23 Aug 2021

23 Aug 2021

Review status: this preprint is currently under review for the journal ESSD.

An inventory of supraglacial lakes and channels across the West Antarctic Ice Sheet

Diarmuid Corr1, Amber Leeson1,2, Malcolm McMillan1,3, Ce Zhang1,2,4, and Thomas Barnes1 Diarmuid Corr et al.
  • 1Centre of Excellence in Environmental Data Science, Lancaster Environment Centre, Lancaster University, Lancaster, LA1 4YQ, U.K.
  • 2Data Science Institute, Lancaster University, Lancaster, LA1 4YW, U.K.
  • 3UK Centre for Polar Observation and Modelling, Lancaster University, Lancaster LA1 4YW, UK
  • 4UK Center for Ecology & Hydrology, Library Avenue, Lancaster, LA1 4AP, U.K.

Abstract. Quantifying the extent and distribution of supraglacial hydrology, i.e. lakes and streams, is important for understanding the mass balance of the Antarctic ice sheet, and its consequent contribution to global sea level rise. The existence of meltwater on the ice surface has the potential to affect ice shelf stability and grounded ice flow, through hydrofracturing and the associated delivery of meltwater to the bed. In this study, we systematically map all observable supraglacial lakes and streams in West Antarctica, by applying a semi-automated Dual-NDWI (Normalised Difference Water Index) approach to > 2000 images acquired by the Sentinel-2 and Landsat-8 satellites during January 2017. We use a K-Means clustering method to partition water into lakes and streams, which is important for understanding the dynamics and inter-connectivity of the hydrological system. When compared to a manually-delineated reference dataset on three Antarctic test sites, our approach achieves average values for sensitivity (85.3 % and 77.6 %), specificity (99.1 % and 99.7 %) and accuracy (98.7 % and 98.3 %) for Sentinel-2 and Landsat-8 acquisitions, respectively. In total, we identified 10,478 supraglacial features (10,223 lakes and 255 channels) on the West Antarctic Ice Sheet (WAIS) and Antarctic Peninsula (AP), with a combined area of 119.4 km2 (114.7 km2 lakes, 4.7 km2 channels). 27.3 % of feature area was found on grounded ice, 17.8 % of feature area comprised lakes which crossed the grounding line, while 54.9 % of feature area was found on floating ice shelves. New continental-scale inventories such as these, the first produced for WAIS and AP, are made possible by the recent expansion in satellite data provision. The inventories provide a baseline for future studies and a benchmark to monitor the development of Antarctica’s surface hydrology in a warming world, and thus enhance our capability to predict the collapse of ice shelves in the future. The dataset is available at (Corr et al., 2021).

Diarmuid Corr et al.

Status: open (until 18 Oct 2021)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on essd-2021-257', Anonymous Referee #1, 08 Sep 2021 reply
  • RC2: 'Comment on essd-2021-257', Anonymous Referee #2, 20 Sep 2021 reply

Diarmuid Corr et al.

Data sets

Supraglacial lakes and channels in West Antarctica and Antarctic Peninsula during January 2017 Diarmuid Corr; Amber Leeson; Mal McMillan; Ce Zhang; Thomas Barnes

Model code and software

Sentinel-2 and Landsat-8 SGL and Channel Classifier Diarmuid Corr

Diarmuid Corr et al.


Total article views: 309 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
213 88 8 309 1 5
  • HTML: 213
  • PDF: 88
  • XML: 8
  • Total: 309
  • BibTeX: 1
  • EndNote: 5
Views and downloads (calculated since 23 Aug 2021)
Cumulative views and downloads (calculated since 23 Aug 2021)

Viewed (geographical distribution)

Total article views: 247 (including HTML, PDF, and XML) Thereof 247 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
Latest update: 23 Sep 2021
Short summary
We identify 119 km2 of meltwater area over West Antarctica in January 2017. In combination with Stokes et al., 2019, this forms the first continent-wide assessment, helping to quantify the mass balance of Antarctica, and its contribution to global sea level rise. We apply thresholds for meltwater classification to satellite images, mapping the extent and manually post-process to remove false positives. Our study provides a high-fidelity dataset to train and validate machine learning methods.