Snow megadunes in Antarctica: Sedimentary structure and genesis

M. FrezzottiEnte per le Nuove Tecnologie, l’Energia e l’Ambiente, Roma, Italy

S. GandolfiDipartimento di Ingegneria delle Strutture, dei Trasporti, delle Acque, del Rilevamento, del Territorio, University of Bologna,Bologna, Italy

S. UrbiniIstituto Nazionale di Geofisica e Vulcanologia, Dipartimento per lo Studio del Territorio e delle sue Risorse, University ofGenova, Genova, Italy

Received 26 March 2001; revised 20 September 2001; accepted 2 January 2002; published 18 September 2002.

[1] Megadune fields occupy large areas in the interior of the East Antarctic ice sheet andare the result of unusual snow accumulation and redistribution processes. They thereforeare important to surface mass balance and ice core interpretation. Field observations (GPS,GPR, and surface measurements) have provided a detailed description of megadunesedimentation and morphology over a 70 km2 area, located 200 km east of Dome C. Acombination of remote sensing analysis (using Landsat and satellite radar altimetry) andfield measurements indicate that slope in the prevailing wind direction (SPWD) andclimatic conditions play a crucial role in megadune genesis. The megadune areas tend tobe characterized by slightly steeper regional slope and the presence of highly persistentkatabatic winds. The megadunes represent 2 to 4 m amplitude waves of 2 to 5 kmwavelength formed by variable net accumulation, ranging between 25% (leeward faces) to120% (windward faces) of the accumulation in adjacent nonmegadune areas. Leewardfaces are characterized by glazed, sastrugi-free surfaces and extensive depth hoarformation. Windward faces are covered by large rough sastrugi up to 1.5 m inheight. INDEX TERMS: 1827 Hydrology: Glaciology (1863); 1863 Hydrology: Snow and ice (1827);

1894 Hydrology: Instruments and techniques; 3322 Meteorology and Atmospheric Dynamics: Land/

atmosphere interactions; 5470 Planetology: Solid Surface Planets: Surface materials and properties;

KEYWORDS: snow dune, aeolian morphology, mass balance, ice core, katabatic wind, Antarctica

Citation: Frezzotti, M., S. Gandolfi, and S. Urbini, Snow megadunes in Antarctica: Sedimentary structure and genesis, J. Geophys.

Res., 107(D18), 4344, doi:10.1029/2001JD000673, 2002.

1. Introduction

[2] The Antarctic impact on global climate change, andthe impact of global climate change on Antarctica, are focalpoints of current international and interdisciplinary expedi-tions. Because of the remoteness of the continent, Antarc-tica is an ideal location to monitor local-to-global scaleclimate changes. However, this remoteness has also pre-vented the collection of instrumental records similar to thosecollected in the Northern Hemisphere. Such records areneeded to assess Antarctica’s role in, and response to,climate change.[3] The snow deposition process is very complicated on

the Antarctic plateau, where the katabatic wind is strong,blowing snow is severe, accumulation is low and snowremains on the surface as aeolian particles [Okuhira andNarita, 1978]. The deposition-erosion process on the snow

surface depends upon meteorological and geomorphologicalconditions and varies according to the surface conditions,yielding a complex surface snow layer. Snow transportchanges topography, and topography in turn alters thekatabatic wind field, which affects transport. It is a feedbacksystem between the cryosphere and atmosphere [Seko et al.,1992].[4] On the local scale, there is continual interaction

among processes such as wind, radiation balance and snowsurface temperature variations; in particular, the surface-energy balance and katabatic wind patterns are highlyinterrelated [e.g., Bintanja, 1999]. Snow drifting is con-trolled not only by wind speed but also by snow surfaceroughness, particle shape of snow, and air temperature[Narita, 1978]. Many types of surface microrelief, such assastrugi, snow dunes and pitted patterns, are distributed onthe surface of the Antarctic ice sheet in various horizontaldimensions (from decimeter to kilometer), and occur as aresult of interactions between the air and the ice sheetsurface [e.g., Watanabe, 1978]. Smooth and glazed surfaces

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 107, NO. D18, 4344, doi:10.1029/2001JD000673, 2002

Copyright 2002 by the American Geophysical Union.0148-0227/02/2001JD000673

ACL 1 - 1

should also be considered a type of surface feature, althoughas a type, it differs from other features that are generallydistributed in a zonal form with specific dimensions andorientations [Watanabe, 1978]. Sastrugi and dunes orienta-tions are a factor of wind direction and are therefore usefulindicators of prevailing wind direction [Mather, 1962]. Ithas long been known that slope and curvature may play animportant part in the topographic effect on accumulationrate, as concave depressions gain accumulation at theexpense of convex rises [Black and Budd, 1964; Whillans,1975; Pettre et al., 1986; Liston et al., 2000].[5] Macrorelief aeolian surfaces were first described by

Swithinbank [1988], who termed these features megadunesbased on their likeness to transverse sand megadunes. Thesefeatures are extensive on the East Antarctic plateau andoccupy more than 500,000 km2. They are oriented perpen-dicular to the regional katabatic wind direction, amplitudesare small (about 4 m), wavelengths range from 2 to over 4km, and megadune crests are nearly parallel and 10 to 100km in length [Fahnestock et al., 2000].[6] Megadunes also occur close to deep ice core drilling

sites such as at Vostok station (Russia), which contains thelongest climatic record presently available [Petit et al.,1999]. Interpretation of climatic records from ice cores isnot straightforward, as measurements are a convolution ofboth large-scale and small-scale processes acting over aspectrum of temporal and spatial scales [van der Veen et al.,1999]. Spatial and temporal snow accumulation variabilityreflects local surface processes that modulate deposition andpreservation. In order to obtain an accurate interpretation ofcore stratigraphy, it is important to recognize the various

contributions to the record, and to assess how representativethe core is for the site and for the larger region in general. Inmass balance studies, accurate data on snow accumulationdistribution and variability are needed to reach correctconclusions. However, it is often difficult to obtain accuratefield data on snow accumulation because the accumulationrate may vary substantially over a short distance [Richard-son et al., 1997; Liston et al., 2000]. Because of logistic andbudgetary constraints, drilling at a single location is oftenmandatory in order to obtain a climatic record that ishopefully representative of a much wider area.[7] This paper combines field measurements (Global

Positioning System, Ground Penetration Radar, and surfacemeasurements) and remote sensing data (using Landsat andsatellite radar altimetry) to provide the first detailed descrip-tion of megadune sedimentation and morphology in an areaof East Antarctica 200 km east of Dome C (Figure 1). Thefield measurements represent the first dedicated groundsurvey of megadunes since their recent characterizationfrom satellite imagery by Fahnestock et al. [2000]. Theresulting analysis provides the basis for a study of theinteraction between atmosphere-cryosphere and surfaceslope that produces the megadune formation and evolution.This study provides new information about the process(ablation/accumulation) of their formation and implicationsof snow accumulation, distribution, and variability. Theseprocesses have important consequences concerning thechoice of sites for ice coring, since orographic variationsof few meters per kilometer have a significant impact on thesnow accumulation process. These new detailed field datarepresent a new ground truth and foundation of knowledge

Figure 1. Schematic map of the ITASE 1998/99 showing: traverse route, GPS-GPR survey, themegadune field, the Landsat ETM+coverage, and drill sites.

