The AR algorithm was run for the period November 2018 to February 2019 for the longitudinal sector spanning from 60°W (approximately at the Antarctic Peninsula/western edge of the Weddell Sea) to 90°E. In total, ten AR events with landfall within this extended DML sector were detected during November 2018 to February 2019 (period of YOPP-SOP-SH). There were four ARs in November, three in December, two in January, and one in February. Only two AR events affected Syowa’s location. The other eight events had their landfall in the western part of DML, from which only two were in the vicinity of Neumayer. For our analysis, we have chosen the two most prominent AR events (lasting for at least one day) that affected the Neumayer and Syowa sites and had minimal data gaps (which are still present due to severe weather conditions during ARs).
Table 1 lists the reanalysis and radiosonde timings corresponding to each identified AR step. The spatial distribution of IWV together with mean sea level pressure (MSLP), and their temporal evolution together with the AR contours, are shown in Fig. 2. The AR landfall longitude shows the mean longitude at which the AR crossed the Antarctic coast, together with its extension expressed as minimum and maximum longitudes (Table 1). In the table we also provide the corresponding IVT and IWV (from near the surface up to 300 hPa) calculated based on the radiosonde profiles as well as ERA-Interim and ERA5 reanalysis data.
(a) AR event affecting Syowa on 16−17 November 2018 AR events timings AR landfall longitude (mean/east/west)/latitude Radiosonde times IWV (kg m−2) IVT (kg m−1 s−1) 1200 UTC 16 Nov 34°E/32°E/36°E/68°S 1100 UTC 16 Nov 7/7/7 43/40/50 1800 UTC 16 Nov 43°E/32°E/54°E/67°S no data −/7/8 −/82/102 0000 UTC 17 Nov 39°E/31°E/48°E/68°E 2300 UTC 16 Nov 9/9/9 192/163/183 0600 UTC 17 Nov 40°E/30°E/50°E/68°E no data −/9/9 −/185/195 1200 UTC 17 Nov 32°E/26°E/37°E/69°S no data −/9/9 −/154/184 (b) AR event affecting Neumayer on 18 December 2018 AR events timings AR landfall longitude (mean/east/west)/latitude Radiosonde times IWV (kg m−2) IVT (kg m−1 s−1) 0000 UTC 18 Dec 20°W/34°W/6°W/74°S 2300 UTC 17 Dec 12/12/13 177/184/199 0600 UTC 18 Dec 21°W/38°W/4°W/74°S 0400 UTC 18 Dec 13/13/14 221/222/235 1200 UTC 18 Dec 22°W/40°W/3°W/74°S no data −/14/14 −/268/283 1800 UTC 18 Dec 21°W/40°W/2°W/74°S 1700 UTC 18 Dec 15/15/16 340/330/343
Table 1. List of the AR time steps for two prominent AR events with landfall at the East Antarctic coast within the 60°W−90°E longitudinal sector identified during the YOPP SOP-SH period at the sites of Syowa and Neumayer. The columns give (1) AR event timings based on six-hourly time steps of the ERA-Interim reanalysis data, (2) AR landfall location—mean longitude and total extent (east/west longitudes) and mean latitude, (3) timings of the radiosonde launches corresponding to the AR events [the launch time is normally 1 h prior to the reanalysis hour given in column (1)], (4) corresponding IWV, and (5) IVT for Syowa and Neumayer calculated from radiosondes/ERA-Interim/ERA5.
Figure 2. Maps of IWV (colors; units: kg m-2) and mean sea level pressure (contours; units: hPa) at each time step of the AR event with landfall (a) near Syowa station from 1200 UTC 16 November to 1200 UTC 17 November 2018, and (b) near Neumayer from 0000 UTC to 1800 UTC 18 December 2018. The fields are based on ERA-Interim reanalysis data. Blue contours show the AR boundaries defined using the AR algorithm (see section 3.2). Blue stars show the mean landfall location of the AR at the Antarctic coast. Red circles show the locations of Syowa and Neumayer stations.
