Rigorous 3D change determination in Antarctic Peninsula glaciers from stereo WorldView-2 and archival aerial imagery

This paper presents detailed elevation and volume analysis of 16 individual glaciers, grouped at four locations, spread across the Antarctic Peninsula (AP). The study makes use of newly available WorldView-2 satellite stereo imagery to exploit the previously untapped value of archival stereo aerial photography. High resolution photogrammetric digital elevation models (DEMs) are derived to determine three-dimensional glacier change over an unprecedented time span of six decades with an unparalleled mean areal coverage of 82% per glacier. The use of an in-house robust surface matching algorithm ensured rigorous alignment of the DEMs to overcome inherent problems associated with processing archival photography, most notably the identification and correction of scale error in some datasets. The analysis provides insight into one of the most challenging and data-scarce areas on the planet by expanding the spatial extent north of the AP to include previously un-studied glaciers located in the South Shetland Islands. 81% of glaciers studied showed considerable loss of volume over the period of record. The mean annual mass loss for all glaciers yielded 0.24±0.08m.w.e.peryear, with a maximum mass loss of up to 62m.w.e. and frontal retreat exceeding 2.2km for Stadium Glacier, located furthest north on Elephant Island. Observed volumetric loss was broadly, though not always, correlated with frontal retreat. The combined mass balance of all 16 glaciers yielded −1.862±0.006Gt, which corresponds to −0.005mm sea level equivalent (SLE) over the 57year observation period. •3D change results for 16 AP glaciers, including examples on South Shetland Islands•Predominance of ice loss over gain is observed in the majority of studied glaciers•On average, 82% glacier areal coverage is achieved over a six-decade period•WorldView-2 imagery allows higher resolution and more reliable comparison than before•Rigorous in-house surface matching algorithm corrects scale errors in archival data