Marine air intrusion and subsequent cloud formation plays a dominant role in the energy budget and mass balance of the Antarctic. However, the intrusion is very difficult to understand using the ground-based measurements alone. In this paper we present simulations of marine air intrusion into the Adelie Land, East Antarctica, using the Modèle Atmosphérique Régional (MAR), for July 1994 and January 1995. The model is nested into the European Centre for Medium-Range Weather Forecasts (ECMWF) analyses. The simulations show a strong influence of large-scale disturbances, over the ocean, which helped in the penetration of marine air into the interior and the formation of clouds. Each marine air intrusion episode resulted in cloud formation in July 1994. Blocking anticyclones have also been found to be responsible for much of the moisture transport far into the interior elevated locations. MAR simulations, as well as ECMWF analyses, show influence of cyclones in strengthening and prolonging the surface layer flow. The study also indicated that the influence of depressions on surface winds is pronounced during the period when the depression is approaching the Adelie Land coast.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Marine air intrusion into the Antarctic interior cannot be discussed or introduced without discussing transient eddies or cyclones. The topography of the continent is such that it acts like a barrier to any local penetration. This is particularly true for East Antarctica, where coastal slopes are of the order of 10−1 to 10−2. Cyclones can be regarded as forces helping the air mass to rise against topography and penetrate inland. Advection of warm marine air is the only source of moisture into the continent and is very important for the mass balance of the Antarctic ice sheet. Connolley and King  reported that the main contribution to water vapor into the continent is provided by transient eddies. Clouds in the coastal region are associated with major low-pressure systems [Schwerdtfeger, 1984; King and Turner, 1997]. Warm air masses advected by cyclones into the interior of the continent encounter topography, rise adiabatically (because of elevation), cool, condense, and result in cloud formation, and often, precipitation in the coastal regions. Extratropical cyclones also advect cloud cover ahead of them as they track eastward. The presence of clouds greatly effects the reflected radiation of the atmosphere. In summer their main impact is to decrease the short wave radiation coming to Earth, while in winter, when the solar radiation is not there, they reflect the long wave radiation back to Earth, and therefore result in temporary heating near the surface. This heating occasionally destroys inversions. The rise in temperature is more pronounced in winter months than in summer months. Thus clouds play an important role, besides inversions and katabatic winds, in the radiation budget of the Antarctic boundary layer, in addition to their significant role in precipitation and mass budget of the continent.
Webster and Stephens  underlined the influence of clouds on key quantities such as Earth radiation budget, the heating of the ocean, and the diabatic heating of the atmosphere; they concluded that of all climate parameters, clouds have the largest possible effect on the radiation field. Stone et al.  studied the short-term variation in the surface radiation balance and temperature associated with cloudiness at the South Pole. They reported that clouds tend to heat the surface in snow-covered polar regions, and the heating increases monotonically with increasing cloud cover and/or cloud optical depth. Ramanathan et al.  presented satellite-based remote sensing results from the Earth Radiation Budget Experiment, showing that, whereas clouds generally have a net cooling effect on the surface of the Earth, they contribute to anomalous surface warming in high-latitude snow-covered regions. According to them the radiative forcing due to clouds potentially far exceeds that due to gaseous green house warming. Even small changes in the frequency, type, or spatial distribution of clouds result in significant climate change on a regional scale [Stone and Kahl, 1991]. Neff  emphasized that the energy balance of the South Polar Plateau depends critically on the structure of the surface temperature inversion and its control by synoptic scale processes.
 In coastal regions near 70°S, clouds occur throughout the year with little variability and only a small decrease during winter months [King and Turner, 1997]. This might be due to the formation of sea ice, which reduces the flux of moisture inland and, to the reduced cyclonic activity in winter [van Loon, 1967]. It is easier and natural for the moist air to penetrate far inland during summer when there is no sea ice and the warmer ocean is nearby. In winter it should be more difficult because the sea ice is almost up to its maximum extent and the open sea is far north from the continent. The warm marine air moving southward toward the continent gets cooled while traveling over the sea ice. Therefore this air is cold to rise adiabatically in winter months. For the marine air to reach the continent and to penetrate far inland against the ice slopes, the operating mechanism has to be strong to advect the air across the sea ice into the interior, so the travel time of the warm air over the sea ice is less. As we go into the interior, the amount of cloud cover decreases. Most of the cloud formation is due to transient eddies, and because these eddies cannot penetrate far inland, the amount of cloud cover in the interior is very low. However, some well-organized depressions do penetrate far into the interior even in winter months [Neal, 1972; King and Turner, 1997; Pook and Cowled, 1999]. Synoptically quiet or high-pressure situations also can cause clouds and precipitation in the coastal region [King and Turner, 1997, p. 111; Turner et al., 1995].
