Foehn Winds in the Mcmurdo Dry Valleys, Antarctica: the Origin of Extreme Warming Events*

1JULY 2010 S P E I R S E T A L . 3577

Foehn Winds in the McMurdo Dry Valleys, Antarctica: The Origin of Extreme Warming Events*

JOHANNA C. SPEIRS Climate Research Group, School of Geography, Planning and Environmental Management, The University of Queensland, St. Lucia, Queensland, Australia

DANIEL F. STEINHOFF Polar Meteorology Group, Byrd Polar Research Center, and Atmospheric Sciences Program, Department of Geography, The Ohio State University, Columbus, Ohio

HAMISH A. MCGOWAN Climate Research Group, School of Geography, Planning and Environmental Management, The University of Queensland, St. Lucia, Queensland, Australia

DAVID H. BROMWICH Polar Meteorology Group, Byrd Polar Research Center, and Atmospheric Sciences Program, Department of Geography, The Ohio State University, Columbus, Ohio

ANDREW J. MONAGHAN Research Applications Laboratory, National Center for Atmospheric Research, Boulder, Colorado

(Manuscript received 13 August 2009, in ﬁnal form 5 January 2010)

ABSTRACT

Foehn winds resulting from topographic modification of airflow in the lee of mountain barriers are frequently experienced in the McMurdo Dry Valleys (MDVs) of Antarctica. Strong foehn winds in the MDVs cause dramatic warming at onset and have significant effects on landscape forming processes; however, no detailed scientific investigation of foehn in the MDVs has been conducted. As a result, they are often mis- interpreted as adiabatically warmed katabatic winds draining from the polar plateau. Herein observations from surface weather stations and numerical model output from the Antarctic Mesoscale Prediction System (AMPS) during foehn events in the MDVs are presented. Results show that foehn winds in the MDVs are caused by topographic modification of south-southwesterly airflow, which is channeled into the valleys from higher levels. Modeling of a winter foehn event identifies mountain wave activity similar to that associated with midlatitude foehn winds. These events are found to be caused by strong pressure gradients over the mountain ranges of the MDVs related to synoptic-scale cyclones positioned off the coast of Marie Byrd Land. Analysis of meteorological records for 2006 and 2007 finds an increase of 10% in the frequency of foehn events in 2007 compared to 2006, which corresponds to stronger pressure gradients in the Ross Sea region. It is postulated that the intra- and interannual frequency and intensity of foehn events in the MDVs may therefore vary in response to the position and frequency of cyclones in the Ross Sea region.

* Byrd Polar Research Center Contribution Number 1391. 1. Introduction Foehn winds result from topographic modiﬁcation of Corresponding author address: Johanna Speirs, Climate Re- ﬂow in the lee of mountain barriers. They are a climato- search Group, School of Geography, Planning and Environmental Management, The University of Queensland, Brisbane, QLD, logical feature common to many of the world’s midlati- 4072, Australia. tude mountainous regions, where they can be responsible 2 E-mail: [email protected] for wind gusts exceeding 50 m s 1 (Brinkmann 1971;

DOI: 10.1175/2010JCLI3382.1

Ó 2010 American Meteorological Society Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3578 JOURNAL OF CLIMATE VOLUME 23

McGowan and Sturman 1996a) and warming at foehn suggested that the winds are foehn and result from onset of 1288C (Math 1934). Intensive monitoring ex- the topographic modification of southwesterly airflow periments in midlatitude regions such as the Alpine Ex- (Thompson et al. 1971; Thompson 1972; Riordan 1975; periment (ALPEX) (Kuettner 1986; Seibert 1990) and the Keys 1980; Bromley 1985; Clow et al. 1988; McKendry Mesoscale Alpine Programme (MAP; Bougeault et al. and Lewthwaite 1990, 1992). 2001) in the Alps have detailed the complex atmospheric Episodes of strong and warm westerly airflow in the processes that occur during foehn by use of high-density MDVs have generally been attributed to adiabatic observational networks, aircraft, satellite imagery, and warming of descending air, however, the disagreement mesoscale numerical modeling. The classic mechanism lies with the exact forcing mechanisms. Katabatic ter- used to describe foehn from midlatitude studies involves minology was initially used by Bull (1966) to explain the forced orographic lifting of moist air over a mountain westerly winds in surface-based meteorological obser- barrier. As the air rises it cools at the dry-adiabatic lapse vations in the MDVs. Based on the thermal character- rate (DALR) of 9.88Ckm21. If saturation occurs, then istics of the westerly flow, Thompson et al. (1971) and further lifting will result in the air cooling at the slower Thompson (1972) reject a katabatic origin of these winds saturated adiabatic lapse rate of 48–68Ckm21.Descending and suggest they are foehn resulting from the aero- air on the lee side of the barrier compresses and warms at dynamic deflection of strong southerly-to-westerly mid- the DALR, also reducing the relative and absolute hu- tropospheric flow to the valley floor. Similarly, Riordan midity of the air (Barry and Chorley 2003). However, as (1975), Keys (1980), and Bromley (1985) all note that the discussed by Seibert (1990), the classic foehn mechanism westerlies are more likely of foehn origin rather than does not always fit with observations because moisture katabatic. Clow et al. (1988) also identified strong west- removal and cloud formation over mountains is not nec- erly winds as foehn and noted that the MDVs do not lie in essary for foehn development. Foehn winds can develop as a katabatic convergence zone; however, they consider the air is forced to descend over lee slopes, resulting from foehn winds to be instigated by katabatic surges draining large-amplitude mountain waves (e.g., Klemp and Lilly from the continental interior. The only research on the 1975; Flamant et al. 2002; Jiang et al. 2005), and by the atmospheric structure during westerly airflow was by blocking of low-level winds upwind of mountain barriers, McKendry and Lewthwaite (1990, 1992) in the Wright causing subsequent descent of air over lee slopes from near Valley during summer using observations from pilot the ridge top (e.g., Parish 1983; Doyle and Shapiro 2000; balloons, airsondes, an acoustic sounder, and a network of Gohm et al. 2004; Jiang et al. 2005). Generally, for the surface-based weather stations. These authors concluded development of foehn, the flow needs to be directed within that the westerly winds are foehn, caused by strong upper- 308 of perpendicular to the ridgeline with a steep pressure level flow deflected down into the valley and related to gradient across the mountain (Durran 1990). The strongest a synoptic situation characterized by low surface pressure foehn winds are associated with flow nearly normal to the in the Ross Sea. They note an atmospheric structure ridgeline (Za¨ngl 2003). similar to foehn winds elsewhere with a stagnant layer and Studies of foehn in high latitudes and polar regions, a midtropospheric capping inversion above the strong however, are rare. Analysis of meteorological records near-surface winds. Despite these observations, McKendry from the McMurdo Dry Valleys (MDVs), a unique ice- and Lewthwaite (1990) did not dismiss the alternate free area of the Antarctic, has identified that foehn winds explanation that westerlies are associated with katabatic are responsible for unprecedented temperature changes of surges from the polar plateau. In more recent studies, .1408C. The resulting warm, dry, and gusty winds are Doran et al. (2002) and Nylen et al. (2004) classified all suspected to have significant effects on landscape forming strong down-valley, westerly winds in the MDVs as kata- processes in the MDVs, including glacial melt (Welch batic winds draining from the polar plateau, although et al. 2003; Doran et al. 2008), rock weathering (Selby et al. no investigations into the forcing mechanisms were un- 1973), aeolian processes (Ayling and McGowan 2006; dertaken. They did, however, document important spatial Speirs et al. 2008), and biological productivity (Fountain and annual statistics of these ‘‘warm’’ winds, such as the et al. 1999; Foreman et al. 2004). increase in frequency in the western sections of the valleys Despite the significance of foehn in this region, no with the highest frequency of events occurring in winter. detailed scientific investigation of foehn has been con- Two coauthors (J. C. Speirs and H. A. McGowan) have ducted in the MDVs. As a result, the cause of these warm personally observed lenticular clouds in the MDVs during winds remains equivocal, with some studies invoking the warm westerly wind events in summer, which are in- adiabatic warming of katabatic winds descending from dicative of mountain wave activity and deep transbarrier the polar plateau to explain their occurrence (Bull 1966; flow rather than shallow katabatic drainage from the polar Doran et al. 2002; Nylen et al. 2004), while others have plateau.

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3579

FIG. 1. (a) Map of the Ross Sea region of Antarctica. (b) Inset of black box in (a) showing the MDVs region. (c) The MDVs AWS network: VV, WV, Lake Brownworth (WB), BV, TTa, TB, TH, THo, TCa, TF, TCo, and TE. Landsat Enhanced Thematic Mapper (ETM1) image captured 21 Nov 2001.