The Syowa AR event lasted from 1200 UTC 16 November until 1200 UTC17 November, with a mean landfall longitude ranging from 32°E to 43°E (Table 1, Fig. 2a). Radiosonde measurements showed an increase in the IWV up to 9 kg m−2 and an IVT up to 192 kg m−1 s−1 (at 0000 UTC 17 November), compared to the monthly median values (5.7 kg m−2 and 45 kg m−1 s−1, respectively) and exceeding the 98th percentiles for the month of November (7.4 kg m−2 and 180 kg m−1 s−1, respectively). The AR event at Neumayer was identified during 18 December 2018 from 0000 UTC to 1800 UTC, with a mean landfall longitude at 20°−22°W (Table 1, Fig. 2b). Radiosonde measurements showed a maximum increase in IWV up to 15 kg m−2 and an IVT up to 340 kg m−1 s−1. This is a significant increase compared to the monthly median values (4.1 kg m−2 and 37.5 kg m−1 s−1, respectively), exceeding the 98th percentiles for the month of December (12.6 kg m−2 and 270 kg m−1 s−1, respectively).
The relatively weak peak values of IVT during these AR events at the two Antarctic stations point to a lower atmospheric moisture-holding capacity and problematic use of AR algorithms based on the absolute threshold of IVT [typically at least 250 kg m−1 s−1; see Shields et al. (2018)] and fixed AR scales (Ralph et al., 2019) when applied in polar regions. At the same time, the minimum IVT threshold of 100 kg m−1 s−1 used in the global algorithm by Guan and Waliser (2015) is a reasonable cut-off value for the two studied events.
The AR event affecting Syowa on 16−17 November 2018 was characterized by a corridor of moisture stretching from the western part of the Indian Ocean sector of the Southern Ocean (Fig. 2a). The AR contour shows an extent reaching far into the lower latitudes (as far as 38°S). While the mean longitude of the AR landfall was centered around 32°−43°E, the AR landfall spanned 166−950 km along the coast, depending on the time step (Table 1, Fig. 2a). The AR was associated with a deep low-pressure system centered at (60°S, 25°−30°E) and was blocked to the east by a pronounced high-pressure ridge.
During the AR event affecting Neumayer on 18 December 2018, a corridor with high IWV values was found stretching from the southern Atlantic Ocean (AR contour extending as far as 48°S) towards the eastern part of the Weddell Sea (Fig. 2b). The mean longitude of the AR landfall was centered at 20°−22°W, with the AR landfall spanning 850−1150 km along the coast (Table 1, Fig. 2b). As in the Syowa case in November, the AR affecting Neumayer in December was formed in association with a deep low-pressure system [centered in the northern Weddell Sea; (68°−70°S, ~40°E)], blocked by high pressure to the east.
A three-day backward trajectory analysis of the air parcels released at the locations of Neumayer and Syowa stations (at 500 m MSL) at the time of the AR landfall (figures not shown; see setup details in section 3.3) confirmed that the majority of the air parcels had a long-distance origin (south of 45°−50°S). This is consistent with the preferential direction of the two ARs extending into the subtropics (Fig. 2).
The changes in the vertical profiles before, during and after the AR landfall in the vicinity of Neumayer and Syowa stations during the two prominent AR events are shown in Figs. 3 and 4.
Figure 3. Evolution of the vertical profiles (from 1000 to 500 hPa) at Neumayer’s location during and after the AR event on 18 December 2018: zonal (MTu, blue, positive eastward, negative westward) and meridional (MTv, red, positive northward, negative southward) component vectors of moisture transport, and total moisture transport magnitude (MT, black) (left-hand panels); air temperature (black) and relative humidity with respect to ice (RHi, blue) (center panels); specific humidity (blue), wind speed (black), and wind direction (red) (right-hand panels). Variables are derived from radiosonde measurements at Neumayer station (OBS, solid lines), and from ERA-Interim (ERA-I, dashed) and ERA5 (dashed with crosses) reanalysis data for the closest grid to the station. The radiosonde launch time (Obs) and the time step as in the reanalyses (ERA), together with the relevance to the AR event evolution, is indicated above each plot.
Figure 4. As in Fig. 3 but for Syowa for the AR case on 16−17 November 2018.
