Over the past 50 years, the maritime west coast of the Antarctic Peninsula has been one of the most rapidly warming parts of the planet, with the largest warming occurring in the winter season (Vaughan et al., 2001; King et al., 2004; Turner et al., 2005a, 2009; Steig et al., 2009). These different studies find warming rates that are 1 order of magnitude greater than the mean rate of global warming as reported by the Intergovernmental Panel on Climate Change (IPCC, 2007)). Recently performed comprehensive analyses of synoptic observations of cloud parameters recorded at the Antarctic base Faraday/Vernadsky for the period 1960–2005 have shown that one effect of the warming is a significant increase in the annual mean of the total cloud cover. The strongest and most significant positive seasonal trend was found in winter, and positive tendencies are observable in all seasons (Kirchgäßner, 2010). A possible cause for the increase in cloud cover could be the increasing capacity of air to hold water vapour with rising temperatures. Reduced sea ice cover baring open ocean may contribute by providing a source for the necessary moisture. Another phenomenon observed in connection with the warming around the Antarctic Peninsula is an increased cyclonic activity in the region of the Amundsen and Bellingshausen Sea which could also be a source of increased cloud cover over Faraday/Vernadsky. The overall causality between the winter warming and the sea ice reduction on the western side of the Antarctic Peninsula are not well understood yet.
The aim of this paper is to investigate whether this increase in total cloud cover has had a direct impact on the characteristics of the precipitation at Faraday/Vernadsky. An increase in the number of precipitation events and days in this region has been reported (Turner et al., 1997, 2005b), but there are currently no studies available that investigate the phase of precipitation. The phase of the precipitation directly determines the albedo and is thus an important factor in the temperature-albedo feedback mechanisms. Rain is also thought to accelerate the retreat of glaciers by physically corroding the ice. Non-frozen precipitation falling on frozen ground cannot be stored and thus will not be available for the land-based ecosystem. Instead it will run off, thereby increasing the fresh water influx into the ocean. Last but not least is precipitation the main positive term in the mass balance of the Antarctic ice sheet. Thus, changes in precipitation can directly affect global sea level.
The results of this study add important information to our overall understanding about how the observed warming has already affected other factors defining atmospheric conditions in this region. They will thereby improve insight into how climate conditions in the Antarctic Peninsula will change in the future.
The first permanent research base on the Argentine Islands (65°15′S and 64°16′W) was built in 1947 on the site that had previously been occupied by a research base during the British Graham Land Expedition. On 6 February, 1996, the station—then known as ‘Faraday’–was officially handed over to the Ukraine and is since then known under its present name ‘Vernadsky’. Figure 1 shows the location of the base in Antarctica, and in more detail its position on Galindez Island.
Information on precipitation are part of the routine synoptic observations as records of ‘Present weather’ or ‘Past weather’. While ‘Present weather’ refers to any weather phenomenon at the time of the observation, ‘Past weather’ includes any weather phenomena that has occurred since the last observation. These observations, though, only allow statements on the occurrence of precipitation events, and to some extent on the character of the precipitation, but not on the precipitated amount. Hence, this study only looks at the frequency of precipitation events and at the frequency of days on which precipitation events were recorded, from hereon referred to as precipitation days.
Observations are recorded according to the World Meteorological Organization's Manual on Codes (WMO, 1995). In 1982, WMO introduced several changes in the coding procedure (WMO, 1988). One of these changes instructs observers to set ww (‘Present weather’) to solidus (‘/’) if present weather is either ‘not available’ or ‘observed phenomena were not of significance’. This has obviously reduced the overall number of recorded ‘Present weather’ phenomena. But as it can rightfully be assumed that actual precipitation events are ‘of significance’ this change will therefore not influence the absolute number of precipitation events, which is analysed in this present study. It would only influence the data analysis were numbers of precipitation events in relation to number of ‘present weather’ recordings studied.
Synoptic observations began at Faraday/Vernadsky in 1947. For consistency with the analysis of cloud observations by Kirchgäßner (2010), it was intended to use precipitation records from the same period, i.e. years from 1960 to 2005. Investigations into the data quality though showed that probably due to a change in observational routine on base (no change was introduced by WMO), precipitation records after 1999 should not be included in the study.
Data analyses were carried out based on 3-hourly observations. Analyses carried out for the section on precipitation events only take records of ‘Present weather’ into account, while results for precipitation days also include observations of ‘Past weather’ in accordance with the definition of precipitation days used by Turner et al. (2005b).
