Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3549

Home , Casey Station, Vanderford Glacier, Vincennes Bay

3548 MONTHLY WEATHER REVIEW—SPECIAL SECTION VOLUME 133

Identifying the Characteristics of Strong Southerly Wind Events at Casey Station in East Antarctica Using a Numerical Weather Prediction System

NEIL ADAMS Australian Bureau of Meteorology, and Antarctic Climate and Ecosystems, Cooperative Research Centre, Hobart, Tasmania, Australia

(Manuscript received 22 December 2004, in final form 23 June 2005)

ABSTRACT

Casey Station in East Antarctica is not often subject to strong southerly flow off the Antarctic continent but when such events occur, operations at the station are often adversely impacted. Not only are the dynamics of such events poorly understood, but the forecasting of such occurrences is difficult. The fol- lowing study uses model output from a 12-month experiment using the Antarctic Limited-Area Prediction System (ALAPS) to advance the understanding of the dynamics of such events and postulates that what are often described as katabatic wind events are more likely to be synoptic in scale, with mid- and upper-level tropospheric dynamics forcing the surface layer flow. Strong surface layer flows that have a katabatic signature commonly develop on the steep Antarctic escarpment but rarely extend out over the coast in the Casey area, most probably as a result of cold air damming. However, the development of a strong south- southwesterly jet over Casey provides a mechanism whereby the katabatic can move out off the coast.

1. Introduction nantly from the northeast, off Law Dome, East Ant- arctica (Fig. 1). A meteorological event that is poorly forecast in the The katabatic flow off the Vanderford Glacier is of- Casey Station area in East Antarctica is the onset of ten visible from Casey, with a gray “smudge” on the strong to gale-force southerly flow, often accompanied southern horizon, indicative of blowing snow advecting by clear-sky conditions. The events have the appear- out to sea in the strong south-southeasterly outflow. ance of a strong katabatic flow moving up the coast Given the common occurrence of the outflow off the from the Vanderford Glacier situated to the south of Vanderford Glacier, but the rare strong southerly flow Casey (Fig. 1). However, it is not a common occurrence at Casey, it has been difficult to identify the ambient to see the katabatic wind pushing as far north as Casey, conditions that lead to the katabatic wind reaching as or of such strength, despite the Vanderford Glacier be- far north as Casey. Simply having a strong flow off the ing only 30 km to the south, and an area that regularly glacier is not enough to predicate the strong flow reach- experiences strong katabatic winds (south-southeast- ing Casey, casting some doubt as to whether the strong erly) flow. For example, Fig. 2 shows a comparison of southerly flow at Casey is in fact a true katabatic wind, time series wind data from Casey and Haupt Nunatak, given that ambient conditions obviously need to be East Antarctica, some 34 km south-southeast of Casey. right for the strong southerly flow to reach as far north The nunatak is right on the northeast edge of the as Casey. In this context a “true” katabatic wind is Vanderford Glacier and experiences a very consistent defined as a surface wind resulting from gravitational Ϫ1 southeasterly wind (140°), often around 20 m s . forcing of cold air masses on inclined terrain (Schwerdt- Whereas, coincident with these strong wind events, Ca- feger 1984), but with the more strict interpretation im- Ϫ1 sey often experiences wind as little as 5 m s . Casey posed by Phillpot (1997) whereby a speed decrease oc- Ϫ generally experiences only light outflow, and predomi- curs from fairly high near-surface values to 5 m s 1 or less by about 1 km, or 850 hPa. There are many landmark studies of the Antarctic Corresponding author address: Neil Adams, Bureau of Meteo- katabatic wind regime, from the early work of Parish rology, GPO Box 727, Hobart, Tasmania 7001, Australia. and Bromwich (1987), through to more recent work by E-mail: [email protected] Parish and Cassano (2003) and van den Broeke and van

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3549

FIG. 1. Map of the Casey local area detailing the orography and the location of Casey Station and the surrounding significant locations.