The first time step of the AR landfalling near Neumayer on 18 December 2018 at 0000 UTC was characterized by increased total MT magnitude (see Appendix) in the extended layer between 950 and 750 hPa (up to 70 g kg−1 m s−1) driven by relatively large specific humidity values [up to 3 g kg−1, which is frequently attained in summer; see Fig. 5a and Lenaerts et al. (2010)] and strong wind speeds in the same layer (up to 30 m s−1) veering from northeasterly near the ground to north-northeasterly above 800 hPa (Fig. 3a). The annual mean values for specific humidity and wind speed reported for Neumayer station are 1.2 g kg−1 and 9 m s−1 (van den Broeke et al., 2010). Near the ground (below 900 hPa), both zonal and meridional moisture flux components contributed to the total MT, while above 750 hPa the meridional (onshore) flux dominated (Fig. 3a). The observed profiles were well represented by both reanalyses (Fig. 3a, dashed and cross-dashed lines), with ERA5 showing a slightly stronger meridional MT component in the 900−850 hPa layer (Fig. 3a, right panel, red cross-dashed line). During this time at Neumayer, the IWV attained 12 kg m−2 and IVT 177 kg m−1 s−1 (according to radiosonde profiles), with IVT overestimated by both ERA-Interim and ERA5 (Table 1).
Figure 5. Temporal evolution of the vertical profiles (from the first radiosonde measurement near the surface to 300 hPa) for (a) specific humidity, (b) wind speed, and (c) moisture transport during December 2018 at Neumayer, based on four radiosondes per day—at launch times of 0500, 1100, 1700, and 2300 UTC. Missing profiles are shown in white (see section 3.1.2 about data gaps). The red dashed lines show the periods when Neumayer experienced AR landfall.
While the AR was persisting from 0000 to 1800 UTC on 18 December 2018 (Fig. 2b), the IWV and IVT at Neumayer reached their maximum values (15 kg m−2 and 340 kg m−1 s−1) at 1800 UTC (Table 1, Fig. 3b). This peak IWV was well represented by both reanalyses, while IVT was slightly underestimated by ERA-Interim (330 kg m−1 s−1) and well captured by ERA5 (343 kg m−1 s−1) (Table 1). The peak IWV and IVT values at 1800 UTC are associated with an increase in the near-surface wind speed up to 32 m s−1, with the maximum values at about 950 hPa and northeasterly in direction (Fig. 3b). This wind speed maximum is strongly underestimated by both reanalyses, which showed a maximum of only 22 m s−1 (Fig. 3b). As noted in section 2, easterly to northeasterly winds of high magnitude at Neumayer are typically associated with cyclonic disturbances and warm/moist advection, steered by the topography (Figs. 1a-c). The specific humidity profile was characterized by a strong inversion (up to 4 g kg−1) between 900 and 850 hPa (above the LLJ) (Fig. 3b). At the same level, a temperature inversion can also be noted, indicating heat advection as it shows decoupling from the surface (Fig. 3b, left panel). The elevated humidity and temperature inversions are not well captured by ERA-Interim, while ERA5 represents the inversions quite closely, albeit slightly overestimating both temperature and humidity values near the surface (Figs. 3b). These moisture and temperature inversions can be explained by the enhanced onshore advection from the ocean, as it was found in the layers where the wind veers more towards the northerly direction (Fig. 3b). The wind direction in the moist layer is in the northeasterly quadrant, turning towards northerly with height, with the meridional (onshore) flux component dominating the total MT, attaining 100 g kg−1 m s−1 between 900 and 850 hPa (Fig. 3b). While the zonal MT component is well captured by both reanalyses, the meridional and the total MT are well captured only by ERA5, being underestimated by ERA-Interim (Fig. 3b).
