• Barry, R. G., , and A. H. Perry, 2001: Synoptic climatology and its applications. Synoptic and Dynamic Climatology, R. G. Barry and A. M. Carleton, Eds., Routledge, 547–603.

    • Search Google Scholar
    • Export Citation
  • Bonner, W. D., 1968: Climatology of a low level jet. Mon. Wea. Rev., 96 , 833850.

  • 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.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., 1989a: An extraordinary katabatic wind regime at Terra-Nova Bay, Antarctica. Mon. Wea. Rev., 117 , 688695.

  • Bromwich, D. H., 1989b: Satellite analyses of Antarctic katabatic wind behavior. Bull. Amer. Meteor. Soc., 70 , 738749.

  • Bromwich, D. H., 1992: A satellite case-study of a katabatic surge along the Transantarctic Mountains. Int. J. Remote Sens., 13 , 5566.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , and Z. Liu, 1996: An observational study of the katabatic wind confluence zone near Siple Coast, West Antarctica. Mon. Wea. Rev., 124 , 462477.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , Y. Du, , and T. R. Parish, 1994: Numerical simulation of winter katabatic winds from West Antarctica crossing Siple Coast and the Ross Ice Shelf. Mon. Wea. Rev., 122 , 14171435.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , J. J. Cassano, , T. Klein, , G. Heinemann, , K. M. Hines, , K. Steffen, , and J. E. Box, 2001: Mesoscale modeling of katabatic winds over Greenland with the Polar MM5. Mon. Wea. Rev., 129 , 22902309.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , A. J. Monaghan, , K. W. Manning, , and J. G. Powers, 2005: Real-time forecasting for the Antarctic: An evaluation of the Antarctic Mesoscale Prediction System (AMPS). Mon. Wea. Rev., 133 , 579603.

    • Search Google Scholar
    • Export Citation
  • Buzzi, A., , R. Cadelli, , and P. Malguzzi, 1997: Low-level jet simulations over the Southern Ocean in Antarctica. Tellus, 49A , 263276.

  • Carrasco, J. F., , and D. H. Bromwich, 1993: Satellite and automatic weather station analyses of katabatic surges across the Ross Ice Shelf. 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, 93–108.

    • Search Google Scholar
    • Export Citation
  • Cassano, J. J., , J. E. Box, , D. H. Bromwich, , L. Li, , and K. Steffen, 2001: Evaluation of Polar MM5 simulations of Greenland’s atmospheric circulation. J. Geophys. Res., 106 , 3386733889.

    • Search Google Scholar
    • Export Citation
  • Cassano, J. J., , P. Uotila, , and A. Lynch, 2006: Changes in synoptic weather patterns in the polar regions in the twentieth and twenty-first centuries. Part 1: Arctic. Int. J. Climatol., 26 , 10271049.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and M. A. Shapiro, 1999: Flow response to large-scale topography: The Greenland tip jet. Tellus, 51A , 728748.

  • Gallée, H., , and G. Schayes, 1994: Development of a three-dimensional meso-γ primitive equation model: Katabatic winds simulation in the area of Terra Nova Bay, Antarctica. Mon. Wea. Rev., 122 , 671685.

    • Search Google Scholar
    • Export Citation
  • Glickman, T., 2000: Glossary of Meteorology. 2nd ed. Amer. Meteor. Soc., 855 pp.

  • Guo, Z. C., , D. H. Bromwich, , and J. J. Cassano, 2003: Evaluation of Polar MM5 simulations of Antarctic atmospheric circulation. Mon. Wea. Rev., 131 , 384411.

    • Search Google Scholar
    • Export Citation
  • Hewitson, B. C., , and R. G. Crane, 2002: Self-organizing maps: Applications to synoptic climatology. Climate Res., 22 , 1326.

  • Kohonen, T., 2001: Self-Organizing Maps. Springer, 501 pp.

  • Kohonen, T., , J. Hynninen, , J. Kangas, , and J. Laaksonen, 1996: SOM_PAK: The Self-Organizing Map Program Package. Laboratory of Computer and Information Science (CIS). [Available online at http://www.cis.hut.fi/research/som-research.].

  • Li, J., , and Y. L. Chen, 1998: Barrier jets during TAMEX. Mon. Wea. Rev., 126 , 959971.

  • Lim, E-P., , and I. Simmonds, 2007: Southern Hemisphere winter extratropical cyclone characteristics and vertical organization observed with the ERA-40 data in 1979–2001. J. Climate, 20 , 26752690.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., , and D. H. Bromwich, 1993: Acoustic remote-sensing of planetary boundary-layer dynamics near Ross Island, Antarctica. J. Appl. Meteor., 32 , 18671882.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., , and D. H. Bromwich, 1997: Dynamics of the katabatic wind confluence zone near Siple Coast, West Antarctica. J. Appl. Meteor., 36 , 97118.

    • Search Google Scholar
    • Export Citation
  • Michaelides, S. C., , F. Lisassidou, , and C. N. Schnizas, 2007: Synoptic classification and establishment of analogues with artificial neural networks. Pure Appl. Geophys., 164 , 13471364. 10.1007/s00024-007-0222-7.

    • Search Google Scholar
    • Export Citation
  • Murphy, B. F., , and I. Simmonds, 1993: An analysis of strong wind events simulated in a GCM near Casey in the Antarctic. Mon. Wea. Rev., 121 , 522534.

    • Search Google Scholar
    • Export Citation
  • 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 , 137150.

    • Search Google Scholar
    • Export Citation
  • Parish, T. R., 1982: Surface air-flow over East Antarctica. Mon. Wea. Rev., 110 , 8490.

  • Parish, T. R., , and D. H. Bromwich, 1986: The inversion wind pattern over West Antarctica. Mon. Wea. Rev., 114 , 849860.

  • Parish, T. R., , and D. H. Bromwich, 1987: The surface windfield over the Antarctic Ice Sheets. Nature, 328 , 5154.

  • Parish, T. R., , and D. H. Bromwich, 1989: Instrumented aircraft observations of the katabatic wind regime near Terra Nova Bay. Mon. Wea. Rev., 117 , 15701585.

    • Search Google Scholar
    • Export Citation
  • Parish, T. R., , and D. H. Bromwich, 1998: A case study of Antarctic katabatic wind interaction with large-scale forcing. Mon. Wea. Rev., 126 , 199209.

    • Search Google Scholar
    • Export Citation
  • Parish, T. R., , and J. J. Cassano, 2003: Diagnosis of the katabatic wind influence on the wintertime Antarctic surface wind field from numerical simulations. Mon. Wea. Rev., 131 , 11281139.

    • Search Google Scholar
    • Export Citation
  • 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 , 19611973.

    • Search Google Scholar
    • Export Citation
  • 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:10.1029/2005JD006185.

    • Search Google Scholar
    • Export Citation
  • Powers, J. G., , A. J. Monaghan, , A. M. Cayette, , D. H. Bromwich, , Y. H. Kuo, , and K. W. Manning, 2003: Real-time mesoscale modeling over Antarctica: The Antarctic Mesoscale Prediction System. Bull. Amer. Meteor. Soc., 84 , 15331545.

    • Search Google Scholar
    • Export Citation
  • Schwerdtfeger, W., 1975: The effect of the Antarctic Peninsula on the temperature regime of the Weddell Sea. Mon. Wea. Rev., 103 , 4551.

    • Search Google Scholar
    • Export Citation
  • Schwerdtfeger, W., 1984: Weather and Climate of the Antarctic. Elsevier, 276 pp.

  • Seefeldt, M. W., , G. J. Tripoli, , and C. R. Stearns, 2003: A high-resolution numerical simulation of the wind flow in the Ross Island region, Antarctica. Mon. Wea. Rev., 131 , 435458.

    • Search Google Scholar
    • Export Citation
  • Seefeldt, M. W., , J. J. Cassano, , and T. R. Parish, 2007: Dominant regimes of the Ross Ice Shelf surface wind field during austral autumn 2005. J. Appl. Meteor. Climatol., 46 , 19331955.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , K. Keay, , and E-P. Lim, 2003: Synoptic activity in the seas around Antarctica. Mon. Wea. Rev., 131 , 272288.

  • Steinhoff, D. F., , D. H. Bromwich, , M. Lambertson, , S. L. Knuth, , and M. A. Lazzara, 2008: A dynamical investigation of the May 2004 McMurdo Antarctic severe wind event using AMPS. Mon. Wea. Rev., 136 , 726.

    • Search Google Scholar
    • Export Citation
  • Uccellini, L. W., , and D. R. Johnson, 1977: Coupling of upper and lower tropospheric jet streaks and implications for development of severe convective storms. Bull. Amer. Meteor. Soc., 58 , 686687.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Geographic location map for the Ross Ice Shelf, Ross Sea, and surrounding regions. Topography contours are in 250-m intervals.