McKendry and Lewthwaite (1992, p. 596) concluded 4800 km2, the largest ice-free area in Antarctica. Large that ‘‘further work is required to clarify the interactions mountain ranges rising over 2000 m above sea level sepa- between synoptic-scale flow and the rather unusual to- rate the valleys, which have a polar desert climate because pographic setting, and to explain the exact mechanism of their location in a precipitation shadow of the Trans- by which upper level flow is deflected into the [Wright] antarctic Mountains (Monaghan et al. 2005). Annual pre- valley.’’ Since McKendry and Lewthwaite’s statement cipitation is ,50-mm water equivalent, with precipitation almost two decades ago, no comprehensive study into decreasing away from the coast (Fountain et al. 2009). the meteorology of foehn wind events of the MDVs has Mean annual air temperature from seven valley floor been undertaken and the cause of these winds remains AWSs range between 214.88 and 2308C, depending on a subject of considerable debate. This study presents site location and the period of measurement (Doran et al. findings resulting from the ongoing collaborative re- 2002). The wind regime of the MDVs is strongly domi- search between The University of Queensland and The nated by either up- or down-valley topographically Ohio State University combining observational and channeled airflow. During summer, thermally generated model data to broaden the understanding of the com- easterly valley winds dominate. This circulation develops plex terrain mountain meteorology in the MDVs. Model due to differential surface heating between the low- output from the Antarctic Mesoscale Prediction System albedo valley floors and the high-albedo glacier surfaces (AMPS) project is combined with data from a network to the east, analogous to sea-/lake-breeze circulations of automatic weather stations (AWSs) to 1) detail the elsewhere (McKendry and Lewthwaite 1990). In winter, typical meteorological conditions and synoptic forcing wind direction is typically more variable. Cold air pools mechanisms associated with a foehn wind event, and 2) associated with light winds and minimum temperatures provide a 2-yr synoptic climatology of foehn wind events ,2508C often occupy topographic low points of the during the calendar years of 2006 and 2007. valleys during winter (Doran et al. 2002). Topographi- cally channeled southwesterly wind events that we report here as foehn winds are frequently recorded throughout 2. Physical setting the year in the MDVs (Thompson 1972; Keys 1980; Clow The MDVs are situated in the Transantarctic Moun- et al. 1988; McKendry and Lewthwaite 1990; 1992; Ayling tains, bounded by the McMurdo Sound/Ross Sea to the and McGowan 2006; Speirs et al. 2008). east and the east Antarctic ice sheet to the west (Fig. 1). 3. Methods The MDVs consist of three large northeast–southwest- trending ice-free valleys (the Victoria, Wright, and Taylor Meteorological data presented here were obtained from Valleys), which collectively cover an area of approximately AWSs operated by the McMurdo Dry Valleys Long-Term

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3580 JOURNAL OF CLIMATE VOLUME 23

TABLE 1. MDV AWS information.

ID Location Station Lat, lon Elev (m ASL) VV Victoria Valley Lake Vida 77.37788S, 161.80068E 351 WV Wright Valley Lake Vanda 77.51688S, 161.66788E 296 WB Lake Brownworth 77.43358S, 162.70368E 279 TE Taylor Valley Explorers Cove 77.58878S, 163.41758E26 TF Lake Fryxell 77.61098S, 163.16968E19 TH Lake Hoare 77.62548S, 162.90048E78 TB Lake Bonney 77.71448S, 162.46418E64 TTa Taylor Glacier 77.74028S, 162.12848E 334 THo Howard Glacier 77.67158S, 163.07918E 472 TCa Canada Glacier 77.61278S, 162.96348E 264 TCo Commonwealth Glacier 77.56378S, 163.28018E 290 BV Beacon Valley Beacon Valley 77.82808S, 160.65688E 1176

Ecological Research (LTER) program (Doran et al. 1995). Meteorology Group at the Byrd Polar Research Center, Table 1 lists the location and station identification for the The Ohio State University (OSU) in support of U.S. AWSs used here. The configuration of these stations is Antarctic Program (USAP) operations. AMPS employs a detailed online (see http://www.mcmlter.org/queries/met/ version of the fifth-generation Pennsylvania State Uni- met_home.jsp; see also Doran et al. 2002). Measurements versity (PSU)–NCAR Mesoscale Model (MM5; Grell are collected at 3 m above the surface except for the etal.1994),PolarMM5,whichisoptimizedforusein Canada Glacier (TCa), where air temperature and rela- polar regions by OSU (Bromwich et al. 2001; Cassano tive humidity measurements are made 2 m above the et al. 2001). Polar MM5 includes a modified parameteri- surface. zation for the prediction of ice cloud fraction, improved A selection criterion was developed to identify foehn cloud–radiation interactions, an optimal stable boundary wind events in the MDVs AWS records similar to layer treatment, improved calculation of heat transfer studies of Northern Hemisphere foehn, such as Richner through snow and ice surfaces, and the addition of a frac- et al. (2006) and Gaffin (2007). Foehn onset in the tional sea ice surface type. MDVs was identified by an increase of wind speed AMPS output used in this case study is at 20-km res- above 5 m s21 from a southwesterly direction, a warm- olution, on a grid domain covering Antarctica and much ing of at least 118Ch21, and a decrease of relative hu- of the surrounding Southern Ocean, and at 2.2-km res- midity of at least 5% h21. Glacier AWSs were excluded olution on a grid encompassing the Ross Island area, from the foehn identification criteria because topograph- extending into the MDVs. There are 31 vertical half- ically controlled glacier winds experienced at these sta- sigma levels, with 11 levels in the lowest 1000 m to cap- tions have a westerly component and, in addition to weak ture the complex interactions in the planetary boundary warming associated with mixing, glacier winds can obscure layer. The lowest half-sigma level is about 13 m above the onset and cessation of foehn winds. Because of the tran- surface. For comparisons with AWS observations at 3 m sient nature of some foehn events, an additional criterion above the surface, AMPS winds were interpolated loga- of a ‘‘foehn day’’ was developed. A foehn day at an AWS rithmically from the lowest model level. Air temperature station is defined as a day that experiences 6 h or more of was interpolated linearly between the surface and the continuous foehn conditions with wind speed .5ms21 lowest model level, while relative humidity was estimated from a consistent southwesterly direction. We accept that with the interpolated temperatures, mixing ratios, and the classification of a foehn day excludes weak and brief calculated pressure. periods (,6 h) of foehn winds, which are more difficult AMPS Polar MM5 is initialized twice daily at 0000 to distinguish from, for example, local glacier winds. To and 1200 UTC. The initial and boundary conditions are quantify temporal and spatial trends of foehn events in derived from the National Centers for Environmental the AWS observations and link to model circulation data, Prediction (NCEP) Global Forecasting System (GFS) a criterion such as the foehn day is necessary. model. AMPS uses three-dimensional variational data Numerical forecast model products presented here were assimilation (3DVAR; Barker et al. 2004). The observa- obtained from AMPS (Powers et al. 2003). AMPS is tions available for assimilation into AMPS include re- a collaboration between the Mesoscale and Microscale ports from radiosondes, surface synoptic (SYNOP) reports, Meteorology (MMM) division of the National Center AWS observations, ship and buoy reports, atmospheric for Atmospheric Research (NCAR) and the Polar motion vector winds from satellites, and GPS radio

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3581

show that the same variables are well resolved at synoptic time scales. Monaghan et al. (2005) reviewed the climate of the McMurdo region (including the MDVs) in the 3.3-km-grid domain and show that AMPS captured important temporal and spatial aspects of the region’s climate with skill. Additionally, Steinhoff et al. (2008) demonstrated that the 3.3-km domain is valuable on an event basis in the analysis of a downslope windstorm at Ross Island. In terms of numerical modeling of foehn events elsewhere, MM5 has been successfully used to model foehn dynamics in the European Alps as part of MAP (Gohm et al. 2004; Za¨ngl 2003). Comparisons of AMPS time series and AWS data by this study shows that the 2.2-km domain performs well in the MDVs. The model points chosen for this comparison were those closest to the AWS location and elevation. In this paper, AMPS products for the 2.2-km domain are used to determine the regional ﬂow characteristics in which the inﬂuence of these near-surface effects is markedly reduced. Subsets of the 20-km domain are also uti- lized to examine synoptic-scale circulation characteristics during foehn events.

