. To this end a cyclone frequency index I for each winter and region is defined by

 
formula

where f is the area averaged cyclone frequency of a particular winter and 〈f〉 its 1980 to 2010 average. The index is normalized by the standard deviation σ of the seasonal averages. Composites are then calculated for years when cyclones are particularly frequent (I > 0.5) or rare (I < −0.5) in the respective region. Each sample contains about one-third of the years in the study period.

The composite differences of CAO frequency, cyclone frequency, and turbulent heat flux are shown in Fig. 15. Winters with large cyclone frequency in the ABS are accompanied by an increase of the frequency of CAOs in the Amundsen Sea by more than 10% (Fig. 15a). This implies a strong signal in the turbulent heat flux, which is larger by more than 40 W m−2 in years with a high cyclone frequency (Fig. 15c). Interestingly, a slightly increased CAO frequency is also found in the Bellingshausen Sea immediately off the sea ice edge. Even though in stormy winters warm air advection from the north occurs more often in the Bellingshausen Sea, as then the mean circulation is more northerly, also more cyclones propagate into the very eastern parts of the Bellingshausen Sea. Associated northward advection of cold air on the western side of these systems can trigger additional CAOs. The weaker variability of RS cyclone frequency translates into a more modest increase of CAO frequency (Fig. 15b) and a smaller amplification of the turbulent heat flux (Fig. 15d) in winters with frequent cyclones in the RS. This suggests that the larger variability of the turbulent heat flux in the ABS as compared to the RS (see Fig. 2c) is caused by the different strength of interannual variations of cyclone frequency in the two regions.

Fig. 15.

Composite difference of CAO frequency with θthresh = 4 K for winters with I > 0.5 and winters with I < −0.5 in (a) the ABS and (b) RS regions. (c),(d) The composite differences of the turbulent heat flux are shown. Gray contours show the composite difference of cyclone frequency in intervals of 4% with positive values solid and negative values dashed. The sea ice boundary (50% sea ice concentration) is indicated by the solid black line.

Fig. 15.

Composite difference of CAO frequency with θthresh = 4 K for winters with I > 0.5 and winters with I < −0.5 in (a) the ABS and (b) RS regions. (c),(d) The composite differences of the turbulent heat flux are shown. Gray contours show the composite difference of cyclone frequency in intervals of 4% with positive values solid and negative values dashed. The sea ice boundary (50% sea ice concentration) is indicated by the solid black line.

The frequencies of cyclones in the RS and the ABS are not independent. An increase of cyclone frequency in the RS is accompanied by a decrease in the ABS of similar magnitude, which accordingly leads to the observed dipole patterns of CAO frequency and turbulent heat fluxes. In winters with frequent cyclones in the RS, fewer cyclones propagate into the Amundsen Sea but instead become stationary in the RS and decay there.

The cyclone-induced variations of CAO frequency have strong implications for freshwater fluxes. Over most regions of the Southern Ocean the variability of freshwater fluxes is due to variations of precipitation associated with cyclones and their fronts (Papritz et al. 2014). While sufficiently far from the sea ice edge in the South Pacific, cyclone precipitation exceeds evaporation by 0.4 to 0.7 mm day−1 (Papritz et al. 2014), evaporation in cyclone-induced CAOs compensates for much of the additional freshwater input due to cyclone precipitation close to the sea ice edge. This is underpinned by the composites of precipitation and evaporation based on cyclone frequency in the ABS (Fig. 16). In the Amundsen and the western Bellingshausen Seas the additional precipitation and evaporation associated with cyclone-induced CAOs in cyclone-rich winters are of a similar magnitude (between 0.6 and 1 mm day−1) and compensate each other, causing an anomalous freshwater flux close to neutrality. In the eastern Bellingshausen Sea and off the coast of South America, however, the influence of CAOs is weak and therefore the imprint of cyclones in the freshwater flux remains mostly determined by precipitation.

Fig. 16.

Composite difference of wintertime (JJA) (a) precipitation and (b) evaporation for winters with anomalously frequent storms and winters where storms are rare in the ABS. Gray contours show the composite difference of cyclone frequency in intervals of 4% with positive values solid and negative values dashed. The sea ice boundary (50% sea ice concentration) is indicated by the solid black line.

Fig. 16.