  • View in gallery

    Domains for the SOM analysis of column-averaged wind speed. The region A is the subset domain from the AMPS 30-km archive. The domain for the column-averaged wind speed SOM is B. The region for the low-level wind speed vertical profile is C. Additional polygons (labeled 1–10) represent the regions for the node-averaged vertical profiles. Topography contours are in 250-m intervals.

  • View in gallery

    Vertical profile of average wind speed by sigma level from the 2001–05 AMPS archive for the 30-km domain over the low-level average wind speed region (see Fig. 2).

  • View in gallery

    Self-organizing map for column-averaged wind speed (lowest 12 sigma levels) from the AMPS 30-km domain for 2001–05. The node reference is indicated in brackets and the frequency of occurrence for each node is indicated in parentheses to the left of each node (panel).

  • View in gallery

    Sea level pressure averaged for each node of the column-averaged wind speed SOM from the AMPS 30-km 2001–05 archive. Isobars are in intervals of 2 hPa.

  • View in gallery

    Node-averaged vertical profile of potential temperature (K; left side of x axis in each panel) and wind speed (m s−1; right side) for selected regions. Wind barbs are plotted at the height of each sigma level. The selected regions are indicated in Fig. 1 and the plotted nodes are the corners: [6, 1], [1, 1], [6, 4], and [1, 4].

  • View in gallery

    (columns 1, 3, 5) Wind speed and (columns 2, 4, 6) wind directional constancy overlaid with wind direction vectors for (columns 1, 2) the 1st, (columns 3, 4) 5th, and (columns 5, 6) 10th, lowest sigma levels for the (top) high-plateau wind speed corner [1, 1], (center) Ross Sea cyclone corner [6, 4], and (bottom) the high-wind-dominant corner [1, 4].

  • View in gallery

    Comparative vertical profiles of potential temperature (K; left side of x axis in each panel) and wind speed (m s−1; right side) for (left) a region next to the Transantarctic Mountains and (right) a region away from the barrier for nodes (top) [1, 1], (middle) [6, 4], and (bottom) [1, 4]. The solid line represents the region next to the barrier and the dashed line is the region away from the barrier.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 20 20 3
PDF Downloads 7 7 3

An Analysis of Low-Level Jets in the Greater Ross Ice Shelf Region Based on Numerical Simulations

View More View Less
  • 1 Cooperative Institute for Research in Environmental Sciences, Department of Atmospheric and Oceanic Sciences, University of Colorado, Boulder, Colorado
© Get Permissions
Full access

Abstract

An analysis of the presence and location of low-level jets (LLJs) across the Ross Ice Shelf region in Antarctica is presented based on the analysis of archived output from the real-time Antarctic Mesoscale Prediction System (AMPS). The method of self-organizing maps (SOMs) is used to objectively identify different patterns in column-averaged wind speed (over the approximately lowest 1200 m of the atmosphere) as an identifier to the location of LLJs. The results indicate three primary LLJs in the region. The largest and most dominant LLJ is along the Transantarctic Mountains by the Siple Coast and the southern end of the Ross Ice Shelf. The second LLJ extends from the base of Byrd Glacier and curves to the north passing by the eastern extremes of Ross Island. The third LLJ extends from the base of Reeves Glacier and curves to the north across the western Ross Sea. A strong seasonality is observed in the frequency and intensity of the LLJs with the highest values for wind speed and the size of the LLJ at a maximum during the winter and spring months.

Corresponding author address: Mark W. Seefeldt, University of Colorado, 216 UCB, Boulder, CO 80309. Email: mark.seefeldt@colorado.edu

Abstract

An analysis of the presence and location of low-level jets (LLJs) across the Ross Ice Shelf region in Antarctica is presented based on the analysis of archived output from the real-time Antarctic Mesoscale Prediction System (AMPS). The method of self-organizing maps (SOMs) is used to objectively identify different patterns in column-averaged wind speed (over the approximately lowest 1200 m of the atmosphere) as an identifier to the location of LLJs. The results indicate three primary LLJs in the region. The largest and most dominant LLJ is along the Transantarctic Mountains by the Siple Coast and the southern end of the Ross Ice Shelf. The second LLJ extends from the base of Byrd Glacier and curves to the north passing by the eastern extremes of Ross Island. The third LLJ extends from the base of Reeves Glacier and curves to the north across the western Ross Sea. A strong seasonality is observed in the frequency and intensity of the LLJs with the highest values for wind speed and the size of the LLJ at a maximum during the winter and spring months.

Corresponding author address: Mark W. Seefeldt, University of Colorado, 216 UCB, Boulder, CO 80309. Email: mark.seefeldt@colorado.edu

1. Introduction

The surface wind field of the Ross Ice Shelf region (Fig. 1) has been widely studied. Parish and Bromwich (1987) indicate the presence of confluence zones of katabatic winds in regions above the major glacier valleys in the Transantarctic Mountains as well as the Siple Coast of West Antarctica. The confluence zones are locations of enhanced katabatic winds due to the pattern of the underlying topography of the Antarctic Plateau. A reexamination of the near-surface wind field is presented by Parish and Bromwich (2007) using the archived output of the Antarctic Mesoscale Prediction System (AMPS). Katabatic winds at Reeves Glacier have been studied by Bromwich (1989a) using automatic weather station (AWS) observations, by Bromwich (1989b) using thermal infrared satellite imagery and AWS observations, by Parish and Bromwich (1989) using instrumented aircraft observations, and through numerical simulations by Gallée and Schayes (1994). Katabatic winds along the Transantarctic Mountains, including Byrd Glacier, have been studied by Breckenridge et al. (1993) through the use of thermal infrared satellite imagery and AWS observations. The Siple Coast confluence zone has been studied in the past with observations and numerical simulations by Bromwich et al. (1994), Bromwich and Liu (1996), and Liu and Bromwich (1997). Synoptic forcing has been indicated to have a significant role in the characteristics and modulation of the surface wind field in the Antarctic. Murphy and Simmonds (1993) indicate the relative roles of synoptic and katabatic forcing near Casey, Antarctica, and Parish and Cassano (2003) diagnose katabatic and ambient synoptic forcing on the winds of Adelie Land. Katabatic winds propagating large distances across the Ross Ice Shelf under the influence of synoptic-scale forcing are presented by Bromwich (1992), Carrasco and Bromwich (1993), and Parish and Bromwich (1998). The presence of barrier winds along the Transantarctic Mountains is discussed by Schwerdtfeger (1984), O’Connor et al. (1994), and Steinhoff et al. (2008). The identification of the dominant wind regimes in the Ross Ice Shelf region in AWS observations is discussed by Seefeldt et al. (2007).

Through all of these studies there has been little to no mention of the wind field in the lower levels above the surface. This has been primarily the result of limited above-surface observations. An instrumented aircraft was used to make observations of the katabatic winds in the Reeves Glacier region (Parish and Bromwich 1989). Boundary layer winds were studied with the use of a sodar near Ross Island to gain an understanding of the complex flow in that region (Liu and Bromwich 1993). A month long observational study was conducted on the Siple Coast katabatic region using conventional surface, upper-air, as well as remote sensing observations of this preferred katabatic region to gain a better understanding of the low-level wind field (Bromwich and Liu 1996). The observations indicated the presence of a low-level jet (LLJ) approximately 200 m above ground level (AGL) with increasing intensity toward the Transantarctic Mountains. The increased use of numerical simulations has provided more insight into the low-level wind field. Seefeldt et al. (2003) discuss the complex structure of the low-level wind field in the widely varying topography of the northwest Ross Ice Shelf. Numerical simulations of LLJs near the northern west coast of the Ross Sea due to synoptic, barrier wind, and topographic forcing is presented by Buzzi et al. (1997). The Ross Ice Shelf airstream (RAS) has been identified as a dominant low-level feature, which has been classified as a northward-moving airstream in the lower atmosphere over the Ross Ice Shelf. Parish et al. (2006) studied the RAS using a 1-yr study of AMPS simulations and concluded that the RAS is best defined in the lowest levels with maximum winds typically found around 200–500 m AGL.

The purpose of this study is to qualitatively and quantitatively provide an understanding of the presence of LLJs in the Ross Ice Shelf region based on the analysis of archived real-time numerical weather prediction output. The LLJs are indicated as a dominant feature of the low-level wind field and are likely associated with significant northward atmospheric mass transport as a part of the RAS. The method of self-organizing maps (SOMs) will be used to provide an objective analysis to identify the locations of the major LLJs across the region. One of the advantages of using SOMs is that the results will span the entire continuum of the data providing full coverage as to the presence and location of LLJs. The frequency, seasonality, structure, and associated synoptic environment for the LLJs are presented.