After the AR’s passing, high IVT values persisted at Neumayer until 1200 UTC 19 December (reaching 286 and 216 kg m−1 s−1 at 0600 UTC and 1200 UTC, respectively), with humidity inversion still present between 930 and 875 hPa and the near-surface LLJ weakening (Fig. 3c). Both reanalyses were still underestimating the LLJ and the peak in MT magnitude (Fig. 3c). During the entire AR event, the lower troposphere remained saturated, with RHi showing a small oversaturation with respect to ice, being slightly underestimated by reanalyses (Figs. 3a-c). One day after the AR event (at 1800 UTC 19 December), a strong drying of the layers between 950 and 800 hPa occurred, while the temperature profile changed only slightly (Fig. 3d). This drying could be associated with the dry-air intrusion from near the tropopause frequently associated with the extratropical cyclone cold sector, as described by Browning (1997). MT suddenly dropped to below 50 g kg−1 m s−1 at 1800 UTC 19 December, with moisture inversion trapped near the surface and a dry layer present above 950 hPa (Fig. 3d). All these rapid changes were well captured by both reanalyses (Fig. 3d).
During the AR event that affected Syowa on 16−17 November 2018, the enhanced MT was more influenced by a prominent LLJ between 900 and 825 hPa, while specific humidity attained maximum values near the surface (Figs. 4a and b). The IVT calculated using radiosonde measurements showed a maximum at 0000 UTC 17 November (192 kg m−1 s−1; Table 1), when the total MT increased up to 100 g kg−1 m s−1 at about 920 hPa with equal contribution from zonal and meridional MT components (Fig. 4b), and coincident with the peak in the LLJ wind speed (Fig. 4b). Neither the increase in humidity near the surface nor the LLJ at higher levels were well captured by ERA-Interim and ERA5, both showing significant underestimation (Fig. 4b). This leads to significant underestimation of both the zonal and meridional MT components, and the total MT magnitude (Fig. 4b, left). The wind direction near the surface and in the lower troposphere had a persistent northeasterly direction throughout the AR event (Figs. 4a-c), indicating onshore advection from the ocean steered by the topography (and equally strong meridional and zonal MT components; Fig. 4b). This lasted until a sudden change to the southerly wind direction below 900 hPa at 1200 UTC 18 November (Fig. 4d), typical of the katabatic flow from the plateau (see section 2 and Fig. 1d for wind regimes at Syowa). ERA-Interim and ERA5 show much stronger drying in the near-surface layers compared to radiosonde observations after the AR’s passing (Figs. 4c and d). At the same time, as discussed in section 3.1.1, radiosondes are prone to wet bias in the dry layers situated above the humid layers. The wind direction shifting to the southerly direction is well captured by ERA5, while ERA-Interim instead shows a northerly direction (Fig. 4d).
Figures 5 and 6 demonstrate the temporal evolution of the vertical profiles of specific humidity, wind speed and MT at Syowa and Neumayer stations during November and December 2018, respectively, using frequent radiosonde observations during SOP-SH. AR events at Neumayer (indicated by red dashed lines in Fig. 5) show that the strong increase in specific humidity associated with ARs occupies a deep layer affecting almost the entire troposphere, with values peaking up to 4 g kg−1 between 800 and 925 hPa during at least one and a half days. At the same time, the wind speed increase is concentrated near the surface, with the LLJ directed along the topography in a northeasterly direction turning to northerly (onshore) when approaching 500 hPa (as shown by individual profiles in Fig. 3). The layers just above the surface are typically characterized by barrier jets (e.g., van den Broeke and Gallée, 1996) and katabatic flow (e.g., van den Broeke and van Lipzig, 2003), which many studies have found to be persistent and amplified during cyclonic disturbances and precipitation events (Parish, 1983; Parish and Bromwich, 2007; Seefeld and Cassano, 2008; van Wessem et al., 2015; Yamada and Hirasawa, 2018; Vignon et al., 2019). Enhanced easterly winds along the topography at the coastal and escarpments regions of DML are also associated with cyclonic disturbances and moisture influx from the lower latitudes, as opposed to the katabatic flow where the southerly component is present (König-Langlo and Loose, 2007; Gorodetskaya et al., 2013; Souverijns et al., 2018). The onshore moist advection occupies higher levels, with MT values much higher compared to the background state during the month (Fig. 5c), and compared to the median values as discussed in section 4.4 (Figs. 8 and 9).
