Streamlines of the mean annual near-surface winds over the Antarctic continent suggest a confluent channeling of the drainage flows off the ice sheets and onto the Ross Ice Shelf. A persistent cyclonic circulation to the north of the ice shelf supports a large-scale pressure field that reinforces the continental drainage flows. Owing to these two processes, an enhanced low-level airflow is present along the southern and western sections of the Ross Ice Shelf. The resulting air stream, known as the Ross Ice Shelf air stream (RAS), is one of the persistent and prominent low-level wind features seen in the Antarctic. Real-time mesoscale simulations of the Antarctic atmosphere and high southern latitudes using a modified version of the Pennsylvania State University/National Center for Atmospheric Research Mesoscale Modeling System have been ongoing since the 1999–2000 austral field season. Model results from the 1-year period November 2001 to October 2002 have been analyzed to investigate the mean structure and modulation of the Ross Ice Shelf air stream. Analyses of model results show the low-level air stream over the western Ross Ice Shelf has a wind speed maxima that is linked to the steep topography to the west. Individual cases of strong wind events appear to contain a significant barrier wind component that arises from cold air damming against the Transantarctic Mountains. Cyclones that frequently form in the Ross Sea are shown to establish conditions that promote barrier wind dynamics and thus significantly modulate the intensity of the RAS.
 The Ross Sea and the adjacent Ross Ice Shelf to the south serve as one of two large embayments that disrupt the zonal symmetry of the great Antarctic ice sheets. Two factors conspire to influence the climate of the Ross Sea sector of Antarctica. First, the topography, dominated by the Transantarctic Mountains and elevated East Antarctic ice sheet immediately inland, provides a constraining barrier to air motion along the southern and western sides of the Ross Ice Shelf extending north to approximately 72°S. Second, the Ross Sea to the north of the ice shelf is among the most active cyclogenetic regions in the world [Schwerdtfeger, 1984; King and Turner, 1997]. A semi-permanent cyclonic circulation exists in the eastern Ross Sea, which helps shape the atmospheric circulation over the entire region including the ice shelf to the south. The meteorology of the Ross Ice Shelf region is coupled to more northerly latitudes by processes acting over a variety of time scales. Previous work [e.g., Parish and Bromwich, 1998] has shown that the Ross Sea region is an important transport corridor that links Antarctica and the rest of the Southern Hemisphere during periods of extratropical cyclone forcing. Hines and Bromwich  show significant El Nino-Southern Oscillation teleconnections exist between the Ross Sea sector and more northerly latitudes.
Figure 1 serves to illustrate the key geographical features of the Ross Sea sector. The general low-level flow pattern over the western Ross Ice Shelf near Ross Island has been recognized for nearly 100 years. Schwerdtfeger  notes that the meteorological observations made by Simpson during the Scott and Shackleton expeditions early last century have provided strong evidence for persistent mountain-parallel flow south of Ross Island and flow separation around Ross Island [see also O'Connor and Bromwich, 1988; O'Connor et al., 1994]. Parish and Bromwich  used the diagnostic model of Ball  to infer the mean wintertime streamlines of the surface wind field over the Antarctic interior. More recently, Van Lipzig et al.  show similar flow fields based on a regional atmospheric model. Zones of streamline convergence or “confluence zones” are found over the continental hinterland upslope from the Ross Ice Shelf and continental sources of cold air reach the Ross Ice Shelf as a large-scale drainage from the south. Extensive drainage flows emerge onto the ice shelf also through the various glacier valleys along the Transantarctic Mountains to the west. In addition, the circulation is influenced by the pressure field associated with cyclones to the north that results in a mean horizontal pressure gradient force across the ice shelf directed to the east. Owing to the combination of terrain-induced and synoptic forcing, winds are enhanced along the western Ross Ice Shelf and a low-level wind regime persists for the entire year. This broad wind system has been coined the “Ross Ice Shelf air stream” or “RAS”.