 The motivation of this paper is to understand the mechanism behind marine air intrusions into the Antarctic interior. In this study we present observations and simulations of marine air intrusions using the Modèle Atmosphérique Régional (MAR) (regional climate model) [Gallée and Schayes, 1994; Gallée, 1995], over the Adelie Land quadrant of East Antarctica. Analyses of the European Centre for Medium-Range Weather Forecasts (ECMWF) are used to initialize the model and force its lateral boundaries. For this study we have used July 1994, 15 October to 15 November, 1994 and, January 1995, representing three Special Observing Periods (SOPs) of the Antarctic First Regional Observing Study of the Troposphere (FROST) project [Turner et al., 1996]. Details of the FROST project and earlier work done can be found in the works of Turner et al. [1996, 1999] and in the special issue on the FROST project in Weather and Forecasting, volume 14, December 1999. Estimation of 500 hPa geopotential heights from station-level observations of pressure and temperature using Phillpot  technique were also a part of the FROST project and have been included in the analyses. This work started with the objective to utilize the comprehensive data sets assembled during the three SOPs. Out of the three SOPs, the first SOP showed two very interesting examples of marine air intrusion, more pronounced than the other SOPs. Both cases took place with different synoptic conditions. In this work we present simulations from the first and last SOP, i.e., SOP1 and SOP3, representing winter and summer months, respectively.
2. Model Description
 In this study we present simulations using MAR (Modèle Atmosphérique Régional). MAR is a hydrostatic, primitive equation model. Model details can be found in the works of Gallée and Schayes  and Gallée . However, a brief description is given here and in the following sections. In the present version the model uses 30 sigma levels to represent the vertical structure of the atmosphere (sigma = 0.998, 0.997, 0.994, 0.990, 0.980, 0.963, 0.933, 0.886, 0.821, 0.744, 0.667, 0.600, 0.544, 0.497, 0.456, 0.420, 0.385, 0.352, 0.319, 0.287, 0.255, 0.224, 0.194, 0.165, 0.137, 0.111, 0.088, 0.067, 0.048, 0.0321). The lowest sigma level is ∼10 m above the surface, and the top level is at 0.1 hPa. In the top six sigma levels, horizontal diffusion is more pronounced so as to damp the waves horizontally, i.e., to smooth the vertically propagating gravity waves. This is introduced as an extra term in the horizontal eddy exchange coefficient for the upper levels. This viscous damping layer minimizes the reflection from the upper boundary. MAR has been used for studies of the Antarctic atmosphere [Gallée and Schayes, 1994; Gallée, 1995, 1996; Gallée et al., 1996]. In previous studies, MAR was initialized using soundings. In the present work it is initialized from ECMWF fields for Antarctica [Brasseur, 2001]. High-resolution limited area models nested within large-scale models have been successfully used for climate studies [Giorgi et al., 1993a, 1993b; Walsh and McGregor, 1996]. The present study will determine the capability of MAR as a nested version in simulating the Antarctic atmosphere.
 ECMWF analyses have been used to initialize lateral boundaries of the model. Boundary conditions are updated every 6 hours from ECMWF analyses and interpolated in each direction using a bicubic horizontal interpolation. The horizontal resolution used in MAR is 40 km by 40 km.
2.1. MAR Moist Physics
 The hydrological cycle includes a cloud microphysical model, with conservation equations for cloud droplet, raindrop, cloud ice crystal, and snowflake concentrations. Blowing snow particles are considered as snowflakes. The representation of the cloud microphysical processes is essentially based on the Kessler  parameterization. In particular, the sublimation of snowflakes is a function of relative humidity of the air, according to Lin et al. [1983, relation (13), p.1072].