Lipzig (2003). Parish and Cassano (2003) used the fifth- dynamics occurring in the Casey area under a southerly generation Pennsylvania State University–National wind regime the Australian Bureau of Meteorology’s Center for Atmospheric Research (PSU–NCAR) Me- Antarctic Limited-Area Prediction System (ALAPS) soscale Model (MM5), to model the different forces was used to investigate such occurrences over the 12- acting on the surface flow, concluding that the persis- month period from July 2001 to June 2002. tency in wind direction is not necessarily indicative of a radiatively forced katabatic wind regime, but rather a 2. Model description and data analysis result of topographic adjustment of all pressure gradient forces. Van den Broeke and van Lipzig (2003), used The ALAPS model is a modified version of the Aus- a medium-resolution regional atmospheric model to in- tralian Bureau of Meteorology’s Limited-Area Predic- vestigate the momentum budget of the Antarctic sur- tion System (LAPS). LAPS is a globally relocatable face layer and concluded that the near-surface wind limited-area gridpoint model employing full data as- field could be explained in terms of the katabatic pres- similation. The system uses a latitude–longitude hori- sure gradient force, the large-scale pressure gradient zontal grid and sigma coordinates in the vertical. A full force, and the thermal wind effect. The thermal wind description of the model can be found in Puri et al. effects were found to be significant in areas where weak (1998), but in essence the governing equations are the large-scale forces allowed cold air to build up over sea multilevel primitive equations for momentum, mass, ice or ice shelves and often opposed the katabatic pres- temperature, and moisture, written in advective form, sure gradient force. It is possible that this effect plays a except for the mass equation, which is in flux form. The role in modulating the strong southerly outflow in the model runs on an Arakawa A grid, and in the current Casey area. Murphy and Simmonds (1993), analyzed study, employed fully explicit Miller–Pearce time dif- strong wind events simulated in a GCM, in the Casey ferencing. High-order spatial differencing was used area, and looked at the relative roles of the katabatic wherever possible to ensure accuracy to at least that of flow and synoptic situation, and concluded that very second-order C-grid models. The physical parameter- strong katabatic flow appeared to be related to the pro- izations used in the model were the same as those em- duction of cold air inland of Casey by stronger-than- ployed in the Australian Global Assimilation and Pre- average surface temperature inversions a few days be- diction System (GASP) and described in Puri et al. fore the strong wind event. To further investigate the (1998). The analysis system used in the assimilation

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC 3550 MONTHLY WEATHER REVIEW— SPECIAL SECTION VOLUME 133

Ϫ1 FIG. 2. Time series data detailing the wind direction and speed (m s ) from (top two panels) Haupt Nunatak and (bottom two panels) Casey Station for the period 1200 UTC 9 Nov 2003–1200 UTC 13 Nov 2003. cycle was a limited-area adaptation of the global mul- the model should have captured any such southerly tivariate statistical interpolation (MVSI) used in the events during the 12-month trial period (July 2001– GASP system as described by Seaman et al. (1995). Modi- June 2002), and so provide a chronology of the devel- fications to the LAPS system for running over Antarctica oping dynamics associated with the events, and give were minor, including subtle changes in the sea ice zone valuable clues as to what precursors to the development to better represent surface fluxes, and fixes to defining may be observed in the Casey observations. During the surface temperatures over the Antarctic continent. 12-month verification period of the ALAPS system, The ALAPS domain, in this study, had a resolution seven strong to gale-force south-southeast flows were of 0.25° of latitude ϫ 0.50° of longitude, giving an ap- observed at Casey, with six of the seven events having proximate horizontal resolution of 27.5 km, with model occurred during periods of ambient light northeast– boundaries from 0°–180° to 80°–35°S. Twenty-nine ver- southeast flow at the station, and with one event di- tical sigma levels were used, ranging from 0.9988 near rectly preceding a gale-force easterly storm. the surface (approximately 8 m), to 0.05 (approximately In this study a strong wind is defined as one in which Ϫ 50 hPa), at the model upper boundary, with a concen- the wind speed exceeds 13.0 m s 1 and a gale where the Ϫ tration of levels in the planetary boundary layer. A full wind speed exceeds 17.0 m s 1. Table 1 details how description of the ALAPS system may be found in successful ALAPS was at forecasting the gale events at Adams (2004). The model was initialized twice daily, varying time steps throughout each model run, from the at 1100 and 2300 UTC, and run out to ϩ96 h. If the analysis to the ϩ48 h forecast. The number of false ALAPS dynamics and resolution were sufficient then alarms is also listed, where a false alarm is defined as an

