The floating ice shelves and glacier tongues which fringe the
Antarctic continent are important because they help buttress ice flow from
the ice sheet interior. Dynamic feedbacks associated with glacier calving
have the potential to reduce buttressing and subsequently increase ice flow
into the ocean. However, there are few high temporal resolution studies on
glacier calving, especially in East Antarctica. Here we use ENVISAT ASAR
wide swath mode imagery to investigate monthly glacier terminus change
across six marine-terminating outlet glaciers in Porpoise Bay
(76
Iceberg calving is an important process that accounts for around 50 % of total mass loss to the ocean in Antarctica (Depoorter et al., 2013; Rignot et al., 2013). Moreover, dynamic feedbacks associated with retreat and/or thinning of buttressing ice shelves or floating glacier tongues can result in an increased discharge of ice into the ocean (Rott et al., 2002; Rignot et al., 2004; Wuite et al., 2015; Fürst et al., 2016). At present, calving dynamics are only partially understood (Benn et al., 2007; Chapuis and Tetzlaff, 2014) and models struggle to replicate observed calving rates (van der Veen, 2002; Astrom et al., 2014). Therefore, improving our understanding of the mechanisms driving glacier calving and how glacier calving cycles have responded to recent changes in the ocean–climate system is important in the context of future ice-sheet mass balance and sea level.
Calving is a two-stage process that requires both the initial ice fracture and the subsequent transport of the detached iceberg away from the calving front (Bassis and Jacobs, 2013). In Antarctica, major calving events can be broadly classified into two categories: the discrete detachment of large tabular icebergs (e.g. Mertz Glacier tongue: Massom et al., 2015) or the spatially extensive disintegration of floating glacier tongues or ice shelves into numerous smaller icebergs (e.g. Larsen A and B ice shelves; Rott et al., 1996; Scambos et al., 2009). Observations of decadal-scale changes in glacier terminus position in both the Antarctic Peninsula and East Antarctica have suggested that despite some degree of stochasticity, iceberg calving and glacier advance/retreat is likely driven by external climatic forcing (Cook et al., 2005; Miles et al., 2013). However, despite some well-documented ice-shelf collapses (Scambos et al., 2003; Banwell et al., 2013) and major individual calving events (Masson et al., 2015) there is a paucity of data on the nature and timing of calving from glaciers in Antarctica (e.g. compared to Greenland: Moon and Joughin, 2008; Carr et al., 2013), and particularly in East Antarctica.
Following recent work that highlighted the potential vulnerability of the East Antarctic Ice Sheet in Wilkes Land to ocean–climate forcing and marine ice-sheet instability (Greenbaum et al., 2015; Aitken et al., 2016; Miles et al., 2013, 2016), we analyse the recent calving activity of six outlet glaciers in the Porpoise Bay region using monthly satellite imagery between November 2002 and March 2012. In addition, we also observe the start of a large calving event in 2016. We then turn our attention to investigating the drivers behind the observed calving dynamics.
MODIS image of Porpoise Bay, with glacier velocities overlain (Rignot et al., 2011). The hatched polygon represents the region where long-term 25 km resolution SMMR/SSM/I sea ice concentrations were extracted. The non-hatched polygon represents the region where the higher resolution (6.25 km) AMSR-E sea ice concentrations were extracted.
Porpoise Bay (76
Porpoise Bay is 150 km wide and is typically filled with landfast multi-year
sea ice (Fraser et al., 2012). In total, six glaciers were analysed, with
glacier velocities (from Rignot et al., 2011) ranging from
Glacier terminus positions were mapped at approximately monthly intervals between November 2002 and March 2012, using Envisat Advanced Synthetic Aperture Radar (ASAR) Wide Swath Mode (WSM) imagery across six glaciers, which were identified from the Rignot et al. (2011) ice-velocity data set (Fig. 1). Additional sub-monthly imagery between December 2006 and April 2007 were used to gain a higher temporal resolution following the identification of a major calving event around that time. During the preparation of this manuscript we also observed the start of another large calving event with Sentinel-1 imagery (Table 1).