Composite difference of wintertime (JJA) (a) precipitation and (b) evaporation for winters with anomalously frequent storms and winters where storms are rare in the ABS. Gray contours show the composite difference of cyclone frequency in intervals of 4% with positive values solid and negative values dashed. The sea ice boundary (50% sea ice concentration) is indicated by the solid black line.

6. Conclusions

A novel climatology of CAOs in the RS and the ABS has been presented in this study. The contribution of CAOs to sensible and latent heat fluxes and their variability has been quantified. Based on our results, we conclude that cyclone-induced CAOs are an integral part of the freshwater and turbulent heat flux forcing of the high-latitude South Pacific. The key findings are as follows:

  1. Two distinct regions with frequent CAOs from autumn to spring are located in the RS and the ABS immediately off the sea ice edge, whereby the highest frequency is found in the RS. These two regions were previously identified by Kolstad (2011) as the locations in the Southern Hemisphere where CAOs are particularly frequent and the formation of polar lows is most likely to occur. In summer a significant CAO frequency is found only off the Ross Ice Shelf.

  2. A large fraction of the air masses leading to CAOs in the RS (48%) and the ABS (40%) originate from inside the Antarctic continent and descend substantially within the four days before the CAO. The Ross Ice Shelf corridor is the major pathway for these descending flows. Descending CAO air masses differ in their thermodynamic characteristics from nondescending ones by a much lower specific humidity, stronger diabatic cooling and much more intense adiabatic warming due to descent during the four days before a CAO. Both categories have a very similar PV evolution. The enhanced cyclonic PV values with respect to the climatology that are observed before the CAO quickly reduce during the incipient CAO due to the destabilizing effect of the sensible heat flux from the ocean.

  3. More than 75% of the CAOs are induced by the circulation associated with extratropical cyclones and thus the variability of the frequency of CAOs is strongly linked to the variability of the strength of the storm track (i.e., the cyclone frequency). The persistent cyclonic circulation associated with stationary cyclones in the eastern RS is the dominant mechanism by which the air masses descending from Antarctica are advected from the Ross Ice Shelf corridor toward the ABS.

  4. From autumn to spring CAOs strongly contribute to the net turbulent heat flux off the sea ice edge and the turbulent heat flux in CAOs is strongly enhanced compared to the climatological mean. The sensible heat flux scales linearly with the strength of CAOs, whereas the latent heat flux is limited by saturation specific humidity with respect to the SST. While in the climatological mean the magnitude of the latent heat flux associated with CAOs exceeds the sensible heat flux, the opposite is true for very strong CAOs, where the sensible heat flux is considerably more intense. Between 70% and 90% of year round extremes of the sensible heat flux (>95th percentile) occurs in CAOs and the intensity of the extremes is shaped by CAOs. The latter applies to a lesser extent to extremes of the latent heat flux, but nonetheless CAOs contribute between 50% and 80% to the extreme events. The frequency of CAOs with θSKTθ850 ≥ 2 K is a reliable predictor for the seasonal net turbulent heat flux in the regions considered. Furthermore, the strong variability of the wintertime turbulent heat flux in the ABS is due to the large interannual variations of cyclone-induced CAOs, which are linked to the pole of variability of cyclone frequency in that region.

  5. In the ABS evaporation in cyclone-induced CAOs largely compensates for much of the additional input of freshwater due to cyclone precipitation during winter. This consequently modifies the predominantly negative signal in the freshwater flux associated with the passage of cyclones observed over most regions of the Southern Ocean (Papritz et al. 2014). Thus, close to the sea ice edge cyclones play a dual role in terms of the freshwater flux. Poleward advection of warm, moist air from midlatitudes leads to ascent and precipitation. This is compensated by a strong latent heat flux of the same magnitude induced by the equatorward advection on the rearward side of the cyclones of cold air masses across the ice edge.

The present study focused on the relevance of CAOs for the turbulent heat flux from the ocean into the atmosphere, their climatological characteristics, and the mechanisms leading to their formation. However, the dynamical impacts of CAOs are also of considerable interest. CAOs not only provide the preferential environment for the genesis of intense and potentially hazardous weather systems such as polar lows (e.g., Rasmussen and Turner 2003), but they also contribute to midlatitude baroclinicity, which influences the track and intensification of extratropical cyclones—the actual “engine” that transforms the radiatively cooled polar air masses into a CAO. The investigation of the mutual interaction of extratropical cyclones and CAOs could provide another piece of the puzzle to the still incomplete understanding of the high-latitude storm track in the South Pacific.