2.2. MAR Boundary Conditions
 The temperature, specific humidity, horizontal wind speed, and surface pressure at the lateral boundaries of MAR are specified from the analyses. Furthermore, MAR variables are relaxed toward the analyses over five grid points near the lateral boundaries, using both Newtonian and diffusive relaxation [e.g., Davies, 1983]. Sea surface temperatures (SSTs) are prescribed from the analyses. Sea ice is assumed in each grid point where the SST is lower than the seawater freezing temperature (271.2K).
3. Results of Simulations
 In this section, model output and ECMWF analyses for SOP1 and SOP3, i.e., July 1994 and January 1995, are presented. Figure 1 shows the model domain. The height contours are drawn at 500 m intervals. From the figure it is clear that the model is able to resolve topographical details of the continent to a good approximation. The model domain includes the sea ice, the nearby ocean, and the part of the continent from where winds originate, all at sufficient distance from the domain boundaries. This consideration ensures that the simulation of cyclones and katabatic winds are unbiased from large-scale boundary forcings, which are provided at the boundaries and slowly relaxed inward up to five grid points (after which inward actual model simulation prevails). We will first present the main results of the two 1-monthlong simulations (the two SOPs) to show the model's ability to simulate the circumpolar trough correctly. Later, we will concentrate on the periods exhibiting marine air intrusion.
3.1. Long-Term Simulations
Figures 2 and 3 show the time-averaged 500 hPa level geopotential height and mean sea level pressure (mslp) fields for the model simulations and the difference between the simulated and the analyzed mslp fields, for SOP1 and SOP3, respectively. Results are plotted over the model domain shown in Figure 1. The 500 hPa geopotential gradients are weak over the continent as compared to over the ocean in both SOPs (Figures 2a, 3a). The difference between the simulated and the analyzed heights are largest along the coastline of the Ross Ice Shelf in both SOPs (Figures 2b, 3b), where the model has underestimated the magnitudes. Nevertheless, the difference does not exceed 65 m. Similar underprediction of 60 m has been reported by Hines et al.  for the June 1998 averaged 500 hPa simulation by the MM4 model. The 500 hPa heights are well represented in the rest of the domain, the difference never exceeding 20 m. Coming to the mslp plots (Figures 2c, 3c), the circumpolar trough is well represented by MAR. The mslp has been overpredicted in simulations by 4 hPa over the ocean in both SOPs (Figures 2d, 3d). The maximum overprediction by MAR is in the interior near the coast, being 10 and 4 hPa for SOP1 and SOP3, respectively. The mslp is underpredicted by MAR over the coastline along the Ross Ice Shelf.
Figures 4 and 5 present the temperature and horizontal wind, time averaged over SOP1 and SOP3, respectively, for the lowest model sigma level. Temperatures are overpredicted by MAR in the interior by 6° in July 1994 (Figure 4b) and underpredicted over the Ross Ice Shelf area by 6°. Temperatures are similar to the ECMWF data over the rest of the continent. In January 1995 the positive bias in temperature is 2° (Figure 5b). A negative bias exists along much of the coastline, being a maximum (−5°) over the Ross Ice Shelf coastline. Over the ocean the simulated values are similar to the analyses. MAR-simulated wind speeds are maximum along the coastline (Figures 4c, 5c), which is as expected since the ice slopes are maximum there. The wind speeds are similar for both SOPs over much of the domain (Figures 4d, 5d). The difference in wind speeds between MAR and ECMWF is noticeable only along the coastline, being maximum along the Ross Ice Shelf coast. Surface winds along the coastal area, especially along the Victoria Land coast, are underestimated by ECMWF.
 From Figures 2, 3, 4, and 5, we see the high negative bias in 500 hPa, mslp and temperature fields and a high positive bias in wind speeds between MAR and ECMWF fields over the coastline along the Ross Ice Shelf. Table 1 gives the mean monthly averaged meteorological parameters for Enigma Lake (EL) (74.43°S, 164.02°E, 210 m above sea level (asl)) and Cape Philips (CP) (73.03°S, 169.36°E, 550 m asl) sites over the Ross Ice Shelf. MAR grid points for these sites are (74.7°S, 164.1°E,234 m asl) and (72.5°S; 169.8°E, 569 m asl), respectively, making them 31 m and 26 m above the actual elevation, respectively. Average monthly pressures are similar for both MAR and ECMWF for July 1994 and differ by a maximum of 4 hPa from observations. For January 1995, average pressure predicted by MAR is less than about 3 hPa compared to ECMWF for both sites. Temperature values are underestimated compared to observations by MAR and ECMWF for both SOPs and for both sites. MAR-simulated average winds are higher as compared to ECMWF, which predicts extremely low winds at both sites. However, MAR-simulated winds are closer to observations for both SOPs. In general, MAR simulations for July 1994 are closer to observations than the ECMWF ones.