4. Case study: Foehn meteorology 20–27 May 2007 a. Synoptic environment The synoptic-scale meteorological situation associated with a winter foehn event is presented here. This event is representative of other events in the MDVs, including those in summer [synoptic composites presented in the following section; see also McKendry and Lewthwaite (1990) and Speirs et al. (2008) for field observations of summer foehn events]. Using the foehn identification criteria we isolated a strong foehn event in the MDVs AWS records with onset on 21 May 2007. This event lasted 5 days with wind gusts up to 38.9 m s21 and induced warming of up to 148.58C at the valley floor. Figure 2 FIG. 2. AMPS (a) SLP (and near-surface wind vectors) and shows the synoptic scale meteorology at 0000 UTC (b) 500-hPa geopotential height analyses prior to foehn event 19 May 2007 (prior to foehn onset). The sea level pressure (0000 UTC 19 May 2007). Star denotes location of MDVs. Note SLP (SLP) charts at this time (Fig. 2a) show a cyclonic de- data above 500 m is masked because of uncertainties in calculating pression north of the Ade´lie Land coast with a minimum SLP over the cold and high-elevation Antarctic continent. central pressure of 962 hPa, a broad area of low pressure over the Amundsen Sea, and weak synoptic pressure occultation soundings. AMPS ingests sea ice data daily gradients over the western Ross Sea region. Sea level from the National Snow and Ice Data Center for its pressure calculation over the Antarctic interior is erro- fractional sea ice depiction. neous because of the height of the topography, and is not Guo et al. (2003) evaluate Polar MM5 performance over shown. The 500-hPa charts are more accurate approxi- Antarctica for a 1-yr period (1993) on a 60-km-resolution mations of the lower-level synoptic-scale forcing and domain and show that the intra- and interseasonal vari- geostrophic wind direction over the Antarctic interior ability in pressure, temperature, wind, and moisture are (Phillpot 1991) and show weak pressure gradients over well resolved. Bromwich et al. (2005) evaluate 2 yr of East Antarctica and along the Transantarctic Mountains AMPS Polar MM5 forecasts on the 30-km domain and (Fig. 2b). Surface airflow over the study area in the days

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3582 JOURNAL OF CLIMATE VOLUME 23

By 0000 UTC 23 May 2007, the cyclone evident in Fig. 2a had tracked eastward and strengthened with a minimum central pressure of 950 hPa (Fig. 3a). During this time, a large ridge developed over east Antarctica as seen in the 500-hPa chart in Fig. 3b. The cyclonic system then slowed and remained relatively stationary in a position north of the coast of Marie Byrd Land between the Ross and Amundsen Seas (Fig. 3) until 26 May 2007. This synoptic setting produced a baroclinic zone (not shown) with a strong southeast–northwest pressure gradient and strong southwesterly airflow across the western Ross Sea and Transantarctic Mountains region for several days (Fig. 3b). The cyclonic system then began to weaken on 26 May 2007 and the region of low pressure propagated eastward around the West Antarctic coast. b. Local meteorological observations AWS observations and AMPS model forecasts (for grid points nearest and most representative of the AWS locations) during the foehn event are shown in Fig. 4 for the western MDVs region and Fig. 5 for the eastern Taylor Valley region. Model performance is described further in the following section. Prior to foehn onset, meteorological conditions on the floors of the MDVs were cold and calm, while at the higher-elevation glacier stations, cold and moist downslope winds approached 8ms21 [e.g., Howard Glacier (THo); see Fig. 5c). These winds are localized glacier or slope winds from the surrounding mountain ranges with flow at Taylor Glacier (TTa), TCa, Commonwealth Glacier (TCo), and THo (Fig. 1c) directed toward the valley floor. Cold air draining to the valley floor accumulates at the topographic low points resulting in stable cold air pools with near-surface (3 m) air temperatures below 2408C and relative humidity .80%. Coldest air temperatures are recorded at Lake Vida (VV; 253.58C), which reflects the relative strength of cold pool formation in the Victoria Valley compared to the Wright and Taylor Valleys (Doran et al. FIG. 3. AMPS (a) SLP (and near-surface wind vectors) and 2002). It is difficult to accurately estimate the inversion (b) 500-hPa geopotential height analyses during the foehn event strength and depth of the near-surface cold pools in the (0000 UTC 23 May 2007). MDVs because vertical profile observations during winter are unavailable. Whiteman et al. (2004) demonstrate that valley sidewall temperatures can be used in place of free- prior to foehn onset was dominated by katabatic winds air vertical profile measurements under clear and stable draining from the east Antarctic ice sheet and localized conditions. Accordingly, comparisons of temperatures on cold air drainage winds in the MDVs. Surface wind vec- the valley sidewalls in the MDVs to those on the valley tors at the lowest model sigma level (approximately 13 m floor would suggest that cold pooling may be in the order above surface; Fig. 2a) show katabatic divergence west of of ;108C in the Taylor Valley [Lake Fryxell (TF); Lake the MDVs with winds draining out the large glacial valleys Bonney (TB)], ;158C in the Wright Valley [Lake Vanda south (Byrd, Mulock, and Skelton Glaciers) and north (WV)], and ;208C in the Victoria Valley (VV). Doran (David and Reeves Glaciers) of the MDVs. Higher- et al. (2002) suggest that the greater cold pool strength resolution streamlines (not shown) also display this kata- near Lake Vida is related to the valley’s bowl-shaped batic divergence west of the MDVs. topography. The exposed yet closed topography would

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3583

FIG. 4. Foehn meteorological observations and AMPS model output 19–28 May 2007 for the western MDVs region: (a) TB (Taylor Valley), (b) WV (Wright Valley), and (c) VV (Victoria Valley). AWS data were resampled at 30-min averages centered on the hour to match AMPS output. result in more intense radiative cooling and formation of a characteristic also noted by Nylen et al. (2004), al- a stronger temperature inversion (e.g., Clements et al. though they referred to the winds as katabatic. Onset of 2003). The stably stratified inversion in the valley floors strong foehn winds first occurs on 21 May 2007 at TTa may decouple from winds above and explain why drain- (0315 UTC) and TB (0945 UTC), followed by WV age winds recorded at the glacial stations are not observed (2315 UTC) in the adjacent Wright Valley (Table 2). at the valley floor stations in the days leading to foehn Onset is characterized by an immediate increase in wind onset. speed .10 m s21 from a consistent southwest direction, Between 0930 and 1000 UTC 20 May 2007, a gradual an increase in air temperature, and a decrease in relative warming commenced at all MDV valley floor AWSs humidity. It is almost 24 h after initial foehn onset in the (Figs. 4 and 5). This warming characterizes the ‘‘prefoehn western Taylor Valley when strong winds are recorded at conditions’’ of foehn events in the MDVs, as noted in the eastern stations (TE, TF, TH, TCa, TCo, and THo). other case studies by McGowan and Speirs (2008), and is Lake Vida was the last station to identify foehn onset at believed to be associated with the gradual erosion of the 1700 UTC 22 May 2007. This station is often the last to stably stratified cold air pool in the valley floors from record foehn winds, possibly because of the longer time above by the foehn. Warming of approximately 108C required to erode the cold air pool that forms over the was observed over the 24 h prior to onset of strong Lake Vida depression and the more open nature of the foehn winds. Coupling of foehn winds to the surface valley topography. As the cold air pool was eroded by shows significant spatial complexity through the Victoria, foehn winds, a warming of 129.78C in 3 h was recorded at Wright, and Taylor Valleys (Table 2). Foehn conditions foehn onset and 148.58C over the course of the event. were initially monitored in the western Taylor Valley, While the most dramatic warming occurred on the valley

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3584 JOURNAL OF CLIMATE VOLUME 23

FIG. 5. Foehn meteorological observations and AMPS model output 19–28 May 2007 for the eastern Taylor Valley region: (a) Lake Hoare (TH), (b) Explorer’s Cove (TE) and (c) Howard Glacier (THo). AWS data were resampled at 30-min averages centered on the hour to match AMPS output.

floors, valley sidewall stations still exhibited significant foehncessationatTE,TF,TH,TCa,andTCointhe warming with temperature changes of 124.88, 124.58, Taylor Valley; however, at THo on the southern val- and 126.28C observed at THo, TCo, and TCa respectively ley wall (Kukri Hills), foehn conditions prevailed until during the foehn event. 0200 UTC 26 May when it was replaced by the onset of A break in westerly foehn winds was monitored in the light southerly drainage winds. Foehn conditions ceased eastern Taylor Valley (TE, TF, TH, TCa, and TCo) on at all remaining western valley floor stations [TB, WV, 23 May 2007 and continued until early 24 May 2007. During VV, and Beacon Valley (BV)] by 0745 UTC 26 May with this time, easterly winds intruded from McMurdo Sound the return to pre-foehn conditions dominated by light with speeds ,10 m s21 and relative humidity .90% and cold drainage winds. Foehn cessation at these stations caused air temperatures to fall below 2258C (Figs. 5a,b). was marked by an immediate drop in wind speed, a The influence of these easterlies decreased at sites further gradual decrease in air temperature, and an increase in west in the MDVs, with Lake Hoare (TH) only briefly relative humidity. Postfoehn air temperatures remained experiencing a break in foehn. Foehn conditions then elevated (compared to prefoehn) for several days fol- became reestablished at all stations following the break lowing the event, a feature also noted by Nylen et al. when maximum wind gusts were recorded (Table 2). (2004). Foehn cessation at Taylor glacier (TTa) was not Foehn cessation was first observed in the eastern Taylor clearly identifiable because of the strong westerly glacier Valley on 2130 UTC 24 May with a strengthening of winds experienced at this station; however, temperature cool and moist easterly winds with initial gusts up to began to decrease and relative humidity increased at 26.2 m s21. Wind speed later decreased to below 10 m s21 1000 UTC 26 May, which may signify cessation of foehn on 0900 UTC 25 May. Onset of strong easterlies caused winds.

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3585

TABLE 2. Foehn characteristics at MDV AWSs.