Figure 6. As in Fig. 5 but for Syowa during 15−30 November 2018, based on three radiosondes per day—at launch times of 0500, 1100, and 2300 UTC. Missing profiles are shown in white (see section 3.1.2 about data gaps). The red dashed lines show the period when Syowa experienced AR landfall.
Figure 8. Composite profiles from 1000 to 500 hPa (median and interquartile range) for all enhanced moisture transport events at Neumayer during 2009−19 for (a) temperature (T; units: °C), (b) specific humidity (q; units: g kg−1), (c) wind speed (WS; units: m s−1), (d) wind direction (WD—N = northerly, NE = northeasterly, E = easterly, SE = southeasterly), (e) moisture transport magnitude (MT; units: g kg−1 m s−1), (f) MT zonal vector component (MTu; units: g kg−1 m s−1), (g) MT meridional vector component (MTv; units: g kg−1 m s−1). Variables are derived from radiosonde measurements at Neumayer station from the IGRA2 archive (OBS, cyan), from ERA-Interim (ERA-I, green) and ERA5 (red) for the closest grid to the station. The thick gray line on the composite profiles is the median based on the entire radiosonde period (2009−19). The radiosonde IGRA2 original profiles are interpolated to 5-hPa intervals from the first layer near the surface to 500 hPa. Differences between the reanalyses and observations (for ERA-Interim in green, for ERA5 in red) for respective variables are given in panels (h−n).
Figure 9. As in Fig. 8 but for Syowa station.
The LLJ during the AR events is accompanied by a strong increase in the upper-level jet stream at the 300 hPa level (Fig. 5b), which can be an indication of a cyclonically induced LLJ when the low-level pressure decreases below the upper-level divergence in the left-hand exit region of a jet streak, which causes development of the ageostrophic component and horizontal acceleration, in turn causing the LLJ (e.g., Burrows et al., 2019). These LLJs assist in the poleward MT towards Antarctica during the ARs. The peak in the MT during the two AR cases affecting Neumayer station in December 2018 was found between 800 and 900 hPa, with significant values also below 900 hPa. Neumayer location on the ice shelf makes it sensitive to strong onshore MT not only at the peak level above 900 hPa, but also occurring closer to the surface—even if the flow follows the topography, it is initially associated with the moisture advection from the ocean.
The AR event at Syowa during 16−18 November 2018 also showed a strong and rapid change in the wind and humidity profiles compared to the rest of the month (Fig. 6). However, the AR signatures are different than those at Neumayer (Fig. 5). Firstly, both at Neumayer and Syowa stations there are elevated specific humidity values near the surface trapped in the stable boundary layer, with values up to 2−2.5 g kg−1, during the entire months of November and December. At Neumayer, during the AR events, specific humidity values increased up to 4 g kg−1, extending to higher levels, while at Syowa humidity values showed only a slight increase near the surface, with the height of the moist layer extending much less compared to Neumayer (Fig. 6a). At the same time, there was a much stronger increase in the wind speed below 900 hPa. This LLJ was accompanied by a very strong (up to 50 m s−1) upper-level jet stream (Fig. 6b). The MT increase was not as strong compared to Neumayer, but still showed a significant anomaly up to 100 kg m−1 s−1 below 900 hPa (Fig. 6c).
In order to put the AR events observed during YOPP SOP-SH into context over a longer period, we used radiosonde data at the two stations available via IGRA2 during a 10-yr period (2009−19). Analysis of the AR events detected during YOPP SOP-SH showed that they were all coincident with an enhanced MT signature in the vertical profiles, with IVT exceeding 100 kg m−1 s−1 and the maximum value in the MT profile exceeding 50 g kg−1 m s−1. Distribution analysis applied to the maximum MT within the profile, and the IVT calculated based on IGRA2 radiosonde profiles, showed that these thresholds correspond to the 95th percentile based on the 2009−19 radiosonde measurements (Fig. 7).