ACL 1 - 2 FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA

for remote sensing data of the interior of the East Antarcticplateau.[8] As a part of the ITASE project [Mayewski and Good-

win, 1999], the Italian Programma Nazionale di Ricerchein Antartide (PNRA) undertook a traverse (Figure 1) fromthe Terra Nova Station (164�060E 74�410S) to Dome C(123�230E 75�060S, 3232 m). The scientific objectives ofthe traverse program were to develop a high resolutioninterpretation and 3-D map documenting the last 200–1000years of climate, atmospheric and surface conditions (snowaccumulation, air temperature, atmospheric circulation) overthe eastern Dome C drainage area. Along the traverse, theparty carried out several tasks (drilling, glaciological, andgeophysical exploration, etc.).[9] The traverse started from GPS1 (160�48.030E 74�

50.040S) on 19 November 1998 and reached Dome C on5 January 1999, a distance of 1300 km (Figure 1). Thetraverse route was selected and surveyed in Italy fromgeoreferenced satellite image analysis (Landsat TM, ERSSAR, and AVHRR) and using a Digital Elevation Model(DEM) derived from an ERS1 Radar Altimeter [Remy et al.,1999]. The pre-analysis of the Landsat TM image indicateda megadune surface morphology along the traverse between129�560E–75�270S and 129�140 E–75�250S, close to the D6site. At this site, GPS, GPR and a detailed snow surfacesurvey have been carried out in an area of about 70 km2,with about 100 km of GPS profile and 60 km of GPR.

2. Materials and Methods

2.1. Geophysical Survey

[10] Surface elevation profiles and local topographyalong the traverse were measured by Global PositioningSystem (GPS), whereas regional surface topography wasanalyzed using a DEM of Antarctica provided by Remy etal. [1999]. The DEM was created with a 1 km grid sizeusing a radar altimeter from ERS satellite. The DEM hasbetter than 1 m accuracy over the portions of East Antarc-tica where slope gradients are less than 0.5%. GPS surveywas performed along the traverse using dual-frequencyGeodetic Trimble 4000 SSE receivers and two fixed stationslocated at the beginning (Terra Nova Bay Station) and at theend of the traverse (Concordia Station). During the mega-dune survey a fixed station at D6 (129�48.530E 75�26.850S;3024 m WGS84 height) was used. The receivers (Masterand Rover) sampling rate was fixed at 5 s (one coordinateevery 10–15 m). The rover receivers, equipped with geo-detic antennas, were installed on vehicles and were used forthe kinematic surveys to perform altimetric profiles and tocorrect GPR acquisition as a function of both ellipsoidalheight and surface coordinates [Urbini et al., 2001]. Theprocessing of the GPS data was performed using GPS-Geotracer (V.1.03, V2.25, and V.2.28) software and analgorithm implemented on a kinematic program [Urbini etal., 2001]. The accuracy of the altimetric profile along thetraverse is related to the distance between the fixed androver stations and ranges between less than 1 to 3 m at thefarthest point. However, the accuracy of the altimetricprofile performed at the megadune area, using the D6master station, is as much as 10 cm.[11] Ground Penetration Radar (GPR) along continuous

profiles provides detailed information on spatial variability

in snow accumulation [Richardson et al., 1997; Richardsonand Holmlund, 1999; Urbini et al., 2001]. GPR method isbased on the reflection of the electromagnetic waves gen-erated from an antenna, due to one or more discontinuitiesin the media dielectric properties. The dielectric propertiesof dry snow areas are affected by various physical andchemical parameters (snow density, crystal geometry andfabric, conductivity, concentration and composition of ionsand microparticles), of which the most important is density,which can affect the dielectric constant of snow [e.g.,Kovacs et al., 1995; Mazler, 1987]. In this work variationsin snow accumulation mapped by electromagnetic stratig-raphy in the upper 15–20 m of the snowpack will be shown.Data acquisition was performed with a GSSI SIR10B unitequipped with one monostatic antenna with a centralfrequency of 400 MHz. Principal acquisition parameterswere 150 ns for the vertical investigation range and from 1to 5 scan s�1 for the acquisition rate. Vehicle speed rangedfrom 8 to 12 km h�1, meaning about one scan every 1–3 m(with the acquisition rate at 1 scan s�1) and 0.4–0.7 m (withthe acquisition rate at 5 scan s�1).[12] ‘‘In situ’’ experiments by Kovacs et al. [1995]

determined the relationship between snow dielectric proper-ties and firn density:

e0 ¼ 1þ 0:845 rð Þ2;

where r is the specific density [see also Sihvola et al., 1985;Mazler, 1987]. This empirical equation is valid for snow,firn, and ice. Using this equation, local core density wasused to calculate an initial electromagnetic wave velocityfunction for converting reflection arrival times to meters ofdepth. For electromagnetic wave speed calculations thedepth-density relation for the snowpack was establishedusing the density profile of 4 firn cores (ranging in-depthfrom 12 to 52 m), retrieved with an electromechanicaldrilling system (diameter: 100 mm) at the D4 (135�49.890E,75�35.790S) and D6 sites (Figure 1). Density data werefitted by exponential function obtaining a determinationcoefficient of about 0.83. According to the methoddescribed by Richardson and Holmlund [1999] and Urbiniet al. [2001], the two-way travel times picked out from radardata were converted to meters of depth using depth-density,depth-wave speed and depth-travel time relations whichwere fitted by both exponential and power functions. Usingan average value of density for the upper 20 m depth insteadof a variable function involves a layer positioning maximumerror of about 50 cm. Digital enhancement of the GPR dataincluded horizontal high-pass, vertical low and high-passfilters (FIR type), migration, and gains optimization [e.g.,Glover and Rees, 1992; Richardson et al., 1997]. From theenhanced data, the arrival times of the strongest reflectionswere picked out as reference layers. We assumed each ofthese represented a specific event or short time period, as doVaughan et al. [1999b] and that layers yielding a strongradar reflection are isochronous. Variations in snowthickness between these reference layers provide informa-tion on snow accumulation variability [Richardson et al.,1997; Urbini et al., 2001].[13] Using the GPS result as a geographic reference,

electromagnetic stratigraphies can be positioned not in

FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA ACL 1 - 3

respect to the surface (as usual), but in respect to absoluteellipsoidal height (WGS84). The approach used to combineGPS and GPR solutions is illustrated in Figure 2.[14] In order to obtain a three-dimensional representation

of the megadune area, an interpolation of the GPS and GPRsurvey has been performed. The contour line and the three-dimensional view (the surface and three internal layers at

different depths; Figure 3) of an area of about 30 km2 with a30 m grid spacing was created using a kriging interpolationprocedure [Capra et al., 2000].