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3551

TABLE 1. Southerly strong wind events at Casey. The third event, around 29 November 2001 was significant, in as much as the wind speed reached during ALAPS ϩ12 h ϩ24 h ϩ36 h ϩ48 h Event date analysis forecast forecast forecast forecast the southerly event was high, and the duration of the gale-force wind relatively long, as evident in Fig. 3, 24 Aug 2001 Yes Yes Yes Yes Yes where the surface observational data from Casey are 30 Oct 2001 No No No No No 29 Nov 2001 Yes Yes Yes Yes Yes plotted as a thin black line. Analysis of the ALAPS 14 Dec 2001 Yes Yes Yes Yes Yes model output during this event from the model run 19 Mar 2002 Yes Yes Yes Yes Yes initiated at 1100 UTC 26 November 2001 (thick, 2 Apr 2002 Yes Yes No No No marked lines in Fig. 3), highlighted a remarkably good 5 May 2002 Yes No No No No fit between the model and observations with only a False alarms 0 2111 slight lag in the model onset of the gale, and a nonfore- cast of the secondary peak in wind speed at 2000 UTC 29 November, some 84 h into the ALAPS model inte- ALAPS forecast of a southerly gale event but for which gration. there was no supporting observations of a southerly A comparison of the upper-air data from the Casey gale at Casey. It should be noted that by ϩ48 h the radiosonde flights from before and after the gale event model occasionally suffered a lag of between 12 and 18 with ALAPS model wind profiles highlighted reason- h in onset timing of the strong southerly flow. able model performance. Figure 4a shows a comparison

FIG. 3. Time series of Casey surface wind direction, wind speed, surface pressure, and temperature (thin continuous lines) for the period 1100 UTC 26 Nov 2001–1100 UTC 30 Nov 2001, overlaid with 96 h of ALAPS model data from a run initiated at 1100 UTC 26 Nov 2001 (thick marked lines).

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC 3552 MONTHLY WEATHER REVIEW— SPECIAL SECTION VOLUME 133

FIG. 4. Vertical profiles of wind direction and speed from the Casey radiosonde observations (thin lines) overlaid with ALAPS model output (thick marked lines): (a) 2323 UTC 27 Nov 2001 and ϩ36 h ALAPS model output valid at 2300 UTC 27 Nov 2001, (b) 1125 UTC 28 Nov 2001 and ϩ48 h ALAPS model output valid at 1100 UTC 28 Nov 2001, (c) 2343 UTC 29 Nov 2001 and ϩ84 h ALAPS model output valid at 2300 UTC 29 Nov 2001. of the radiosonde profile taken at 2323 UTC 27 No- flight at 1100 UTC 28 November 2001 (Fig. 4b), and as vember 2001 (thin line), just prior to the onset of the the storm was beginning to ease, with a successful flight southerly gale with the concurrent ALAPS ϩ36 h fore- at 2343 UTC 29 November 2001 (Fig. 4c). The ϩ84 h cast (thick marked line). The comparison showed very ALAPS profile concurrent with the later radiosonde good agreement in the wind speed profile but with the flight is also shown in Fig. 4c. The near-surface flow was modeled wind direction more easterly than that ob- very well modeled by the ALAPS ϩ84 h forecast with served. Although it should be noted that wind speeds the strong near-surface jet well captured. ALAPS did Ϫ were generally less than 15 m s 1 through out the tro- overforecast the midatmospheric wind speeds by 5–10 Ϫ posphere. By 1100 UTC 28 November 2001, the near- ms 1 but showed very good agreement with the ob- surface southerly wind had begun to increase at Casey served wind direction profile. The comparison of near- and wind speeds throughout the troposphere had also surface parameters (Fig. 3), and upper profiles (Fig. 4) begun to strengthen. A comparison of the radiosonde show enough similarity between model output and profile at this time with the concurrent ϩ48 h ALAPS available observations to warrant using the ALAPS profile (Fig. 4b) showed good agreement through the model output from this run to further investigate the mid- and upper atmosphere, but with the model under- development and dynamics of the southerly storm at forecasting the wind strength in the lower troposphere, Casey. and being more southerly in direction than southeast- The synoptic situation in the presoutherly storm en- erly as observed. These differences were most likely vironment [taken from the National Centers for Envi- associated with the small lag in the onset timing of the ronmental Prediction (NCEP)–NCAR reanalyses] was southerly gale forecast by ALAPS (Fig. 3). One prob- dominated by an intense low pressure system, with a lem during strong wind events at Casey, and in particu- central pressure of 955 hPa, near 59°S, 112°E at 0000 lar, strong southerly wind events, is the difficulty in UTC 25 November 2001 (Fig. 5a), and which was track- launching radiosondes from the station. It is almost im- ing steadily eastward. A ridge of high pressure along possible to successfully launch during very strong wind 58°E was also evident at the time. By 1200 UTC 27 events due to strong eddying of the surface wind November, some 12–15 h prior to the onset of the around the balloon launch site, and as a result, radio- strong southerly, the low had moved away to the east of sonde flights are often aborted with a resulting lack of Casey and weakened to around 982 hPa. The ridge of upper-air data during these events. On the occasion of high pressure had moved into near 95°E, and begun to the November 2001 storm event, upper-wind data were strengthen (Fig. 5b). By 1200 UTC on 29 November available only during the buildup of the storm, with a 2001, at the height of the southerly storm, the low was