The term “low-level jet” has been used in a variety of contexts in atmospheric literature (Bonner 1968; Uccellini and Johnson 1977; Li and Chen 1998). The term LLJ for this study is based on definitions by the Glossary of Meteorology (Glickman 2000). The LLJ is defined as a jet stream that is typically found in the lower 2–3 km of the troposphere. A jet stream is defined as relatively strong winds concentrated within a narrow stream in the atmosphere. Emphasis in this study will be placed on locations with relatively strong winds concentrated within a narrow stream of air in identifying and characterizing LLJs.

A description on the background of the data, SOM methodology, and use of column-averaged wind speeds is included in the second section. The third section discusses the result of the SOM in identifying three primary LLJs across the Ross Ice Shelf region. Included in the third section is a description on the frequency, seasonality, and associated synoptic patterns for the different LLJs. A summary and suggestions for additional study are provided in the final section.

2. Data sources and methodology

a. Model data—Antarctic Mesoscale Prediction System

AMPS is used as the source of data to characterize the LLJs. AMPS uses a version of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) that has been modified for use in the polar regions, referred to as the Polar MM5. Bromwich et al. (2001) and Cassano et al. (2001) provide a detailed description of the Polar MM5 model and an evaluation of simulations over the Greenland ice sheet. Guo et al. (2003) indicate that the Polar MM5 is reasonably accurate in simulating the atmospheric state over the Antarctic continent on the synoptic scale. The version of AMPS used in this study consists of six model domains with resolutions of 90, 30, 10 (3 domains), and 3.3 km. Powers et al. (2003) and Bromwich et al. (2005) provide a detailed description of the configuration and operation of AMPS.

This study uses the output from the AMPS real-time forecasts as extracted from the NCAR Mass Storage System (MSS). The 30-km domain is used to characterize the atmosphere as it is the highest-resolution domain, which covers the entire Ross Ice Shelf, Ross Sea, and the surrounding regions. The model topography is indicated in Fig. 2. AMPS uses a terrain-following sigma vertical coordinate with five sigma levels in approximately the lowest 200 m, and 10 sigma levels in approximately the lowest 700 m AGL across the Ross Ice Shelf. The AMPS archive from 2001 to 2005 is used in this study. January 2001 is the beginning of the AMPS archive and December 2005 is the end of the 90-, 30-, 10-, and 3.3-km AMPS resolutions (in September 2005 AMPS started running with 60-, 20-, 6.6-, and 2.2-km resolutions over the same domains).

The analysis presented here uses the AMPS forecasts valid 12, 15, 18, and 21 h after the model initialization time. The model is initialized every 12 h (0000 and 1200 UTC) providing a continuous series of 3-hourly model fields. Each forecast of AMPS model fields is referred to as a time slice and it represents a snapshot of the conditions in 3-hourly intervals. The initial 12 h of each AMPS forecast are not analyzed to provide the model atmospheric state 12 h to adjust from the coarse-resolution initial fields [provided by the Global Forecast System (GFS) analysis] to the higher-resolution AMPS grids and topography. This is similar to the methodology in Guo et al. (2003) and Bromwich et al. (2005). For example, the 3-hourly forecasts from 12– 21 h from the 0000 UTC 25 March 2005 model run are used to represent the 1200–2100 UTC 25 March 2005 conditions. Gaps in the AMPS archive exist due to missing forecasts or more likely, they were not originally run because of hardware and/or software problems. The 24–33-h forecasts from the previous model run, or if needed the 36–45-h forecasts from two previous model runs, are used to fill in the missing forecasts. A gap of more than three model runs results in missing time slices in the model archive time series. The total time series comprises 14 273 (98.5%) time slices out of the possible 14 488 three-hour time slices from 1200 UTC 5 January 2001 to 0900 UTC 21 December 2005 (i.e., the entire duration of the 30-km archive).

b. Column-averaged wind speed

The column-average wind speed for each model grid point is the variable of interest. This value is calculated by averaging the wind speed at each vertical level, from the lowest MM5 sigma level to the height of the average wind speed minimum above the surface as described below. A vertical limit provides the best representation of a LLJ in the column-averaged wind speeds as it confines the averaging to the expected locations of the LLJ. The upper limit is defined based on the average wind speed profile in the preferred LLJ region (Fig. 2—polygon C). Figure 3 shows a vertical profile of average wind for each vertical level over this region. The average wind speed increases from the lowest sigma level (approximately 13 m AGL) to a relative maximum at the fifth sigma level above the surface (approximately 157 m AGL). The average wind speed then decreases to a relative minimum at the 12th sigma level above the surface (approximately 1180 m AGL). This 12th sigma level is considered to be the upper limit to the presence of LLJs. The lowest 12 sigma levels are thus used when calculating the column-averaged wind speed for this study of LLJs over the Ross Ice Shelf. A comparison of column-average wind speed (not shown) with a three-dimensional isotach analysis of a selected LLJ event was made to verify the utility of using column-averaged wind speed. The two plots showed favorable correlations in the position, width, and intensity of the LLJ. A similar comparison was made to column-maximum wind speed (not shown) and a close correspondence of the LLJ position and intensity was also shown. The column-averaged wind speed is thus a good representation of the three-dimensional wind field, and the location of LLJs, while being a much simpler field to analyze than the full three-dimensional wind field. In particular, the two-dimensional column-averaged wind speed lends itself to analysis using the method of SOMs.

c. Self-organizing maps—Background

The method of SOMs was chosen as the analysis tool for this study on LLJs as represented by the AMPS forecasts. SOMs are one of several techniques that can be used to stratify large volumes of data into a small number of recurring patterns on a physically meaningful basis. Barry and Perry (2001) provide a detailed overview of the field of synoptic climatology and applications of cluster analysis.

The SOM methodology uses an unsupervised and objective classification procedure to group events into common patterns or clusters, known as nodes. The patterns are then displayed as a two-dimensional array of nodes (Fig. 4), which is known as a map. Used in this manner the SOM technique is similar to other cluster analysis methods in that it seeks to define common patterns in the input data. SOMs have the advantage over other cluster analysis techniques in that the patterns identified by the SOM analysis are not dependent on an expected function or distribution of the data and cover the entire continuum of the input data (Hewitson and Crane 2002). Furthermore, the map resulting from the SOM training is organized such that similar patterns are located in the same portion of the map, allowing simplified analysis of similar patterns. This organization of the patterns on the map leads to this map being referred to as a self-organizing map. Here the results from the SOM training will be referred to as a map or SOM interchangeably. A theoretical discussion on the SOM technique is provided by Kohonen (2001). The use of SOMs for synoptic climatology studies is described by Hewitson and Crane (2002). Cassano et al. (2006) provides a thorough description of similar synoptic climatology methods as well as an in-depth discussion on the application of the SOM technique to atmospheric data. A practical application of using SOMs for synoptic pattern classification is presented by Michaelides et al. (2007).

The method of SOMs was chosen for this study because of its distinct strengths. One strength is that the resultant patterns from the SOM training cover the entire continuum of events depicted in the 3-hourly AMPS model output. This provides the confidence that the study is capturing the entire spectrum of the occurrences of high column-averaged wind speed distributed across the Ross Ice Shelf region from the AMPS 2001–05 archive. Another strength is that the SOM training places similar nodes next to each other and dissimilar nodes apart resulting in an easy way to analyze similar and dissimilar patterns on the map. This resulting distribution of nodes thus provides both node-by-node analysis of the depicted events as well as area-by-area analysis if desired. Very different synoptic states map to opposite corners and edges of the SOM map. Last, more nodes are clustered in regions with a higher density of time slices across the data space. Each resulting node represents an approximation of the mean of the time slices composing that node.

d. Application of the SOM technique to column-averaged wind speed over the Ross Ice Shelf

The domain for the SOM analysis of column-averaged wind speed is defined such that it is large enough to include all possible locations of LLJ in the Ross Ice Shelf region and small enough so that the SOM trains on primarily the LLJ features. The region of interest (Fig. 2—rectangle B) for the SOM training and analysis is based on results of the previous research on barrier winds, katabatic winds, and the RAS (see the introduction for references) as well as experience in viewing individual analyses from the AMPS output. The southern limit of this region is to the south of the Transantarctic Mountains (see Fig. 1 for geographic names), which will allow the analysis to capture the airflow, which wraps around the southern end of the Transantarctic Mountains and flows onto the Ross Ice Shelf. The northern limit is south of Cape Adare to limit the influence of high wind speeds associated with the frequent passage of cyclones in the northern Ross Sea. The western limit is to the west of the Transantarctic Mountains to allow the SOM analysis to capture any influences of katabatic drainage through the glacier valleys. The eastern limit is east of the Siple Coast and slightly below the peak elevations of the West Antarctica ice sheet, ensuring that flow from West Antarctica onto the Ross Ice Shelf is captured. The resulting domain from the AMPS 30-km grid is 52 points × 60 points.