Min prefoehn Max foehn air Min Mean foehn Max foehn air temperature temperature foehn wind gust speed AWS Onset (UTC) Cessation (UTC) (8C) (8C) RH (%) direction (8) (m s21) TTa 0315 UTC 21 May 2007 1000 UTC 26 May 2007 234.2 25.1 10.1 224.8 33.3 TB 0945 UTC 21 May 2007 0630 UTC 26 UTC 2007 241.4 23.5 6.9 247.9 31.9 TH 0300 UTC 22 May 2007 2315* UTC 25 May 2007 236.1 23.9 12.8 237.5 29.6 TF 0500 UTC 22 May 2007 0245* UTC 25 May 2007 244.1 27.3 13.8 234.3 34.5 TE 0700 UTC 22 May 2007 2130* UTC 24 May 2007 240.8 28.1 17.0 238.2 29.9 TCa 0300 UTC 22 May 2007 0800* UTC 25 May 2007 231.6 25.4 12.5 234.8 35.7 TCo 0445 UTC 22 May 2007 2300* UTC 24 May 2007 232.0 27.5 13.2 245.8 38.9 THo 0500 UTC 22 May 2007 0200 UTC 26 May 2007 230.9 26.1 10.9 215.6 34.8 WV 2315 UTC 21 May 2007 0745 UTC 26 May 2007 246.5 24.6 12.1 251.6 32.3 VV 1700 UTC 22 May 2007 0715 UTC 26 May 2007 253.5 25.0 7.9 232.8 27.9 BV 1615 UTC 22 May 2007 0345 UTC 26 May 2007 236 210.0 7.0 195.2 32.7

* Noncontinuous foehn conditions between onset and cessation.

During this event strong southerly foehn winds were in AMPS with strengthening (weakening) of wind speed monitored in the Beacon Valley in the southwest MDV at foehn onset (cessation) within 12 h of the observed region (see Fig. 1). The Beacon Valley is sheltered by increase (decrease) in the AWS records from all sta- mountain ranges in all directions except to the northeast tions. Mean wind speeds during foehn were overes- where it opens to the upper Taylor Glacier. Given these timated in the eastern Taylor Valley (12.9 m s21 at TE, topographic constraints it is unlikely negatively buoyant TF, and TH) while they were underestimated at the katabatic drainage from the east Antarctic ice sheet western surface stations (22.5 m s21 at TB, WV, and could enter the Beacon Valley. VV). Temperature errors were markedly reduced during foehn with mean bias of 24.08C at valley floor stations. On the valley sidewalls (THo, TCa, and TCo) c. Model performance in the MDVs mean temperature errors were reduced to 21.88C dur- Prior to foehn onset, AMPS has difficulty representing foehn conditions. The break in foehn conditions and ing the near-surface inversion associated with local cold intrusion of easterly winds in the eastern Taylor Valley air pooling observed at the valley floor AWSs. At in- is also represented in the AMPS model at Explorer’s dividual grid points, AMPS shows light winds and a Cove (TE) and TF. AMPS output suggests that these warm/dry bias while near calm conditions are observed at easterlies are caused by forced deflection of southerly the valley floor AWSs (Figs. 4 and 5). The errors appear winds by Ross Island (not shown). to increase with strength of cold air pooling with prefoehn Evidently, even with a relatively finescale spatial temperature bias (model-observed) of 112.78Conthe resolution (2.2 km), AMPS has difficulties reproducing valley floor in the Taylor Valley (TE, TF, TH, and TB), point observations of the near-surface conditions. This 122.88C in the Wright Valley (WV) and 126.48Cinthe emphasizes the complexity of the local-scale mixing Victoria Valley (VV). These issues could be due to the processes in the MDVs environment, particularly during smooth modeled topography which may not accurately weak-mixing conditions. However, the AMPS–AWS detail the topographic low points in the valleys and also comparisons presented here do demonstrate that AMPS due to the PBL physics in AMPS. Monaghan et al. (2003) resolves the onset and cessation of foehn winds on the and Steinhoff et al. (2009) both note similar problems valley floors and reasonably replicates meteorological with AMPS temperature and humidity during stable parameters on the valley sidewalls, away from the influ- conditions and attributed these issues to the PBL scheme. ence of localized effects. It is especially encouraging that Temperature and humidity model errors in the MDVs AMPS recognizes the onset and cessation of strong wind are only substantial during weak mixing conditions on the events in the MDVs considering the variation of timing valley floor. At glacier stations away from the influence of of foehn onset between stations. The ability of AMPS to localized cold air pooling, the temperature bias (14.68C approximate the observed foehn events at fine scales at THo, TCa, and TCo) is markedly reduced and AMPS suggests that the regional and large-scale forcing in the closely follows AWS observations (e.g., THo in Fig. 5c). outer AMPS domains is accurate. Therefore, we con- AMPS performance is greatly improved during foehn clude that AMPS is suitable for analyzing the regional conditions where stronger forcings are present. Onset airflow and synoptic conditions during foehn in the and cessation of strong westerly winds are identified well MDVs, the primary focus of this study.

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3586 JOURNAL OF CLIMATE VOLUME 23

FIG. 6. AMPS 6-h backward trajectories for air parcels arriving at MDV AWS station sites (at approximately 13 m above the modeled surface) at 0600 UTC 25 May 2007. Dashed line shows the location of the cross section in Fig. 7. Wind vectors are for the lowest model level (13 m above the surface). d. Modeled foehn dynamics layer of ;23 K (Fig. 7b). Upstream winds are forced by the strong synoptic pressure gradient; however, given that Backward air parcel trajectories from 0600 UTC 25 May an inversion is present, a small component of this flow 2007 are presented in Fig. 6 in order to help identify may also be buoyancy-related flow. As these strong winds the origin of air arriving at model points (lowest vertical arrive to the MDVs, they interact with topography and level) best representing the AWSs in the MDVs during are forced to cross the mountain ranges rather than the foehn event. The trajectories are three-dimensional simply draining into the upper reaches of the valleys with both vertical and horizontal motions considered and from the west. In response, a prominent large-amplitude were run backward for 6 h prior to 0600 UTC 25 May. mountain wave pattern develops with vertical propaga- Figure 7 shows cross sections of wind speed and potential tion to levels at least 8 km above sea level (Fig. 7b). The temperature (at 0000 UTC 20 May and 0600 UTC 25 May) downward displacement of the isentropes in the valleys for the dashed line evident in Fig. 6 and highlight the implies a foehn mechanism. Similar mountain wave ac- marked difference in atmospheric structure associated tivity is commonly associated with foehn winds in moun- with foehn onset in the MDVs. tainous midlatitude regions (e.g., Beer 1976; Durran 1990; Prior to foehn onset, light winds and stable stratifica- Seibert 1990; Za¨ngl 2003). Wave development appears to tion of the lower troposphere are evident (Fig. 7a). commence at the Taylor Dome region with flow relatively During the foehn event, near-surface wind vectors (Fig. 6) laminar over the ice sheet west of the MDVs before show evidence of flow divergence on either side of the reaching this rise in topography. Wave amplitude in- Royal Society Range (peak elevation . 4000 m) and, to creases above the MDVs as strong flow crosses the chain a lesser extent, Taylor Dome (peak elevation of 2450 m). of mountain ridges. Large regions of lower wind speeds A layer ,700 m above the Taylor Dome region is char- appear above the valley centers associated with ascending acterized by wind speeds .30 m s21, a shallow near- air to the mountain wave crests, while wind speed maxima neutral layer adjacent to the surface capped by an inversion occur at the wave troughs near the north-facing valley

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3587

examine katabatic jump behavior (e.g., Yu et al. 2005), but it can be applied here to demonstrate the potential for downslope flow acceleration associated with mountain waves when flow is supercritical (Fr . 1). Clearly, Fr values are .1 for most the MDVs region during foehn with peaks in the calculations on the lee slopes in the Wright Valley and the upper reaches of the Taylor Valley demonstrating significant flow acceleration (Fig. 7). Deflection of airflow is evident along the valley axis, particularly in the Wright Valley (Fig. 6). This is a combination of forced channeling owing to the westerly component of the upstream flow but also pressure-driven channeling (see Whiteman and Doran 1993) associated with the strong horizontal pressure gradient across the MDVs from west to east. During the strongest winds late on 24 May 2007, wind direction at TTa turned almost southerly, demonstrating that airflow overcame topographic controls as it descended the slopes of the Kukri Hills (See Fig. 1). The trajectory analysis was performed when foehn winds ceased in the eastern Taylor Valley stations (TF and TE) and moist easterlies intruded from McMurdo Sound. The wave pattern evident in Fig. 7b remains relatively stationary through the course of the foehn event. Weak- ening of the synoptic cyclone off Marie Byrd Land and associated pressure gradients later on 26 May 2007 damp- ened the wave pattern with a return to near-stable stratifi- cation by 27 May 2007.