In 1987, an Automatic Weather Station was installed and put into operation. Due to this change, a different coding scheme was used to record ‘Past weather’ since then, which had to be taken into account when analysing this parameter for precipitation days. This change from manual to automatic recording of meteorological parameters also means that nighttime meteorological observations no longer required a trained meteorologist to be on duty. Thus, since then nocturnal weather observations are carried out by the night watch rather than by a fully trained meteorologist. This could have led to an obvious change in the characteristics of the data set after 1987; such a change was not found though. The claim that ‘Past weather’ observations had become less reliable could not be confirmed.
All trends presented in this study are calculated using a standard least-squares method. The methodology used to calculate the significance levels is based upon Santer et al. (2000).
The data set covers 14 610 days, so—looking at 3-hourly data—this means records potentially comprise 116 880 recordings. Until 1982 no observation was missed, afterwards on average 1224 observations are coded as ‘null’ per year. This is certainly a consequence of the change in recording practice regarding non-significant weather mentioned above, and not a sign of ‘negligence’ by the observers. Seasons are defined as follows: winter [June, July, August (JJA)], spring [September, October, November (SON)], summer [December, January, February (DJF)] and autumn [(March, April, May (MAM)].
3. Precipitation events
The following section looks firstly at the overall number of events when some form of precipitation was recorded, and then at the records of non-frozen precipitation in particular. Both types of information are derived from the observations under the label ‘Present weather’. Observations of ‘Past weather’ are not included in this section to keep results comparable with those presented by Turner et al. (1997).
Annual and seasonal mean numbers of precipitation events and the according standard deviations are listed in Table I. On average, 923 precipitation events were recorded each year. This means that some form of precipitation is recorded roughly with every third observation. The annual cycle calculated from the average monthly number of precipitation events shows maxima of precipitation events in March and October (Figure 2), which agrees with results by Turner et al. (1997). The obvious difference between the two cycles is due to the fact that Turner et al. (1997) based their study on 6-hourly records and this present study uses 3-hourly observations.
Table I. Annual and seasonal mean and standard deviation of all precipitation events, non-frozen precipitation events and precipitation days
All precipitation events
Non-frozen precipitation events
923 ± 86
139 ± 46
260 ± 14
198 ± 41
75 ± 31
59 ± 9
236 ± 36
44 ± 19
67 ± 7
230 ± 43
6 ± 5
65 ± 8
259 ± 31
18 ± 10
69 ± 5
The interannual variability in the annual number of precipitation events is relatively low, with a standard deviation of 86, i.e. less than 10%. In winter (JJA), an average number of 230 ± 43.1 (19%) events is recorded; in summer (DJF), on average 198 ± 41.2 (21%) events are observed and in autumn (MAM) 236 ± 36.1 (15%) events. In spring (SON), the interannual variability is lowest with an average number of 259 events and a standard deviation of ± 31.1 (12%).
Trend analysis of the data set shows that only the number of precipitation events recorded in wintertime has increased significantly by 13 events per decade (significant at 5% level). Over the entire period, this has led to an increase by 52 events. On the basis of 3-hourly observations, this corresponds to the equivalent of 6 1/2 days more precipitation. By contrast, annual records as well as records of the other seasons do not show significant changes in the frequency of precipitation events (Table II). These findings agree with results published by Turner et al. (1997) who found a significant increase in winter season precipitation events for the period 1956–1993 of 10 events per decade and no significant trends for the other seasons or the annual number of precipitation events.
Table II. Annual and seasonal trends of all and of non-frozen precipitation events at Faraday/Vernadsky between 1960 and 1999
Over the period under investigation, a total of 5563 events of non-frozen precipitation were observed, namely, 26 cases of rain showers, 1013 cases of drizzle and 4524 cases of rain. Interestingly, these cases of non-frozen precipitation occur throughout the year and occur at air temperatures down to − 5 °C. This can, for example, occur during inversion when near surface temperatures are lower than those higher up. The trends in the frequency of non-frozen precipitation events are given in Table II.