 Aside from the early observations of Simpson, discussion of persistent southerly winds along the Transantarctic Mountains can be found in Schwerdtfeger . The author notes that the significantly different meteorological conditions on the east and west sides of the Ross Ice Shelf played a pivotal role in the outcome of Amundsen and Scott expeditions to reach the south pole nearly a century ago. A study by Parish and Bromwich  suggested that a major source of the southerly winds along the Transantarctic Mountains is from the confluence of drainage flows from West Antarctica that emerge onto the southern portions of the ice shelf. Bromwich  used satellite imagery to first document a case study of enhanced surges of strong southerly flow over the western Ross Ice Shelf, noting that the wind regime was continuous over a horizontal distance in excess of 1000 km. Breckenridge et al.  and Carrasco and Bromwich  have also reported on surges in the lower atmosphere over the western Ross Ice Shelf that appear on infrared satellite imagery. Numerical simulations of the West Antarctic and Ross Ice Shelf regions were first conducted by Bromwich et al.  based on the model described by Parish  to show that continental drainage of katabatic flows can contribute to the wind regime over the western Ross Ice Shelf. Work by O'Connor et al.  showed the influence of cyclones in supporting winds parallel to the Transantarctic Mountains. The first appearance of the term “RAS” can be seen in documents relating to the Antarctic Regional Interactions Meteorology Experiment (Antarctic RIME [see Parish and Bromwich, 2002]) where it was recognized that this air stream is one of the prominent boundary layer features of the Antarctic lower atmosphere.
 The purpose of this study is to examine characteristics of the RAS using recent numerical weather prediction model output validated with surface observations. In addition, an understanding as to how the RAS is influenced by the large-scale meteorology of the region will be sought. Available observational data by which an examination into the mean conditions associated with the low-level airflow can be made are minimal. During the past two decades a series of automatic weather stations (AWS) have been deployed about the Ross Ice Shelf [e.g., Keller et al., 1994]. At present there exist about a dozen stations that collect surface information over the ice shelf. The only measure of upper level conditions is from the radiosonde ascents at McMurdo Station, situated at the southern end of Ross Island. No record of upper level information is available on the Ross Ice Shelf. Therefore, the primary analyses conducted for this study are based on gridded output from mesoscale modeling. Real-time mesoscale forecasts for the Antarctic atmosphere have been conducted since 1999 [Powers et al., 2003]. A version of the fifth-generation Pennsylvania State University/National Center for Atmospheric Research Mesoscale Model (MM5) has been used for this task. A number of physical parameterization schemes for the surface, radiative processes, and cloud microphysics required modification for use in the polar regions [Bromwich et al., 2001; Cassano et al., 2001; Bromwich et al., 2003; Guo et al., 2003]. The first real-time forecasts for the Antarctic using a polar-specific numerical weather prediction model were produced at the Byrd Polar Research Center using the Polar MM5. The importance of such numerical products for logistical operations became obvious and in September 2000 the Antarctic Mesoscale Prediction System (AMPS), based on the Polar MM5, began producing numerical forecasts using a triply-nested grid of 90-, 30- and 10-km horizontal resolution that is centered over the Ross Island region. The 30-km domain covers the entire Antarctic continent [Powers et al., 2003], providing sufficient horizontal resolution to capture detailed topographic forcing over nearly the entire continent with the exception of highly complex terrain such as along the Transantarctic Mountains.
 Analyses shown here are taken from the 30-km horizontal resolution grid for the 1-year period from November 2001 through October 2002. This period was chosen so as to maintain a consistent grid structure; the configuration of the nested grids in AMPS changed in early November 2002. AMPS is based on a sigma coordinate representation with the top of the model set at 100 hPa. The vertical grid structure in AMPS consists of 31 levels and is designed with the highest resolution in the lower atmosphere. Grids in the lowest kilometer over the Ross Ice Shelf are situated at levels approximately 13, 40, 70, 110, 150, 220, 300, 400, 520, 700 and 960 meters above the surface.