5. Foehn climatology 2006/07 a. Temporal and spatial variability Foehn events such as the 20–27 May 2007 event are a common occurrence in the MDVs. Figure 8 shows the frequency of foehn days for 2006 and 2007 at MDV valley floor AWSs. Most foehn events last less than 5 days, al- FIG. 7. AMPS cross section of wind speed (shaded) and potential though occasional events exceed a week in length. In the temperature (solid lines, contour interval 3 K) for (a) 0000 UTC two calendar years of 2006 and 2007, foehn days occurred 20 May 2007 and (b) 0600 UTC 25 May 2007. Location of cross on 28% of all days at TH, 27% at WV, and 10% of days at section shown in Fig. 6. Line plot above (b) is the dimensionless VV. Data suggest that a slightly higher frequency occurs at Froude number along the cross section for 0600 UTC 25 May 2007. TB in the western Taylor Valley region, although in- complete AWS data prevent a complete analysis. The walls. To examine the potential for downslope flow ac- highest frequency of foehn days occurred in winter (JJA, celeration associated with mountain wave activity, the 33% of winter days) followed by autumn (MAM, 21%), Froude number (Fr) was calculated, which can be rep- spring (SON, 18%), and summer (DJF, 16%). Because of resented as the high frequency of winter foehn events, nonfoehn u À1/2 conditions may only prevail for 1–2 days before onset Fr 5 u gh , (1) Du of another foehn event. Winter events frequently see temperature changes .408C, and on occasions cause where u is the wind speed, h is the height of the inversion temperatures to rise above 08C as occurred on 3 August layer, u is the average potential temperature of the layer, 2007 during a 5-day foehn event (10.38C at WV). Sum- Du is the potential temperature deficit (inversion strength), mer foehn events also regularly cause temperatures to and g is gravity. This definition of Fr is often used to rise above freezing and, in combination with intense

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3588 JOURNAL OF CLIMATE VOLUME 23

FIG. 8. Foehn days for (a) 2006 and (b) 2007 at MDVs valley floor AWS stations (no. denotes missing data). solar radiation, temperatures can exceed 1108C. The pressure is also more confined and centered closer to the highest temperatures during 2006 and 2007 were ach- Ross Sea in 2007. Figure 9c shows the sea level pressure ieved during a 3-day event in January 2007, which re- differences between 2007 and 2006 and statistical sig- sulted in a maximum temperature of 18.68C(WV). nificance (using a two-tailed Student’s t test) while Fig. 9d Figure 8 highlights significant spatial variability and demonstrates the pressure gradient differences using a complexity in the onset and duration of foehn events first-order centered finite difference of the averaged through the MDVs. Sites within the Taylor Valley (TB, pressure gradient of 2006 and 2007. The direction of the TH, TF, and TE) generally monitor foehn conditions pressure gradient (from low to high pressure) is shown concurrently; however, on occasion, foehn winds can be by an arrow at a sample location on the Ross Ice Shelf. experienced at either the eastern or western end of the Positive statistically significant differences in synoptic valley only. All events at Lake Vida in the Victoria Valley pressure gradients are evident across the Ross Ice Shelf occur simultaneously with events in the Taylor and Wright and near the MDVs between 2006 and 2007 and is likely Valleys. associated with the increase of foehn winds in 2007. Foehn wind frequency on the valley floors increases To isolate the synoptic configuration associated with substantially from an average of 16% foehn days in 2006 foehn in the MDVs, composites of SLP field and wind to 26% in 2007. This increase in foehn days is observed vectors for 2006 and 2007 were constructed based on 172 at all stations and during all seasons and represents an in- foehn days recorded at three or more valley surface crease in foehn event frequency in addition to an increase stations, compared to the 172 nonfoehn days. There in the duration of events. As a result of increased foehn were 398 total nonfoehn days through 2006 and 2007, (and warming) in 2007, annual average air temperatures at 172 were selected to match the sample number of foehn valley floor stations in the MDVs were 2.38–3.38Cwarmer days. Nonfoehn days were purposely excluded if they than 2006. This increase is in accordance with Nylen et al. occurred on either side of a foehn event to reduce the (2004), who suggest that an increase of katabatic [foehn] chance of prefoehn and postfoehn conditions being wind frequency in the MDVs would increase mean annual included in the nonfoehn analyses. AMPS backward temperatures. trajectories were produced for these 172 foehn days (Fig. 10) and display south-southwesterly, cross-barrier b. Synoptic climatology flow similar to the case study of 20–27 May 2007. The Annual mean sea level pressure and near-surface trajectories of VV are more consistent near the AWS wind vector composites presented in Figs. 9a,b demon- site than TB, with air being channeled to VV along the strate a slight tightening of isobars in the western Ross Balham and McKelvey Valleys in the upper reaches of Sea region in 2007 compared to 2006. The area of low Victoria Valley. At TB, the air parcels either cross the

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3589

FIG. 9. AMPS SLP and near-surface wind vector annual composites for (a) 2006 and (b) 2007, (c) SLP difference (2007 2 2006), and (d) pressure gradient difference (2007 2 2006). In (c) and (d) stippling is for positive differences, hatching is for negative differences. Light stippling/hatching refers to 90% conﬁdence level, heavy stippling/hatching refers to 95% conﬁdence level. Star denotes location of MDVs and arrow in (d) shows the direction of the pressure gradient.

Kukri Hills at a location near TB (similarly to Fig. 6) or pressure gradients can be seen in the Ross Sea region for enter the Taylor Valley farther west and are deflected by the annual (Fig. 11c) and summer (Fig. 12c) foehn com- the Asgard Range and channeled down valley along the posites. Differences in sea level pressure between foehn valley axis. and nonfoehn days are not statistically significant in Mean conditions presented in Figs. 11–13 clearly iden- winter (Fig. 13c) because of the quasi-stationary nature of tify the presence of a strong cyclonic system off the cyclones in the Ross Sea during these months (Simmonds coast of Marie Byrd Land during foehn days in contrast et al. 2003). However, the mean cyclone on winter foehn to weak pressure gradients during nonfoehn days. Sea days has a more closed pressure pattern with tightened level pressure differences and pressure gradient differ- isobars in the Ross Sea area, which would generate the ences were calculated in Figs. 11–13, similar to Fig. 9 statistically significant stronger pressure gradients (Fig. 13d) except subtracting nonfoehn days from foehn days. Sta- and stronger winds (Fig. 14c) along the Transantarctic tistically significant differences in sea level pressure and Mountains.

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3590 JOURNAL OF CLIMATE VOLUME 23

FIG. 10. AMPS backward trajectories for all 172 foehn days at (a) VV and (b) TB. The trajectories represent air parcels arriving to the surface at 1800 UTC on the foehn day.

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3591

FIG. 11. AMPS SLP and near-surface wind vector 2-yr composites (2006 and 2007) for (a) nonfoehn days and (b) foehn days, (c) SLP difference (foehn 2 nonfoehn), and (d) pressure gradient difference (foehn 2 nonfoehn). In (c) and (d) stippling is for positive differences, hatching is for negative differences. Light stippling/hatching refers to 90% conﬁdence level, heavy stippling/hatching refers to 95% conﬁdence level. Star denotes location of MDVs and arrow in (d) shows the direction of the pressure gradient.

The cyclonic system and associated pressure gradients and Bromwich 1998; Parish et al. 2006; Seefeldt and present on MDV foehn days have widespread effects Cassano 2008; Steinhoff et al. 2009). across East Antarctica. Composites of the winter mean 6. Discussion wind speed (Fig. 14) highlight stronger airﬂow across the Ross Ice Shelf and Ross Sea coast during foehn days. The MDVs frequently experience episodes of warm, Bromwich et al. (1993) and Seefeldt et al. (2007) both dry, and gusty winds, which are a dramatic climatologi- note that strong katabatic winds across the Ross Ice cal feature of this snow- and ice-free environment. AWS Shelf occur when pressure gradients are perpendicular observations of foehn events in the MDVs are not con- to the Transantarctic Mountains similar to those shown sistent with observations of katabatic winds elsewhere in in Figs. 11d–13d. A tongue of stronger airﬂow can also Antarctica (e.g., Wendler et al. 1997; Davolio and Buzzi be seen east of Ross Island, which is related to the cli- 2002; Renfrew and Anderson 2006). Cross sections and matological Ross Ice Shelf airstream (RAS; see Parish trajectory analyses from the AMPS 2.2-km-grid domain

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3592 JOURNAL OF CLIMATE VOLUME 23

FIG. 12. As in Fig. 11, except for summer [December–February (DJF)]. confirm that south-southwesterly cross-barrier airflow winter. Dramatic temperature changes on the valley floor is deflected from higher atmospheric levels to the sur- at foehn onset can be explained by the displacement of face as initially proposed by Thompson et al. (1971) and cold, stable air by potentially warmer air from upper McKendry and Lewthwaite (1990). This suggests that a levels, in addition to adiabatic warming as air is brought foehn mechanism is responsible for such strong wind to the surface from above ridge level. South-southwesterly events in the MDVs and that the influence of katabatic upper-level airflow is deflected by the valley walls, ob- surges from the polar plateau triggering events is mini- served as a southwesterly wind at the valley floor. In mal. A near-surface inversion layer is present on the polar addition to forced deflection, a component of the flow plateau upstream of the valleys and a small component of may be driven down valley from high to low pressure with this upstream flow may be buoyancy driven, however, the the strong along-valley pressure gradient. dominant driving mechanism for foehn wind occurrence Although foehn in the MDVs does not follow the ide- in the MDVs is the strong synoptic pressure gradients alized foehn formation mechanism with moisture removal (and strong winds) in the Transantarctic Mountain region on the windward side during upslope flow, the cross sec- caused by cyclonic activity in the Ross Sea. tions of potential temperature illustrate significant moun- Foehn-induced warming frequently exceeds 408Cwithin tain wave activity similar to those commonly associated several hours at valley surfaces in the MDVs during with foehn winds in midlatitude regions (e.g., Durran 1990;

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3593

FIG. 13. As in Fig. 11, except for winter [June–August (JJA)].