Currently, the mechanisms leading to the genesis and intensification of mesocyclones and in particular polar lows are strongly debated, not least due to novel measurements in the environment of polar lows (Linders and Saetra 2010), which question the plausibility of conditional instability as a mechanism for their intensification. As was shown here, the air mass forming a CAO often has PV values of less than −1 PVU, irrespective of the flow being of Antarctic origin or not, and therefore has a strong potential for inducing cyclonic flow. However, when the air mass is advected over open ocean and ultimately becomes a CAO, it is strongly destabilized by the sensible heat flux from below and the cyclonic PV anomaly is quickly reduced. This raises the question of whether and in what circumstances the initially strongly cyclonic PV can contribute to mesocyclogenesis. Our methodology presented here has a great potential for the further improvement of the understanding of the environment in which these systems form.

Acknowledgments

We thank F. A. Haumann (ETH Zürich) for insightful discussions on the interaction of cold air outbreaks, the ocean and sea ice in the RS region, and I. A. Renfrew (University of East Anglia) for valuable comments on the properties of katabatic flows in Antarctica. We are grateful for the comments of three anonymous reviewers. L. Papritz acknowledges support by ETH Research Grant CH2-01 11-1. MeteoSwiss and the ECMWF are acknowledged for providing access to the ERA-Interim reanalysis data. The open-source software package R (http://www.r-project.org/) has been used for the analysis of the trajectories and the creation of some of the figures in this study.