Satellite imagery used in the study
Approximately 65 % of all glacier frontal measurements were made using an automated mapping method. This was achieved by automatically classifying glacier tongues and sea ice into polygons based on their pixel values, with the boundary between the two taken as the terminus position. The threshold between glacial ice and sea ice was calculated automatically based on the image pixel statistics, whereby sea ice appears much darker than the glacial ice. In images where the automated method was unsuccessful, terminus position was mapped manually. The majority of these manual measurements were undertaken in the austral summer (December–February) when automated classification was especially problematic due to the high variability in backscatter on glacier tongues as a result of surface melt. Following the mapping of the glacier termini, length changes were calculated using the box method (Moon and Joughin, 2008). This method calculates the glacier area change between each time step divided by the width of the glacier, to give an estimation of glacier length change. The width of glacier was obtained by a reference box which approximately delineates the sides of the glacier.
Given the nature of the heavily fractured glacier fronts and the moderate
resolution of Envisat ASAR WSM imagery (80 m) it was sometimes difficult to
establish whether individual or blocks of icebergs were attached to the
glacier tongue. As a result, there are errors in precisely determining
terminus change on a monthly timescale (
Sea ice concentrations in Porpoise Bay were calculated using mean monthly
Bootstrap sea ice concentrations derived from the Nimbus-7 satellite and the
Defense Meteorological Satellite Program (DMSP) satellites which offers near
complete coverage between October 1978 and December 2014 (Comiso, 2014;
Daily sea ice concentrations derived from the ARTIST Sea ice (ASI) algorithm
from Advanced Microwave Scanning Radiometer – EOS (AMSR-E) data (Spreen et
al., 2008) were used to calculate daily sea ice concentration anomalies
during the January 2007 sea ice break-up
(
We used the Regional Atmospheric Climate Model (RACMO) V2.3 (van Wessem et al., 2014) to simulate daily surface melt fluxes in the study area between 1979 and 2015 at a 27 km spatial resolution. The melt values were extracted from floating glacier tongues in Porpoise Bay because the model masks out sea ice, equating to seven grid points. The absolute surface melt values are likely to be different on glacial ice, compared to the sea ice, but the relative magnitude of melt is likely to be similar temporally.
In the absence of weather stations in the vicinity of Porpoise Bay we use
the 0.25
Terminus position change of six glaciers in Porpoise Bay between
November 2002 and March 2012. Note the major calving event in January 2007
for five of the glaciers. Terminus position measurements are subject to
Analysis of glacier terminus position change of six glaciers in Porpoise Bay
between November 2002 and March 2012 reveals three broad patterns of glacier
change (Fig. 2). The first pattern is shown by Holmes (West) Glacier, which
advances a total of
Envisat ASAR WSM imagery in January 2007
Envisat ASAR WSM imagery showing the evolution of the 2007 calving event. Red line shows the terminus positions from 11 December 2006 on all panels.
A series of eight sub-monthly images between 11 December 2006 and
8 April 2007 shows the evolution of the 2007 calving event (Fig. 4).
Between 11 December 2006 and 2 January 2007, the landfast sea
ice edge retreats past Sandford Glacier to the edge of Frost Glacier and
there is some evidence of sea ice fracturing in front of the terminus of
Glacier 2 (Fig. 4b). From 2 to 9 January a small
section (
MODIS imagery showing the initial stages of disintegration of
Holmes (West) Glacier in March 2016. On 19 March a large section of
sea ice breaks away from the terminus (circled), initiating the rapid
disintegration process. By 24 March an 800 km
Sentinel-1 imagery showing the evolution of the 2016 calving event. Purple line shows the terminus position from 2 March on all panels.
Mean monthly sea ice concentration anomalies from November 2002 to June 2016. The red line indicates sea ice concentration anomalies in Porpoise Bay and the blue line indicates pack-ice-concentration anomalies.
During the preparation of this manuscript satellite observations of Porpoise
Bay revealed that another large near-simultaneous disintegration of glacier
tongues in Porpoise Bay was currently underway. This event was initiated on
19 March where the edge of the multi-year sea ice retreated to the
Holmes (West) Glacier terminus, removing multi-year sea ice which was at
least 14 years old. By 24 March this had led to the rapid
disintegration of an 800 km
Analysis of mean monthly sea ice concentration anomalies in Porpoise Bay
between November 2002 and June 2016 (Fig. 7) reveals that a major negative sea
ice anomaly occurred between January and June 2007, where monthly sea ice
concentrations were between 35 and 40 % below average. This is the
only noticeable (
Atmospheric circulation anomalies in the months preceding the January 2007 and March 2016 sea ice break-ups reveal contrasting conditions. In the austral summer which preceded the January 2007 break-up there were strong positive SST anomalies and atmospheric-circulation anomalies throughout December 2005 (Fig. 9a). The circulation anomaly was reflected in a strong easterly airflow offshore from Porpoise Bay. This is associated with a band of cooler SSTs close to the coastline and the northward shift of the Antarctic Coastal Current in response to the weakened westerlies (e.g. Langlais et al., 2015). A weakened zonal flow combined with high SST in the South Pacific would allow the advection of warmer maritime air into Porpoise Bay. Consistent with warmer air are estimates of exceptionally high melt values in Porpoise Bay during December 2005 derived from the RACMO2.3, which contrasts with the longer-term trend of cooling (Fig. 10). However, the December 2005 anomaly was short lived and, by January 2006, the wind-field conditions were close to average, although SST remained slightly higher than average (Fig. 9b).