2. Ross Ice Shelf Air Stream
Figure 2 shows the wind vectors and streamlines of the mean wind field over the Ross Sea sector at the lowest sigma level (σ = 0.9981), approximately 7 m above ground level (agl) over the high Antarctic interior to 13 m agl over the ocean, from the AMPS archive for the period November 2001 to October 2002. Analyses are taken from the daily 1200 UTC forecast simulations. Results from the first 12-h of the simulation are not used to allow an ample period of spin up and adjustment to the model topography. Averages have been computed from the three-hourly model output for each 24-hour period of the yearlong record beginning at the 12-h forecast period (i.e., 0000 UTC for each day). The large-scale airflow pattern is confluent over the continent upwind from the southern and western sections of the Ross Ice Shelf. Three regions of streamline confluence at the lowest sigma level can be identified: over the southern edge of the Ross Ice Shelf as the continental air emerges onto the ice shelf, to the southeast of Ross Island and just east of the extreme northern edge of the Transantarctic Mountains near 72°S. Extensive cold air drainage occurs off the Siple Coast region of West Antarctic and from the elevated East Antarctic plateau through major glacier valleys of the Transantarctic Mountains, similar to that depicted in Parish and Bromwich . There seems no doubt that the RAS adjacent to the southern sections of the Ross Ice Shelf contains a katabatic component [Bromwich et al., 1994] but, as will be shown later, this is not the dominant forcing mechanism. There is evidence that the large cyclone in the Ross Sea to the north of the continent supports the airflow over the ice shelf. The pronounced cyclonic circulation depicted in the eastern Ross Sea from the 2002 annual streamline pattern is indicative of the strong cyclogenetic nature of the coastal environment of the Ross Sea sector. A similar cyclonic circulation can be seen on other time-averaged maps [e.g., King and Turner, 1997] and is a climatological feature of the Ross Sea.
 Streamlines of the mean wind from the AMPS archive for the November 2001-October 2002 period shows that the RAS extends at least 1000 m above the ice shelf adjacent to the Transantarctic Mountains. Figure 3 depicts wind vectors and streamlines at sigma levels 0.9575 and 0.8688, corresponding to heights above the ice shelf of approximately 300 m and 960 m, respectively. In each case, the effect of the continental outflow from Siple Coast onto the Ross Ice Shelf and the deep cyclone over the Ross Sea is apparent. Transports through glacier valleys within the Transantarctic Mountains that serve to feed the RAS appear to be shallow, limited to the lowest few hundred meters. Note that while there is evidence of drainage flow through the Transantarctic Mountains on the σ = 0.9981 surface (Figure 2) there is little evidence of drainage flow through the Transantarctic Mountains on the σ = 0.9575 surface in Figure 3. The relatively coarse 30-km resolution of the AMPS grid, and thus inadequate representation of the glacial valleys in the Transantarctic Mountains, may be in part responsible.
 Topographic steering of the wind field by the Transantarctic Mountains to the west of the Ross Ice Shelf is present in each of the maps shown previously. In comparison with Figure 2, the flow at 300 m (σ = 0.9575) above the ice sheet suggests the RAS is aligned in a direction that is more parallel to the western mountainous barrier. This is not surprising since the Transantarctic Mountains and elevated East Antarctic ice sheet situated further west provide a continuous barrier to the flow in the lowest 2–3 km over the ice shelf. Lower levels of the atmosphere over the Ross Ice Shelf are extremely stable throughout the year owing to strong radiative cooling at the surface for much of the year [Schwerdtfeger, 1984; King and Turner, 1997]. Mean vertical temperature profiles over the ice shelf (examples shown later) taken from the AMPS archive for the 1-year period are very stable, showing inversion conditions to prevail in the lowest few hundred meters. Potential temperatures observed at the surface of the ice shelf are among the lowest ever recorded on earth. As this stable air moves across the ice shelf and becomes forced against the mountain range along the south and west, cold air damming or blocking occurs. Schwerdtfeger  first discussed this process, noting that as the cold air impinges against the elevated terrain, it is forced to rise along the windward side and decelerate. This mass convergence along the windward face results in a pressure increase and hence the development of a horizontal pressure gradient force directed away from the barrier. If these conditions persist for several hours, a geostrophic response occurs in the wind field and a mountain-parallel or barrier wind develops in the lower levels of the atmosphere. The low-level character of the RAS adjacent to the Transantarctic Mountains is thus consistent with a barrier wind [Schwerdtfeger, 1975; Parish, 1982, 1983; Bell and Bosart, 1988; O'Connor and Bromwich, 1988; O'Connor et al., 1994].