The number of non-frozen precipitation events shows an annual trend of + 24 events per decade. Looking at individual seasons, this positive trend can be found throughout the year and is significant in all seasons except winter (Figure 3). This is consistent with the increase in air temperature reported by several authors (Vaughan et al., 2001; King et al., 2004; Turner et al., 2005a). This warming—though it is found all year round and is strongest in winter—will especially make a significant difference on the phase of the precipitation in those seasons when temperatures are around the threshold between frozen and unfrozen precipitation, i.e. in spring and in autumn. This relationship is supported by significant correlations between the number of precipitation events and mean air temperature for summer, autumn and winter [0.34 (5%), 0.35 (5%), 0.45 (1%)].
The increase in the number of non-frozen precipitation events becomes even clearer when the proportion of non-frozen precipitation events is analysed in relation to annual or seasonal trends in the overall number of precipitation events.
Table II shows that the percentage of non-frozen precipitation events has increased all year round, with the increase being significant below the 5% level annually and in all seasons except winter. This makes sense, as the proportion of non-frozen precipitation during winter months is marginal.
4. Precipitation days
Another way to analyse the frequency of precipitation events is to merge them into precipitation days as has been done in a study by Turner et al. (2005b). For this section, their definition of precipitation days was adopted and the data used in their study were reprocessed. This definition of precipitation days takes observations recorded under “Present Weather” and “Past Weather” into account. One has to keep in mind that precipitation recorded under past weather at 00UTC is strictly speaking precipitation of the previous day, and therefore—if it is the only precipitation recorded on that date—does not make this date a ‘precipitation day’. In contrast to Turner et al. (2005b), this effect has been taken into account for the present study and precipitation recorded under ‘Past weather’ at 00UTC has been attributed to the previous day.
On average, precipitation was recorded on 260 ± 14 days a year, of which 69 ± 5 were recorded in spring, 59 ± 9 days in summer, 67 ± 7 days in autumn, and on average 65 ± 8 precipitation days were recorded for winter. Mean monthly numbers of precipitation days reveal a similar annual course as the one found for precipitation events, with maxima in March and October. It is rather weak though, so when months are combined to seasons, precipitation days seem to be spread regularly throughout the year, apart from a slight minimum in precipitation days in summer.
Results in Table III show a significant trend of 5.7 days per decade in the annual number of precipitation days for the reprocessed data. Significant increase rates of 1.8 and 2.4 days per decade are found for autumn and winter, respectively (Figure 4). Trends in summer and spring are not significant. Trends are considerably smaller than those found by Turner et al. (2005b) for the period from 1951 to 1999 (about 50%). Using the same procedure as Turner et al. (2005b) but reducing the trend analysis to years from 1960 to 1999 reveals that the differences are mainly due to the selection of the period under investigation (Table III). Irregularities in the observations during the 1950s, which were found in the course of the reprocessing of the data set, lead to incorrectly low numbers of precipitation days during these years and thus artificially increase the trend. For much of these early years, past weather was not recorded at all and present weather observations were only recorded at 6-hourly intervals rather than 3-hourly as used by Turner et al. (2005b) for their analysis. Therefore, only results based on the same period are compared here. Overall the results are quite similar. The annual trends of 5.1 and 5.7 precipitation days per decade are significant at the 10% level and 5% level, respectively. In both analyses, the most significant seasonal trend is found in autumn and these are very close in magnitude. Trends in summer and winter are of the same respective magnitude, in both cases slightly larger than trends for autumn. In the present study, the winter trend is significant at the 10% level but not in the analysis after Turner et al. (2005b). Spring shows no tendency at all in the analysis after Turner et al. (2005b), and a negligible negative tendency in the present study.
Table III. Annual and seasonal trends of precipitation days at Faraday/Vernadsky for different periods of investigation in days per decade
Some basic observations of clouds (cloud cover and cloud type) are part of the routine synoptic observations that also include the records on ‘Present weather’ and ‘Past weather’ which the previous sections are based on. For a more detailed description of the cloud observations used in this section, see WMO (1995) and (Kirchgäßner (2010)). Analyses of these cloud observations found a significant increase in the mean annual total cloud cover over the period from 1960 to 2005 with highest increase rate during the winter season (Kirchgäßner, 2010). This matches the results of this study on precipitation events and even more the results on precipitation days. As presented in the Sections 3 and 4, both these parameters show a significant increase in winter; precipitation days have also increased in frequency on an annual basis and in autumn.
The detrended time series of precipitation days and precipitation events are both significantly correlated with the total cloud cover during all seasons. Correlation coefficients are compiled in Table IV. As an example, the detrended time series for total cloud cover and precipitation days in winter are shown in Figure 5. For the degrees of freedom describing the period under investigation, correlations above 0.31 are significant below the 5% level.