2.2. Surface Morphology

[15] A spatial distribution survey of the meter-scale sur-face features including type, size and orientation was con-

Figure 2. (a) Scheme of methodology used for GPR-GPS data integration. (b) Example of GPRregistration recorded across the megadune area; the vertical timescale corresponds to a maximum depthof 15 m. (c) Interpretation of internal layering of Figure 2b repositioned in respect to the surface asmeasured by GPS.

ACL 1 - 4 FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA

Figure 3. Three-dimensional view of the megadune area (30 km2) results from GPR and GPS surveycarried out during traverse. The top figure shows surface topography, based on GPS measurements; theother figures represent the elevation of three internal layers detected by GPR. Each layer is representedusing (left) contour maps and (right) block diagrams, where the contour interval has been fixed at 1 m andticker lines correspond to 5 m contour line. The surface contour map shows the trace of the GPS and GPRsurvey. These figures are represented in a local reference system in order to obtain better representation;location of the area is shown in Figure 4. Vertical dash lines along the block diagrams represent thesurface culmination of the megadune. Vertical exaggeration 1 : 620.

FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA ACL 1 - 5

ducted along the route [Frezzotti et al., 2002] and amongsome profiles in the megadune area. The timing of thetraverse, in November–December, provided the bestopportunity to investigate microrelief development anddistribution, since the observed features formed duringthe winter when the majority of precipitation occurs andthe strongest katabatic surface winds occur [Goodwin,1990]. Microrelief features were classified according to asystem described by Fujiwara and Endo [1971] andapplied to Japanese and Australian Antarctic ResearchExpedition (JARE and ANARE) traverses inland in theplateau region of Dronning Maud Land and Wilkes Land[Goodwin, 1990]. The mean azimuths of the microrelieffeatures were measured by magnetic compass and con-verted into true bearings by GPS; also, the mean height ofthe features were recorded using a ruler. Microrelieforientation generally follows the wind field streamlinesindicated by Parish and Bromwich [1991]. However, thefield-inferred directions show the surface wind pattern inmuch greater detail and reveal areas of divergence andconfluence as a result of mesoscale topography (Figure 4).The surface microrelief features observed are divided intothree types [Watanabe, 1978; Goodwin, 1990] as follows:(1) depositional features formed from wind-transportedfriable snow (barchanoid, dune, etc.); (2) redistributionfeatures formed from erosion of depositional features(sastrugi, pit, etc.); and (3) erosional features formed fromthe long-term accumulation hiatus and exposure to kata-batic winds (glazed surface).

2.3. Satellite Data

[16] In this study we have integrated the previous datawith satellite images of Landsat TM 4 (84/113, 17 January1992), Landsat ETM+ 7 (81/114, 2 January 2000 and 81/115 2 January 2000) and RADARSAT-1 (9 September to 20October 1997). RADARSAT-1 mosaic images, with aground resolution of 125 m, were obtained at the Web sitehttp://iceberg.mps.ohio-state.edu [Jezek, 1999]. The Landsat4 Thematic Mapper (TM) and 7 ETM+ have a groundresolution of 30 m and seven spectral bands: three in thevisible wavelength region (bands 1, 2, and 3), two in thenear infrared wavelength region (bands 4, 5, and 7) and oneregion of thermal infrared wavelength (band 6) with groundresolution of 120 m for TM and 60 m for ETM+. TheLandsat 7 ETM+ has a more panchromatic band with ahigher ground resolution of 10 m. Landsat 7 ETM+ imageswere georeferenced by means of geographical coordinatesof satellite ephemeris; the geolocation of ETM+ has a one-sigma error of 50 m (R. Bindschadler, personal communi-cation, 2000). RADARSAT-1 SAR mosaic images presentan absolute geolocation better than 100 m [Jezek, 1999].[17] Landsat 4 TM (84/113) and 7 ETM+ (81/114 and 81/

115) satellite images were acquired with sun-elevationangles of 22�, 28.3�, and 27.1� and with a sun azimuthangle of 80�, 70.7�, and 73.9� respectively. The satelliteimages were analyzed, compared and organized into ageographic information system (GIS) using ERDAS andARCINFO software. Image processing included routineprocedures such as radiometric corrections, noise and strip-ing removal, and application of special linear stretches ofindividual TM bands after inspection of grey-value histo-grams to bring out surface features and characteristics.

[18] Snow albedo decreases as grain-size increases at nearinfrared wavelengths, but it is little affected by grain-size invisible wavelengths. Whereas firn and ice albedos generallydecrease passing from the visible to the near infraredwavelengths [e.g., Warren, 1982], glazed surfaces presentspectra intermediate between snow and ice [Frezzotti et al.,2002]. The snow spectral reflectivity on the Antarcticplateau depends on grain-size and shape, depth, surfaceroughness and angle of solar incidence [Orheim and Luc-chitta, 1987]. The difference between snow (white color inFigure 4), and glazed surfaces (blue color in Figure 4) arebetter seen in false color composites of ETM+ bands 2, 3,and 4, than in the individual bands. In false color imageglazed surface look bluish because solar radiation isstrongly absorbed in the red part of the spectrum andreflected in the deeper-penetrating blue part. Remote sens-ing analysis relates these aeolian features to ablation(glazed-surface area) and accumulation processes (dune,sastrugi). Remote sensing analysis also provides informa-tion about wind surface fields (Figure 4) through thesurveying of aeolian surface feature locations, directions,and areal extents [Bromwich et al., 1990; Frezzotti, 1997].