Table 1. Mean Monthly Average Values of Pressure (p), Temperature (T), and Horizontal Wind (u) for Enigma Lake (EL) and Cape Philips (CP)
 In the above discussion and Figures 2, 3, 4, and 5, we see that MAR is able to simulate the synoptic situation and other parameters to a fairly accurate approximation. In the following we will be discussing data obtained from grid points closest to Dumont d'Urville (DdU) (67.7°S, 140°E, 40 m asl), D80 (70.2°S, 134.72°E, 2500 m asl), and Dome C (74.5°S, 123°E, 3280 m asl). The corresponding grid point locations in MAR domain for DdU are 66.4°S, 140.7°E, 49 m asl; for D80 70.0°S, 134.4°E, 2500 m asl; and for Dome C 74.5°S, 122.3°E, 3042 m asl. This makes DdU and D80 grid points, in MAR, 9 and 7 m above the observational sites, respectively, while the model grid point for Dome C is 228 m below the observational site. These sites are marked in Figure 1. These three stations are chosen because they represent the main direction of the surface layer flow in the Adelie Land. Surface winds at DdU mainly originate at Dome C and pass through D80.
3.2. Grid Point Studies
Figure 6 shows MAR-simulated, ECMWF-analyzed, and observed time series of meteorological parameters for DdU and D80 for the two SOPs. While looking into the data and making comparison, it should be kept in mind that model grid points, best representing DdU and D80, represent an average for 40 km by 40 km grid box. Results should, therefore not be expected to be exactly similar to observations. The differences in elevation between the model grid points and the actual observation sites, however, are 9 m for DdU and 7 m for D80. It should be kept in mind that ECMWF analyses are constrained by the radiosonde observations at DdU and can thus be considered as representing the actual conditions at DdU. For both SOPs, Dome C results are different. This might be due to large difference in elevation (228 m) between the actual location and the model grid point (Dome C results are not shown here). For the first SOP, all three data types (observations, simulations, and analyses) show a good correlation except that for D80 and Dome C grid points, ECMWF and MAR overestimate winds in comparison to observations. For SOP3, time series of MAR simulation and ECMWF analyses match very well with observations except that pressure values are overestimated for DdU and Dome C. For D80, observations are available only up to 17 January 1995.
 For SOP1, time series of temperature for all stations show episodes, where a temperature increase of up to 20°C is observed within 2 and 3 days. This temperature anomaly is observed in all three stations, being maximum in DdU. The diurnal temperature cycle is underestimated by MAR. The temperature anomaly in the last week of July 1994 builds up gradually for DdU, while for D80 and Dome C, the temperature increase is abrupt. The increase in temperature in July 1994 is associated with high winds during the same period and can be explained by the vertical mixing provided by strong winds between the cold boundary layer air and the warm free atmospheric air. In January 1995, winds along with the temperature depict a diurnal variation. This variation in wind is broken during the passage of low-pressure systems. The effect can also be seen in the temperature cycle. In both SOPs, each depression is preceded closely by a high-pressure ridge. This high pressure allows strong surface inversions to form and helps in forming cold air supplies feeding the gravity flow during the passage of the depression for longer duration and greater intensity [Murphy, 2000].
Figures 7, 8, and 9 present the time-height variation of wind speed, specific humidity (g/kg), and clouds (g/kg) for two SOPs. Clouds are represented as the sum of ice water and liquid water content. Time-height plots give an idea of the temporal and spatial evolution of the phenomena in each station. Relative magnitudes of each episode in each parameter between stations should be looked upon by considering the distance of the station from the sea and their elevation above sea level. The sea ice extent is approximately up to 63°S in July 1994, which was similar to the maximum extent in September 1994. The sea ice extent made the sites farther away from the open ocean. Sea ice continued to extend to an average of 63°S between 145°E and 170°S, and a polynya (open water) existed in the Ross Ice Shelf area in January 1995.