Seibert 1990; Za¨ngl 2003). It is difficult to draw detailed Hornsteiner (2007) in the Alps, although this requires comparisons between mountain wave patterns in the further examination beyond the scope of this paper. MDVs to those of midlatitude regions given the markedly The lower frequency of foehn events during summer different environments and topographic setup. However, compared to winter, also noted by Nylen et al. (2004), some similarities can be seen between the ‘‘deep foehn’’ may be accounted for by reduced synoptic activity in the in the Innsbruck region of the Alps, which displays com- Ross Sea in summer (Simmonds et al. 2003). Although parable depth of wave disturbances in the atmosphere mean synoptic activity is reduced, synoptic composites (.8 km; see Za¨ngl 2003; Gohm et al. 2004). Deep foehn indicate that the ‘‘mean’’ cyclonic system that develops in the Alps occurs when the large-scale wind direction is in this region during summer events is comparable in approximately perpendicular to the ridge line resulting in size and strength to winter. A well-defined synoptic-scale wind speeds .30 m s21 and warming up to 158Catthe cyclone may be necessary for the formation of foehn surface (Za¨ngl 2003; Gohm et al. 2004). A critical layer in winds in summer due to interactions with the thermally the upper troposphere is likely to play a role in trapping driven easterly circulation. This circulation is extremely and amplification of the wave pattern in the MDVs sim- well developed in terms of its strength, depth, and per- ilar to cases modeled by Za¨ngl et al. (2004) and Za¨ngl and sistence (McKendry and Lewthwaite 1990), and may

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3594 JOURNAL OF CLIMATE VOLUME 23

FIG. 14. AMPS average winter (JJA 2006 and 2007) near-surface wind speed for (a) nonfoehn days, (b) foehn days, and (c) foehn 2 nonfoehn difference. Stippling is for positive differences, hatching is for negative differences. Light stippling/hatching refers to 90% conﬁdence level; heavy stippling/hatching refers to 95% conﬁdence level.

decouple from airflow aloft, preventing the grounding behavior in this unique environment. 1) Down-valley of foehn winds to the valley floors for extended periods stations are sometimes affected before up-valley stations. (.6 h) unless overcome by particularly strong synoptic Foehn onset in the MDVs may occur at down-valley lo- forcing. Vertical wind profiles by McKendry and cations before sites in the upper valleys because of the Lewthwaite (1990) during a foehn event in the Wright topographically modified foehn south-southwesterly en- Valley show evidence of foehn overriding cooler easterly tering the valleys from an acute angle to the valley axis winds. Decoupling of foehn winds from local thermal such as through mountain passes (shown in Fig. 6). Tur- winds (lake breeze) at the surface was also found by bulent motion associated with mountain waves may also McGowan and Sturman (1996b) in the southern Alps, cause foehn onset at different times and locations along New Zealand. When southerly gradient winds are strong the valley axis, while the warm and dry down-valley foehn enough or the easterly circulation weakens (e.g., at night may decouple from the surface on encountering local cold when solar radiation is low) in the MDVs, foehn winds air drainage or cold air pools. The foehn may then ground are able to descend to the surface. farther down valley, possibly in response to local con- Results presented here extend observations of the vective turbulence and associated mixing over the warm temporal and spatial characteristics of foehn in the MDVs, snow-free dry valley floors. 2) Down-valley stations expe- while our modeling study clarifies several aspects of foehn rience strong easterlies during foehn events. As occurred in

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3595 the foehn case study presented here, cool easterly winds that the frequency and intensity of foehn events may vary can penetrate into the eastern section of the Taylor Valley with ENSO and other linked teleconnections, such as the from McMurdo Sound undercutting the foehn and caus- southern annular mode (e.g., Fogt and Bromwich 2006), ing it to decouple from the surface. These easterlies are which influence the position and frequency of these cy- associated with cyclonic low pressure systems and south- clonic systems in the Ross Sea region. An analysis of foehn erly flow parallel to the Transantarctic Mountains. Strong eventsintheMDVsAWSrecordsoveralongertime southerly near-surface flow along the Ross Ice Shelf is period is underway and will further the understanding of deflected by Ross Island and forced to flow in a north- how ENSO and other known drivers of climate variability westerly direction toward the MDVs (O’Connor and such as the southern annular mode are translated into Bromwich 1988; Seefeldt et al. 2003). When foehn winds interannual climate variability in the MDVs. are weak in the eastern sections of the MDVs, easterlies Additionally, variability seen in environmental pro- are able to penetrate into the valleys and are often ob- cesses in the MDVs may be linked to variability in the served immediately prior to foehn onset and/or cessation. foehn wind regime. For instance, Doran et al. (2008) The concept of an easterly ‘‘return flow’’ suggested by suggests that an increase in ‘‘down-valley winds’ (foehn) Nylen et al. (2004) is not plausible with a foehn mecha- in the summer of 2001/02 compared to 2000/01 was re- nism. Foehn winds are synoptically driven unlike circula- lated to significant glacier mass loss and an increase in tionssuchasvalleyswindswhereconservationofmass streamflow. The warmer foehn temperatures and increased infers the existence of an antiwind or return flow. 3) In- glacial melt during the 2001/02 summer also led to an creasing foehn frequency with distance from the coast. Our increase in lake levels, the thinning of the permanent ice analysis does indicate that a higher frequency of foehn covers, and other environmental effects (e.g., Foreman events occurs in the western sections of the MDVs. This et al. 2004; Barrett et al. 2008). Although foehn warming appears to be related to the intrusion of strong synoptically in summer is not as dramatic as in winter, foehn fre- deflected easterly flow (all seasons) or thermally gener- quently induces temperatures to rise above 08C, an im- ated easterly winds (in summer) into the eastern sections portant environmental threshold. of MDVs, thereby reducing the frequency of foehn at the surface in these areas. Lower topography at the eastern 7. Conclusions end of the Kukri Hills may also contribute to reduced frequency of foehn in the eastern Taylor Valley region. This paper presents initial findings from the analysis This research demonstrates how cyclonic systems near of observational records and modeling to further under- the coast of Marie Byrd Land (between the Ross and stand the complex meteorology of the unique MDVs, Amundsen Seas) result in strong gradient winds over the particularly during foehn events. A winter foehn event MDVs that lead to foehn formation. The Ross Sea is examined here presents the spatial and temporal com- a climatologically favored region for a high density of plexity associated with foehn onset and cessation in the cyclonic systems (Simmonds et al. 2003; Uotila et al. MDVs. AMPS output from the AMPS 2.2-km domain 2009), which are crucial in the development of southerly indicates that topographical interaction of synoptically winds over the MDVs and the associated development forced airflow with the Transantarctic Mountains causes of foehn winds. Accordingly, variability in cyclone ac- mountain wave activity that contributes to foehn wind tivity and the development of strong southerly pressure genesis in the MDVs. A climatological analysis of all 2006 gradients in this region will directly influence foehn and 2007 foehn events was performed and illustrates a frequency in the MDVs as shown in this study with 10% strong cyclonic low pressure system off the coast of Marie more foehn events in 2006 than 2007. The MDVs are Byrd Land, and resulting strong pressure gradients over the known to exhibit significant interannual climate vari- mountain ranges of the MDVs are responsible for foehn in ability (Welch et al. 2003; Bertler et al. 2006; Doran et al. the MDVs during all seasons. Importantly, this paper 2008), yet the causal mechanisms have not been explored. clarifies that a foehn mechanism is responsible for such Given the strong influence foehn winds have on the strong warm wind events in the MDVs and the influence of MDVs climate, it is highly likely that variability in the katabatic surges from the polar plateau as an origin or track and intensity of cyclonic systems in the Ross and triggering mechanism of events is minimal. Accordingly, we Amundsen Sea is a major contributor to MDV climate suggest the ‘‘katabatic’’ terminology adapted by researchers variability. Furthermore, the position and intensity of low in the MDVs when referring to these wind events be pressure systems in this region displays one of the most replaced by ‘‘foehn’’ to reflect their correct origin. prominent ENSO signals in the Antarctic (Carrasco and This study has demonstrated that AMPS products are Bromwich 1993; Cullather et al. 1996; Kwok and Comiso an effective tool to assist understanding large-scale cir- 2002; Bromwich et al. 2004). Accordingly, it is postulated culations, regional airflow and local-scale atmospheric