Time series of Frost and Sandford glaciers calving showing that sea ice clears prior to calving and dispersal of icebergs.
Mean monthly ERA-Interim derived wind-field and
sea-surface temperature anomalies in the months preceding the 2007 and 2016
sea ice break-ups:
In December 2006 and January 2007, which are the months immediately before and during the break-up of sea ice, atmospheric conditions were close to average, with very little deviation from mean conditions in the wind field and a small negative SST anomaly (Fig. 9c). However, on 11 January 2007, which is the estimated date of sea ice break-up from AMSR-E data, we note that there were very high winds close to Porpoise Bay (Fig. 11a).
In contrast to the months preceding the January 2007 event, we find little deviation from average conditions prior to the March 2016 break-up event. In the austral summer which preceded the 2016 break-up (2014/2015), there was little deviation from the average wind field and only a small increase from average SSTs (Fig. 9d). In December and January 2015/2016, there was evidence for a small increase in the strength of westerly winds, and cooler SSTs in the South Pacific (Fig. 9e). However, in February and March 2016 there was no change from the average wind field and slightly cooler SSTs (Fig. 9f).We note, however, that there was a low-pressure system passing across Porpoise Bay on 19 March 2016, the estimated date of break-up initiation (Fig. 11b).
Mean RACMO2.3-derived December melt 1979–2015 in Porpoise Bay.
ERA-Interim derived wind fields for the estimated dates of sea
ice break-up.
Through mapping the terminus position in all available satellite imagery (Table 1) dating back to 1963, we are able to reconstruct large calving events on the largest glacier in Porpoise Bay, Holmes (West) (Fig. 12). On the basis that a large calving event is likely during the largest sea ice break-up events, we estimate the date of calving based on sea ice concentrations in Porpoise Bay when satellite imagery is not available. Our estimates suggest that Holmes (West) Glacier calves at approximately the same position in each calving cycle, including the most recent calving event in March 2016.
We report a major, near-synchronous calving event in January 2007 and a
similar event that was initiated in 2016. This resulted in
The majority of sea ice in Porpoise Bay is multi-year sea ice (Fraser et al., 2012), and it is likely that various climatic processes operating over different timescales contributed to the January 2007 sea ice break-up event. Although there are no long-term observations of multi-year sea ice thickness in Porpoise Bay, observations and models of the annual cycle of multi-year sea ice in other regions of East Antarctica suggest that multi-year sea ice thickens seasonally and thins each year (Lei et al., 2010; Sugimoto et al., 2016; Yang et al., 2016). Therefore, the relative strength, stability and thickness of multi-year sea ice over a given period is driven not only by synoptic conditions in the short term (days/weeks), but also by climatic conditions in the preceding years.
In the austral summer (2005/2006) which preceded the break-up event in January 2007, there was a strong easterly airflow anomaly throughout December 2005 directly adjacent to Porpoise Bay (Fig. 9a). This anomaly represents the weakening of the band of westerly winds which encircle Antarctica, and is reflected in an exceptionally negative Southern Annular Mode (SAM) index in December 2005 (Marshall, 2003). This contrasts with the long-term trend for a positive SAM index (Marshall, 2007; Miles et al., 2013). A weaker band of westerly winds combined with anomalously high SST in the southern Pacific (Fig. 9a) would allow a greater advection of warmer maritime air towards Porpoise Bay. Indeed, RACMO2.3-derived surface-melt estimates place December 2005 as the second highest mean melt month (1979–2015) on the modelled output in Porpoise Bay (Fig. 10). To place this month into perspective, we note that it would rank above the average melt values of all Decembers and Januarys since 2000 on the remnants of Larsen B Ice Shelf. Comparing MODIS satellite imagery from before and after December 2005 reveals the development of significant fracturing in the multi-year sea ice (Fig. 13a, b). These same fractures remain visible prior to the break-out event in January 2007 and, when the multi-year sea ice begins to break-up, it ruptures along these pre-existing weaknesses (Fig. 13c). As such, this strongly suggests that the atmospheric-circulation anomalies of December 2005 played an important role in the January 2007 multi-year sea ice break-up and near-simultaneous calving event.