 From the streamline depictions it appears as if the RAS is a continuous, coherent feature along the western Ross Ice Shelf. Streamlines, however, only indicate the direction of the mean wind. Figure 4 illustrates the intensity of the annual mean wind at the σ = 0.9575 surface over the Ross Ice Shelf. This is near the level of the maximum wind speeds associated with the RAS. A gradient in the wind speed is present across the Ross Ice Shelf in the annual mean record with strongest winds adjacent to the Transantarctic Mountains. Three wind speed maxima are seen along the Transantarctic Mountains. The first is located at the point where drainage flows off the West Antarctic ice sheet reach the southern Ross Ice Shelf with strongest winds approximately 20 m s−1. As noted previously, this maximum must contain a katabatic component. A second maximum in excess of 14 m s−1 is situated to the south of Ross Island and appears to be associated with the topographic complex near Ross Island. A speed minimum is shown in the shadow of Ross Island. O'Connor and Bromwich , O'Connor et al. , Seefeldt et al. , and Monaghan et al.  provide a detailed description of the airflow near Ross Island. From the standpoint of the terrain representation in the 30-km AMPS grid, the orographic complex to the south of Ross Island appears as a protrusion into the RAS, similar to the capes and points that act to block the northerly low-level flow along the coast of California [e.g., Rogers et al., 1998]. This wind maxima in Figure 4 is tied directly to the terrain and, similar to that seen in the wind maxima along the California coast, may also reflect hydraulic processes such as expansion fan dynamics [e.g., Winant et al., 1988] in addition to the barrier effects that act to further accelerate the flow.
 A third wind maximum with magnitudes of roughly 12 m s−1 is seen at the far northern edge of the Transantarctic Mountains. It appears that blocking provided by the barrier plays a role in the northern wind maximum as well. All three wind maxima can be seen to be tied to enhanced streamline confluence seen in Figure 2.
 To understand the forcing of the RAS, it is necessary to evaluate the horizontal pressure gradient force over the ice shelf. The AMPS archive does not contain explicit information regarding the forcing terms in the equation of motion and thus interpolation from the sigma grids to isobaric surfaces has been done in attempt at depicting the mean pressure gradient force. An analysis of the mean height contours and geostrophic wind magnitudes of the 950-hPa surface, corresponding roughly to the σ = 0.9575 surface shown in Figure 3a, over the Ross Ice Shelf is shown in Figure 5. No analysis is shown over the elevated ice sheets owing to the uncertainty in reducing the surface pressure to the 950-hPa level. A strong height gradient exists in the annual mean across the ice shelf that is tied to the large cyclone in the Ross Sea. An elongated maximum in heights can be seen stretching northward along the barrier, suggesting that the Transantarctic Mountains and elevated continental topography inland constrain the horizontal pressure field and that blocking is occurring. One signature of blocking and hence barrier winds is for a maximum horizontal pressure gradient force adjacent to the mountains. Geostrophic wind magnitudes as derived from the 950-hPa height field are largest adjacent to the western boundary of the Ross Ice Shelf. This confirms the barrier influence on the RAS as represented in the AMPS archive. Maximum geostrophic winds in Figure 5 correspond to the locations of maximum wind speeds seen in Figure 4. From this analysis, the RAS seems to be approximately in geostrophic balance.
 To illustrate the fine-scale vertical structure of the RAS as represented in the AMPS archive from November 2001–October 2002, cross sections of wind speed and potential temperature across the Ross Ice Shelf at latitudes 78 and 84°S are shown in Figures 6a and 6b. The 78° latitude roughly corresponds to the second wind maximum alluded to previously that is situated to the southeast of Ross Island; the 84° latitude is along the southern Ross Ice Shelf where flows from the continental interior converge and is near the first wind maximum where the continental flows reach the ice shelf. In each case the maximum wind speeds are found in the lowest 1 km. Wind speeds adjacent to the Ross Island show a jet-like structure with a maximum at approximately 200–400 m above the surface. Maximum winds along 84°S are found at 500 m. Wind directions over the ice shelf can be inferred from the streamline maps and suggest that the wind regime is controlled by the large-scale Antarctic topography. The orientation of the Transantarctic Mountains changes from a southeast-northwest direction at the southern end of the Ross Ice Shelf to a south-north direction near the Ross Island complex and winds as shown in Figures 2 and 3 reflect this topographic constraint.