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3553

FIG. 5. NCEP–NCAR reanalysis of mean sea level pressure valid at (a) 0000 UTC 25 Nov 2001, (b) 1200 UTC 27 Nov 2001, and (c) 1200 UTC 29 Nov 2001. well to the east of Casey, with the ridge of high pressure port for both the low pressure system and developing having moved to 100°E and strengthened substantially, high pressure ridge. The 0000 UTC 500-hPa analysis directing a strong southerly airstream over Casey (Fig. from the NCEP–NCAR reanalysis project for 25 No- 5c). In the midlevels, at 500 hPa, there was strong sup- vember 2001 (Fig. 6a), had a 4910-m minimum located

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC 3554 MONTHLY WEATHER REVIEW— SPECIAL SECTION VOLUME 133

FIG. 6. Same as in Fig. 5, but for 500-hPa geopotential height. near 63°S, 91.5°E, to the west of Casey. By 1200 UTC weak 500-hPa geopotential height ridge evident along on 27 November 2001 (Fig. 6b), the upper low had 85°E. By 1200 UTC 29 November 2001 (Fig. 6c), the moved to near 65°S, 126°E out to the east–northeast of 500-hPa minimum had tracked inland to near 69.5°S, Casey, supporting the surface low in that area, with a 147°E, and deepened to around 4920 m, while the

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3555

FIG.7.North–south cross section along 109°E from 76° to 60°S, and the surface to 300 hPa, showing in plane wind speed (grayscale Ϫ shading, m s 1), wind barbs (␷, w), and potential temperature contours from (a) ALAPS ϩ9 h forecast valid at 2000 UTC 26 Nov 2001, (b) ALAPS ϩ36 h forecast valid at 2300 UTC 27 Nov 2001, (c) ALAPS ϩ48 h forecast valid at 1100 UTC 28 Nov 2001, and (d) ALAPS ϩ54 h forecast valid at 1700 UTC 28 Nov 2001.

500-hPa ridge had sharpened and moved to near 100°E ment in the Casey area was dominated by very light on the coast, and significantly tightened the midlevel flow at all levels over the continent, with the receding south-southwesterly gradient over Casey. low pressure system having advected warm maritime Cross sections from the ALAPS model output were air well up onto the plateau (Fig. 7a), and decaying the constructed in an attempt to define the time evolution normal low-level temperature inversion found over the of the three-dimensional airflow in the Casey area, with continent. Figure 7a was derived from the ALAPS cross sections taken along 109.0°E, some 65 km to the ϩ9 h model output valid at 2000 UTC 26 November west of Casey, over the western side of the orographic 2001, some 34 h prior to the onset of the strong south- basin carrying the Vanderford Glacier, and along a erly flow. The cross section is along 109.0°E, from 76° southeast–northwest transect running through Casey to 60°S, including the two-dimensional airflow in the Station, from 70°S, 116.4°Eto61°S, 104.9°E. Associ- vertical plane, comprising wind barbs and a grayscale- ated with each cross section is a map of the two- shaded plot of the wind speed (derived from the me- dimensional near-surface flow, with the location of the ridional flow and vertical motion field), accompanied vertical cross sections marked. The prestorm environ- by a contour analysis of potential temperature. Figure

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC 3556 MONTHLY WEATHER REVIEW— SPECIAL SECTION VOLUME 133

Ϫ1 FIG. 8. Same as in Fig. 7, but for near-surface wind speed in the Casey area (grayscale shading, m s ) and wind barbs. The location of Casey is marked in each frame by a cross enclosed by a circle.