The software package used for the SOM analysis in this study is freely available (see online at http://www.cis.hut.fi/research/som-research) and an in-depth description of the software is provided by Kohonen et al. (1996). This software was applied to a total of 14 273 time slices of column-averaged wind speed from the AMPS 30-km grid to create 4 × 3, 5 × 4, 6 × 4, 6 × 5, 7 × 5, and 9 × 7 maps. This range of sizes (from 12 to 63 patterns) were created to determine the best size map to represent a variety of LLJ distributions and also not being too selective to result in nodes representing a limited number of events. The different map sizes were analyzed among themselves and evaluated in terms of value gained by going to more nodes in relation to the increasing complexity of a larger number of nodes. The 6 × 4 SOM was selected based on this evaluation.

Once the SOM is selected, individual time slices can be mapped to the SOM by identifying the node that has the smallest cumulative squared difference in column-averaged wind speed, over all grid points in the analysis domain, to the data sample of interest. This is repeated for all time slices and results in all time slices being associated with a single node on the SOM. From this the frequency of occurrence of each node in the dataset can be calculated. Furthermore, averages of other variables of interest (i.e., sea level pressure) can be calculated for each node by averaging that variable for all time slices that map to a particular node.

3. Analysis

a. SOM map of column-averaged wind speed

The resulting SOM of column-averaged wind speed for the 12 lowest sigma levels is shown in Fig. 4. The node reference is indicated in brackets and the frequency of occurrence for each node is indicated in parentheses above each node. There are three pronounced regions of LLJs across the Ross Ice Shelf region. The most intense LLJ is present along the Transantarctic Mountains, descending from West Antarctica across the Siple Coast and onto the southern Ross Ice Shelf. Based on previous research results (Bromwich et al. 1994; Bromwich and Liu 1996; Liu and Bromwich 1997) it is known that this region and the associated LLJ involve complex dynamics involving katabatic winds, the influence of synoptic-scale and mesoscale cyclones, possible barrier wind influences, and topographic blocking. This LLJ will be referred to as the Siple LLJ. Another LLJ extends from Byrd Glacier across the Ross Ice Shelf and around the eastern extremes of Ross Island. The airflow associated with this LLJ has a katabatic origin, is enhanced with the presence of nearby cyclones, and is influenced by the complex terrain of the northwest Ross Ice Shelf. This LLJ will be referred to as the Byrd LLJ. A less obvious LLJ extends from Reeves Glacier and turns northward over the Ross Sea. This pattern is similar to the Byrd LLJ but it does not involve flow around complex terrain. This LLJ will be referred to as the Reeves LLJ. A fourth region of high column-averaged wind speeds is present near Cape Colbeck. The high column-averaged wind speeds are indicated in only a few nodes and appear to be the result of frequent strong cyclones making landfall in this area. This is in contrast to the expected LLJ structure of a narrow stream of relatively high winds.

Additional insight into the conditions, forcing mechanisms, and characteristics of the LLJs across the Ross Ice Shelf region is provided by the corresponding sea level pressure analyses for each node (Fig. 5). The sea level pressure analyses were created by averaging the AMPS 30-km sea level pressure field for every time slice mapped to the matching node. High-elevation sea level pressure fields, such as over the East Antarctic Plateau, are generally considered suspect because of the difficulties involved with making sea level pressure adjustments in the Antarctic environment where sharp and shallow inversions are common near the surface. However, the indicated high-elevation sea level pressure fields are consistent with the lower-elevation fields and are included to provide additional information for the plateau of a qualitative nature.

The advantage of similar patterns being placed close to each other and dissimilar patterns far apart is evident in the SOM. There are four distinct patterns present in the four corners. The upper-right corner indicates light wind speeds throughout the domain. The sea level pressure analysis indicates a weak pressure gradient across the entire region. This corner will be referred to as light-wind dominant. The lower-right corner has patterns with high wind speed values over the Ross Ice Shelf and Ross Sea. The corresponding sea level pressure analysis shows a strong cyclone located in the Ross Sea. This corner will be referred to as the Ross Sea cyclone corner. The upper-left corner is dominated by high column-averaged wind speeds over the eastern East Antarctic Plateau including strong katabatic activity through many of the glacier valleys in the Transantarctic Mountains. The wind speeds across the Ross Ice Shelf and Ross Sea are light to moderate. The matching sea level pressure analysis shows a weak to moderate cyclone east of the Ross Ice Shelf with a pressure gradient supporting flow away from the glaciers adjacent to the Ross Ice Shelf. This corner will be referred to as the high-plateau wind speed corner. The lower-left corner has the most pronounced column-averaged wind speeds throughout the entire region. The eastern East Antarctic Plateau and glacier valleys have high wind speeds similar to the high-plateau wind speed corner. The western Ross Ice Shelf and Ross Sea have the highest column-averaged wind speeds with the appearance of an extreme LLJ extending along the Transantarctic Mountains from the Siple Coast northward past Ross Island and over the western Ross Sea. The related sea level pressure analysis indicates a strong cyclone on the northeastern edge of the Ross Ice Shelf, near Cape Colbeck. A strong pressure gradient is oriented perpendicular to the mountains, which supports strong northward geostrophic airflow along the Transantarctic Mountains. This corner will be referred to as the high-wind-dominant corner. The Ross Sea cyclone, high-plateau wind speed, and high-wind-dominant corners are the three primary corners of interest for this study.

b. Seasonal dependency

A strong seasonal dependency of the different nodes is evident in looking at a distribution of the frequencies of occurrence for each node. Table 1 provides the node frequencies for the entire year as well as by season. The seasons for this analysis are defined as December–January (DJ), February–April (FMA), May–August (MJJA), and September–November (SON) for the summer, fall, winter, and spring seasons. A two-month summer and a four-month winter were chosen as this distribution corresponds more closely to the annual progression of temperature and radiation in the Antarctic. This phenomenon has been referred to as the coreless winter (Schwerdtfeger 1984). For each season the left number indicates the frequency of occurrence (in percent) of that node for the indicated season. The sum of the values for all of the nodes for a given season is 100%. For example during DJ the node [3, 2] occurs 5.22% of the time. The values in parentheses indicate the percentage of occurrences for that season for that particular node. The sum of all the values in parentheses for every season for a node is 100%. For example out of all the occurrences of node [3, 2], 23.35% of those take place in DJ. The shading indicates nodes with a seasonal percentage significantly greater than or less than the normal distribution expected for that season (i.e., the expected distribution for DJ is 16.51% or 2356 time slices for DJ divided by 14 273 for the 5-yr time series). For example node [6, 1] occurs 57.23% of the time in DJ which is more than 200% of the expected 16.51% therefore it is shaded dark. The light-wind-dominant corner—[4, 1], [5, 1], and [6, 1]—occurs predominantly in summer (i.e., 52.1%–calculated by summing the time slices in the individual nodes by season and dividing by the total time slices for the season) and almost entirely summer or fall (80.1%). The other three corners occur primarily in winter and spring. The Ross Sea cyclone corner—[4, 6], [5, 4], and [6, 4]—occurs primarily in winter and spring (72.3%) but occurs with some regularity in the other seasons. The nodes making up the high-wind-dominant corner—[1, 3], [2, 4], [1, 4], and [2, 4]—occur 52.7% of the time in winter and 85.2% of the time in winter or spring. The nodes on the left-side of the map (high-plateau wind speeds and high wind dominant) rarely occur in the summer (less than 5%). The nodes located in the center of the SOM occur with more regularity during all of the seasons. The distribution of the nodes with a strong seasonal dependence agrees with what would be expected. The katabatic forcing is related to the strength of the inversion on the polar plateau and it is strongest during the long polar night during the winter months. The temperature contrasts between continental interior and the southern ocean are strongest during the winter months. This produces a stronger baroclinic environment resulting in an increase in the frequency and intensity of cyclone activity in the Ross Ice Shelf region (Lim and Simmonds 2007). More intense cyclones result in a stronger synoptic pressure gradient producing regions of stronger winds. The summer months are often accompanied with weak cyclones, generally light and varying wind across the Ross Ice Shelf, and minimal katabatic activity due to the limited net radiational cooling on the plateaus. The spring and autumn transitional seasons provide a mix of the winter and summer conditions producing a wide variety of low-level wind conditions. The spring has patterns that are somewhat more representative of the winter season because of the greater extent of sea ice around the continent at this time of year compared to the conditions in the fall when the sea ice is starting to regrow after the late summer minimum extent. This seasonality of cyclones is in agreement with a review of synoptic activity and the seasonality of synoptic features around Antarctica by Simmonds et al. (2003).