Figure 7. Yearly distribution (as of kernel density estimation, KDE) of (a) maximum horizontal moisture transport (MTmax) and (b) integrated water vapor transport (IVT) within a profile related to enhanced moisture transport events at Neumayer. (c, d) As in (a, b) but for Syowa. Vertical dotted lines represent yearly 95th percentiles for the period 2009−19; the solid black line represents the median of all yearly 95th percentiles (value given in the legend); and the solid red line represents the applied thresholds (values in the legend).
Enhanced MT profiles were isolated according to the above mentioned criteria using IGRA2 radiosonde data during 2009−19 at Neumayer and Syowa. Figures 8 and 9 show the median and the interquartile range (from the 75th to 25th percentiles) for the composites of the enhanced MT profiles at both stations during the 10-yr period. It should be noted that these composites include the entire year of measurements, while YOPP SOP-SH was conducted during the austral summer period.
The composite profiles of the enhanced MT events at Neumayer show the peak values (median up to 58 g kg−1 m s−1 and 75th percentile up to 67 g kg−1 m s−1) between 950 and 900 hPa (Fig. 8e). This peak is dominated by the zonal MT component (Fig. 8f), while the meridional MT component peaks at the higher levels between 900 and 850 hPa (Fig. 8g). Both ERA-Interim and ERA5 capture well the peak in total and zonal MT (Fig. 8e and f), with ERA-Interim showing smaller errors compared to ERA5 in total MT (Fig. 8l and m). Both reanalyses underestimate the peak in the meridional MT (Figs. 8g and n). The peak in the MT corresponds to the maximum wind speed within the same layer between 950 and 900 hPa, with median values up to 27 m s−1 and the 75th percentile exceeding 32 m s−1 (Fig. 8c). Both reanalyses underestimate the median and the 75th percentile values of the wind peak (Figs. 8c and j). The location of the LLJ, its strength (up to 32 m s−1), and the underestimation by both reanalyses are similar to the AR case on 18 December 2018 (Fig. 3b). At the same time, the composites do not show a pronounced elevated moisture inversion found during the December 2018 event (Fig. 8b and Fig. 3b), but rather show a small near-surface moisture inversion with median values up to 3 g kg−1 (Fig. 8b). This could be related to the low temporal resolution of regular radiosonde observations and reduced vertical resolution of IGRA2 archive data, which allows large-scale features to be captured whilst being unsuitable for analyzing small-scale changes in the tropospheric profile, including details of the humidity inversions and LLJs.
At Syowa, the composite of the enhanced MT events during 2009−19 (Fig. 9) shows similar signatures to the AR event observed during 16−17 November 2018 (Fig. 4). Firstly, the IGRA2 radiosonde-based composite shows a strong near-surface maximum in specific humidity, with the median peaking at 3.2 g kg−1 and 75th percentile up to 3.5 g kg−1 (Fig. 9b). This near-surface trapped humidity inversion is more pronounced compared to Neumayer (Fig. 8b). The near-surface humidity increase coincides with the warm layer, where the air temperature increase is up to 0°C (Fig. 9a), which is important in determining the precipitation phase and potential for surface melt. This finding is in line with earlier work by Kurita et al. (2016) showing that surface air temperature frequently exceeds 0°C during warm events at Syowa. This near-surface humidity inversion is completely missed by both reanalyses, underestimating the median value compared to the radiosonde composite by up to 0.7 g kg−1 (Fig. 9b, i).
A prominent increase in the wind speed is present in the composite median, with a peak between 925 and 875 hPa (Fig. 9c) and a persistent northeasterly direction throughout the lower troposphere (Fig. 9d). The intensity of this LLJ is strongly underestimated by the ERA-Interim composites, while ERA5 shows only a small underestimation (Fig. 9c, j). MT shows a moderate peak exceeding 60 g kg−1 m s−1 (with the 75th percentile reaching 50 g kg−1 m s−1) within the heights of the wind speed maximum (Fig. 9e). The AR event affecting Syowa on 16−17 November 2018 showed an increase in MT up to 100 g kg−1 m s−1 (Fig. 4b), exceeding the 75th percentile of the 10-yr time series of enhanced MT events. The total MT and its both components are strongly underestimated by both reanalyses, with ERA5 showing a smaller but still strong bias (Figs. 9e−g, and l−n).