Images from the ends of the Earth

By Lucy Clarke (@DrLucyClarke)

Thinking about Antarctica conjures up images of a remote ice covered wilderness; it’s the coldest, windiest and driest continent and the only one to not have permanent residents living on it. It is the last terrestrial frontier on Earth that we haven’t yet fully conquered. So when I got the chance to work at the British Antarctic Survey (BAS) in Cambridge it felt like the opportunity of a lifetime – I would get to work on this distant continent… Sadly my current project doesn’t involve fieldwork, but I do get access to a huge archive of aerial photographs so I can explore large areas of Antarctica remotely, plus I’m not giving up hope of heading ‘South’ just yet!

Lots of research and the recent Intergovernmental Panel on Climate Change (IPCC) report published last year have all highlighted the retreat of Antarctica’s ice sheets and glaciers, and are concerned with the impact of future melting on global sea levels. My research, in collaboration with colleagues at the University of Newcastle, will contribute to this debate by quantifying glacier change during the 20^th and 21^st Century.

Photograph of Ryder Bay (left) and the Sheldon Glacier (right) on the Antarctic Peninsula (Photographs courtesy of the BAS Photo Repository)

Antarctica can be divided into 3 areas: the West Antarctic ice sheet, the East Antarctic ice sheet and the Antarctic Peninsula… it is the latter that is the focus of my research. The Antarctic Peninsula is situated on the north-western tip of Antarctica and unlike the rest of the continent it isn’t completely covered by ice sheets, it is a mountainous area and there are many glaciers feeding into ice sheets and the surrounding sea. 

Map of Antarctica showing the 3 ice sheets, with an inset highlighting the Antarctic Peninsula (Source: Antarctic Digital Database)

The Antarctic Peninsula has over 400 glaciers and current thinking is that climate change is causing a rapid reduction of these, however there is very little detailed long-term information to support this. Most of the glaciers are inaccessible thereby preventing collection of measurements in the field and so remote techniques have to be used to determine how these may have altered. Satellite imagery has been used to reconstruct glacial change since the 1990s but the impact of change over the 20^th Century is still unknown for the majority of glaciers. Fortunately we do have an archive of aerial photography of the Antarctic Peninsula at BAS dating back to the 1940s.  We can use this data source to not only visually compare differences in glacier extent during this period but also calculate the volumetric change using photogrammetry.  My December blog post covers use of this technique: What’s in a photo?

Normally photogrammetry requires: (1) two overlapping photographs of an area, (2) details on the camera used, and (3) some identifiable points on the ground that you know the co-ordinates of, to create a 3D model of the overlap area that can then be used to take measurements. In the case of the Antarctic Peninsula we don’t have any ground measurements for large areas, so using the standard technique wasn’t possible and therefore we had to come up with a new way to undertake photogrammetry with no available ground control.

The BAS Twin Otter that the aerial photography is flown from (left) and the camera and storage set up inside the plane (right) (Photos courtesy of the BAS Photo Repository)

New aerial photography in Antarctica is flown by the BAS using a digital camera mounted in a Twin Otter plane with the camera set into a holding that records the exact position, height and rotation of the camera at the instant that each photograph is taken (shown in the pics above). Using this information I can create a high resolution digital elevation model, or 3D model of the surface, using photogrammetry without the need for ground control measurements, thus allowing us to undertake this research in even the most inaccessible areas. The accuracy of this technique (with the potential for 40 cm resolution, so every pixel in the image equates to 40 cm on the ground) far exceeds that offered by current satellite imagery (with a resolution of 15 m). This results in clearly definable features on the subsequently processed photography. I can therefore look at the modern digital elevation model and identify co-ordinates for rock outcrops and mountain peaks that won’t have changed through time. This can then be reverse engineered to create ground control points for the historic aerial photos without ever having to set foot on the glacier! So as long as I have contemporary aerial photography of a glacier I can use this to process older photography from the same area, allowing us to fully utilise the rich archive of historic air photography stored in the BAS archives.

The Moider glacier on the Antarctic Peninsula in (a) 1947 and (b) 2005 showing the thinning and retreat at the glacier front, and (c) the digital elevation model produced from the 2005 imagery.

Preliminary results show dramatic mass change in the study glaciers over the last few decades, and I am currently processing these results in further detail and extending the study sites. I will be blogging about these results in the near future so watch this space…

This research is part of the NERC funded grant: Ref NE/K004867/1: “The spatial and temporal distribution of 20^th Century Antarctic Peninsula glacier mass change and its drivers”