c. Analysis of vertical profiles

A more in-depth understanding of the LLJs and the different SOM patterns is accomplished by analyzing node-averaged vertical profiles of potential temperature and wind speed for selected regions (Fig. 6). Ten different regions (see Fig. 2) of grid points were identified across the domain in an attempt to characterize specific features (e.g., region 3 covers the Byrd LLJ and region 4 covers a portion of the East Antarctic Plateau serving as the source region for the Byrd LLJ). Figure 6 provides node-averaged profiles for six selected regions for the four corner patterns. The top row is for the light-wind-dominant corner. The six represented regions have minimal to no indications of an inversion in the lowest levels. The wind speed profiles indicate consistent light wind speeds. The six regions are all indicating wind speeds less than 10 m s−1 at all levels. The light-wind-dominant corner is a predominantly summer and fall feature (greater than 82%) and the Ross Ice Shelf region experiences minimal inversions during that time of year. The other three corners indicate an inversion in the profiles for the East Antarctic Plateau (region 4). These three corners have a high rate of occurrence during the winter and spring months during which inversions occur frequently on the plateau. The wind speed profiles for the East Antarctic Plateau region show a strong increase in the wind speed from the surface through the lowest levels and then decreases. The wind speed for this region is highest for the high-plateau wind speed (Fig. 6—second row) and high-wind-dominant corners (Fig. 6—fourth row). Region 8 shows the profiles for the LLJ along the Transantarctic Mountains on the southern Siple Coast. The wind speeds are highest at all levels for the high-wind-dominant corner with relatively large values for the high-plateau wind speed and Ross Sea cyclone corners.

d. A detailed analysis of individual low-level jets

A closer analysis of the corner nodes provides a more detailed characterization and understanding of the LLJs. Plots of wind speed and wind directional constancy overlaid with wind direction vectors have been created to provide more detail into the structure of the LLJs. Wind directional constancy is the ratio of vector mean wind to mean wind speed. A value of 1 indicates that wind direction is constant for all measurements. Past studies (Parish 1982; Bromwich 1989a; Bromwich et al. 1994) have shown that high-wind directional constancies are a good measure to identify a topographically influenced wind field. Figure 7 shows the node-averaged wind speed and wind constancy overlaid with wind direction vectors for the 1st, 5th, and 10th sigma levels for the high-plateau wind speed node [1, 1], Ross Sea cyclone node [6, 4], and high-wind-dominant corners node [1, 4]. The 1st sigma level is a close representation of the near-surface wind field. The fifth sigma level represents the level of the highest-averaged wind speed. The 10th sigma level provides insight into the characteristics of the flow above the primary LLJ wind speed maxima. Across the Ross Ice Shelf and Ross Sea the 5th lowest sigma level is at approximately 150 m AGL and the 10th lowest sigma level is at approximately 630 m AGL.

1) Siple low-level jet

The Siple LLJ is located alongside the southern Transantarctic Mountains and extends from approximately 84°S, 110°W to 82°S, 170°E. The Siple LLJ can be identified for all three corner patterns at all three levels shown in Fig. 7. The northern extent of the LLJ is less defined as it is dependent on the intensity of the LLJ, the associated synoptic forcing, and it sometimes merges with an additional high wind region along the western Ross Ice Shelf, occasionally merging with the Byrd LLJ. The Siple LLJ is a persistent feature throughout the year. There is at least some hint of the Siple LLJ in all of the SOM patterns and in many patterns it is the dominant feature. There are two maxima regions typically found with the Siple LLJ. The first is at 85°S, 130°W. This appears to be the central location of the West Antarctica confluence zone (Parish and Bromwich 1986, 1987) as well as at the end of a region of localized relatively steep topography. The presence of this LLJ corresponds to the LLJ observed by Bromwich and Liu (1996) and Liu and Bromwich (1997). The second region is at 84°S, 170°W and is at the northern extent of the Queen Maud Mountains. This second localized maxima is likely a tip jet and will be explained in more detail below.

The horizontal wind vectors (levels 1 and 5 in Fig. 7) indicate that the Siple LLJ is the result of the confluence of airflow from East Antarctica and West Antarctica. The airflow from East Antarctica reaches the LLJ through the glacier valleys in the southern Transantarctic Mountains as well as being wrapped around the southern extremes of the Transantarctic Mountains. The airflow from West Antarctica is the result of frequent cyclones over the eastern Ross Sea and Ross Ice Shelf as well as the confluence of katabatic drainage from West Antarctica. This complex dynamic of airflow from West Antarctica and East Antarctica is in agreement with the results of the Bromwich and Liu (1996) observational study as well as the analysis of the dynamics of the region by Liu and Bromwich (1997). The locations of high wind speed are also locations of high wind constancy for the 1st, 5th, and 10th sigma levels (Fig. 7).

The node-averaged vertical profiles (Fig. 6) for region 8 provides an understanding of the profile of potential temperature and wind speed for the Siple LLJ. The wind direction at all levels of region 8 shows the same wind direction for the three primary corners. The Siple LLJ has a maximum at the fourth sigma level for the high-plateau wind speed and Ross Sea cyclone patterns, and for the high-wind-dominant pattern the maximum is at the sixth sigma level.

The node-averaged vertical profile (Fig. 6) for the region upstream of the Siple LLJ and toward West Antarctic (region 10) indicates a moderate inversion in the lowest seven sigma levels. Parish and Cassano (2003) estimate the inversion strength by using a linear interpolation of the ambient potential temperature profile (the profile above the seventh sigma level) and extending it to the surface. This method results in an inversion strength of approximately 7 K for the high-plateau wind speed pattern and 10 K for the Ross Sea cyclone pattern. This indicates that there is some component of katabatic forcing for the LLJs.

The synoptic forcing for the Ross Sea cyclone corner—Fig. 5 node [6, 4]—results in the LLJ pulling away from the Transantarctic Mountains and extending more toward the center of the Ross Ice Shelf (Fig. 7). In the first and fifth sigma levels the LLJ is directed more northward along the 180° longitude. In the 10th sigma level the LLJ stays close to the Transantarctic Mountains and the southern extent of the LLJ is relatively weaker.

The Liu and Bromwich (1997) dynamical study indicates greatly increased wind speeds for the LLJ when the region is under the influence of a strong large-scale pressure gradient. The results of the AMPS modeled large-scale pressure gradient correspond favorably with the high-wind-dominant pattern—Fig. 5 node [1, 4]. The Siple LLJ has its most northward extent in the high-wind-dominant pattern with the LLJ extending to around 82°S and nearly merging with the Byrd LLJ. This is most apparent in the 10th sigma level. The extension of the Siple LLJ across the Ross Ice Shelf for the high-wind-dominant corner nodes is similar to what has been observed in satellite imagery by Bromwich (1992) and Carrasco and Bromwich (1993). The most pronounced high wind speeds for the lower levels occur with this pattern (Fig. 6). The vertical profile shows an extreme profile with average wind speed values greater than 30 m s−1 for a layer nearly 500 m deep with a maximum value near 35 m s−1. The potential temperature profile for the region upstream of the Siple LLJ (region 10) indicates a 5-K inversion in the lowest seven sigma levels indicating some katabatic forcing is present, but it is smaller than for the two other primary patterns.

2) Byrd low-level jet

The Byrd LLJ starts at the base of Byrd Glacier and extends to the northeast and then curves to the north, e.g. Fig. 7 node [1, 4]. As it flows toward the north it passes to the eastern extremes of the complex terrain of the northwest Ross Ice Shelf. A confluence zone is indicated to be upstream of the Byrd Glacier in the Parish and Bromwich (1987) study. This LLJ is a combination of airflow extending from Byrd Glacier as well as air flowing from the southern Ross Ice Shelf and being deflected around the area of the northwest Ross Ice Shelf. There are indications of the Byrd LLJ in all of the SOM patterns. This frequent occurrence of flow down Byrd Glacier has been studied by Breckenridge et al. (1993) where they concluded that over 50% of all available thermal infrared satellite imagery showed signatures of katabatic winds in one or more of the Skelton, Mulock, and Byrd Glaciers. The study indicated that this percentage was likely a low-end value.

The location of the wind speed maxima within the LLJ is dependent on the pattern and the associated forcing. In the light-wind patterns the LLJ only minimally meets the definition of relatively strong winds in a narrow stream with a weak wind speed maximum present. For the high-plateau wind speed pattern the maximum occurs to the northeast of the base of Byrd Glacier. For the Ross Sea cyclone pattern the LLJ is barely noticeable at the base of Byrd Glacier and it has its maximum on the eastern extremes of Ross Island. The maximum for the high-wind-dominant pattern is to the northeast of Byrd Glacier and it extends to the east of Ross Island.