 Time-height plots of MAR simulations for all sites show a good similarity with ECMWF analyses. To see marine intrusion episodes, specific humidity and cloud time-height plots present a good indication. Various episodes of cold air outbreak (Figure 7) and warm air intrusion (Figures 8 and 9) are well represented by MAR. The magnitude of each episode is either overestimated or underestimated a little by MAR in comparison to ECMWF analyses. In the case of a cold air outbreak, MAR magnitudes are overestimated for DdU in July 1994 (Figure 7b), while marine intrusion magnitudes (high-humidity episodes) are slightly underestimated for both grid points and for both SOPs (Figure 8). Nevertheless, each episode is clearly identified by MAR. Of notice is the presence of clouds each time an intrusion took place (Figure 9). This is particularly visible in SOP1 (Figures 9a, 9b), where each small marine air intrusion episode in time and space domain is able to cause cloud formation. In the case of SOP3 (Figures 9c, 9d), episodes of marine air intrusion are not able to cause cloud formation as effectively as during July 1994 episodes.
 In the time series plots (Figures 6, 7, 8, 9) we notice certain distinct locations in time, where cold air outbreak and warm air intrusion episodes took place. From time-height plots it is clear that the warming shown by time series of temperature in July 1994 in Figure 6 is due to the combined influence of the vertical mixing provided by strong winds, warm marine air intrusions, and the subsequent heating provided by clouds. For instance, the first week and the last week of July 1994 show distinct episodes of marine air intrusion. Similarly, there are some events of marine air intrusions in SOP3: around 3, 10, and 30 January 1995. In the following section we will focus on two warming/intrusion episodes that occurred in the first and in the last week of July 1994 and three episodes of SOP3. We will see later that synoptic situations in both episodes of SOP1 are totally different.
3.3. Marine Intrusion Episodes
3.3.1. First week of SOP1
Figure 10 shows the composite of mslp fields and cloud optical depth (OD) (g/m3) simulated by MAR for 2, 4, and 5 July 1994, respectively. Wind vectors are plotted at the first sigma level, i.e., 10 m above the surface. Results are plotted over the model domain. The scale for the optical depth is kept similar for all plots to allow comparison between episodes and seasons.
 On 2 July 1994, a low-pressure trough influenced the coast. Surface winds increased on 2 July 1994 due to the presence of this small depression near the DdU coast. Strong winds were observed for almost 3 days. This small depression merged with the main depression after 3 July 1994, and together they helped in the persistence of strong winds. The system moved past the DdU site after 5 July 1994. This indicates the influence of cyclones in strengthening the already existing downslope winds by increasing the pressure gradient between the continent and the coast. During this episode (Figures 7a– 7d), first the cold air outbreak took place and then the warm air penetrated. From time- height plots (Figures 7a, 7b, 7c, 7d and Figures 8a, 8b, 8c, 8d), we notice that initially, the wind was high and the moisture penetration took place above the boundary layer. At the same time a cloud appeared above DdU around 2 July (Figure 10b), which slowly penetrated inland and a little later appeared over D80 (Figures 9a, 9b10d), and finally, on 5 July, clouds appeared over Dome C (Figure 10f). Figures 11a and 11b present the composite infrared satellite images for 2 and 4 July 1994 on which clouds can be seen advected by the depression. The clouds predicted by MAR in Figures 10b and 10d are similar to those in satellite images. Because of the vertical mixing provided by strong winds, between the cold boundary layer air and the warm free atmospheric air, warming took place near the surface. Warming is, in part, also due to warm air advection and clouds. The presence of clouds increased the long wave radiation back to the surface, resulting in an increase in surface temperature. With the combined effect the surface temperature rose noticeably, inversions were destroyed, and finally, winds ceased on 5 July 1994. Cessation of winds is also due to the building up of pressure over the Adelie Land coast, after the passage of the depression to the east of it.