Table IV. Correlation between total cloud cover and precipitation events/days as observed at Faraday/Vernadsky from 1960 to 1999 (detrended time series)
The correlation is highest in summer (0.56 for precipitation events, 0.73 for precipitation days) and autumn (0.50 for precipitation events, 0.63 for precipitation days). The increase in total cloud cover is therefore causing the rising number of precipitation events. The causality between total cloud cover and precipitation days is less clear as the increase in precipitation days in autumn is not matched by an increase in total cloud cover.
Table V combines the annual number of precipitation events with concurrent observations of clouds at different levels. These different classes were further divided according to the seasons and analysed for trends (not shown).
Table V. Number of precipitation events (all and non-frozen only) sorted according to concurrent cloud observations
Concurrent cloud observations
All precipitation events (Σ 36 919)
Non-frozen precipitation events (Σ 5 563)
Low clouds only
17 449 (47%)
2 517 (45%)
Medium level clouds only
9 487 (26%)
1 644 (30%)
Low and medium level clouds
4 095 (11%)
No observations available
5 331 (14%)
High clouds? (residual)
The annual number of non-frozen precipitation events that were recorded in connection with medium clouds is highly correlated with the overall number of non-frozen precipitation events (r = 0.71) and hence shows a similar annual and seasonal increase over the period under investigation. This correlation is much stronger than that between the overall number of non-frozen precipitation events and those recorded in connection with only low clouds present (0.46). This is most likely due to the fact that the cloud type most often recorded in connection with precipitation events in general as well as with non-frozen precipitation events (‘Altostratus opacus or Nimbostratus’) is always to be recorded as medium level cloud, even though its cloud base is often in the range of low clouds.
The low cloud types most often recorded during precipitation events are ‘Stratocumulus other than Stratocumulus cumulogenitus’ and ‘Stratus fractus or Cumulus fractus […] usually below Altostratus or Nimbostratus’. ‘Stratocumulus other than Stratocumulus cumulogenitus’ is observed together with 23% of precipitation events, whether frozen or non-frozen. ‘Stratus fractus or Cumulus fractus below Altostratus or Nimbostratus’ is recorded with 16% of all precipitation events and with 18% of the non-frozen precipitation events. The medium level cloud type ‘Altostratus opacus or Nimbostratus’ is consequently the one most often recorded together with precipitation events in general (32%) as well as with non-frozen ones (42%).
This is especially interesting in combination with the analyses of cloud observations by Kirchgäßner (2010) that showed an increase in the total cloud amount at Faraday/Vernadsky over the past 50 years. This is neither reflected in the records of low cloud amount nor the frequency with which low height, medium height or high cloud type were recorded (records for medium height cloud amount and high cloud amount are not stipulated). More specifically, it was even found that high cloud type was recorded less often, which either could be due to an actual decrease in the occurrence of high clouds or to an increase in the cover of medium clouds, which would conceal any high clouds. The present results indicate that this apparent increase in medium height cloud amount is constituted by an increase in the medium height cloud type ‘Altostratus opacus or Nimbostratus’. A significant trend towards more frequent record of this particular cloud type is not distinguishable in the data set at hand.
6. Large-scale circulation patterns
6.1. Southern annular mode
The southern annular mode (SAM) or Antarctic Oscillation is the principal mode of variability of the extra-tropical atmospheric circulation, and typically describes ∼30% of total Southern Hemisphere climate variability. Essentially, it is an annular structure with synchronous pressure anomalies of opposite sign in mid- and high latitudes: when pressures are below (above) average over Antarctica the SAM is said to be in its high (low) index or positive (negative) phase. In its positive phase, SAM strengthens the circumpolar westerlies. Over the past few decades, SAM has shown significant positive trends during autumn and summer, and a non-significant tendency towards a more positive polarity in winter (Thompson et al., 2000; Marshall, 2003).
Lubin et al. (2008) have found that the frequency of mesoscale cyclones in the western Antarctic Peninsula (WAP) region during 1991–1994 is correlated with SAM, most strongly during winter and spring. Faraday/Vernadsky lies within the region of these cyclone tracks. Indeed winter is the only season where there is a significant correlation between the detrended time series of SAM index and number of precipitation events. The underlying mechanism is most likely very straight forward: during winters with a high SAM index the circumpolar westerlies strengthen, and more mesoscale cyclones enter the WAP region. These can potentially advect moist air from ice free regions, which could lead to an increase in cloud cover and more precipitation events at Faraday/Vernadsky.