Region 3 (see Fig. 2) was identified to represent the conditions in the Byrd LLJ and region 4 was selected to represent a region on the East Antarctic Plateau, which is upstream of the Byrd LLJ. The vertical profile of wind direction for region 3 shows more of a westerly component for the high-plateau wind speed pattern—node [1, 1]—than for the other two primary patterns. This corresponds with the stronger airflow on the plateau flowing down the glacier and extending over the Ross Ice Shelf. The plots of wind direction for the first and fifth sigma levels of the high-plateau wind speed pattern show the LLJ extending farther to the east than the other patterns. With minimal synoptic activity the flow through Byrd Glacier is able to flow more eastward. The Ross Sea cyclone and high-wind-dominant patterns result in a strong southerly component of the Byrd LLJ on the Ross Ice Shelf. There is a strong inversion indicated in the source region of the Byrd LLJ (region 4) for all three primary patterns. The estimated values are 21 K for the high-plateau wind speed pattern, 19 K for the Ross Sea cyclone pattern, and 18 K for the high-wind-dominant pattern. These values imply a large katabatic component for the wind flowing toward Byrd Glacier, but the difference between the patterns’ inversion values is not enough to account for the difference in wind speeds.

The Byrd LLJ is barely noticeable at the 1st sigma level for the high-plateau wind speed pattern and yet it is very apparent at the 5th and 10th sigma levels. This indicates that the automatic weather station sites on the Ross Ice Shelf near Byrd Glacier are likely not fully indicating the presence of the Byrd LLJ as it appears to be above the near-surface layer. The Byrd LLJ barely extends much beyond the northern edge of the Ross Ice Shelf for the high-plateau wind speed pattern at the 1st, 5th, and 10th sigma levels. The corresponding sea level pressure plot (Fig. 5) indicates weak synoptic forcing on the Ross Ice Shelf aids in its northward progression.

When there is additional synoptic forcing, as seen with the Ross Sea cyclone and high-wind-dominant patterns, the LLJ is able to extend northward beyond the end of the Ross Ice Shelf and sometimes merge with the Reeves LLJ. There is minimal indication of the LLJ at the base of Byrd Glacier, but it is clearly indicated for the high-wind pattern. For both of these patterns the Byrd LLJ is present east of Ross Island at all levels (Fig. 7). At the 10th sigma level the high-plateau wind speed pattern indicates air flowing from Byrd Glacier into the LLJ. For the other two primary patterns the air flowing into the LLJ is from the southern portions of the Ross Ice Shelf flowing alongside the Transantarctic Mountains. This is likely another difference between the patterns with weak synoptic forcing and those with strong synoptic forcing.

3) Reeves low-level jet

The Reeves LLJ extends from the base of Reeves Glacier out over the Ross Sea and then curves northward. The northern end of this LLJ appears to lead into the LLJ near Cape Adare, immediately to the north of the SOM domain, as studied by Buzzi et al. (1997). The presence of the Reeves LLJ is the weakest of the three discussed in this research. It is barely detectable with SOM patterns in the upper-right corner that have light column-averaged wind speeds across the Ross Sea. The Reeves LLJ is most noticeable with the more extreme patterns located on the bottom and left-hand side of the SOM.

Vertical profiles of potential temperature and wind speed in area of the Reeves LLJ (region 1) and over an area of the East Antarctic Plateau, which is upstream of the Reeves LLJ (region 2) were created to analyze the Reeves LLJ. The vertical profiles (not shown) exhibit similar patterns as to what was present with regions 3 and 4 for the Byrd LLJ. Region 1 had little to no inversion present; meanwhile region 2 has an inversion strength of approximately 15 K for the three primary patterns. The high-plateau wind speed profile for region 1 has strong westerly winds through most all of the lower levels. The profiles for the Ross Sea cyclone and high-wind-dominant patterns have primarily southerly winds by the fifth sigma level.

The Reeves LLJ is most clearly defined with the high-plateau wind speed pattern at the 1st, 5th, and 10th sigma levels. This indicates that the Reeves LLJ is more dependent on the flow down the Reeves Glacier as strong synoptic forcing makes the narrow stream of air extending from the base of Reeves Glacier less defined. The wind constancy plots at the 1st, 5th, and 10th sigma levels show the presence of the Reeves LLJ as good as or better than any other field. This LLJ associated with the katabatic wind regime at Reeves Glacier and the propagation of the airstream for hundreds of kilometers is in agreement with the Bromwich (1989a, b) studies and the observations from Parish and Bromwich (1989).

4) Queen Maud Mountains tip jet

The region at approximately 85°S, 170°W consistently has the strongest relative column-averaged wind speed values across all of the SOM patterns. The Queen Maud Mountains are present in this region and they extend northward into the LLJ flowing along the Transantarctic Mountains and over the southern Ross Ice Shelf. The protruding mountains deflect the stable airflow resulting in an acceleration of the air, necessary to travel around the mountains. This pattern is similar to what has been referred to as a tip jet off the southern tip of Greenland. Doyle and Shapiro (1999) perform idealized and case study simulations of tip jets for the southern tip of Greenland and they determined that the structure of the tip jet was associated with the Bernoulli function during orographic descent down the lee side as well as the importance of the acceleration of a parcel as it is deflected around the tip. A similar pattern has been indicated in a numerical simulation study of a LLJ near northern coast of the western Ross Sea (near Cape Adare) by Buzzi et al. (1997). Similar to this Queen Maud Mountains tip jet, the steep topography of the Transantarctic Mountains near Cape Adare extend out into the strong southerly flow resulting in significant lateral and vertical deviations and a localized region of high wind speeds. In their study, simulations were conducted without the topography and no LLJ formation was present. Based on a modest amount of study related to this project and the 30-km resolution of model data it appears that it is the acceleration of an air parcel around the Queen Maud Mountains that plays the more significant role in this localized jet. This conclusion is not definitive and further study involving higher-resolution simulations is necessary in order to fully analyze the dynamics of the LLJ in this region, and is beyond the scope of this paper.

e. Additional comments

There are some indications of a barrier wind component in the upper regions of the LLJs along the Transantarctic Mountains. Schwerdtfeger (1984) describes a barrier wind as a cold low-level wind blowing parallel to a barrier. Figure 8 is a comparison of node-averaged vertical profiles of potential temperature and wind speed between regions along the barrier and away from the barrier. Such an analysis can be used to evaluate the barrier wind forcing component. The analysis uses region 6 (see Fig. 2 for region locations) for along the barrier (Transantarctic Mountains) on the Ross Ice Shelf, region 5 for away from the barrier on the Ross Ice Shelf, region 8 along the barrier near Siple Coast, and region 9 for away from the barrier. The solid lines represent the region near the barrier and the dashed lines are for the region away from the barrier. The barrier wind will have a relatively colder column of air near the barrier in relation to the region away from the barrier with strong winds blowing parallel to the barrier.

The LLJ extending along the Transantarctic Mountains from the Siple Coast and across the western Ross Ice Shelf has a predominantly barrier parallel wind direction indicated at the 5th and 10th sigma levels above the surface (Fig. 7). The wind directional constancy and wind speed, at primarily the 10th sigma level, have relatively high values near the barrier. Cold air against the barrier is most noticeable with the high-wind-dominant corner—node [1, 4]—on the Ross Ice Shelf. The region near the barrier (region 6) is colder than the region away from the barrier (region 5) at 500 m and continues to be colder through 1500 m. The wind speed near the barrier is more than 5 m s−1 greater than away from the barrier starting at 400 m and continuing through 1500 m. This is an example of a possible barrier signature in the profiles. The Ross Sea Cyclone corner—node [6, 4]—shows similar, but less pronounced, indications of a barrier wind component on the Ross Ice Shelf. The profile of potential temperature is colder near the barrier than away from the barrier above the 400-m level. A higher wind speed near the barrier than away from the barrier is present, but is not as great as the high-wind-dominant corner. There are minimal indications of a barrier wind component for the Ross Sea Cyclone corner—node [1, 1]—for the Ross Ice Shelf and for the Siple Coast for all three corners. The vertical profiles have little difference in potential temperature for these cases. The vertical profile of wind speed shows dramatically higher wind speeds at nearly all levels near the barrier than away from the barrier. This is likely an indication of a higher wind speed near the barrier as the result of a katabatic confluence zone and/or topographic blocking.

The causation of the accumulation of a pool of cool air near the Transantarctic Mountains is not able to be determined through the technique of SOMs. The reason for the accumulation of cold air near the barrier may be related to cold stable air being directed orthogonally toward the barrier, as described by Schwerdtfeger (1975, 1984) and O’Connor et al. (1994), the result of katabatic drainage of lower potential temperature air into the regions, referred to as a katabatic barrier wind by Bromwich et al. (1994), or the result of advection of lower potential air from the plateau as the result of cyclonic forcing. To determine the method to which the cold air accumulates a time series analysis preceding the barrier wind conditions would need to be studied, which is not easily available with the method of SOMs. The selection and analysis of case studies matching the different corners would provide such information. High-resolution (greater than 10 km) numerical simulations of the case studies would provide even greater insight into the proper diagnosis and treatment of the barrier wind components of the LLJs.