REFERENCES

REFERENCES
Baines
,
P. G.
, and
K.
Fraedrich
,
1989
:
Topographic effects on the mean tropospheric flow patterns around Antarctica
.
J. Atmos. Sci.
,
46
,
3401
3415
, doi:.
Bieli
,
M.
,
S.
Pfahl
, and
H.
Wernli
,
2014
:
A Lagrangian investigation of hot and cold temperature extremes in Europe
.
Quart. J. Roy. Meteor. Soc.
, doi:, in press.
Bracegirdle
,
T. J.
,
2012
:
Climatology and recent increase of westerly winds over the Amundsen Sea derived from six reanalyses
.
Int. J. Climatol.
,
33
,
843
851
, doi:.
Bracegirdle
,
T. J.
, and
S. L.
Gray
,
2008
:
An objective climatology of the dynamical forcing of polar lows in the Nordic seas
.
Int. J. Climatol.
,
28
,
1903
1919
, doi:.
Bracegirdle
,
T. J.
, and
E.
Kolstad
,
2010
:
Climatology and variability of Southern Hemisphere marine cold-air outbreaks
.
Tellus
,
62A
,
202
208
, doi:.
Bracegirdle
,
T. J.
, and
G. J.
Marshall
,
2012
:
The reliability of Antarctic tropospheric pressure and temperature in the latest global reanalyses
.
J. Climate
,
25
,
7138
7146
, doi:.
Breckenridge
,
C. J.
,
U.
Radok
,
C. R.
Stearns
, and
D. H.
Bromwich
,
1993
: Katabatic winds along the Transantarctic Mountains. Antarctic Meteorology and Climatology: Studies Based on Automatic Weather Stations, D. H. Bromwich and C. R. Stearns, Eds., Antarctic Research Series, Vol. 61, Amer. Geophys. Union, 69–92.
Bromwich
,
D. H.
,
1991
:
Mesoscale cyclogenesis over the southwestern Ross Sea linked to strong katabatic winds
.
Mon. Wea. Rev.
,
119
,
1736
1753
, doi:.
Bromwich
,
D. H.
,
J. F.
Carrasco
, and
C. R.
Stearns
,
1992
:
Satellite observations of katabatic-wind propagation for great distances across the Ross Ice Shelf
.
Mon. Wea. Rev.
,
120
,
1940
1949
, doi:.
Bromwich
,
D. H.
,
J. F.
Carrasco
,
Z.
Liu
, and
R.-Y.
Tzeng
,
1993
:
Hemispheric atmospheric variations and oceanographic impacts associated with katabatic surges across the Ross Ice Shelf, Antarctica
.
J. Geophys. Res.
,
98
,
13 045
13 062
, doi:.
Bromwich
,
D. H.
,
J. P.
Nicolas
, and
A. J.
Monaghan
,
2011a
:
An assessment of precipitation changes over Antarctica and the Southern Ocean since 1989 in contemporary global reanalyses
.
J. Climate
,
24
,
4189
4209
, doi:.
Bromwich
,
D. H.
,
D.
Steinhoff
,
I.
Simmonds
,
K.
Keay
, and
R.
Fogt
,
2011b
:
Climatological aspects of cyclogenesis near Adélie Land Antarctica
.
Tellus
,
63A
,
921
938
, doi:.
Carleton
,
A. M.
, and
Y.
Song
,
1997
:
Synoptic climatology, and intrahemispheric associations of cold air mesocyclones in the Australasian sector
.
J. Geophys. Res.
,
102
,
13 873
13 887
, doi:.
Carrasco
,
J. F.
, and
D. H.
Bromwich
,
1994
:
Climatological aspects of mesoscale cyclogenesis over the Ross Sea and the Ross Ice Shelf regions of Antarctica
.
Mon. Wea. Rev.
,
122
,
2405
2425
, doi:.
Cerovečki
,
I.
,
L. D.
Talley
, and
M. R.
Mazloff
,
2011
:
A comparison of Southern Ocean air–sea buoyancy flux from an ocean state estimate with five other products
.
J. Climate
,
24
,
6283
6306
, doi:.
Chenoli
,
S. N.
,
J.
Turner
, and
A. A.
Samah
,
2013
:
A climatology of strong wind events at McMurdo station, Antarctica
.
Int. J. Climatol.
,
33
,
2667
2681
.
Comiso
,
J. C.
,
2012
: Bootstrap sea ice concentrations from Nimbus-7 SMMR and DMSP SSM/I-SSMIS. Version 2 [1980–2010]. National Snow and Ice Data Center, Boulder, CO, digital media. [Available online at http://nsidc.org/data/docs/daac/nsidc0079_bootstrap_seaice.gd.html.]
Connolley
,
W. M.
,
1997
:
Variability in annual mean circulation in southern high latitudes
.
Climate Dyn.
,
13
,
745
756
, doi:.
Dee
,
D.
, and Coauthors
,
2011
:
The ERA-Interim reanalysis: Configuration and performance of the data assimilation system
.
Quart. J. Roy. Meteor. Soc.
,
137
,
553
597
, doi:.
Drüe
,
C.
, and
G.
Heinemann
,
