East Antarctic landfast sea-ice distribution and variability

Landfast sea ice (sea ice which is held fast to the coast or grounded icebergs, also known as fast ice) is a pre-eminent feature of the Antarctic coastal zone, where it forms an important interface between the ice sheet and pack ice/ocean to exert a major influence on high-latitude atmosphere-ocean interaction and biological processes. It is highly vulnerable to climate variability and change, given that its formation and breakup are intimately associated with oceanic and atmospheric forcing, yet is not currently represented in global climate models or coupled atmosphere-ocean-ice models. Fast ice forms a key breeding habitat for a number of iconic species, including Weddell seals and Emperor penguins, and plays a crucial role in the breeding success and foraging behaviour of Ad¬¨¬•elie penguins. Recent work further suggests that fast ice may stabilize floating glacier tongues and ice shelves, to affect iceberg calving and ultimately the mass balance of the ice sheet plus the drift rates of icebergs. Moreover, fast ice has a major impact on the logistics of the resupply of Antarctic bases. While Antarctic sea ice extent and variability has been the focus of considerable recent research, fast-ice extent and variability are currently poorly understood. This is in large part due to the difficulty associated with discriminating fast ice from pack ice on a large scale in satellite data. Snapshot‚ÄövÑvp analyses are unable to discriminate between the ice types (i.e., fast ice has a non-unique signature), and ice motion techniques have various problems, including persistent cloud cover at visible and infrared wavelengths and low spatial resolution for passive microwave sensors. Furthermore, Synthetic Apterture Radar (SAR) imagery is inherently difficult to interpret over the sea-ice zone. This thesis, presented in a thesis-by-publication‚ÄövÑvp style, overcomes the problems associated with remotely sensing fast ice at visible and infrared wavelengths to produce cloud-free time-averaged images of the surface from March 2000 to December 2008, enabling discrimination between pack and fast ice. From these, the first East Antarctic (10‚Äöv=¬¿ W - 172‚Äöv=¬¿ E) high spatio-temporal resolution (2-km, 20-day) maps of fast-ice extent are created. This allows the first detailed time series analysis of the seasonal to inter-annual variability of East Antarctic fast ice. Fast-ice growth and breakout events are then related to large scale and local atmospheric forcing parameters. In addition to presenting the first near decade-long, high spatio-temporal resolution time series of fast-ice maps in East Antarctica, the main findings of this thesis are as follows. Using MODerate resolution Imaging Spectroradiometer (MODIS) visible and thermal infrared data, quality cloud-free composite images of the high-latitude surface can generally be constructed using 20 days' raw imagery. The compositing technique developed here involves using a modified MODIS cloud mask to select cloud-free pixels, which are then composited together over number of days. The MOD35 MODIS cloud mask performs well during daytime, when shortwave tests could be included into the cloud masking algorithm. However, during night-time, cloud mask performance is insufficient to create quality cloud-free composite images. Spatial filtering on the cloud mask is required to produce high-quality composite images. With a sufficiently long compositing interval (i.e., 20 days), pack ice motion acts to blur‚ÄövÑvp the pack ice zone in the composite imagery, while the fast/pack ice shear zone remains sharply defined. This property of the compositing process is very useful for discriminating between pack and fast ice. Despite the success of the compositing technique, persistent cloud cover and inaccurate cloud masking are found to lower composite image quality at times. Thus, a technique is developed to augment lower quality composite images with Advanced Microwave Scanning Radiometer - EOS (AMSR-E) and additional MODIS data. Fast-ice maps are generated from the resulting imagery. An error analysis shows that fast-ice extent can be retrieved to within ‚Äöv†¬¿ ¬¨¬±3% for over 80% of the 159 consecutive fast-ice maps comprising the 8.8-year time series, with the remainder retrieved to within ‚Äöv†¬¿ ¬¨¬±9%. Analysis of the 8.8-year East Antarctic fast-ice time series shows a statistically significant increase in fast-ice extent (1.43 ¬¨¬± 0.30% yr‚Äöv†v¿1), albeit based upon a relatively short period. Regionally, there is a stronger increase observed in the Indian Ocean sector (20 - 90‚Äöv=¬¿ E) of 4.07 ¬¨¬± 0.42% yr‚Äöv†v¿1, compared to a slight (nonsignificant) decrease in the Western Pacific Ocean sector (90 - 160‚Äöv=¬¿ E) of -0.40 ¬¨¬± 0.37% yr‚Äöv†v¿1. In the Indian Ocean sector, a slightly decreasing trend in fast-ice extent changes to a strongly increasing trend from 2004 - 2008. An analysis of the timing of maximum and minimum fast-ice extent shows high variability compared to that of overall sea ice. Analysis of the shape of the mean annual fast-ice extent cycle reveals a limit to maximum fast-ice extent, apparently related to the locations of grounded icebergs. Ten fast-ice regimes were identified across the coast, relating to bathymetry, coastal configuration, and prevailing atmospheric conditions. The percentage of fast ice comprising overall sea-ice extent varies seasonally from ‚Äöv†¬¿19% during the summer minimum extent to ‚Äöv†¬¿3.8% during the winter maximum extent. Nine case studies of anomalous fast-ice breakout or growth are conducted in four regions around the East Antarctic coast. These anomalous fast-ice extents are observed in conjunction with locally anomalous wind speeds and directions, surface air temperatures, and pack ice conditions. On a hemisphere-wide scale, a reasonably strong correlation is found between the Southern Oscillation Index (SOI) and fast-ice extent in the Indian Ocean sector (R=+0.45). No strong correlation is observed in the Western Pacific Ocean sector, and, in contrast to previous work, the correlation between SAM and fast-ice extent is weak in both sectors. This work has greatly improved our knowledge of fast-ice distribution and variability around the East Antarctic coast, and provides benefits for many areas of current research. It has also provided an important new climatic dataset that is directly comparable to, and complements, the widely-used passive microwave-derived time series of overall sea-ice extent.