3.3.2. Last week of SOP1
 Around 26 July the wind speed was high over the whole depth of the troposphere (Figures 7a, 7b, 7c 7d). Figure 12 shows the composite of MSLP fields and the cloud optical depth simulated by MAR for 26 and 27 July 1994, respectively. During this episode the unusual feature, which added to the influence of the depression and the high-pressure system over the interior, is the presence of an anticyclone to the northeast of Adelie Land (Figures 12a and 12c). This high-pressure system remained stationary for some time and provided a blocking to the usual eastward motion of the depression to the west of DdU. Because of this blocking anticyclone, the movement of the cyclone was reduced; it intensified and started to penetrate inland. Pressure charts show a high-pressure ridge that builds up significantly diagonally over the center of the domain, extending from the NE of the domain, from the high pressure over the ocean, onto the SE into the interior of the continent. Looking into the pressure chart for 27 July 1994, the wind is moving from the high over the ocean located toward the northeast of DdU to the low located over the northwest of it and is continuing into the continent. The penetration of warm marine air into the interior along the ridge caused thick clouds to form during 26–28 July 1994. The optical thickness plot for 26 and 27 July 1994 (Figures 12b and 12d) shows two maxima around 120°E in the interior of the continent (latitude and longitude over the domain are presented in Figure 1). These maxima correspond approximately to two small systems that penetrated into the interior of the continent around that time (see DMSP images of cyclonic vortices over around Casey Station at approximately the same time [Pook and Cowled, 1999, Figures 7a, 7b]. Clouds simulated by MAR (Figures 12b and 12d) match very well with DMSP images shown by Pook and Cowled  and composite infrared images created from NOAA AVHRR passes (Figures 11c and 11d). The clouds represented by the satellite images and MAR-simulated cloud optical depth plots are along the high- pressure ridge. We see that the marine intrusion and cloud formation for this episode have been due to an entirely different synoptic situation than for the previous episode. It is important to notice that during this episode, in spite of the presence of a blocking high, off the coast toward the east, winds were high and were influenced by the presence of the depression on to the west of Adelie Land.
3.3.3. Episodes of SOP3
Figure 13 shows mslp charts and cloud optical depth simulated by MAR for 3, 10, and 30 January 1995. The high-humidity episode visible on the time-height plots (Figure 8) around 3 January 1995 is able to generate clouds only over DdU and not over the interior (Figures 9c, 9d). The marine air could not penetrate far into the interior to have widespread clouds (Figure 13b). The episode around 10 January 1995 could only have clouds only along the coast and not into the interior (Figure 13d). However, if we study the pressure chart for 10 January (Figure 13c), the synoptic situation is similar to the one for the second episode of July 1994 (around 26 July 1994, Figure 12a). On 10 January the high in the interior is relatively weak, and the center of the high pressure over the ocean to the east of Adelie Land is not inside the domain. However, in this situation, also wind speeds are high for the whole depth of the troposphere (Figures 7e, 7f, 7g, 7h), and the marine air intrusion was maximum as compared to all episodes of July 1994 and January 1995 (Figures 8e, 8f).
 The last episode of January 1995 (30 and 31 January) resulted in thick widespread clouds. During this episode the winds are high for the whole depth of the troposphere (Figures 7e, 7f, 7g, 7h), particularly in the upper levels. The warm air penetration is significant over a relatively greater depth for all three grid points (Figures 8e, 8f, 8g, 8h). Clouds are thin over DdU and significant over the interior and over the ocean (Figure 13f). The synoptic situation, once again, shows three systems (Figure 13h), one low to the west and one high to the east of the Adelie Land coast and one high in the interior of the continent. The penetration of marine air and the clouds took place because of the presence of the high-pressure ridge similar to the one that occurred during the 26 and 27 July 1994 episode (Figure 12). Winds remained high in spite of the presence of the blocking high over the ocean to the east of DdU, forcing the low-pressure center to the west to slow down and intensify. The clouds simulated by MAR are similar to the ones shown by the composite infrared satellite images (Figure 14) for all the three cases in January 1995.
4. Discussion and Conclusions
 In this study we presented the mechanisms behind marine air intrusions into the interior using observations and model simulations. The regional model MAR is initialized from ECMWF fields. Simulated parameters are in good agreement with ECMWF fields in the whole domain except for a small coastal stretch around the Ross Ice Shelf. The comparison with observations for Enigma Lake and Cape Philips automatic weather station (AWS) sites in this region indicates that the MAR simulations are relatively closer to observations than the ECMWF analyses along this stretch.