But strongest trends in SAM are observed in autumn and summer while precipitation events have only significantly increased in frequency in winter. Should SAM be the cause behind the increase in precipitation events this would require a lag of one to two seasons in the reaction. Such a lagged correlation between SAM and precipitation events was not found.
6.2. El Niño-Southern Oscillation
Largely facing the Pacific Ocean, west Antarctic seas and coasts are likely the Antarctic regions most directly exposed to El Niño-Southern Oscillation (ENSO). Tropospheric wave trains (Trenberth and Carron, 2000; Kidson and Renwick, 2002) have been identified, which connect the tropical Pacific to the higher latitudes at the ENSO pace. Several studies have investigated the influence of ENSO on various precipitation related parameters in the regions of West Antarctica, the Antarctic Peninsula and the surrounding seas (Cullather et al., 1996; Genthon et al., 2003, 2005; Genthon and Cosme, 2003; Turner, 2004).
Model runs analysed by Cullather et al. (1996) show that the variability in precipitation in the South Pacific Sector is correlated to ENSO between 1980 and 1990. Their finding that this relationship turns into an anticorrelation for years after 1990 is an indication that this connection is not temporally stable.
This is confirmed by ERA-40 (European Centre for Medium Range Weather Forecast Re-analysis 1957–2001) data analysed by Genthon and Cosme (2003), who identify a robust main mode of the precipitation variability in West Antarctica, which contains an intermittently strong ENSO signature. However, this high correlation with ENSO appears infrequent, and, unlike suggested by e.g. Cullather et al. (1996), the correlation, when significant, does not appear to change sign in time.
The ITASE (International Trans-Antarctic Scientific Expedition) ice core data analysed by Genthon et al. (2005) partially confirm the spatial structure of the ENSO signature in the West Antarctica precipitation identified from ERA-40 and other model output. Studying the surface mass balance, they find, though, that small-scale perturbations can almost fully obscure any large-scale component of variability within this parameter.
During periods of positive SAM index polarity, there is a shift in the storm tracks to favour more east-bound trajectories, consistent with stronger circumpolar westerlies (Lubin et al., 2008). These strengthened westerlies may act in two ways. They may guide more mesoscale cyclones towards the northern Antarctic Peninsula and advect more relatively warm moist maritime air from sea ice free areas to this region, thereby increasing the precipitation. While this is a plausible mechanism, the seasonal pattern of trends in SAM and precipitation do not match.
Studies by Orr et al. (2008) and Van Lipzig et al. (2008) have shown that stronger summer westerly winds connected to SAM reduce the blocking effect of the Antarctic Peninsula. As Marshall et al. (2006) have found, this has led to a higher frequency of air masses being advected eastward over the orographic barrier of the northern Antarctic Peninsula. The connected warming on the eastern side of the Antarctic Peninsula due to Föhn events has contributed significantly to the collapse of the northern sections of the Larsen Ice Shelf between 1995 and 2002. This reduced blocking effect could be the reason for the relatively weaker increase of precipitation events in summer as potential precipitation is transported closer to or even across the Antarctic Peninsula. It could therefore be that an all year round increase in precipitation (due to a so far unknown cause) is actually counteracted by the positive trend in the summer SAM and the reduced blocking.
Genthon et al. (2003) have used Empirical Orthogonal Functions to identify the main modes of Antarctic tropospheric circulation (500 hPa geopotential height and precipitation) in meteorological analyses and climate model results. The first one is related to SAM, the second one to ENSO. While both modes have a strong pole of variability in West Antarctica, the relative strength of the two signals appears to change in time.
For the present study, various parameters from the monthly ERA-40 data sets were spatially correlated with the annual and seasonal numbers of precipitation events at Faraday/Vernadsky. Figure 6(a) and (b) shows spatial correlations between the 500 hPa geopotential height and precipitation events in summer and winter, respectively. Only years from 1979 to 1999 have been used, as reanalysis data have been found not to be reliable in Antarctica during non-summer months prior to the modern satellite era (Bromwich et al., 2007). Areas of significant correlation (≥95%) are shown by white contours.