3. Discussion

3.1. Microrelief

[19] Topographic and morphological surveys were col-lected across the megadune and show a typical surface’smicro relief distribution (Figures 3, 4, and 5). Glazedsurfaces are located on the leeward slope while severesastrugi (up to 1.5 m in height) are located on the uphillslope (windward). Alternation of sastrugi (up to 40 cm) andglazed surfaces are located at the bottom of the interdunearea. The distribution of glazed surfaces and sastrugi alongthe megadune are confirmed by Landsat ETM+ images anddetailed GPS surveys.[20] The microrelief directions (40�–45� True) measured

in the field are parallel to the sastrugi-glazed surface fieldinferred by satellite image (50�–55� True) and aligned withthe predicted (40�–50� True) katabatic wind field surface[Parish and Bromwich, 1991].[21] Glazed surfaces were one of the common features

observed in the megadune area and consist of thin (0.1 to2 mm) films of regelated ice. When the wind capacity forsnow transportation exceeds the snow supply the snowsurface is scoured clean and flat and suffers neither erosionnor deposition [Gow, 1965; Alley, 1988]. The regelated icefilms form on the surface following the kinetic heating bysaltant drift snow under constant katabatic wind flow[Goodwin, 1990]. Most of the cracks observed (up to2 cm) were located on these glazed surfaces and werepatterned in polygonal form. Glazed surfaces with cracksare linked to a long-term accumulation hiatus surfacemorphology form [Watanabe, 1978]. A trench cut on theleeward slope under a glazed surface showed the depth hoarlayer (up to 2 m) inserted by iced crust with a very coarsesnow grain size (up to 2 mm). This field observation agreeswith digital image analysis of snow grains collected alongthe traverse [Gay et al., 2002] that show large grains at sitescharacterized by glazed surfaces and smaller grains insastrugi areas. Under strongly developed glazed surfacesthe depth hoar layer clearly indicates prolonged sublimation

ACL 1 - 6 FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA

Figure 4. RADARSAT (grey) and Landsat ETM+ in false color (red band 4, green band 3, blue band 2)images of investigated area (location in Figure 1). The insert (lower left) is a detail of the image showingthe pattern of sastrugi-glazed surface field (oblique white strips) and megadune. False color compositionenables discrimination of the glacial areas where accumulation processes are prevalent, and consists ofsnow (white color in the images) from the ablation or hiatus areas, which are made up of glazed surfaces(blue color in the image). The arrows indicate the direction of wind surveyed by field measurements, byLandsat ETM+ image and modeled for general surface wind field. Contour lines have been derived fromDEM [Remy et al., 1999].

FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA ACL 1 - 7

due to a hiatus in accumulation and therefore a long,multiannual, steep temperature-gradient metamorphism[Gow, 1965]. The sublimation and upward transport ofwater vapor within to the subsurface snow layer cause thecondensation of vapor (recrystallization) on the lower partof the ice crust [Fujii and Kusunoki, 1982]. Prolongedsublimation and upward transport of water vapor ultimatelyaffect chemical and isotope composition of the snow.Postdepositional losses by reemission into the atmosphere(HCl, HNO3, MSA) and isotopic diffusion (d18O) have beenobserved along the traverse TNB-DC at M2 site. Chemicaland isotopic analysis of cores and trenches show a concen-tration decreasing abruptly from the surface to deeperlayers, at sites characterized by the presence of glazedsurface and depth hoar layers [Proposito et al., 2002].[22] Long-term hiatus forms do not allow the burial of

snow layers by accumulation in subsequent years. Thedepth hoar layer under strongly developed glazed surfacesclearly indicates prolonged sublimation due to a hiatus inaccumulation and therefore a long, multiannual, steeptemperature-gradient metamorphism [Gow, 1965]. Goodwinet al. [1994] pointed out that in eastern Kemp Land thelarger grain sizes are associated with more numerous glazedsurfaces that are produced from wind exposure during ahiatus in accumulation. The authors found that the mini-mum microwave emissivity occurs where accumulationrates are very low and depth hoar is present, with a crystalsize greater than 1.5 mm. Dunes and sastrugi are made by

deposition and redistribution processes of aeolian particles(snow remains), and thus their surface snow grain sizes arelinked mainly to the aeolian process driven by katabaticwind.[23] Relatively high winds and light or no snowfall favors

the formation of sastrugi. Surface irregularity interruptssnow transport and allows the deposition of a dune (typi-cally several meters long and up to 50 cm high). Theselongitudinal dunes (thereafter called dune in the text) mayremain pinned to an obstacle (sastrugi, pit, etc.) and couldmigrate leeward. When wind speeds decrease, such dunessinter rapidly into hard features of positive relief. Aftersintering, wind erosion of the dune typically producesscoops and hollows with the formation of sastrugi [Alley,1988]. Dunes and sastrugi are believed to ablate faster thanadjacent flat areas, because microrelief surfaces are moreexposed to wind; thus both erosion and deposition tend tosmooth the snow surface [Gow, 1965; Weller, 1969].

3.2. Megamorphology

[24] Remote sensing analysis integrated with field dataillustrate features of larger aeolian morphology (Figure 4) ofthe sastrugi-glazed surface field (parallel to the microreliefdirection) and the megadune (transverse to the microreliefdirection). The megadune field occurs along the traverseroute only between 129�560E–75�270S and 129�140E–75�250S. Other transverse megadunes have been surveyedalong the traverse at the bottom of the slope [Frezzotti et al.,2002], where they are perpendicular to the slope and similarto those reported by Black and Budd [1964] in Wilkes Land.The megadune and sastrugi-glazed surface field has adistribution of several km and it was also possible to mapthem by satellite image (Figure 4).[25] Glazed surface-sastrugi fields have the typical texture

of a seasonal surface, with a relatively flat horizontal planeand a thin and soft glazed surface. The surface condition ischaracterized by the alternating occurrence of a wide,smooth (glazed) surface and a wide, rough surface (sastrugizone) with a distinct boundary (see inset in Figure 4). Thesefields are characterized by the alternation of sastrugi fieldswith longitudinal dunes, having a mean height up to 1 m, andflat glazed surfaces with some sastrugi. The field boundariesare roughly linear, with the line of elongation parallel to thesastrugi and longitudinal dune direction and therefore to theprevailing wind direction. These fields were typically somekm long and 100–200 m wide, have an extension of severalhundred km2 and are similar to longitudinal sand dunes.Glazed surface-sastrugi fields have been surveyed frequentlyalong the traverse, and they are the largest morphologicalstructure observed in the plateau area [Frezzotti et al., 2002].In the megadune area, the glazed surface-sastrugi fields havebeen observed in the field and on satellite images of themegadune field (Figure 4). As for longitudinal sand dunes[Houbolt, 1968], pressure gradients exist between the axes ofthe interdune (glazed surface) and the crest of the dunes(sastrugi-dune field); these pressure gradients are caused bythe sastrugi’s resistance to the wind. This results in theformation of ‘‘wind cells’’ in the glazed areas, which givethe wind an overall spiral motion directed outward at groundlevel towards the sastrugi, with transportation of snowparticles (due to erosion of the dune or/and sastrugi) fromthe glazed surface to the sastrugi field.