 Validation of Polar MM5 simulations and the AMPS archive has been attempted by a number of authors [e.g., Guo et al., 2003; Monaghan et al., 2004; Bromwich et al., 2005]. Specifically, Guo et al.  evaluated Polar MM5 simulations for 1993 and found an average wind speed bias in the model simulation of less than 1 m s−1 over 28 observational sites across the Antarctic. At the Ferrell AWS site the wind speed bias in the Polar MM5 simulations was 0.3 m s−1 and the wind direction bias was −8.3 degrees. In the work of Guo et al.  the wind speed from the Polar MM5 simulations was interpolated to the observational height based on static stability correction terms from the model output.
 Since the mid-1980s, an array of AWS has been deployed about the Ross Ice Shelf to the south of Ross Island. Comparison of the AMPS output has been made to the corresponding AWS sites that were operating during the period November 2001 to October 2003 [e.g., Keller et al., 1994]. AWS data, of course, only provide information at a level roughly 3 m above the surface as compared to the AMPS first sigma level that is typically 13 m above the surface of the ice shelf. Table 1 summarizes the comparison of the AMPS annual output with six AWS operating over the Ross Ice Shelf. AMPS wind speeds have been corrected to a 3-m level using an adiabatic profile with a representative roughness length of 0.001 m. From this comparison, mean annual wind speeds from AMPS are typically greater than those measured by the AWS, especially at the Gill and Elaine sites. Wind directions between the AWS and AMPS archive match reasonably well except again at the Gill and Elaine sites. Both these sites were experiencing trouble, especially during the winter months, and consequently their records may be suspect. Multi-year averages for both stations (not shown) indicate that both wind speeds and directions are closer to the AMPS simulated winds. It appears from this brief summary and previous AMPS and Polar MM5 validation studies that results from the AMPS archive are representative of observed surface wind conditions.
Table 1. Comparisons Between Wind Speed and Direction Averages From AWS Deployed on the Ross Ice Shelf for the Period 1 November 2001 to 31 October 2002 With Results From AMPS for Grid Points at the Same Locationa
V, m s−1
AMPS V, m s−1
Here % refers to the percentage of missing data in the AWS record, V is the AWS mean wind speed, dd is the AWS mean wind direction, AMPS V and AMPS dd are the wind speeds and wind directions, respectively, from the AMPS archive. AMPS wind speeds corrected to 3-m level assuming an adiabatic profile with a roughness length of 0.001 m.
Ferrell (78.0S, 170.8E)
Linda (78.5S, 168.4E)
Gill (80.0S, 179.0W)
Schwerdtfeger (79.9S, 169.8E)
Marilyn (80.0S, 165.0E)
Elaine (83.1S, 174.5E)
3. Seasonal Variations of the RAS
 Streamline maps shown in Figures 2 and 3 as well as previous studies such as Bromwich et al.  and O'Connor et al.  suggest that the RAS owes its existence in part to the cold air drainage off the continental ice sheets. Given the seasonal dependence of katabatic winds as evidenced by variations in wind speed at coastal stations fully exposed to flow off the continent [e.g., Parish and Cassano, 2003], one can surmise that the RAS should show similar behavior if the katabatic drainage was the dominant component. Comparison of wind vectors and streamlines from the AMPS archive at the σ = 0.9981 level for the summertime period December 2001 to February 2002 with those for the midwinter months from June to August 2002 (Figure 7), however, shows surprisingly little difference over the Ross Ice Shelf. Close inspection shows that the wind speeds are somewhat greater in the wintertime case but the differences are not large. The key features described in Figure 2 are present in the seasonal representations and the streamline patterns are essentially the same. This suggests that the broad circulation over the Ross Sea and the RAS itself are robust features of the boundary layer. Even during summer, cyclonic circulation is situated in the eastern Ross Sea and confluence of airflow off the Siple Coast region of West Antarctica is present with a pronounced RAS along the Transantarctic Mountains.