Although the analysis for summer months shows a clear SAM-related signature, with an annular structure of high positive correlation coefficients around the Antarctic continent and high negative correlation coefficients over the continent itself, this pattern is not at all visible in the correlation of annual data (not shown) or for the winter season (Figure 6(b)). In winter, the spatial correlation shows more similarity with an ENSO-related pattern, i.e. highly negative correlation over West Antarctica, and areas of highly positive correlation on either side, towards the South Atlantic and the south Pacific. It is conceivable that the trend towards a more positive phase observed for SAM in summer becomes apparent in the seasonal difference between the spatial correlations.
This study used 3-hourly observations of ‘Present weather’ and ‘Past weather’ from 1960 to 1999 to investigate the number of precipitation events and precipitation days. Additionally, precipitation events were classified into frozen or non-frozen precipitation.
The number of precipitation events recorded at Faraday/Vernadsky during winter seasons has increased significantly by 13 events per decade. Although summer and autumn records also show positive tendencies, the annual number of precipitation events only shows an increase that is not significant. This agrees well with results presented by Turner et al. (1997) who found a significant increase in wintertime precipitation events at this station between 1956 and 1993 of 10 events per decade.
Analysing the type of precipitation event, it is found that the proportion of non-frozen precipitation events has increased significantly on an annual basis, but also in all seasons except for winter.
In regard to precipitation days, this study found a significant increase in the annual number of precipitation days for the period from 1960 to 1999 (5.7 days per decade, significant at the 5% level), with the highest seasonal increase rates in autumn (1.8 days per decade) and in winter (2.4 days per decade). Although it shows a lower increase rate, the autumn trend is significant below the 5% level compared with the winter trend, which is significant below the 10% level. This is generally consistent with results published by Turner et al. (2005b) who found a significant increase in the annual number of precipitation days of 12.4 days per decade and the highest seasonal increase rates in autumn and winter. The discrepancy between the two studies is caused by missing observations in the early 1950s which lead to incorrectly low numbers of precipitation days during these years and thus artificially increase the trend in Turner et al. (2005b).
Regarding the synthesis with cloud observations, a simple correlation analysis has proven the general connection between total cloud cover and precipitation events and days, especially on the seasonal scale, and for precipitation days also on an annual basis. The change observed in total cloud cover, one of the main incentives for this study, is well reflected in the increase in precipitation days presented here, as both parameters show a significant increase annually and in winter.
Both the main large-scale circulation patterns that are important in the Southern Hemisphere influence the frequency of precipitation events at Faraday/Vernadsky. The spatial correlation of the number of precipitation events at Faraday/Vernadsky during summer months with mean sea level pressure shows a clear pattern associated with SAM. This is not the case for winter months when the large-scale influence appears to be more related to ENSO.
This study has confirmed the hypothesis that the observed increase in total cloud cover at Faraday/Vernadsky has led to a general increase of the number of precipitation events and precipitation days. In the Coupled Model Intercomparison Project, all models but one predict a concurrent precipitation increase (Genthon et al., 2009). About three-quarters of this rise originate from the marginal regions (less than 250 km from the coast) of the Antarctic ice sheet with surface elevation below 2250 m above sea level. These characteristics apply to Faraday/Vernadsky and therefore the results agree well with the observed increase in precipitation events there.
Especially the increasing proportion of non-frozen precipitation can be clearly attributed to the observed temperature increase in the area of the Antarctic Peninsula. Several studies (Vaughan et al., 2001; King et al., 2004; Turner et al., 2005a) have shown a warming of nearly 3 °C for this area in the past 50 years. Although the warming is found all year round and is strongest in winter, it will make the biggest difference on the phase of the precipitation in those months when the air temperature is around freezing point. At Faraday/Vernadsky, this is certainly not only the case for the summer season but also for late spring and early autumn. This is reflected in the seasonal differences of the trends and their significance.
The increase in total cloud cover reported by Kirchgäßner (2010) obviously has a direct effect on the number of precipitation events at Faraday/Vernadsky, which is reflected in the concurrent seasonality of the trends. One conclusion drawn in the paper is that the observed increase in total cloud amount is caused by an increased amount of medium height clouds. Of the cloud types recorded, the medium height cloud type ‘Altostratus opacus or Nimbostratus’ is the one that is reported in connection with more precipitation events (32% of all precipitation events, 42% of non-frozen precipitation events) than any other low or medium height cloud type. This not only indirectly verifies the increase in medium height cloud amount but also affirms the connection between cloud cover and precipitation events.