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3596 JOURNAL OF CLIMATE VOLUME 23 dynamics in this region of the Antarctic where few ob- ——, A. J. Monaghan, and Z. Guo, 2004: Modeling the ENSO servations exist. Further research is in progress detailing modulation of Antarctic climate in the late 1990s with the the complex atmospheric structure in the MDVs. Model Polar MM5. J. Climate, 17, 109–132. ——, ——, K. W. Manning, and J. G. Powers, 2005: Real-time simulations with high horizontal and vertical resolution forecasting for the Antarctic: An evaluation of the Antarctic using a polar-modified version of the Weather Research Mesoscale Prediction System (AMPS). Mon. Wea. Rev., 133, and Forecasting model (WRF; Skamarock et al. 2005), 579–603. tailored for use in the MDVs, will be run to explore Bull, C., 1966: Climatological observations in ice-free areas of meteorological features in greater detail. Important Southern Victoria Land, Antarctica. Studies in Antarctic Me- teorology, M. J. Rubin, Ed., Antarctic Research Series, Vol. 9, model validation by field research is planned in 2011 and Amer. Geophys. Union, 177–194. 2012. Carrasco, J. F., and D. H. Bromwich, 1993: Mesoscale cyclogenesis dynamics over the southwestern Ross Sea, Antarctica. J. Geo- Acknowledgments. The authors thank the McMurdo phys. Res., 98, 12 973–12 995. LTER Program for use of automatic weather station Cassano, J., J. Box, D. Bromwich, L. Li, and K. Steffen, 2001: data. This research is supported by the National Science Evaluation of Polar MM5 simulations of Greenland’s atmo- Foundation via NSF Grant ANT-0636523 and UCAR spheric circulation. J. Geophys. Res., 106, 33 867–33 889. Clements, C. B., C. D. Whiteman, and J. D. Horel, 2003: Cold-air- Subcontract S01-22901. We also acknowledge the ongo- pool structure and evolution in a mountain basin: Peter Sinks, ing financial support of the School of Geography, Plan- Utah. J. Appl. Meteor., 42, 752–768. ning and Environmental Management, The University of Clow, G. D., C. P. McKay, G. M. Simmons, and R. A. Wharton, Queensland. Finally, we are grateful for the constructive 1988: Climatological observations and predicted sublimation suggestions from the anonymous reviewers. This work rates at Lake Hoare, Antarctica. J. Climate, 1, 715–728. Cullather, R. I., D. H. Bromwich, and M. L. Van Woert, 1996: was completed while Speirs was a visiting research fellow Interannual variations in Antarctic precipitation related to at the Byrd Polar Research Center, The Ohio State El Nin˜ o–Southern Oscillation. J. Geophys. Res., 101, 19 109– University. 19 118. Davolio, S., and A. Buzzi, 2002: Mechanisms of Antarctic katabatic REFERENCES currents near Terra Nova Bay. Tellus, 54A, 187–204. Doran, P. T., G. L. Dana, J. T. Hastings, and R. A. Wharton, 1995: Ayling, B. F., and H. A. McGowan, 2006: Niveo-aeolian sediment McMurdo Dry Valleys Long-Term Ecological Research deposits in coastal south Victoria Land, Antarctica: Indicators (LTER): LTER automatic weather network (LAWN). An- of regional variability in weather and climate. Arct. Antarct. tarct. J. U.S., 30, 276–280. Alp. Res., 38, 313–324. ——, C. P. McKay, G. D. Clow, G. L. Dana, A. G. Fountain, Barker, D. M., W. Huang, Y.-R. Guo, A. J. Bourgeois, and T. Nylen, and W. B. Lyons, 2002: Valley floor climate obser- Q. N. Xiao, 2004: A three- dimensional variational data as- vations from the McMurdo Dry Valleys, Antarctica, 1986– similation system for MM5: Implementation and initial results. 2000. J. Geophys. Res., 107, 4772, doi:10.1029/2001JD002045. Mon. Wea. Rev., 132, 897–914. ——, ——, A. G. Fountain, T. Nylen, D. M. McKnight, C. Jaros, Barrett,J.E.,R.A.Virginia,D.H.Wall,P.T.Doran,A.G.Fountain, and J. E. Barrett, 2008: Hydrologic response to extreme warm K. A. Welch, and W. B. Lyons, 2008: Persistent effects of a dis- and cold summers in the McMurdo Dry Valleys, East Ant- crete warming event on a polar desert ecosystem. Global Change arctica. Antarct. Sci., 20, 499–509. Biol., 14, 2249–2261. Doyle, J., and M. Shapiro, 2000: A multi-scale simulation of an Barry, R. G., and R. J. Chorley, 2003: Atmosphere, Weather and extreme downslope windstorm over complex topography. Climate. 8th ed. Routledge, 123 pp. Meteor. Atmos. Phys., 74, 83–101. Beer, T., 1976: Mountain waves. Sci. Prog., 63, 1–25. Durran, D. R., 1990: Mountain waves and downslope winds. At- Bertler,N.A.N.,T.R.Naish,H.Oerter,S.Kipfstuhl,P.J.Barrett, mospheric Processes over Complex Terrain, W. Blumen, Ed., P. A. Mayewski, and K. Kreutz, 2006: The effects of joint ENSO- Amer. Meteor. Soc., 59–81. Antarctic Oscillation forcing on the McMurdo Dry Valleys, Flamant, C. P., and Coauthors, 2002: Gap flow in an alpine valley Antarctica. Antarct. Sci., 18, 507–514. during a shallow south fohn event: Observations, numerical Bougeault, P., and Coauthors, 2001: The MAP Special Observing simulations and hydraulic analogue. Quart. J. Roy. Meteor. Period. Bull. Amer. Meteor. Soc., 82, 433–462. Soc., 128, 1173–1210. Brinkmann, W. A. R., 1971: What is a foehn? Weather, 26, 230–239. Fogt, R. L., and D. H. Bromwich, 2006: Decadal variability of the Bromley, A. M., 1985: Weather observations, Wright Valley, ENSO teleconnection to the high-latitude South Pacific gov- Antarctica. New Zealand Meteorological Service Information erned by coupling with the Southern Annular Mode. J. Cli- Publ. 11, 37 pp. mate, 19, 979–997. Bromwich, D. H., J. F. Carrasco, Z. Liu, and R. Y. Tzeng, 1993: Foreman, C. M., C. F. Wolf, and J. C. Priscu, 2004: Impact of Hemispheric atmospheric variations and oceanographic im- episodic warming events on the physical, chemical and bio- pacts associated with katabatic surges across the Ross Ice logical relationships of lakes in the McMurdo Dry Valleys, Shelf, Antarctica. J. Geophys. Res., 98, 13 045–13 062. Antarctica. Aquat. Geochem., 10, 239–268. ——, J. J. Cassano, T. Klein, G. Heinemann, K. M. Hines, Fountain, A. G., and Coauthors, 1999: Physical controls on the K. Steffen, and J. E. Box, 2001: Mesoscale modeling of kata- Taylor Valley ecosystem. Bioscience, 49, 961–971. batic winds over Greenland with the Polar MM5. Mon. Wea. ——, T. H. Nylen, A. J. Monaghan, A. J. Basagic, and D. H. Rev., 129, 2290–2309. Bromwich, 2009: Snow in the McMurdo Dry Valleys,

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 1JULY 2010 S P E I R S E T A L . 3597