 To examine the seasonal variation in the intensity of the RAS, monthly means from the AMPS archive from November 2001 to October 2002 have been computed. Figure 8a depicts the monthly magnitude of the RAS at the σ = 0.9575 level for grid points 78°S, 170°E and 84°S, 172.5°W, corresponding to positions near the maximum wind speeds as shown in Figure 4. As suggested in Figure 7, there is no pronounced annual cycle of wind speed for the grid points shown. Rather, the strongest winds tend to occur during the autumn months from March through May for both locations. Relatively weaker winds are found at these grid points during midwinter. The RAS wind speeds for 78°S, 170°E show some evidence of a semi-annual oscillation with maximum intensity during the equinoctial months. This period has been discussed previously [e.g., Schwerdtfeger, 1984; King and Turner, 1997] and reflects the intensification of the meridional pressure gradients associated with the circumpolar trough about Antarctica. The 1-year AMPS record is too short to depict this cycle, yet the lack of a distinct annual cycle again suggests that katabatic processes are not the dominant forcing mechanism and emphasizes that ambient large-scale forces are significant in forcing the RAS. King and Turner  have noted that cyclogenesis is at a maximum over the Ross Sea during equinoctial months; given the importance of the cyclonic circulation on the RAS as illustrated in the streamlines, such a connection between the large-scale environment and the RAS seems evident.
Figure 8b illustrates the seasonal changes in the vertical RAS profile at 84°S, 172.5°W from the AMPS archive. Well defined jet profiles are present for all months. There is a tendency for the level of the maximum winds to increase as wind speeds increase. The strong winds during the autumn period are apparent in the vertical profiles. It is notable that the weakest RAS intensity for this grid point is found during midwinter. Such profiles again suggest the importance of the large-scale horizontal pressure field in forcing the RAS.
4. Synoptic Modulation of the RAS
 Streamline maps shown earlier suggest that the RAS is a persistent boundary layer feature over the western Ross Ice Shelf with little detectable connection between the seasonal cycle of the katabatic winds based on the 2002 AMPS archive. Given the premise, it is appropriate to ask how and to what degree the RAS intensity varies with transient synoptic-scale events. An examination has been made into the RAS sensitivity to the large-scale environment from the AMPS archive during the month of April 2002, a time of strong cyclone forcing.
 To begin, Figure 9 illustrates the three-hourly wind speeds from the 2002 AMPS archive at four grid points (78°S, 170°E; 80°S, 165°E; 82°S, 167.5°E; 84°S, 172.5°W) spaced at regular intervals adjacent to the Transantarctic Mountains at the σ = 0.9575 level to show that the RAS varies significantly over synoptic time scales. RAS wind speeds along the Transantarctic Mountains display some coherence between grid points with similar temporal trends seen in wind speeds. The time series of pressure (not shown) suggests an active cyclonic month with variations in excess of 30 hPa. Parish and Cassano  have noted that for katabatic-prone stations on the coast of East Antarctica a negative correlation is seen between wind speed and pressure, suggesting that winds are enhanced during cyclone episodes. Such a relationship between wind speed and pressure is not seen for the selected RAS grid points. This suggests that the RAS responds to cyclone forcing differently than surface winds over the coastal margin of East Antarctica and would not be unexpected if barrier processes are significant.