2001
:
Airborne investigation of Arctic boundary-layer fronts over the marginal ice zone of the Davis Strait
.
Bound.-Layer Meteor.
,
101
,
261
292
, doi:.
Fogt
,
R. L.
,
A. J.
Wovrosh
,
R. A.
Langen
, and
I.
Simmonds
,
2012
: The characteristic variability and connection to the underlying synoptic activity of the Amundsen-Bellingshausen Seas low. J. Geophys. Res.,117, D07111, doi:.
Gallée
,
H.
,
1995
:
Simulation of the mesoscale activity in the Ross Sea, Antarctica
.
Mon. Wea. Rev.
,
123
,
2051
2069
, doi:.
Grossman
,
R. L.
, and
A. K.
Betts
,
1990
:
Air–sea interaction during an extreme cold air outbreak from the eastern coast of the United States
.
Mon. Wea. Rev.
,
118
,
324
342
, doi:.
Haumann
,
A.
,
2011
: Dynamical interaction between atmosphere and sea ice in Antarctica. M.Sc. thesis, Utrecht University, 131 pp.
Heinemann
,
G.
, and
T.
Klein
,
2003
:
Simulations of topographically forced mesocyclones in the Weddell Sea and the Ross Sea Region of Antarctica
.
Mon. Wea. Rev.
,
131
,
302
316
, doi:.
Holland
,
P. R.
, and
R.
Kwok
,
2012
:
Wind-driven trends in Antarctic sea-ice drift
.
Nat. Geosci.
,
5
,
872
875
, doi:.
Irving
,
D.
,
I.
Simmonds
, and
K.
Keay
,
2010
:
Mesoscale cyclone activity over the ice-free Southern Ocean: 1999–2008
.
J. Climate
,
23
,
5404
5420
, doi:.
Klein
,
T.
, and
G.
Heinemann
,
2002
:
Interaction of katabatic winds and mesocyclones near the eastern coast of Greenland
.
Meteor. Appl.
,
9
,
407
422
, doi:.
Knuth
,
S. L.
, and
J. J.
Cassano
,
2014
:
Estimating sensible and latent heat fluxes using the integral method from in situ aircraft measurements
.
J. Atmos. Oceanic Technol.
,
31
,
1964
1981
, doi:.
Kolstad
,
E. W.
,
2011
:
A global climatology of favourable conditions for polar lows
.
Quart. J. Roy. Meteor. Soc.
,
137
,
1749
1761
, doi:.
Linders
,
T.
, and
Ø.
Saetra
,
2010
:
Can CAPE maintain polar lows?
J. Atmos. Sci.
,
67
,
2559
2571
, doi:.
Lindsay
,
R.
,
M.
Wensnahan
,
A.
Schweiger
, and
J.
Zhang
,
2014
:
Evaluation of seven different atmospheric reanalysis products in the Arctic
.
J. Climate
,
27
,
2588
2606
, doi:.
Nicolas
,
J. P.
, and
D. H.
Bromwich
,
2011
:
Precipitation changes in high southern latitudes from global reanalyses: A cautionary tale
.
Surv. Geophys.
,
32
,
475
494
, doi:.
Nigro
,
M. A.
, and
J. J.
Cassano
,
2014
:
Identification of surface wind patterns over the Ross Ice Shelf, Antarctica, using self-organizing maps
.
Mon. Wea. Rev.
,
142
,
2361
2378
, doi:.
O’Connor
,
W. P.
,
D. H.
Bromwich
, and
J. F.
Carrasco
,
1994
:
Cyclonically forced barrier winds along the Transantarctic Mountains near Ross Island
.
Mon. Wea. Rev.
,
122
,
137
150
, doi:.
Papritz
,
L.
,
S.
Pfahl
,
I.
Rudeva
,
I.
Simmonds
,
H.
Sodemann
, and
H.
Wernli
,
2014
:
The role of extratropical cyclones and fronts for Southern Ocean freshwater fluxes
.
J. Climate
,
27
,
6205
6224
, doi:.
Parish
,
T. R.
, and
D. H.
Bromwich
,
1991
:
Continental-scale simulation of the Antarctic katabatic wind regime
.
J. Climate
,
4
,
135
146
, doi:.
Parish
,
T. R.
, and
D. H.
Bromwich
,
1998
:
A case study of Antarctic katabatic wind interaction with large-scale forcing
.
Mon. Wea. Rev.
,
126
,
199
209
, doi:.
Parish
,
T. R.
, and
J. J.
Cassano
,
2003a
:
Diagnosis of the katabatic wind influence on the wintertime Antarctic surface wind field from numerical simulations
.
Mon. Wea. Rev.
,
131
,
1128
1139
, doi:.
Parish
,
T. R.
, and
J. J.
Cassano
,
2003b
:
The role of katabatic winds on the Antarctic surface wind regime
.
Mon. Wea. Rev.
,
131
,
317
333
, doi:.
Parish
,
T. R.
, and
D. H.
Bromwich
,
2007
:
Reexamination of the near-surface airflow over the Antarctic continent and implications on atmospheric circulations at high southern latitudes
.
Mon. Wea. Rev.
,
135
,
1961
1973
, doi:.