8a shows the near-surface wind pattern in the Casey ment (Fig. 7c). By this stage (1100 UTC 28 November), area, coincident with the cross section in Fig. 7a, high- the approaching high pressure ridge was evident, with lighting the low pressure system on the eastern bound- deep and strong southerly outflow evident over the ary, and the very light wind regime in Vincennes Bay. continental interior (Fig. 7c). However, the very strong, As the low moved farther to the east, radiative cool- low-level flow over the steep escarpment to the west of ing of the continent gradually returned the surface in- Casey was still confined to the lowest few hundred version to its more normal state, highlighted by the meters. Low-level wind in the Casey area was still very tightening vertical gradient of potential temperature by light at this stage (Fig. 8c). By ϩ54 h into the model run ϩ36 h into the model run (Fig. 7b). Associated with the the very strong surface wind, which had been confined redevelopment of the surface inversion was the devel- to the steep coastal escarpment west of Casey, was be- oping low-level katabatic flow, over the steep escarp- ginning to push out to sea (Fig. 8d), although the flow ment to the west of Casey, near 68.0°S (Fig. 7b). A was no longer a pure katabatic wind, as strong upper- low-level synoptic-scale southwesterly flow was also de- level southerly flow was developing at all levels over veloping at the same time, well to the west of Casey, the continent, and moving northward toward the coast with the approaching high pressure ridge (Fig. 8b). The (Fig. 7d). What was also significant was the mainte- very dry nature of the stream, and associated clear nance of an area of very light surface flow over Casey skies, enhanced the radiative cooling of the Antarctic Station and to the northeast (Fig. 8d). During the de- boundary layer such that by ϩ48 h the developing sur- velopment of the surface katabatic wind over the steep face katabatic flow over the continent had rapidly be- escarpment to the west of Casey, a similar development came more pronounced on the steep coastal escarp- was occurring on the escarpment inland of Casey, al-

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3557

FIG. 9. Northwest–southeast cross section through Casey from 70° to 60°S, and the surface to 300 hPa, showing in plane wind speed Ϫ (grayscale shading, m s 1), wind barbs (u, ␷, w), and potential temperature contours from (a) ALAPS ϩ54 h forecast valid at 1700 UTC 28 Nov 2001 and (b) ALAPS ϩ66 h forecast valid at 0500 UTC 29 Nov 2001. (c) North–south cross section along 109°E from 76° to Ϫ 60°S, and the surface to 300 hPa, showing in plane wind speed (grayscale shading, m s 1), wind barbs (␷, w), and potential temperature contours from ALAPS ϩ66 h forecast valid at 0500 UTC 29 Nov 2001. (d) Near-surface wind speed in the Casey area (grayscale Ϫ shading, m s 1) and wind barbs from the ALAPS ϩ66 h forecast. though the flow was no where near as well developed at cier, and extending an appreciable distance out off the ϩ54 h (Fig. 9a), with the strongest surface flow remain- coast, with a light wind area still evident off the coast to ing well inland, despite the fact that by this time the the northeast of Casey. A comparison of the cross sec- katabatic flow at 109°E had already moved offshore tion, along 109°Eatϩ66 h (Fig. 9c) and near Casey (Fig. 7d). (Fig. 9b), showed significant differences in flow The cross section in Fig. 9a was oriented northwest– strength, with the low-level flow to the west appreciably southeast, to highlight the development of the inland stronger than near Casey. surface flow directly upwind of Casey Station. As the Comparison of the Casey radiosonde flight with the intensifying ridge of high pressure approached Casey ALAPS ϩ48 h forecast flow, at the height of the strong the southerly flow strengthened throughout the tropo- wind (Fig. 4c), showed excellent model performance sphere, with the low-level wind maxima also strength- but also highlighted the fact that the wind profile was ening appreciably, and by ϩ66 h the low-level maxima not indicative of a pure katabatic wind, as the back- had moved down off the escarpment, over the station, ground flow in the lowest few hundred meters of atmo- Ϫ and out into Vincennes Bay (Fig. 9b). The surface wind sphere was on the order of 15 m s 1. However, the at ϩ66 h (Fig. 9d), showed a “channel” of stronger wind rather distinct low-level jet had its development in the blowing down the valley carrying the Vanderford Gla- surface layer, with the development dominated by the