4. Conclusions

The method of SOMs has been used to develop an understanding of the location, frequency, and characterization of low-level jets in the Ross Ice Shelf region. The AMPS real-time data archive was used as the data source for the analysis of LLJs. An LLJ has been defined as a concentrated narrow stream of relatively strong winds in the lower 2–3 km of the atmosphere. To identify the horizontal location of the LLJs a column-averaged wind speed has been calculated for each AMPS grid point. A total of 14 273 time slices at 3-hourly time intervals from 2001 to 2005 are used to create the SOM. The SOM methodology uses an unsupervised and objective classification procedure to group events into common patterns.

The column-averaged wind speed SOM (Fig. 4) produced four distinct patterns in the four different corners, based on the 5-yr AMPS archive for the Ross Ice Shelf region. The light-wind-dominant corner is in the upper right of the SOM and has light winds through the domain. The high-plateau wind speeds corner is in the upper left of the SOM and is characterized by high wind speed values on the East Antarctic Plateau and light wind speeds over the Ross Ice Shelf and Ross Sea. The Ross Sea cyclone corner is in the lower-right corner, it has high wind speeds over the Ross Sea and Ross Ice Shelf, and has a corresponding strong cyclone in the Ross Sea. The high-wind-dominant corner is in the lower-left corner and it has high wind speeds throughout the region especially along the Transantarctic Mountains over the Siple Coast and the southern Ross Ice Shelf.

A strong seasonal dependency was revealed for the different SOM patterns. The patterns associated with the light-wind corner occur primarily during the summer months of DJ and rarely occur during the winter months. The patterns in the three other corners occur primarily during the winter months of MJJA and the spring months of SON. The spring patterns are reasonably similar to the winter months with many of the extreme patterns occurring with regularity. The patterns along the left side of the SOM rarely occur in the summer. The patterns over the center of the map occur with regularity during all of the seasons.

There are three prominent LLJ features that are apparent in the resultant SOM. The Siple LLJ starts in the Siple Coast confluence zone alongside the southern Transantarctic Mountains and extends over the southern Ross Ice Shelf. The Siple LLJ is persistent throughout the year and is the overall most dominant feature of the low-level wind field across the Ross Ice Shelf region. There are two maxima with the Siple LLJ. The more southern maximum appears to be at the central location of the West Antarctica confluence zone. The second region is at the northern extent of the Queen Maud Mountains and appears to be a tip jet and is the result of the wind flowing around the mountains extending into the LLJ. The Siple LLJ is a complex feature comprising katabatic flow from East Antarctica in the lowest levels, katabatic flow from West Antarctica, topographic blocking due to the Transantarctic Mountains, and significant enhancement as the result of synoptic-scale cyclones over the Ross Sea and northern Ross Ice Shelf.

The Byrd LLJ starts at the base of the Byrd Glacier and extends to the northeast and then to the north as it goes around the eastern extremes of Ross Island. The primary source of the Byrd LLJ is the katabatic flow down Byrd Glacier. The outflow extends farther to the east on the Ross Ice Shelf if there is a limited pressure gradient across the western Ross Ice Shelf. The outflow from the Byrd Glacier reaches the Ross Ice Shelf and is then redirected to the north if there is a stronger synoptic-scale pressure gradient. As the Byrd LLJ extends to the north it encounters the complex topography of the northwest Ross Ice Shelf. The complex topography redirects much of the Byrd LLJ to the eastern extremes of this region resulting in an enhancement of the wind speeds across this area. The airflow with the Byrd LLJ past the eastern extremes of Ross Island is especially pronounced when associated with a strong pressure gradient across the western Ross Ice Shelf associated with a synoptic-scale cyclone, mesocyclone, or barrier winds over the western Ross Ice Shelf.

The Reeves LLJ starts at the base of the Reeves Glacier, moves over the Ross Sea, and turns toward the north traveling over the western Ross Sea. The Reeves LLJ is the least defined of the three LLJs. The Reeves LLJ is strongly dependent on the katabatic flow down Reeves Glacier and is the least influenced by synoptic-scale cyclones in the region.

Brief analysis was presented on the influence of barrier wind forcing with the LLJs. The corresponding analysis fields of wind speed and wind constancy indicate an expected pattern that would be observed with barrier wind forcing. Comparisons of the potential temperature profile from along the barrier to away from the barrier indicate at least a component of barrier wind forcing, especially on the Ross Ice Shelf. A more thorough study on barrier winds is necessary to properly understand the influence of barrier wind forcing on the LLJs.

Future work involving the study of LLJs across the Ross Ice Shelf region can focus on case studies and an airborne field program. Case studies for each of the primary patterns can provide a wealth of information as to the characterization and associated forcing mechanisms for the three primary LLJs and to verify the model based results presented here. The dates and times corresponding to each pattern are known based on the mapping of each time slice to a specific pattern. A typical event for each primary pattern can be selected from the corresponding dates and times and a detailed case study can be conducted. The case study can use the associated fields from the AMPS archive, data from automatic weather stations installed in the region, and satellite imagery. Numerical modeling studies of the entire region using a high-resolution grid (10-km resolution or greater) may also provide useful information as it would better resolve the complex topography of the Transantarctic Mountains than the 30-km resolution used from the AMPS archive. Airborne studies for the region are needed in order to verify the results from this modeling study. The results presented in this study provide a good starting point as to the locations and focus of such an airborne study. The seasonal frequencies indicate that an airborne study during the spring months of SON can provide events that are comparable to the extreme conditions often seen during the long polar winter.

Acknowledgments

This work was supported in part by NSF Grants OPP-0229645, ANT-0636811, and ATM-0404790. Seefeldt was supported by a University of Colorado–Cooperative Institute for Research in the Environmental Sciences Graduate Research Fellowship during the research. The AMPS data was retrieved courtesy of the National Center for Atmospheric Research–Computational and Information Systems Laboratory. Thanks to Thomas R. Parish for discussing and sharing his past experience during the formative stages of this project.

REFERENCES

  • Barry, R. G., , and A. H. Perry, 2001: Synoptic climatology and its applications. Synoptic and Dynamic Climatology, R. G. Barry and A. M. Carleton, Eds., Routledge, 547–603.

    • Search Google Scholar
    • Export Citation
  • Bonner, W. D., 1968: Climatology of a low level jet. Mon. Wea. Rev., 96 , 833850.

  • 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.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., 1989a: An extraordinary katabatic wind regime at Terra-Nova Bay, Antarctica. Mon. Wea. Rev., 117 , 688695.

  • Bromwich, D. H., 1989b: Satellite analyses of Antarctic katabatic wind behavior. Bull. Amer. Meteor. Soc., 70 , 738749.

  • Bromwich, D. H., 1992: A satellite case-study of a katabatic surge along the Transantarctic Mountains. Int. J. Remote Sens., 13 , 5566.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , and Z. Liu, 1996: An observational study of the katabatic wind confluence zone near Siple Coast, West Antarctica. Mon. Wea. Rev., 124 , 462477.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , Y. Du, , and T. R. Parish, 1994: Numerical simulation of winter katabatic winds from West Antarctica crossing Siple Coast and the Ross Ice Shelf. Mon. Wea. Rev., 122 , 14171435.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , J. J. Cassano, , T. Klein, , G. Heinemann, , K. M. Hines, , K. Steffen, , and J. E. Box, 2001: Mesoscale modeling of katabatic winds over Greenland with the Polar MM5. Mon. Wea. Rev., 129 , 22902309.

    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., , A. J. Monaghan, , K. W. Manning, , and J. G. Powers, 2005: Real-time forecasting for the Antarctic: An evaluation of the Antarctic Mesoscale Prediction System (AMPS). Mon. Wea. Rev., 133 , 579603.

    • Search Google Scholar
    • Export Citation
  • Buzzi, A., , R. Cadelli, , and P. Malguzzi, 1997: Low-level jet simulations over the Southern Ocean in Antarctica. Tellus, 49A , 263276.

  • Carrasco, J. F., , and D. H. Bromwich, 1993: Satellite and automatic weather station analyses of katabatic surges across the Ross Ice Shelf. 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, 93–108.

    • Search Google Scholar
    • Export Citation
  • Cassano, J. J., , J. E. Box, , D. H. Bromwich, , L. Li, , and K. Steffen, 2001: Evaluation of Polar MM5 simulations of Greenland’s atmospheric circulation. J. Geophys. Res., 106 , 3386733889.

    • Search Google Scholar
    • Export Citation
  • Cassano, J. J., , P. Uotila, , and A. Lynch, 2006: Changes in synoptic weather patterns in the polar regions in the twentieth and twenty-first centuries. Part 1: Arctic. Int. J. Climatol., 26 , 10271049.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and M. A. Shapiro, 1999: Flow response to large-scale topography: The Greenland tip jet. Tellus, 51A , 728748.