Reconstruction of the calving cycle of Holmes (West) Glacier. All observations are represented by black crosses. The estimated terminus position is then extrapolated linearly between each observation. In periods without observations the date of calving is estimated by negative sea ice concentration anomalies.
The break-up of landfast sea ice has been linked to dynamic wind events and ocean swell (Heil, 2006; Ushio, 2006; Fraser et al., 2012). Thus, it is possible that the wind anomalies in December 2005 may have been important in initiating the fractures observed in the sea ice in Porpoise Bay, through changing the direction and/or intensity of oceanic swell. However, this mechanism is thought to be at its most potent during anomalously low pack-ice concentrations because pack ice can act as a buffer to any oceanic swell (Langhorne et al., 2001; Heil, 2006; Fraser et al., 2012). That said, we note that pack-ice concentrations offshore from Porpoise Bay were around average during December 2005 (Fig. 7). This may suggest that there are other mechanisms that were important in the weakening of the multi-year sea ice in Porpoise Bay in December 2005.
In the Arctic, sea ice melt-ponding along pre-existing weaknesses has been widely reported to precede sea ice break-up (Ehn et al., 2011; Petrich et al., 2012; Landy et al., 2014; Schroder et al., 2014; Arntsen et al., 2015). Despite its importance in the Arctic, it has yet to be considered as a possible factor in landfast sea ice break-up in coastal Antarctica. As a consequence of the high melt throughout December 2005, the growth of sea ice surface ponding would be expected, in addition to surface thinning of the sea ice. High-resolution cloud-free optical satellite coverage of Porpoise Bay throughout December 2005 is limited, but ASTER imagery in the vicinity of Frost Glacier on the 4 and 31 December 2005 shows surface melt features and the development of fractures throughout the month (Fig. 13d, e), similar to those observed elsewhere in East Antarctica (Kingslake et al., 2015; Langley et al., 2016). High-resolution imagery from 16 January 2006 (via Google Earth) shows the development of melt ponds on the sea ice surface (Fig. 13f). Therefore, it is possible that surface melt had some impact on the fracturing of landfast sea ice in Porpoise Bay. This may have caused hydro-fracturing of pre-existing depressions in the landfast ice or surface thinning may have made it more vulnerable to fracturing through ocean swell or internal stresses. Additionally, the subsequent refreezing of some melt ponds may temporally inhibit basal ice growth, potentially weakening the multi-year sea ice and predisposing it to future break-up (Flocco et al., 2015). It is important to note that the atmospheric circulation anomalies which favoured the development of fractures in the multi-year sea ice in December 2005 were short-lived. By January 2006, atmospheric conditions had returned close to average (Fig. 9b) and remained so until the austral winter, where sea ice break-up is less likely. This may explain the lag between the onset of sea ice fracturing in December 2005 and its eventual break-up in the following summer (January 2007).
Daily sea ice concentrations and RACMO2.3 derived melt during January 2007 in Porpoise Bay. Sea ice concentrations start to decrease after the melt peak on 11 January.
Consistent with the notion that the multi-year sea ice was already in a weakened state prior to its break-up in 2007, is that the break-up occurred in January, several weeks before the likely annual minima in multi-year sea ice thickness (Yang et al., 2016; Lei et al., 2010) and landfast ice extent (Fraser et al., 2012). Additionally, atmospheric circulation anomalies indicate little deviation from average conditions in the immediate months preceding break-up (Fig. 9b, c), suggesting that atmospheric conditions were favourable for sea ice stability. Despite this, a synoptic event is still likely required to force the break-up in January 2007. Daily sea ice concentrations in Porpoise Bay in January 2007 show a sharp decrease in sea ice concentrations after 12 January, representing the onset of sea ice break-out (Fig. 14). This is preceded by a strong melt event recorded by the RACMO2.3 model, centred on 11 January, which may represent a low-pressure system. Indeed, ERA-Interim estimates of the wind field suggest strong southeasterly winds in the vicinity of Porpoise Bay (Fig. 11a). Unlike in December 2005, pack-ice concentrations offshore of Porpoise Bay were anomalously low (Fig. 7). Therefore, with less pack-ice buttressing, it is possible that the melt event, high winds, and associated ocean swell may have initiated the break-up of the already weakened multi-year sea ice in Porpoise Bay.