Antarctica. Int. J. Climatol., 30, 633–642, doi:10.1002/ O’Connor, W. P., and D. H. Bromwich, 1988: Surface airflow joc.1933. around Windless Bight, Ross Island Antarctica. Quart. J. Roy. Gaffin, D. M., 2007: Foehn winds that produced large temperature Meteor. Soc., 114, 917–938. differences near the southern Appalachian Mountains. Wea. Parish, T. R., 1983: The influence of the Antarctic Peninsula on the Forecasting, 22, 145–159. wind field over the western Weddell Sea. J. Geophys. Res., 88, Gohm, A., G. Za¨ngl, and G. Mayr, 2004: South foehn in the Wipp 2684–2692. Valley on 24 October 1999 (MAP IOP 10): Verification ——, and D. H. Bromwich, 1998: A case study of Antarctic kata- of high-resolution numerical simulations with observations. batic wind interaction with large-scale forcing. Mon. Wea. Mon. Wea. Rev., 132, 78–102. Rev., 126, 199–209. Grell, G. A., J. Dudhia, and D. R. Stauffer, 1994: A description ——, J. J. Cassano, and M. W. Seefeldt, 2006: Characteristics of the of the fifth-generation Penn State/NCAR mesoscale model Ross Ice Shelf air stream as depicted in Antarctic Mesoscale (MM5). NCAR Tech. Note NCAR/TN-3981STR, 122 pp. Prediction System simulations. J. Geophys. Res., 111, D12109, Guo, Z., D. Bromwich, and J. Cassano, 2003: Evaluation of Polar doi:10.1029/2005JD006185. MM5 simulations of Antarctic atmospheric circulation. Mon. Phillpot, H., 1991: The derivation of 500 hPa height from automatic Wea. Rev., 131, 384–411. weather station surface observations in the Antarctic conti- Jiang, Q., J. Doyle, and R. Smith, 2005: Blocking, descent and nental interior. Aust. Meteor. Mag., 39, 79–86. gravity waves: Observations and modelling of a MAP Powers, J. G., A. J. Monaghan, A. M. Cayette, D. H. Bromwich, northerly fohn event. Quart. J. Roy. Meteor. Soc., 131, 675– Y.-H. Kuo, and K. W. Manning, 2003: Real-time mesoscale 701. modeling over Antarctica: The Antarctic Mesoscale Pre- Keys, J. R., 1980: Air pressure, wind, precipitation and atmospheric diction System. Bull. Amer. Meteor. Soc., 84, 1533–1545. humidity in the McMurdo region. Victoria University of Renfrew, I., and P. Anderson, 2006: Profiles of katabatic flow Wellington, Department of Geology Publ. 17 (Antarctic Data in summer and winter over Coats Land, Antarctica. Quart. Series 9), 52 pp. J. Roy. Meteor. Soc., 132, 779–802. Klemp, J. B., and D. K. Lilly, 1975: The dynamics of wave-induced Richner, H. K., and Coauthors, 2006: Unstationary aspects of foehn downslope winds. J. Atmos. Sci., 32, 320. in a large valley part I: Operational setup, scientific objectives Kuettner, J., 1986: The aim and conduct of ALPEX. Scientific and analysis of the cases during the special observing period of Results of the Alpine Experiment, GARP Publications Series the MAP subprogramme FORM. Meteor. Atmos. Phys., 92, 27, WMO/TD 108, 3–14. 255–284. Kwok, R., and J. C. Comiso, 2002: Southern Ocean climate and Riordan, A. J., 1975: The climate of Vanda Station, Antarctica. sea ice anomalies associated with the Southern Oscillation. Climate of the Arctic, G. Weller and S. A. Bowling, Eds., J. Climate, 15, 487–501. University of Alaska Geophysical Institute, 268–275. Math, F. A., 1934: Battle of the chinook wind at Havre, Mont. Mon. Seefeldt, M. W., and J. Cassano, 2008: An analysis of low-level jets Wea. Rev., 62, 54–57. in the greater Ross Ice Shelf region based on numerical sim- McGowan, H. A., and A. P. Sturman, 1996a: Regional and local ulations. Mon. Wea. Rev., 136, 4188–4205. scale characteristics of foehn wind events over the South Is- ——, G. J. Tripoli, and C. R. Stearns, 2003: A high-resolution nu- land of New Zealand. Meteor. Atmos. Phys., 58, 151–164. merical simulation of the wind flow in the Ross Island region, ——, and ——, 1996b: Interacting multi-scale wind systems within Antarctica. Mon. Wea. Rev., 131, 435–458. an alpine basin, Lake Tekapo, New Zealand. Meteor. Atmos. ——, J. J. Cassano, and T. R. Parish, 2007: Dominant regimes of the Phys., 58, 165–177. Ross Ice Shelf surface wind field during austral autumn 2005. ——, and J. C. Speirs, 2008: The polar foehn of the Victoria Valley, J. Appl. Meteor. Climatol., 46, 1933–1955. McMurdo Dry Valleys, Antarctica. Preprints, 13th Conf. on Seibert, P., 1990: South foehn studies since the ALPEX experi- Mountain Meteorology, Whistler, BC, Canada, Amer. Meteor. ment. Meteor. Atmos. Phys., 43, 91–103. Soc., 12B.1. [Available online at http://ams.confex.com/ams/ Selby, M. J., R. W. P. Palmer, C. J. R. Smith, and R. B. Rains, 1973: pdfpapers/140844.pdf.] Ventifact distribution and wind directions in the Victoria McKendry, I. G., and E. W. D. Lewthwaite, 1990: The vertical Valley, Antarctica. N. Z. J. Geol. Geophys., 16, 303–306. structure of summertime local winds in the Wright Valley, Simmonds, I., K. Keay, and E. P. Lim, 2003: Synoptic activity in the Antarctica. Bound.-Layer Meteor., 51, 321–342. seas around Antarctica. Mon. Wea. Rev., 131, 272–288. ——, and ——, 1992: Summertime along-valley wind variations Skamarock,W.C.,J.B.Klemp,J.Dudhia,D.O.Gill,D.M.Barker, in the Wright Valley Antarctica. Int. J. Climatol., 12, 587– W. Wang, and J. G. Powers, 2005: A description of the Advanced 596. Research WRF Version 2. NCAR Tech Notes-4681 STR, Monaghan, A. J., D. H. Bromwich, H. Wei, A. Cayette, J. Powers, 100 pp. [Available online at http://www.mmm.ucar.edu/wrf/ Y. Kuo, and M. Lazzara, 2003: Performance of weather users/docs/arw_v2.pdf.] forecast models in the rescue of Dr. Ronald Shemenski Speirs, J. C., H. McGowan, and D. T. Neil, 2008: Meteorological from the South Pole in April 2001. Wea. Forecasting, 18, controls on sand transport and dune morphology in a polar- 142–160. desert: Victoria Valley, Antarctica. Earth Surf. Processes ——, J. G. Powers, and K. W. Manning, 2005: The climate of the Landforms, 33, 1875–1891. McMurdo, Antarctica, region as represented by one year of Steinhoff, D. F., D. H. Bromwich, M. Lambertson, S. L. Knuth, and forecasts from the Antarctic Mesoscale Prediction System. M. A. Lazzara, 2008: A dynamical investigation of the May J. Climate, 18, 1174–1189. 2004 McMurdo Antarctica severe wind event using AMPS. Nylen, T. H., A. G. Fountain, and P. T. Doran, 2004: Climatology of Mon. Wea. Rev., 136, 7–26. katabatic winds in the McMurdo dry valleys, southern Victoria ——, S. Chaudhuri, and D. H. Bromwich, 2009: A case study of Land, Antarctica. J. Geophys. Res., 109, D03114, doi:10.1029/ a Ross Ice Shelf Air Stream event: A new perspective. Mon. 2003JD002927. Wea. Rev., 137, 4030–4046.

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC 3598 JOURNAL OF CLIMATE VOLUME 23

Thompson, D. C., 1972: Climate of the Dry Valley area of South Whiteman, C. D., and J. C. Doran, 1993: The relationship between Victoria Land. Proc. Seventh New Zealand Geography Conf., overlying synoptic-scale ﬂows and winds within a valley. J. Appl. Hamilton, New Zealand, 259–265. Meteor., 32, 1669–1682. ——, R. M. F. Craig, and A. M. Bromley, 1971: Climate and surface ——, S. Eisenbach, B. Pospichal, and R. Steinacker, 2004: heat balance in an Antarctic Dry Valley. N. Z. J. Sci., 14, Comparison of vertical soundings and sidewall air temperature 245–251. measurements in a small alpine basin. J. Appl. Meteor., 43, 1635– Uotila, P., A. B. Pezza, J. J. Cassano, K. Keay, and A. H. Lynch, 1647. 2009: A comparison of low pressure system statistics derived Yu, Y., X. Cai, J. C. King, and I. A. Renfrew, 2005: Numerical from a high resolution NWP output and three reanalysis simulations of katabatic jumps in Coats Land, Antarctica. products over the Southern Ocean. J. Geophys. Res., 114, Bound.-Layer Meteor., 114, 413–437. D17105, doi:10.1029/2008JD011583. Za¨ngl, G., 2003: Deep and shallow south foehn in the region of Welch, K. A., and Coauthors, 2003: Climate and hydrological Innsbruck: Typical features and semi-idealized numerical variations and implications for lake and stream ecological simulations. Meteor. Atmos. Phys., 83, 237–261. response in the McMurdo Dry Valleys, Antarctica. Climate ——, and M. Hornsteiner, 2007: Can trapped gravity waves be Variability and Ecosystem Response at Long-Term Ecological relevant for severe foehn windstorms? A case study. Meteor. Research Sites, D. Greenland, D. G. Goodin, and R. C. Smith, Z., 16, 203–212. Eds., Oxford University Press, 174–195. ——, A. Gohm, and G. Geier, 2004: South foehn in the Wipp Wendler, G., C. Stearns, G. Weidner, G. Dargaud, and T. Parish, Valley–Innsbruck region: Numerical simulations of the 1997: On the extraordinary katabatic winds of Ade´lie Land. 24 October 1999 case (MAP-IOP 10). Meteor. Atmos. Phys., J. Geophys. Res., 102, 4463–4474. 86, 213–243.

Unauthenticated | Downloaded 09/24/21 01:00 AM UTC