Figure 5. Surface elevation, microrelief morphology, andinternal layering along the megadune profile.

ACL 1 - 8 FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA

[26] Megadune crests are found to be perpendicular to theprevailing katabatic wind direction, as pointed out byFahnestock et al. [2000], and the angle between the winddirection (40�–50� True) and the direction of general sur-face slope (95�–100� True) at a regional scale is about 50–60� (Figure 4). This clearly shows the katabatic flowdraining from the high plateau and turning to the left underthe action of the Coriolis force.[27] The megadune surveyed along the traverse has a

wavelength of about 3 km and amplitude between 2 and 4 m(Figures 3, 4, and 5), in agreement with the observations ofFahnestock et al.’s [2000] remote sensing analysis and byour Landsat ETM+ analysis. Fahnestock et al. [2000]pointed out that regional slopes play a role in megadunegenesis. Analysis of satellite ETM+ images and DEM showsthat the megadune has subparallel crests ten to one hundredkilometers in length, occasionally branching or merging.Regional surface slope in the megadune area has a value of0.15–0.25% (1.5–2.5 m km�1), less than half of themegadune’s slip-face slope of 0.4–0.5% (4–5 m km�1).Our analysis of surface slope and satellite images (in area ofabout 65,000 km2) pointed out that slope in the prevailingwind direction (SPWD) is the most important topographiccharacteristic of the megadune. Profiles of the prevailingwind direction in the megadune area (Figures 4 and 6) showthat the megadune has been formed at the break-slope of theinterior plateau where the SPWD changes from 0.013%–0.022% (0.13–0.22 m km�1) to 0.1–0.15% (1–1.5 mkm�1). Megadunes are present only in the area where theSPWD is between 0.1% and 0.15%; increases or decreasesin slope along these values do not allow their formation.Due to the angle between the prevailing wind (40�–50�True) direction and regional slope (95�–100� True), thedifferences between SPWD and surface slope could belarge.[28] GPS and GPR profiles along the traverse show the

presence of paleomegadunes buried under the leeward sur-face of the megadune field. 3D GPS-GPR megadune images(Figures 3 and 5) show excellent resolution of sedimentarystructures within the megadunes. The layer reflection

pointed out the presence of several buried megadunes thatnevertheless are not in phase; at about 8–10 m under thecrests, buried crests are present in an oppositional phase tothose on the surface.[29] The megadune area is characterized by extreme

climatic conditions (temperature, snow accumulation, andwind). Meteorological data and continental-scale simulationof the wind field surface [Parish and Bromwich, 1991]show that megadune areas are characterized by constantkatabatic wind flow, with very uniform direction and speedsbetween 6 and 12 m s�1. Measurements at the Vostok-1station show that snow transportation by saltation starts atwind speeds of about 4.7 m s�1. King and Turner [1997]used Ball’s [1960] formula to calculate wind speed as afunction of slope gradient and interlayer potential temper-ature difference (20K); wind speeds increase rapidly fromslopes between 0.1 and 0.2% from 2 to 5 m s�1.

3.3. Megadune Formation Hypothesis

[30] The continuous and homogeneous SPWD (0.1–0.15%), the low accumulation value [Vaughan et al.,1999a], and the strength of surface inversion during wintertemperatures of 15�C and 25�C [Phillpot and Zillman,1970] provide environmental conditions for the develop-ment of a megadune.[31] The buried megadune sedimentary structure (Figures

3 and 5), the presence of long-term hiatus (glazed) surfaceson the downwind faces of megadunes and accumulation-redistribution forms (severe sastrugi) on the uphill facessuggest the surface migration of megadunes uphill (wind-ward). The uphill migration could be related to increasedextension and uphill migration of severe sastrugi burying theslip face (long-term hiatus surfaces) of previous windwardmegadunes. Figure 7 shows the possible scheme of mega-dune migration and burial mechanisms that are the oppositeof sand megadune formation where deposition occurs down-wind of the dune. Mudwaves or antidunes, with wavelengthsup to 6 km and heights up to 100 m, are commonly found onthe deep seafloor where steady, sediment-laden currents arepresent; their internal structure suggests that they migrateupstream with time [Flood, 1988; Normark et al., 1980].

Figure 6. Surface elevation profiles were derived fromDEM [Remy et al., 1999], and regional slope along theprevailing wind direction crossing the megadune area wascalculated (location in Figure 4). Grey lines correspond tothe area where the megadune is present, and numbers in thebox represent the slope value at regional scale. Continuouslines and bold numbers correspond to the area without themegadune.

Figure 7. Summary model of megadune growth process.Uphill migration could be related to increased extension anduphill progadation of severe sastrugi burying the slip face(long-term hiatus surfaces) of previous windward mega-dunes.

FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA ACL 1 - 9

The internal structure of the mudwave is very similar to thatof a megadune. As for the megadune, the mudwave crestappears to be perpendicular to the bottom current and themaximum sedimentation rate occurs on the upstream waveflank where the flow speed is lowest, while the minimumsedimentation rate occurs on the downstream flank wherethe flow speed is highest [Flood, 1988].[32] On the base of stake measurements, the accumulation

at D6 is close to the Dome C value (about 30 kg m2 a�1).Using the Dome C depth/age relationship [Legrand andDelmas, 1987; Udisti et al., 2000], we assume that at about12 m depth (5.54 m water equivalent) the snow layer is 183years old (1816 AD). This layer is present in a megadune ata maximum depth of about 14 m (6.70 m water equivalent)and at a minimum depth of about 3 m (1.15 m waterequivalent). Using a density profile and the 12 m layer atsite D6 we estimate the change of accumulation in themegadune area from 7 to 35 kg m2 a�1. The minimum valuerepresents a decrease of accumulation up to 75%, due to thelong-term hiatus of glazed surface. Classification of theglazed surface and snow using Landsat ETM+ imagesindicates that the glazed surface covers about 20% of themegadune area.[33] Using the distance of 1 km of the buried crest 180

years old and the present crest we can evaluate the wind-ward migration of the megadune crest at about 5 m a�1.Black and Budd [1964] pointed out a series of irregularwavelength undulations of 5 to 15 km and amplitudes of 7to 50 m along the Wilkes traverse. The topographic andaccumulation data show that maximum net accumulationoccurred near the bottom of depressions and a minimumoccurred near the crest of undulations. Black and Budd[1964] inferred an upslope migration at a constant rate ofabout 25 m a�1, with analogous results having beenreported by Whillans [1975], who suggested a migrationof the undulations upstream at about 20 m a�1 in Marie