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC 3558 MONTHLY WEATHER REVIEW— SPECIAL SECTION VOLUME 133

FIG. 10. Vertical profiles of wind direction and speed from the Casey radiosonde observations (thin lines) overlaid with ALAPS model output (thick marked lines) (a) 1118 UTC 30 Oct 2001 and ϩ12 h ALAPS model output valid at 1100 UTC 30 Oct 2001, (b) 2321 UTC 30 Oct 2001 and ϩ24 h ALAPS model output valid at 2300 UTC 30 Oct 2001, and (c) radiosonde flight at 2321 UTC 30 Oct 2001 and ϩ24 h ALAPS model output valid at 2300 UTC 30 Oct 2001 at the closest grid point to the southwest of Casey. radiative cooling of the near-surface layer in the post– wind evident under the inversion and indicative of a low pressure system environment. The potential tem- hydraulic jump sitting well inland on the steep coastal perature in the early cross sections was dominated by escarpment. By ϩ54 h the surface inversion had not slack gradients, but as the near-surface air cooled, the only strengthened, but it had also moved down off the potential temperature gradient near the surface in- plateau and onto the coast (Fig. 7d), suggestive of a creased rapidly with a resulting increase in near-surface hydraulic jump having migrated downslope onto the wind speed on the steep slopes (Ball 1960). The devel- coast. At this time the Froude number from the opment of the strong near-surface flow preceded the ALAPS 0.85 sigma level (approximately the same developing upper-level jet. However, the strong surface height as the summit of Law Dome at 1395 m), just flow was confined to the steep coastal escarpments, un- upslope from Casey, had jumped from 0.9 at the ϩ48 h der the very strong surface inversion, and was strongest time step to 1.2 at the ϩ54 h time step, supporting the on the western side of the broad valley carrying the shift in flow regime indicative of a hydraulic jump. Vanderford Glacier. The stronger flow to the west of Casey would have been partially due to the steeper 3. Discussion orography in that area, but also in response to general orographic forcing, where outflow off the continent is In summary, during the event of late November, Ca- channeled to the western side of orographic depres- sey experienced a dramatic increase in surface south- sions, in a similar fashion to barrier winds, resulting in erly flow, which in essence was a katabatic wind signa- a low-level jet generally confined on the steep slopes of ture, but the flow was in a regime of developing deep the western side of valleys (Ball 1960). The strong southerly flow, with the approach of an intensifying southerly flow appeared to develop in this case because ridge of high pressure, and it was the deep, synoptic- of the strength of the upper-level south-southwest jet, scale, southerly flow that caused the katabatic wind that and the developing southerly flow through the depth of had developed upstream and inland of Casey to advect the troposphere advected the surface jet northward and off the escarpment and over the station. over Casey. A major ALAPS forecast failure that occurred dur- Also of interest was the quite dramatic change in the ing the 12-month trial period occurred late in October depth of the surface inversion layer near the coast west 2001 when Casey had been experiencing an almost con- of Casey throughout the development of the southerly tinuous period of southerly flow, from late on 28 Oc- storm. At ϩ48 h the strong surface inversion was well tober until early on 31 October 2001. During this period Ϫ developed inland, south of 67°S (Fig. 7c), with a strong the southerly wind was strong (around 15 m s 1) for