  • Gallée, H., , and G. Schayes, 1994: Development of a three-dimensional meso-γ primitive equation model: Katabatic winds simulation in the area of Terra Nova Bay, Antarctica. Mon. Wea. Rev., 122 , 671685.

    • Search Google Scholar
    • Export Citation
  • Glickman, T., 2000: Glossary of Meteorology. 2nd ed. Amer. Meteor. Soc., 855 pp.

  • Guo, Z. C., , D. H. Bromwich, , and J. J. Cassano, 2003: Evaluation of Polar MM5 simulations of Antarctic atmospheric circulation. Mon. Wea. Rev., 131 , 384411.

    • Search Google Scholar
    • Export Citation
  • Hewitson, B. C., , and R. G. Crane, 2002: Self-organizing maps: Applications to synoptic climatology. Climate Res., 22 , 1326.

  • Kohonen, T., 2001: Self-Organizing Maps. Springer, 501 pp.

  • Kohonen, T., , J. Hynninen, , J. Kangas, , and J. Laaksonen, 1996: SOM_PAK: The Self-Organizing Map Program Package. Laboratory of Computer and Information Science (CIS). [Available online at http://www.cis.hut.fi/research/som-research.].

  • Li, J., , and Y. L. Chen, 1998: Barrier jets during TAMEX. Mon. Wea. Rev., 126 , 959971.

  • Lim, E-P., , and I. Simmonds, 2007: Southern Hemisphere winter extratropical cyclone characteristics and vertical organization observed with the ERA-40 data in 1979–2001. J. Climate, 20 , 26752690.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., , and D. H. Bromwich, 1993: Acoustic remote-sensing of planetary boundary-layer dynamics near Ross Island, Antarctica. J. Appl. Meteor., 32 , 18671882.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., , and D. H. Bromwich, 1997: Dynamics of the katabatic wind confluence zone near Siple Coast, West Antarctica. J. Appl. Meteor., 36 , 97118.

    • Search Google Scholar
    • Export Citation
  • Michaelides, S. C., , F. Lisassidou, , and C. N. Schnizas, 2007: Synoptic classification and establishment of analogues with artificial neural networks. Pure Appl. Geophys., 164 , 13471364. 10.1007/s00024-007-0222-7.

    • Search Google Scholar
    • Export Citation
  • Murphy, B. F., , and I. Simmonds, 1993: An analysis of strong wind events simulated in a GCM near Casey in the Antarctic. Mon. Wea. Rev., 121 , 522534.

    • Search Google Scholar
    • Export Citation
  • 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 , 137150.

    • Search Google Scholar
    • Export Citation
  • Parish, T. R., 1982: Surface air-flow over East Antarctica. Mon. Wea. Rev., 110 , 8490.

  • Parish, T. R., , and D. H. Bromwich, 1986: The inversion wind pattern over West Antarctica. Mon. Wea. Rev., 114 , 849860.

  • Parish, T. R., , and D. H. Bromwich, 1987: The surface windfield over the Antarctic Ice Sheets. Nature, 328 , 5154.

  • Parish, T. R., , and D. H. Bromwich, 1989: Instrumented aircraft observations of the katabatic wind regime near Terra Nova Bay. Mon. Wea. Rev., 117 , 15701585.

    • Search Google Scholar
    • Export Citation
  • Parish, T. R., , and D. H. Bromwich, 1998: A case study of Antarctic katabatic wind interaction with large-scale forcing. Mon. Wea. Rev., 126 , 199209.

    • Search Google Scholar
    • Export Citation
  • Parish, T. R., , and J. J. Cassano, 2003: Diagnosis of the katabatic wind influence on the wintertime Antarctic surface wind field from numerical simulations. Mon. Wea. Rev., 131 , 11281139.

    • Search Google Scholar
    • Export Citation
  • 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 , 19611973.

    • Search Google Scholar
    • Export Citation
  • 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:10.1029/2005JD006185.

    • Search Google Scholar
    • Export Citation
  • Powers, J. G., , A. J. Monaghan, , A. M. Cayette, , D. H. Bromwich, , Y. H. Kuo, , and K. W. Manning, 2003: Real-time mesoscale modeling over Antarctica: The Antarctic Mesoscale Prediction System. Bull. Amer. Meteor. Soc., 84 , 15331545.

    • Search Google Scholar
    • Export Citation
  • Schwerdtfeger, W., 1975: The effect of the Antarctic Peninsula on the temperature regime of the Weddell Sea. Mon. Wea. Rev., 103 , 4551.

    • Search Google Scholar
    • Export Citation
  • Schwerdtfeger, W., 1984: Weather and Climate of the Antarctic. Elsevier, 276 pp.

  • Seefeldt, M. W., , G. J. Tripoli, , and C. R. Stearns, 2003: A high-resolution numerical simulation of the wind flow in the Ross Island region, Antarctica. Mon. Wea. Rev., 131 , 435458.

    • Search Google Scholar
    • Export Citation
  • Seefeldt, M. W., , J. J. Cassano, , and T. R. Parish, 2007: Dominant regimes of the Ross Ice Shelf surface wind field during austral autumn 2005. J. Appl. Meteor. Climatol., 46 , 19331955.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , K. Keay, , and E-P. Lim, 2003: Synoptic activity in the seas around Antarctica. Mon. Wea. Rev., 131 , 272288.

  • Steinhoff, D. F., , D. H. Bromwich, , M. Lambertson, , S. L. Knuth, , and M. A. Lazzara, 2008: A dynamical investigation of the May 2004 McMurdo Antarctic severe wind event using AMPS. Mon. Wea. Rev., 136 , 726.

    • Search Google Scholar
    • Export Citation
  • Uccellini, L. W., , and D. R. Johnson, 1977: Coupling of upper and lower tropospheric jet streaks and implications for development of severe convective storms. Bull. Amer. Meteor. Soc., 58 , 686687.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Geographic location map for the Ross Ice Shelf, Ross Sea, and surrounding regions. Topography contours are in 250-m intervals.

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Fig. 2.
Fig. 2.

Domains for the SOM analysis of column-averaged wind speed. The region A is the subset domain from the AMPS 30-km archive. The domain for the column-averaged wind speed SOM is B. The region for the low-level wind speed vertical profile is C. Additional polygons (labeled 1–10) represent the regions for the node-averaged vertical profiles. Topography contours are in 250-m intervals.

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Fig. 3.
Fig. 3.

Vertical profile of average wind speed by sigma level from the 2001–05 AMPS archive for the 30-km domain over the low-level average wind speed region (see Fig. 2).

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Fig. 4.
Fig. 4.

Self-organizing map for column-averaged wind speed (lowest 12 sigma levels) from the AMPS 30-km domain for 2001–05. The node reference is indicated in brackets and the frequency of occurrence for each node is indicated in parentheses to the left of each node (panel).

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Fig. 5.
Fig. 5.

Sea level pressure averaged for each node of the column-averaged wind speed SOM from the AMPS 30-km 2001–05 archive. Isobars are in intervals of 2 hPa.

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Fig. 6.
Fig. 6.

Node-averaged vertical profile of potential temperature (K; left side of x axis in each panel) and wind speed (m s−1; right side) for selected regions. Wind barbs are plotted at the height of each sigma level. The selected regions are indicated in Fig. 1 and the plotted nodes are the corners: [6, 1], [1, 1], [6, 4], and [1, 4].

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Fig. 7.
Fig. 7.

(columns 1, 3, 5) Wind speed and (columns 2, 4, 6) wind directional constancy overlaid with wind direction vectors for (columns 1, 2) the 1st, (columns 3, 4) 5th, and (columns 5, 6) 10th, lowest sigma levels for the (top) high-plateau wind speed corner [1, 1], (center) Ross Sea cyclone corner [6, 4], and (bottom) the high-wind-dominant corner [1, 4].

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Fig. 8.
Fig. 8.

Comparative vertical profiles of potential temperature (K; left side of x axis in each panel) and wind speed (m s−1; right side) for (left) a region next to the Transantarctic Mountains and (right) a region away from the barrier for nodes (top) [1, 1], (middle) [6, 4], and (bottom) [1, 4]. The solid line represents the region next to the barrier and the dashed line is the region away from the barrier.

Citation: Monthly Weather Review 136, 11; 10.1175/2008MWR2455.1

Table 1.

Node frequencies for the column-averaged wind speed SOM. “All” represents the frequencies for 2001–05. DJ, FMA, MJJA, and SON represent the frequencies for the respective months (defined in text). The values in parentheses indicate the percentage of time slices for a particular node occurring during that season. The shading indicates nodes with a seasonal percentage significantly less than or greater than the normal distribution expected for the season. Italic font indicates nodes with a seasonal percentage between 25 and 50% or between 150 and 200% of the normal distribution expected for the season. Bold font indicates a seasonal percentage less than 25% or greater than 200% of the expected value for the season.

Table 1.
Save