Byrd Land. The windward migration of the megaduneis related to the buried megadune since the ice sheetsurface flows parallel to the local surface slope and carriesthe undulations downstream at a rate of around 1.5 m a�1

(L. Vittuari, personal communication 2001). Megadunes areprograding windward and the ice is flowing downhill, atroughly the 30% rate with a module with 50�–60� differ-ence. So a century-long ‘‘movie’’ of satellite images wouldshow the megadunes moving sideways across the face ofthe ice sheet. Our results support the Fahnestock et al.[2000] satellite observation that shows an identical patternof branching and megadune expanse in images acquired 34years apart with a maximum possible shift of 60 m a�1.[34] One of the most important pieces of positive feed-

back is the smoothness of the glazed surface compared tothe roughness and albedo of the surrounding snow. Therelatively low albedo of the glazed surface increases theenergy available for sublimation, and the smooth surfacealso prevents drifting snow from becoming attached andcausing the wind to be stronger over the glazed surface thanover the snow [Fujii and Kusunoki, 1982; van den Broekeand Bintanja, 1995]. The mechanism of snow megadunemigration is linked to the glazed surface triggered by windacceleration and the windward progadation (on the inter-megadune area) of sastrugi that buried the glazed surface.[35] We assume that the path followed by the wind across

a series of transverse megadunes has a wave-like form(Figure 8), which climbs parallel or semiparallel to thewindward surface of the megadune and descends again tothe windward edge of the succeeding megadune, leaving azone of dead air in the uphill region between megadunes.Wind vortices develop in this zone of dead air and presum-ably account for the accumulation area close to the foot of aslip-face. On the leeward smooth surface, the increase ofwind velocity is due to the higher slope (0.53%) and thereduction in surface roughness that allows the blowing of

Figure 8. Schematic sketch of the hypothesis of megadune formation.

ACL 1 - 10 FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA

snow and the formation of glazed surface. The leeward sideof the megadune becomes covered with active dune inwhich snow is transported from the mound crest. Subse-quently, dunes are destroyed and pitted by wind andchanges in severe sastrugi.[36] The megadune field is characterized by extreme

climatic conditions (wind, temperature, snow accumulation,etc.), with very uniform wind direction and speed and by thepresence of a relatively abrupt increase in SPWD (from0.013–0.022% to 0.10–0.15%). A permanent feedbacksystem between the cryosphere and atmosphere must bepresumed for the creation and maintenance of the megadunefield.[37] One hypothesis could be that the near-surface air

accelerates down the abrupt change of SPWD in response tothe buoyancy force, triggering a wavelength of about 3–4km and an amplitude of 3–4 m that generates a glazedsurface on the leeward side of the megadune and accumu-lation in the uphill intermegadune area. The wind-waves areformed at the change of slope in strongly stable environ-ments with light winds and could be correlated to a naturalresonance.[38] Fahnestock et al. [2000] and Long and Drinkwater

[1999] pointed out an anomalous signature of the megadunefields in microwave emission measured by the SSM/I andNSCAT sensors, and variation in grain size across themegadune by AVHRR and SAR imagery. The authorscorrelated the presence of profound layering or extremelycoarse firn with the extensively recrystallized snow. Ourobservations confirm this hypothesis, i.e., the leeward sideof the megadune is characterized by a glazed surface andshows the following: (1) a relatively wide area of highbackscatter in Synthetic Aperture Radar (SAR) on ERS1-2and Radarsat images due to the large grain size of the hoarlayer under the glazed surface; (2) a decrease in reflectanceof ETM+ bands 3 (visible red) and 4 (near-infrared) is due tohigh absorption of ice crystals on glazed surfaces in the nearinfrared [Warren, 1982] and possibly contribute to theunderlying large grain size of the hoar layer. (3) in summer,a relatively higher brightness temperature (Landsat 7 ETM+

channel 6) is found on a glazed surface with respect to thesurrounding snow, due to the larger amount of absorbedsolar radiation and higher temperature of glazed surfacethan snow temperature, as pointed out for blue-ice areas[Bintanja, 1999, and references therein].[39] The windward side of the megadune is characterized

by sastrugi and dunes and shows the following: (1) a narrowlow-backscatter strip in SAR images is due to the smallergrain size of younger dunes and sastrugi made by snowremains; and (2) a decrease in reflectance of all visible andnear-infrared bands of Landsat ETM+ with low sun eleva-tion in severe sastrugi areas (up to 1.5 m) is due to surfaceroughness. Shadowing from sastrugi can have a substantialinfluence on the snow’s albedo. Shadow size will depend onthe form and direction of the sastrugi, and solar position,height, and azimuth [Wendler and Kelley, 1988].

4. Conclusion

[40] Megadune fields occupy large areas in the interior ofthe East Antarctic ice sheet and are the result of a permanentand unusual feedback system between the cryosphere and

atmosphere. A combination of remote sensing and fieldobservation analysis indicate that slope in the prevailingwind direction (SPWD) and climatic conditions play acrucial role in megadune genesis. The megadunes represent2 to 4 m amplitude waves of 2 to 5 km wavelength formedby variable net accumulation, ranging between 25% (lee-ward faces) to 120% (windward faces) of the accumulationin adjacent nonmegadune areas. Leeward faces are charac-terized by glazed, sastrugi-free surfaces and extensive depthhoar formation and covers 20% of the megadune area.Windward faces are covered by large rough sastrugi up to1.5 meters in height. Megadune internal structures suggestthat they are prograding windward with time at about 5 ma�1. The ice is flowing downhill, at roughly 30% of that rateand with a module with 50�–60� difference in direction.The mechanism of snow megadune prograding is linked tothe uphill migration on the intermegadune area of sastrugithat buried the glazed surface. The megadune field ischaracterized by extreme climatic conditions (wind, temper-ature, snow accumulation etc.), with very uniform winddirection and speed and are present downwind with arelatively abrupt increase in SPWD (from 0.013–0.022%to 0.10–0.15%). Megadunes appear to be formed by anoscillation in the katabatic air flow leading to a wave-likevariation in net accumulation. The wind-waves are formedat the change of SPWD, in response to the buoyancy force,in strongly stable environments with light winds, and mightbe related to a natural resonance.[41] Megadune fields therefore are important to surface