Direct measurements of precipitation in the Antarctic are scarce, and their quality suffers not only from the general difficulties inherent in precipitation measurements but also from additional complicating factors such as snow drift, melting and sublimation. Snow accumulation data derived from ice cores are the main source of precipitation amount information on climatological time scales. Analyses of snow accumulation data from various sites show an increase in the WAP since the beginning of the 20th century (Thomas et al., 2008), which indicates that the precipitation amount has increased, i.e. the increase in precipitation events has also led to higher annual amounts of precipitation.
The influence of the large-scale circulation on the precipitation in the region of West Antarctica has been reported by several studies (Cullather et al., 1996; Genthon et al., 2003, 2005; Genthon and Cosme, 2003). Both SAM and ENSO signatures are found in the variability of the circulation as well as in the precipitation of the region. Their influence varies with time, though, so that their respective signal, where it becomes apparent, should be treated with caution.
An SAM signature is found in the spatial correlation between mean sea level pressure and the number of precipitation events in summer. But as the seasonal pattern of trends observed in SAM, and seasonal patterns found in the trends of cloud cover, and precipitation do not agree, this is thought to reflect the trend in SAM towards a more positive phase during summer months rather than actually express the influence of SAM on summer precipitation. Orr et al. (2008) and Van Lipzig et al. (2008) found that the strengthening of the circumpolar westerlies caused by this trend in SAM reduces the blocking effect of the Antarctic Peninsula and leads to a higher frequency of air masses being advected eastwards over the orographic barrier of the northern Antarctic Peninsula.
Assuming an altogether different reason behind the increase in cloud cover and precipitation, this reduced blocking could effectively weaken or cancel out any underlying positive trend in precipitation events in summer.
While the influence of large-scale circulation patterns on precipitation in West Antarctica has been shown, it cannot explain the trends found in precipitation events presented in this study. It is thought that a combination of temperature rise and sea ice reduction has led to a significant increase in the cloud cover at Faraday/Vernadsky. The WAP and adjoining southern Bellingshausen Sea regions are the only Antarctic sectors to have undergone a statistically significant decrease in sea ice extent over the past few decades, that is, over the era when satellite passive microwave data have been available (Stammerjohn and Smith, 1997; Zwally et al., 2002). It has been shown that a decrease in sea ice extent in the WAP (King, 1994) is linked to a rise in air temperature. Analyses of synoptic cloud observations at Faraday/Vernadsky and HadISST1 data published by Kirchgäßner (2010) in turn show a highly significant correlation (1% level) between an increase in total cloud cover and decreasing sea ice concentration in winter. There is obviously a link between higher air temperatures and the extent and concentration of sea ice. Especially during winter when the temperature difference between the air and the ocean can easily reach 30 K and high wind speeds are frequently observed, turbulent heat fluxes over open water may be as large as some hundred W/m2, while at the same time over thick sea ice they often amount to only a few W/m2 (Brümmer et al., 2005). Under such conditions even little changes in sea ice cover lead to huge changes in the turbulent fluxes of sensible and latent heat. This is one potential source of additional moisture in the air that has led to the increase in total cloud cover reported by Kirchgäßner (2010).
Of special importance is the increase of non-frozen precipitation observed in spring, summer and autumn. This will not only change environmental conditions locally but will also trigger feedbacks in the climate. The fact that more precipitation falls in some liquid form will reduce the period of time when the ground at and around Faraday/Vernadsky is covered with snow. This will, in consequence, decrease the albedo with all the known feedback mechanisms this has on the radiation budget. As temperatures in spring and autumn are still—despite the rise in air temperature—on average below 0 °C, especially in spring much of this liquid precipitation will fall on (still) frozen ground, hence cannot be stored but will run off and is thus not available for the ecosystem.
Non-frozen precipitation draining through snow accelerates the melting process. In coastal regions, this process may contribute to the ‘erosion’ of glaciers in their calving zone and thus accelerate the retreat of glaciers.
Further research is needed to extend these analyses to a wider region with the aid of remote sensing data. A further topic should be the investigation of parameters that are influenced by the presented changes in precipitation, such as the albedo.
I would like to thank the reviewers for their time and dedication. Their suggestions have helped to improve the manuscript. Thanks go also to my colleagues at BAS for valuable discussions, especially to Tom Lachlan-Cope and Gareth Marshall.