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3559

FIG. 11. Same as in Fig. 9, but from the (a) 27.5-km ALAPS ϩ24 h forecast valid at 2300 UTC 23 Oct 2001 and (b) concurrent Ϫ near-surface wind speed in the Casey area (grayscale shading, m s 1) and wind barbs. Concurrent northwest–southeast cross section through Casey from the (c) 11-km ALAPS run and (d) concurrent near-surface winds from the 11-km run. about 12 h, from the middle of 30 October 2001. Data near surface flow during this event showed, in the initial from the radiosonde flight, launched just at the onset of instant (not shown), quite similar development to the the storm (Fig. 10a) show a reasonably deep layer of November case, with a light wind area over Casey and southwesterly flow, consistent with the Casey observa- to the north and northeast, and a strengthening low- tions, however the model wind strength was too low level flow over the steep coastal escarpment to the west right throughout the lowest 12000 m of atmosphere. and southwest. However, the model never developed At the height of the storm, the model maintained a the strong upper-level southerly outflow evident in the light surface southeasterly flow, which had turned November 2001 case (Fig. 9). However, the model did southwesterly by 1500 m and gradually increased to develop a substantial surface jet some 35 km southwest Ϫ around 32 m s 1 by 7800 m (Fig. 10b). Unfortunately of Casey at 66.5°S, 110.0°E (Fig. 10c), but which never the GPS radiosonde launched from Casey at this time advected off the plateau and over Casey during the was unable to fully sample the atmospheric winds, al- October 2001 model run. though what data were available showed that the model The most significant difference in this case appeared slightly underforecasted the low- to midtropospheric to be the orientation of the upper jet and broadscale wind speed, and completely missed the surface wind tropospheric flow, with the October storm having more maxima. Also of note was the fact that both the ob- of a westerly component than the November 2001 served and modeled flow during this storm was more storm. It would seem that the model atmosphere re- westerly than what occurred during the November 2001 quired a substantial synoptic-scale south-southwesterly storm (Fig. 4c). Perusal of the model cross sections and flow to assist in advecting the low-level wind maxima

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC 3560 MONTHLY WEATHER REVIEW— SPECIAL SECTION VOLUME 133

FIG. 12. Time series of Casey surface wind direction, wind speed, surface pressure, and temperature (thin continuous lines) for the period 2300 UTC 29 Oct 2001–2300 UTC 2 Nov 2001, overlaid with 96 h of 27.5-km ALAPS model data from a run initiated at 2300 UTC 29 Oct 2001 (thick line with cross markers) and 96 h of 11-km ALAPS model data (thick lines with circle markers) initiated at the same time. off the steep escarpment inland of Casey, which was the station. One possible explanation for the model fail- present in the November 2001 run of the model, but not ure in not picking up what appeared to be more of a in the October 2001 example. Although the Casey ra- true katabatic flow at Casey than the November storm, diosonde flight from 2321 UTC 30 October 2001, at the was model resolution. With a horizontal resolution of height of the storm, was incomplete, it suggested that only 27.5 km it was possible that ALAPS was not ad- the actual tropospheric flow was quite westerly with a equately defining the orography in the Casey area. very light outflow (southerly) component right through To test the hypothesis, a higher-resolution (11.0 km) the troposphere, so dynamically there appeared to be run of ALAPS was performed, nested within the 27.5- significant differences between the two storm environ- km resolution run and using five additional sigma levels ments. in the lowest 500 m. The model dynamics and physics The southeast–northwest cross section through Ca- remained unchanged. Figure 11c shows the southeast– sey from the ALAPS ϩ24 h forecast, at the height of northwest cross section through Casey from the higher- the storm (Fig. 11a), showed a low-level jet well inland resolution run at the height of the storm. The low-level of Casey but with little or no upper outflow evident at jet was not only stronger, but deeper and extended all. The near-surface flow in the Casey region (Fig. 11b) downslope and over Casey. The near-surface winds showed a strong outflow off the coast just to the south- from the higher-resolution run (Fig. 11d) clearly west of Casey so the model was picking up what was showed a stronger outflow from the Vanderford Glacial essentially a katabatic signature, but just not reaching basin reaching eastward to Casey. In general, the near-