mass balance and ice core interpretation and provideimportant observations on the boundary layer atmosphereover the East Antarctic ice sheet. Ice cores drilled in themegadune area or one that including a buried megadunewould be expected to present climatic and environmentalproxy information (chemical and isotope composition, gasconcentration, snow accumulation, etc.), periodic a scale ofhundreds of years, that would distort the climate record. Theexceptionally stable environmental and climatic conditionsnecessary to form the megadunes may have been presentduring the glacial period over a larger sector of Antarctica,as suggested by Fahnestock et al. [2000]. Megadunes arepresent in the Vostok station area, and therefore could berecorded in deep ice cores; thus the climate record preservedin the cores may be significantly distorted by megadunesedimentary structures, unrepresentative of past climates.Megadune formation is triggered by slightly change in theSPWD, associated with a constant, moderate katabatic flowand wind direction and therefore ice cores drilled at icedomes (Dome C, Dome Fuji etc.) should be free frommegadunes or buried megadunes.[42] There is a need for further study of the distribution,

genesis and structure of megadune formation. This processhas important consequences concerning the choice of sitesfor ice coring and paleoclimate interpretation.

[43] Acknowledgments. This research was carried out within theframework of a Project on Glaciology of the Programma Nazionale diRicerche in Antartide (PNRA) and was financially supported by ENEAthrough a cooperation agreement with the Universita degli Studi di Milano-Bicocca. This work is a contribution of the Italian branch of the ITASEproject. This work is an associate program to the ‘‘European Project for IceCoring in Antarctica’’ (EPICA), a joint ESF (European Science Foundation)/EC scientific program. The authors wish to thank all members of the traverse

FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA ACL 1 - 11

team, the participants in PNRA 1998/99 who assisted at the Terra Nova andConcordia Stations and all persons in Italy who were involved in thepreparation of the traverse. Thanks are due to T. Scambos and J. Splett-stoesser whose comments and editing helped to improve the manuscript.

ReferencesAlley, R. B., Concerning the deposition and diagenesis of strata in polarfirn, J. Glaciol., 34(118), 283–290, 1988.

Ball, F. K., Winds on the ice slopes of Antarctica, in Antarctic Meteorology,pp. 9–16, Pergamon, New York, 1960.

Bintanja, R., On the glaciological, meteorological, and climatological sig-nificance ofAntarctic blue ice areas,Rev. Geophys., 37(3), 337–359, 1999.

Black, H. P., and W. Budd, Accumulation in the region of Wilkes, WilkesLand, Antarctica, J. Glaciol., 5(37), 3–15, 1964.

Bromwich, D. H., T. R. Parish, and C. A. Zorman, The confluence zone ofthe intense katabatic winds at Terra Nova Bay, Antarctica, as derivedfrom airborne sastrugi surveys, and mesoscale numerical modeling,J. Geophys. Res., 95, 5495–5509, 1990.

Capra, A., R. Cefalo, S. Gandolfi, G. Manzoni, I. E. Tabacco, and L.Vittuari, Surface topography of Dome Concordia (Antarctica) from kine-matic interferential GPS and bedrock topography, Ann. Glaciol., 30, 42–46, 2000.

Fahnestock, M. A., T. A. Scambos, C. Shuman, R. J. Arthern, D. P. Wine-brenner, and R. Kwok, Snow megadune fields on the East AntarcticPlateau: Extreme atmosphere-ice interaction, Geophys. Res. Lett.,27(20), 3719–3722, 2000.

Flood, R. D., A lee wave model for deep-sea mudwave activity, Deep SeaRes., 35(6), 973–983, 1988.

Frezzotti, M., Surface wind field of Victoria Land (Antarctica) from surveysof aeolian morphologic features, Terra Antarct. Rep., 1, 43–45, 1997.

Frezzotti, M., S. Gandolfi, F. La Marca, and S. Urbini, Snow dunes andglazed surface in Antarctica: New field and remote sensing data, Ann.Glaciol., 34, 81–88, 2002.

Fujii, Y., and K. Kusunoki, The role of sublimation and condensation in theformation of ice sheet surface at Mizuho Station, Antarctica, J. Geophys.Res., 87, 4293–4300, 1982.

Fujiwara, K., and Y. Endo, Preliminary report of glaciological studies, inReport of the Japanese Traverse, Syowa-South Pole 1968–69, pp. 71–104, edited by M. Murayama, Polar Res. Cent., Natl. Sci. Mus., Tokyo,Japan, 1971.

Gay, M., M. Fily, C. Genthon, M. Frezzotti, H. Oerter, and J. G. Winther,Snow grain size measurements in Antarctica, J. Glaciol., in press, 2002.

Glover, J. M., and H. V. Rees, Digital enhancement of ground probing radardata, in Ground Penetration Radar, Geol. Surv. Can., 90(4), 187–192,1992.

Goodwin, I. D., Snow accumulation and surface topography in the katabaticzone of Eastern Wilkes Land, Antarctica, Antarct. Sci., 2(3), 235–242,1990.

Goodwin, I. D., M. Higham, I. Allison, and R. Jaiwen, Accumulationvariation in eastern Kemp Land, Antarctica, Ann. Glaciol., 20, 202–206, 1994.

Gow, A. J., On the accumulation and seasonal stratification of snow at theSouth Pole, J. Glaciol., 5, 467–477, 1965.

Houbolt, J. J., Recent sediments in the southern bight of the North Sea,Geol. Mijbouw, 47(4), 245–273, 1968.

King, J. C., and J. Turner, Antarctic Meteorology and Climatology, Atmos.Space Sci. Ser., 408 pp., Cambridge Univ. Press, New York, 1997.

Kovacs, A., A. J. Gow, and R. M. Moorey, The in-situ dielectric constant ofpolar firn revisited, U.S. Army Cold Reg. Res. Eng. Lab., 23, 245–256,1995.

Jezek, K. C, Glaciological properties of the Antarctic ice sheet from RA-DARSAT-1 synthetic aperture radar imagery, Ann. Glaciol., 29, 286–290, 1999.

Legrand, M., and R. J. Delmas, A 220-year continuous record of volcanicH2SO4 in the Antarctic ice sheet, Nature, 327, 671–676, 1987.

Liston, G. E., J. G. Winther, O. Bruland, H. Elvehoy, K. Sand, and L.Karlof, Snow and blue-ice distribution patterns on the coastal AntarcticIce Sheet, Antarct. Sci., 12(1), 69–79, 2000.