Parish
,
T. R.
,
J. J.
Cassano
, and
M. W.
Seefeldt
,
2006
: Characteristics of the Ross Ice Shelf air stream as depicted in Antarctic Mesoscale Prediction System simulations. J. Geophys. Res.,111, D12109, doi:.
Pezza
,
A. B.
,
H. A.
Rashid
, and
I.
Simmonds
,
2012
:
Climate links and recent extremes in Antarctic sea ice, high-latitude cyclones, southern annular mode and ENSO
.
Climate Dyn.
,
38
,
57
73
, doi:.
Phillpot
,
H. R.
, and
J. W.
Zillman
,
1970
:
The surface temperature inversion over the Antarctic Continent
.
J. Geophys. Res.
,
75
,
4161
4169
, doi:.
Rasmussen
,
E.
, and
J.
Turner
,
2003
:
Polar Lows: Mesoscale Weather Systems in the Polar Regions.
Cambridge University Press, 612 pp.
Renfrew
,
I. A.
, and
G. W. K.
Moore
,
1999
:
An extreme cold-air outbreak over the Labrador Sea: Roll vortices and air–sea interaction
.
Mon. Wea. Rev.
,
127
,
2379
2394
, doi:.
Renfrew
,
I. A.
, and
P. S.
Anderson
,
2006
:
Profiles of katabatic flow in summer and winter over Coats Land, Antarctica
.
Quart. J. Roy. Meteor. Soc.
,
132
,
779
802
, doi:.
Renfrew
,
I. A.
,
G. N.
Petersen
,
D. A. J.
Sproson
,
G. W. K.
Moore
,
H.
Adiwidjaja
,
S.
Zhang
, and
R.
North
,
2009
:
A comparison of aircraft-based surface-layer observations over Denmark Strait and the Irminger Sea with meteorological analyses and QuikSCAT winds
.
Quart. J. Roy. Meteor. Soc.
,
135
,
2046
2066
, doi:.
Seefeldt
,
M. W.
, and
J. J.
Cassano
,
2012
: A description of the Ross Ice Shelf air stream (RAS) through the use of self-organizing maps (SOMs). J. Geophys. Res.,117, D09112, doi:.
Shapiro
,
M. A.
,
L. S.
Fedor
, and
T.
Hampel
,
1987
:
Research aircraft measurements of a polar low over the Norwegian Sea
.
Tellus
,
39A
,
272
306
, doi:.
Simmonds
,
I.
,
2014
: Comparing and contrasting the behaviour of Arctic and Antarctic sea ice over the 35 year period 1979–2013. Ann. Glaciol.,56 (69), 18–28, doi:.
Simmonds
,
I.
, and
J. C.
King
,
2004
:
Global and hemispheric climate variations affecting the Southern Ocean
.
Antarct. Sci.
,
16
,
401
413
, doi:.
Simmonds
,
I.
,
K.
Keay
, and
E.
Lim
,
2003
:
Synoptic activity in the seas around Antarctica
.
Mon. Wea. Rev.
,
131
,
272
288
, doi:.
Sodemann
,
H.
, and
A.
Stohl
,
2009
: Asymmetries in the moisture origin of Antarctic precipitation. Geophys. Res. Lett.,36, L22803, doi:.
Stohl
,
A.
, and
H.
Sodemann
,
2010
: Characteristics of atmospheric transport into the Antarctic troposphere. J. Geophys. Res.,115, D02305, doi:.
Talley
,
L. D.
,
2008
:
Freshwater transport estimates and the global overturning circulation: Shallow, deep and throughflow components
.
Prog. Oceanogr.
,
78
,
257
303
, doi:.
Trenberth
,
K. E.
,
J. T.
Fasullo
, and
J.
Mackaro
,
2011
:
Atmospheric moisture transports from ocean to land and global energy flows in reanalyses
.
J. Climate
,
24
,
4907
4924
, doi:.
Turner
,
J.
, and Coauthors
,
2009
: Non-annular atmospheric circulation change induced by stratospheric ozone depletion and its role in the recent increase of Antarctic sea ice extent. Geophys. Res. Lett.,36, L08502, doi:.
Turner
,
J.
,
T.
Phillips
,
J. S.
Hosking
,
G. J.
Marshall
, and
A.
Orr
,
2013
:
The Amundsen Sea low
.
Int. J. Climatol.
,
33
,
1818
1829
, 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:.
Wernli
,
H.
, and
C.
Schwierz
,
2006
:
Surface cyclones in the ERA-40 dataset (1958–2001). Part I: Novel identification method and global climatology
.
J. Atmos. Sci.
,
63
,
2486
2507
, doi:.

Footnotes

1

This measurement campaign took place in February and March 2007 and the operational analyses at ECMWF at this time were obtained from the same model version (Cy31r2) as is used for ERA-Interim (Dee et al. 2011).

2

Downloaded from ftp://sidads.colorado.edu/pub/DATASETS/NOAA/G02202_v2/ in December 2013.