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC DECEMBER 2005 ADAMS 3561 surface winds in the higher-resolution run were stron- tabatic wind usually subsides so quickly on the coast. In ger over the entire coastal escarpment region west and the October 2001 case, no such cold air damming was inland of Casey. evident in the cross sections along 109°E (not shown) A comparison of the time series output (Fig. 12) from nor in the northwest–southeast cross section through both the low- and higher-resolution runs, compared Casey (Fig. 11a), leaving the katabatic force unopposed with the Casey observations (thin unmarked lines), by any thermal wind component. showed the higher-resolution model run (thick lines It would seem likely that the south-southwest flow with circle markers) performed significantly better in evident throughout the troposphere in the November capturing the gale-force southerly event than the lower- 2001 case was strong enough to overcome the cold air resolution run (thick lines with cross markers). The damming and advect the katabatic off the escarpment higher-resolution run may have underforecast the and coastward to Casey. In the October 2001 case, strength of the event by some 40%, but did actually there was no deep tropospheric south-southwest flow to capture the event accurately in onset and cessation assist the surface katabatic. However, there was also no times, where the low-resolution model failed entirely. evidence of coastal cold air damming to oppose the The high-resolution model also provided a better fore- katabatic force, leaving the developing katabatic able cast of surface parameters out into the longer term, to extend off the escarpment and over Casey. captured the pressure trends better, and provided a more realistic diurnal temperature signal. Acknowledgments. This research was undertaken at the University of Tasmania as part of a Ph.D. and the author would like to thank Professor W. F. Budd for his 4. Conclusions guidance and helpful suggestions in undertaking this The analysis of the November 2001 southerly gale study. suggested that Casey Station does not experience a true, strong southerly katabatic flow, as the strong sur- REFERENCES face southerlies appeared to be associated with a Adams, N., 2004: A numerical modeling study of the weather in strong, upper south-southwesterly jet, maintaining a East Antarctica and the surrounding Southern Ocean. Wea. wind profile that was not simply katabatic flow. How- Forecasting, 19, 653–672. ever, the sequence of north–south cross sections Ball, F. K., 1960: Winds on the ice slopes of Antarctica. Antarctic through the Vanderford Glacier region and the south- Meteorology, Pergamon Press, 9–16. Murphy, B. F., and I. Simmonds, 1993: An analysis of strong wind east–northwest cross sections through Casey, detailed events simulated in a GCM near Casey in the Antarctic. Mon. here, do show the low-level development of a katabatic Wea. Rev., 121, 522–534. flow on the steep escarpment prior to the upper south- Parish, T. R., and D. H. Bromwich, 1987: The surface wind field westerly jet development. over the Antarctic ice sheets. Nature, 328, 51–54. In the November 2001 case it appeared that the ad- ——, and J. J. Cassano, 2003: Diagnosis of the katabatic wind influence on the wintertime Antarctic surface wind field from vection of the katabatic wind off the glaciers and out numerical simulations. Mon. Wea. Rev., 131, 1128–1139. over Vincennes Bay was triggered by the development Phillpot, H. R., 1997: Some observationally-identified meteoro- of the upper southwesterly jet and general increasing of logical features of East Antarctica. Meteorological Study No. south-southwesterly flow through the entire tropo- 42, Australian Bureau of Meteorology, Melbourne, Austra- sphere, driven by a fairly deep low to the east of Casey lia, 275 pp. [Available from the Bureau of Meteorology, GPO Box 1289, Melbourne, VIC 3001, Australia.] and a strong, well-developed ridge of high pressure to Puri, K., G. S. Dietachmayer, G. A. Mills, N. E. Davidson, R. A. the west. However, in the October case the tropo- Bowen, and L. W. Logan, 1998: The new BMRC Limited spheric steering flow was more westerly, with no south- Area Prediction System, LAPS. Aust. Meteor. Mag., 47, 203– erly outflow component to assist in advecting the ka- 223. tabatic off the escarpment and over Casey. Another Schwerdtfeger, W., 1984: Weather and Climate of the Antarctic. Vol. 15, Developments in Atmospheric Science, Elsevier, 261 significant difference between the two events was the pp. existence of significant cold air damming over the coast Seaman, R., W. Bourke, P. Steinle, T. Hart, G. Embery, M. in the November 2001 case, as evident in Fig. 9c. The Naughton, and L. Rikus, 1995: Evolution of the Bureau of potential temperature cross section in this figure Meteorology’s Global Assimilation and Prediction System. showed a significant pool of cold air over the coast, Part 1: Analysis and initialisation. Aust. Meteor. Mag., 44, 1–18. creating a thermal wind component opposing the ka- van den Broeke, M. R., and N. P. M. van Lipzig, 2003: Factors tabatic force. Van den Broeke and van Lipzig (2003), controlling the near-surface wind field in Antarctica. Mon. cited cold air damming as a primary reason the ka- Wea. Rev., 131, 733–743.

Unauthenticated | Downloaded 09/25/21 04:33 AM UTC