 Our results indicated clouds being greater and far deeper into the interior in July 1994 than in January 1995. Widespread clouds were observed due to the passage of a series of depressions. Satellite images and the cloud optical depth simulated by MAR indicate the advection of clouds due to intense depressions, and also along the high-pressure ridges. Although we used only a selected short data set representing winter and summer seasons in Antarctica, we found clouds to be relatively thicker and widespread into the interior in July 1994 than in January 1995 (Figures 8, 9). This is envisaged as due to abnormally increased cyclonic activity in July 1994 in spite of this being a winter month. Intense and well-organized depressions helped in pulling the moist air from the ocean to over the sea ice and into the interior of the continent. In normal days the warm air traveling over the ice will cool before arriving near the continent and cannot be expected to be warm to rise against steep ice slopes. Also, the fact that the blocking high pressure holds the depression in one place for a longer period, therefore the air above the sea ice is expected to be warm, so the presence of sea ice becomes less important. Relatively lower air temperatures in July 1994 episodes caused vigorous cooling of the moist marine air and resulted in thick and widespread clouds inland in July 1994 than in January 1995, although the moisture flux was similar during both periods (Figure 8). Widespread clouds far inland were, in particular, observed in July 1994 each time the intrusion took place. Because of lower atmospheric temperatures each passing cyclone could cause clouds in July 1994 (Figure 9). In January 1995 the intensity of depressions were similar to the depressions in July 1994, but the clouds were not observed far inland except for the last episode of January 1995. This is due to the fact that the central pressure of the semipermanant anticyclone over the continent has been higher in July 1994 than in January 1995, except in the last week when the clouds were widespread into the interior. High-pressure ridges helped in the formation of widespread clouds, particularly in the last week of July 1994 and January 1995.
 The presence of depressions well into the vicinity of the Antarctic coast in winter months and their penetration inland have also been reported by Alvarez and Lieske , Neal , Phillpot , and King and Turner . Neal , using the data for the period November 1969 and June 1970, carried out a similar study. He was looking for weather systems in the interior of the continent. He found no developments at all over the continent in November 1969, whereas several cyclogenesis events were found near the South Pole and over Coats Land in June 1970. Phillpot  studied 3 years of observations from the Vostok Station. He found evidence of 42 depressions, having been in the vicinity of the station, and detected these systems by precipitation falling and relatively high surface temperatures. Of the depressions, 75% were found in the winter months of June to October, suggesting significant cyclonic activity over the continent at that time of the year. King and Turner  reported a rare movement of the depression (a major low), during the period 24–26 May 1988, which tracked southward from the Weddell Sea, over Marie Byrd Land and onto the Ross Ice Shelf. Alvarez and Lieske  found one event of inland penetration in May 1957 which brought precipitation to the South Pole and caused the surface temperature to rise by 40° over 4 days as the oceanic air reached the interior. Astapenko  examined the occurrence of depressions over the interior and found that a significant number tracked from the Ross Sea to the Weddell Sea, affecting stations such as Amundsen-Scott. He observed these events in all seasons. Stone , using the radiometersonde and meteorological measurements, found that the winter season clouds at South Pole associated with large-scale synoptic disturbances that reached the South Pole.
Figure 15 shows MAR-simulated cumulative snow for July 1994 and January 1995. In July 1994 the snow accumulation has been larger than in January 1995. However, in both cases the area of maximum snow accumulation has been along the coastal periphery around 120°E longitude. Precipitation is observed to be higher in the region of greater cyclonic activity or where depression becomes slow moving. As already discussed, the large specific humidity and clouds in the last weeks of July 1994 and January 1995 are due to the presence of a high or blocking anticyclone offshore in the eastern side which provided a blocking to the cyclone on the western side of the Adelie Land coast. Because of this blocking anticyclone, the depression became slow moving, intensified, and started to penetrate inland. Pook and Cowled  demonstrated that vortices originating over the Southern Ocean can penetrate the high plateau of East Antarctica and move well inland before decaying. They reported that the development of an intense blocking anticyclone in the Tasman Sea sector appeared to have been critical for the penetration of the depression inland. According to Lejenas  the Pacific Ocean to the east of Australia, between Australia and New Zealand, is the preferred region for blocking anticyclones. His study was based on 500 hPa charts and revealed two annual maxima, one 19% in April and another 20% during June–August. He also reported that the majority of blocks have a mean eastward velocity, although the longer they last the more they move westward. When a blocking high breaks down or moves away, there is a marked tendency for another high to form and intensify at the same location [Taljaard, 1972]. Trenberth and Mo  also reported, using 1000 and 500 hPa charts, that the primary location for blocking anticyclones is the New Zealand sector and that the blocking high is accompanied by a cutoff low.