Long, D. G., and M. R. Drinkwater, Cryospheric applications of NSCATdata, IEEE Trans. Geosci. Remote Sens., 37(3), 1671–1684, 1999.

Mather, K. B., Further observations on sastrugi, snow dunes and the patternof surface winds in Antarctica, Polar Rec., 11(7), 158–171, 1962.

Mayewski, P. A., and I. D. Goodwin, Antarctic’s role pursued in globalclimate change, Eos Trans. AGU, 80, 398–400, 1999.

Mazler, C., Applications of the interaction of microwaves with the naturalsnow cover, Remote Sens. Rev., 2, 259–392, 1987.

Narita, H., Controlling factor of drifting snow, Mem. Natl. Inst. Polar Res.,7, 81–92, 1978.

Normark, W. R., G. R. Hess, D. A. V. Stow, and A. J. Bowen, Sedimentwaves on the Monterey fan levee: A preliminary physical interpretation,J. Sediment. Res., 1(70), 84–93, 1980.

Okuhira, F., and H. Narita, Glaciological studies in Mizuho Plateau, eastAntarctica, 1969–1975, Mem. Natl. Inst. Polar Res., 7, 140–153,1978.

Orheim, O., and B. K. Lucchitta, Snow and ice studies by thematic mapperand multispectral scanner Landsat images, Ann. Glaciol., 9, 109–118,1987.

Parish, T. R., and D. H. Bromwich, Continental scale of the Antarctickatabatic wind regime, J. Clim., 4(2), 135–146, 1991.

Petit, J. R., et al., Climate and atmospheric history of the past 420,000 yearsfrom the Vostok ice core, Antarctica, Nature, 399, 429–436, 1999.

Pettre, P., J. F. Pinglot, M. Pourchet, and L. Reynaud, Accumulation inTerre Adelie, Antarctica: Effect of meteorological parameters, J. Glaciol.,32, 486–500, 1986.

Phillpot, H. R., and J. W. Zillman, The surface temperature inversion overthe Antarctic continent, J. Geophys. Res., 75, 4161–4169, 1970.

Proposito, M., S. Becagli, E. Castellano, O. Flora, R. Gragnani, B. Stenni,R. Traversi, R. Udisti, and M. Frezzotti, Chemical and isotopic snowvariability along the 1998 ITASE traverse from Terra Nova Bay to DomeC (East-Antarctica), Ann. Glaciol., in press, 2002.

Remy, F., P. Shaeffer, and B. Legresy, Ice flow processes derived from theERS-1 high resolution map of the Antarctica and Greenland ice sheets,Geophys. J. Int., 139, 645–656, 1999.

Richardson, C., and P. Holmlund, Regional and local variability in shallowsnow-layer depth from a 500 km continuous radar traverse on the polarplateau, central Dronning Maud Land, east Antarctica, Ann. Glaciol., 29,10–16, 1999.

Richardson, C., E. Aarholt, S. E. Hamran, P. Holmlund, and E. Isaksson,Spatial snow distribution mapped by radar, J. Geophys. Res, 102(B9),20,343–20,353, 1997.

Seko, K., T. Furukawa, and O. Watanabe, The surface condition on theAntarctic Ice Sheet, in Proceedings of the Symposium on Polar Regionsand Climate Change, edited by G. Weller, C. L. Wilson, and B. A. B.Severin, vol. 1, pp. 238–242, Univ. of Alaska, Fairbanks, 1992.

Sihvola, A., E. Nyfors, and M. Tiuri, Mixing formulae and experimentalresults for the dielectric constant of snow, J. Glaciol., 31(108), 163–170,1985.

Swithinbank, C., Antarctica, U.S. Geol. Surv. Prof. Pap., 1386-B, 1988.Udisti, R., S. Becagli, E. Castellano, R. Mulvaney, J. Schwander, S. Torcini,and E. Wolff, Holocene electrical and chemical measurements from theEPICA-Dome C ice core, Ann. Glaciol., 30, 20–26, 2000.

Urbini, S., S. Gandolfi, and L. Vittuari, GPR and GPS data integration:Examples of application in Antarctica, Ann. Geofis., 44(4), 687–702,2001.

Van den Broeke, M. R., and R. Bintanja, The interaction of katabatic windsand the formation of blue-ice areas in East Antarctica, J. Glaciol., 41(38),395–407, 1995.

Van der Veen, C. J., E. Mosley-Thompson, A. J. Gow, and B. G. Mark,Accumulation at South Pole: Comparison of two 900-year records,J. Geophys. Res., 104, 31,067–31,076, 1999.

Vaughan, D. G., J. L. Bamber, M. Giovinetto, J. Russell, and P. R. Cooper,Reassessment of net surface mass balance in Antarctica, J. Clim., 12,933–946, 1999a.

Vaughan, D. G., H. J. F. Corr, C. S. M. Doake, and E. D. Waddington,Distortion of isochronous layers in ice revealed by ground-penetratingradar, Nature, 398(6725), 323–326, 1999b.

Warren, S. G., Optical properties of snow, Rev. Geophys., 20, 67–89, 1982.Watanabe, O., Distribution of surface features of snow cover in MizuhoPlateau, Mem. Natl. Inst. Polar Res., 7, 154–181, 1978.

Weller, G., The heat and mass balance of snow dunes on the central Ant-arctic plateau, J. Glaciol., 8, 277–284, 1969.

Wendler, G., and J. Kelley, On the albedo of snow in Antarctica: Contribu-tion to I.A.G.O., J. Glaciol., 34, 19–25, 1988.

Whillans, I. M., Effect of inversion winds on topographic detail and massbalance on inland ice sheets, J. Glaciol., 14(70), 85–90, 1975.

�����������������������M. Frezzotti, Ente per le Nuove Tecnologie, l’Energia e l’Ambiente, P.O.

Box 2400, 00100 Roma AD, Italy. (frezzotti@casaccia.enea.it)S. Gandolfi, Dipartimento di Ingegneria delle Strutture, dei Trasporti,

delle Acque, del Rilevamento, del Territorio, University of Bologna, VialeRisorgimento 2, 40136 Bologna, Italy.S. Urbini, Istituto Nazionale di Geofisica e Vulcanologia — Dipartimento

per lo studio del Territorio e delle sue Risorse, University of Genova, VialeBenedetto XV 5-16134, Genova, Italy.

ACL 1 - 12 FREZZOTTI ET AL.: SNOW MEGADUNE IN ANTARCTICA