In contrast to January 2007, we find no link between atmospheric-circulation anomalies and the March 2016 sea ice break-up. In the months preceding the March 2016 break-up, wind and SST anomalies indicate conditions close to average conditions favouring sea ice stability (Fig. 9d, e, f). This suggests that another process was important in driving the March 2016 sea ice break-up. A key difference between the 2007 and 2016 event is that the largest glacier in the bay, Holmes (West) Glacier, only calved in the 2016 event. Analysis of its calving cycle (Fig. 12) indicates that it calves at roughly the same position in each cycle and that its relative position in early 2016 suggests that calving was “overdue” (Fig. 12). This indicates that the calving cycle of Holmes (West) Glacier has not necessarily been driven by atmospheric circulation anomalies. Instead, we suggest that as Holmes (West) Glacier advances, it slowly pushes the multi-year sea ice attached to its terminus further towards the open ocean to the point where the sea ice attached to the glacier tongue becomes more unstable. This could be influenced by local bathymetry and oceanic circulation, but no observations are available. However, once the multi-year sea ice reaches an unstable state, break-up is still likely to be forced by a synoptic event. This is consistent with our observations, where ERA-Interim derived wind fields show the presence of a low-pressure system close to Porpoise Bay on the estimated date of sea ice break-up in March 2016 (Fig. 11b). Whilst we suggest that the March 2016 sea ice break-up and subsequent calving of Holmes (West) are currently part of a predictable cycle, we note that this could be vulnerable to change if any future changes in climate alter the persistence and/or strength of the multi-year sea ice, which is usually attached to the glacier terminus.
We identify two large near-simultaneous calving events in January 2007 and March 2016 which were driven by the break-up of the multi-year landfast sea ice which usually occupies the bay. This provides a previously unreported mechanism for the rapid disintegration of floating glacier tongues in East Antarctica, adding to the growing body of research linking glacier tongue stability to the mechanical coupling of landfast ice (e.g. Khazander et al., 2009; Massom et al., 2010). Our results suggest that multi-year sea ice break-ups in 2007 and 2016 in Porpoise Bay were driven by different mechanisms. We link the 2007 event to atmospheric-circulation anomalies in December 2005 weakening multi-year sea ice through a combination of surface melt and a change in wind direction, prior to its eventual break-up in 2007. This is in contrast to the March 2016 event, which we suggest is part of a longer-term cycle based on the terminus position of Holmes (West) Glacier that was able to advance and push sea ice out of the bay. The link between sea ice break-up and major calving of glacier tongues is especially important because it suggests that with predictions of future warming (DeConto and Pollard, 2016) multi-year landfast ice may become less persistent. Therefore, the glacier tongues which depend on landfast ice for stability may become less stable in the future. In a wider context, our results also highlight the complex nature of the mechanisms which drive glacier calving position in Antarctica. Whilst regional trends in terminus position can be driven by ocean–climate–sea ice interaction (e.g. Miles et al., 2013, 2016), individual glaciers and individual calving events have the potential to respond differently to similar climatic forcing.
Porpoise Bay terminus position shapefiles are available upon request, please contact the corresponding author.
Envisat ASAR Wide Swath imagery is available from the European Space Agency. Landsat and ASTER imagery are
available from Earth Explorer (
The authors declare that they have no conflict of interest.
We thank the ESA for providing Envisat ASAR WSM data (project ID: 16713) and Sentinel data. Landsat imagery was provided free of charge by the US Geological Survey Earth Resources Observation Science Centre. We thank M. van den Broeke for providing data and assisting with RACMO. B. W. J. Miles was funded by a Durham University Doctoral Scholarship program. S. S. R. Jamieson was supported by Natural Environment Research Council Fellowship NE/J018333/1. We would like to thank Allen Pope and Ted Scambos for reviewing the manuscript, along with the editor, Rob Bingham, for providing constructive comments which led to improvement of this manuscript.Edited by: R. Bingham Reviewed by: T. A. Scambos and one anonymous referee