 Increased near-surface temperatures in July 1994 during various episodes can be summed up as due to warm marine air intrusion, increased long wave radiation due to the presence of clouds, and the vertical mixing provided by strong winds. These features owe their presence to intense offshore cyclonic activity. This heating is similar to what is reported by Carroll  and Stone and Kahl  for winter warming. Alvarez and Lieske  described a warm advection situation during which the surface temperature at the South Pole rose by nearly 40° in four days during May 1957. Sinclair  reported the abnormally high temperatures at the South Pole, McMurdo, and Vostok associated with the warm air intrusion episode. Schwerdtfeger  reported the surface warming to be due to the long wave heating from clouds formed during the upper level intrusion of moist air. Such intrusions are frequently laden with cloud condensation nuclei reaching the interior, producing a variety of cloud types [Ohtake, 1978]. Bodhaine et al.  observed the enhancement of sea-salt aerosols at the South Pole associated with rapid transport of moist air from coastal regions, during surface warming events. According to Hogan et al. , heat, water vapor, and aerosols are transported poleward at preferred levels in the troposphere when the position of the polar high-pressure ridge shifts from a northeasterly to a more northerly orientation. All of these studies, including the present one, indicate a surface warming, which is due to large-scale circulation patterns.
 The cessation of strong winds during various episodes is, in part, due to the exhaustion of cold air mass at the point of origin but mostly due to the building up of pressure along the Adelie Land coast, after the passage of the depression. Also, the warming produced by the combined effect of clouds, high winds, and warm air advections, which subsequently destroyed inversions, contributed to abrupt wind cessation.
 To conclude, this study confirms conclusions of previous investigations, that cyclones are responsible for much of the moisture transport and cloud formation in the Antarctic continent. The amount of precipitation being maximum in July 1994 is as expected and observed [King and Turner, 1997; Kobayashi, 1980]. The influence of warm air intrusions in increasing the surface temperature was more pronounced in July 1994, while in January 1995, similar intrusions could not effectively increase the surface temperature. In this study we also aimed to test the capability of MAR to simulate marine air intrusions and to test its viability as a nested version to study the Antarctic climate. Results have been very optimistic. The nested version of the model is able to simulate correctly synoptic and climatological features of the Antarctic continent and the surrounding ocean, namely, the circumpolar trough, development of the ridge, surface layer, upper tropospheric features, and clouds. Comparisons with DMSP images [Pook and Cowled, 1999, Figures 7a, 7b], and composite infrared images created from NOAA AVHRR passes have been very good. MAR simulations are in fairly good agreement with observations and ECMWF analyses except for a small stretch along the coast of Victoria Land, where MAR simulations are closer to observations than the ECMWF analyses.
 This work is part of the European Commission Project for Ice Coring in Antarctica (EPICA). One of the authors (JN) is thankful to the Belgian program of the Scientific Research on the Antarctic (Services of the Prime Minister-Federal Office for Scientific, Technical and Cultural Affairs), for giving the opportunity to work in this project. This author is also thankful to G. Furlan, Director TRIL, ICTP, Trieste, Italy, and IFA, Roma, Italy, for granting the leave to join the present contract at UCL. The analyses to force the model boundaries were obtained from ECMWF archives and the meteorological data have been downloaded from the Web sites of the Wisconsin University and British Antarctic Survey. Composite infrared satellite images were provided by Matthew Lazzara of the University of Wisconsin-Madison. Data for DdU site have been the courtesy of Paul Pettre. AWS data for Enigma Lake and Cape Philips were obtained from the Italian National Antarctic Research Programme PNRA, Meteoclimatological Observatory. The help provided by Olivier Brasseur, Alexis Dutrieux, Filip Lefebre, and Phillipe Marbaix during the course of the work is thankfully acknowledged. Thanks are also due to Raoul De Guchteneere, the computer administrator. We are thankful to the helpful comments and suggestions provided by three unknown reviewers.