 The most striking RAS event is seen during the period 26–28 April and is reflected in the AMPS output for all four grid points. Wind speeds in excess of 40 m s−1 are modeled at the σ = 0.9575 level for the 82°S grid point and intense southerly flow components can be found in a mountain-parallel fashion along the western Ross Ice Shelf. To study the forcing of this RAS event, it is logical to examine the horizontal pressure field over the ice shelf. Figure 10 illustrates the height and geostrophic wind speeds at the 950-hPa surface over the Ross Ice Shelf for two periods, 0000 UTC 25 April and 0000UTC 28 April, corresponding to times prior to and during the intense wind event. A large cyclone can be seen over the Ross Sea on 25 April, situated directly north of the Ross Ice Shelf. Strong horizontal pressure gradients and geostrophic wind speeds greater than 30 m s−1 associated with the cyclone are found to the north of the ice shelf. Relatively weak forcing is seen over the ice shelf and along the Transantarctic Mountains with maximum geostrophic wind speeds generally less than 20 m s−1. The cyclone moved to the southeast over the next two days, reaching the coast of West Antarctica by 1200 UTC on 27 April near 130°W, and then commenced a retrograde movement. By 28 April the cyclone had moved along the West Antarctic coast to near 150°W. Streamlines at 0000 UTC 28 April showed that the circulation associated with the cyclone influenced the drainage pattern off the continent including a significant area of the interior of Antarctica where cold air is funneled into the Siple Coast confluence zone and onto the Ross Ice Shelf. The horizontal pressure gradient over the ice shelf increased in intensity during the event such that by 0000 UTC 28 April a pressure difference in excess of 30 hPa was simulated by AMPS. Geostrophic winds along the Transantarctic Mountains at the 950-hPa level were in excess of 50 m s−1 in sections by 0000 UTC 28 April. Strong winds that developed as the continental flows emerge onto the southern section of the Ross Ice Shelf can be seen to respond to the large-scale pressure field. This again suggests that the southern sections of the RAS cannot be simply interpreted as katabatic events.
 Additional evidence that RAS events during April respond to the large-scale synoptic patterns can be obtained by comparing the horizontal pressure differences across the Ross Ice Shelf with wind speeds. Figure 11 illustrates the surface pressure differences over the ice shelf as taken from grid points at 165°E and 170°W along the 80°S parallel and the RAS wind speed at the northern grid point shown in Figure 9. A high correlation is apparent throughout the month and it can be surmised that RAS events require large-scale forcing such as seen in Parish and Bromwich . It can be envisaged that the barrier effect and synoptic forcing are linked during strong wind events. Cyclonic circulation in the Ross Sea acts to force large-scale circulation over the ice shelf that sets up a blocking situation against the Transantarctic Mountains and development of barrier wind dynamics. It appears as though adjustment processes associated with blocking are central to the development of the RAS.
5. Summary and Conclusions
 The Ross Ice Shelf air stream is a prominent climatological feature over the western Ross Ice Shelf. Primary sources for the RAS include continental air that reaches the Ross Ice Shelf by way of confluent drainage patterns over the West Antarctic ice sheet and through prominent glacier valleys along the Transantarctic Mountains. Significant cyclonic forcing is responsible for the spatial patterns and intensity of the RAS. As the stable air becomes forced against the mountain range that extends along the southern and western boundary of the ice sheet, blocking occurs and a mountain-parallel wind becomes established in the lower atmosphere. From the AMPS archive for the 1-year period 1 November 2001 to 31 October 2002, the RAS is best defined in the lowest levels of the atmosphere and maximum winds typically are found at the 300–600 m levels. Minor seasonal variations are present with strongest RAS events possibly occurring during the equinoctial months although the 1-year record is too short to draw definite conclusions. The RAS shows spatial variations in the wind speed with strongest winds situated to the south of the ice shelf, southeast of Ross Island and at the northern extent of the Transantarctic Mountains.
 Large-scale synoptic events significantly modulate the strength of the RAS. To the north of the ice shelf in the Ross Sea is found one of the most cyclogenetic regions on earth. Strong cyclones that form here influence the horizontal pressure gradient over extended regions including onto the high plateau of Antarctica. Resulting topographic modification of the pressure field over the ice shelf provides support for enhanced barrier wind processes that result is significant intensification of the RAS.
 Ongoing work based on existing and newly deployed AWS and proposed airborne observations, as part of the Antarctic Regional Interactions Meteorology Experiment (Antarctic RIME), will be used to describe the three-dimensional structure and forcing mechanisms of the RAS. Additionally, a combination of AMPS forecasts and an enhanced AWS network established during the 2004 and 2005 field seasons will be used to further explore the modulation of the RAS by the synoptic scale pressure field.
 Automatic weather station data were provided by the Antarctic Meteorological Research Center at the University of Wisconsin. This research was supported by the National Science Foundation grants OPP-0229337 and OPP-0229645.