BergCAT - Iceberg census and tracking in the Weddell Sea utilizing satellite data
The Antarctic ice sheet is the world’s largest freshwater reservoir and plays an important role in the climate system.
The ice sheet mass balance, comprising mass gain (accumulation) and loss (e.g. iceberg calving and basal melting) is of
particular interest, because it influences not only the global sea level, but also the ocean currents. In Antarctica iceberg
calving amounts to up to 2/3 of the complete mass loss of the ice sheet. At present, the estimated calving rate is based only
on the contribution of large icebergs (> 10 km edge length) the effect of small icebergs on the mass balance is not known
yet.
Satellite remote sensing is the most effective way to observe iceberg calving, drift and decay in the Polar Regions.
The combination of images with different regional coverage and spatial resolutions enables the monitoring of large
regions and the detection of small icebergs at the same time.
Within the framework of this project we develop a method for detecting small icebergs in SAR images from two test
sites in the Weddell Sea, Antarctica (blue rectangles in Figure 1).
Fig.1: Regions of interest (blue rectangles) in the Weddell Sea. The shaded image is taken from MODIS Mosaic of Antarctica.
To separate icebergs from the surrounding sea ice, statistical analyses of the icebergs and their
surrounding were made. The test site at the tip of the Antarctic Peninsula is completely sea-ice free in the summer months,
which makes a detection of icebergs more complex. The algorithm we developed results in a classification of three pixel
groups: (i) iceberg, (ii) iceberg or background, and (iii) background (Fig. 2).
Fig.2: Result of iceberg detection for two example images. In (a) the result of the detection of icebergs in open-water
conditions is shown and in (b) the corresponding original image. In (c) the detection of icebergs with surrounding sea ice is
shown and in (d) the corresponding original image. The white targets in (a) and (b) are the first class (icebergs), the grey
targets are the second class, and the black areas are the third class (background).
To improve the algorithm, more information about the variability of iceberg and background signature is necessary. The seasonal
variability of icebergs and sea ice is studied in ENVISAT ASAR images at different resolutions in the north of Berkner
Island (Fig. 1).
Extracting the icebergs from the radar images eases their recovery in other images, and by comparing the location of the icebergs,
their drift can be calculated. Using SAR data, the time gap between data acquisitions may be relatively large. An iceberg drift model
including wind and ocean drag, Coriolis force and sea ice concentration is used to predict the location of an iceberg as a function of
time and hence make its tracking and identification in a sequence of SAR images easier. On the other hand, the actually observed
location of an iceberg is used to validate and improve the predictions of the iceberg drift model. With the aid of the drift model,
the paths of icebergs can be traced back to the ice shelf they calved from. The surface area of the icebergs will be determined from
the SAR images, and with a known height of the ice shelf edge, the mass loss caused by these icebergs can be calculated. The results
of our study will improve estimations of the mass balance of Antarctica.
Scientists
Christine Wesche
Wolfgang Dierking
Dr. Hertmut H. Hellmer
Alfred Wegener Institute Bremerhaven
Research areas
Weddell Sea
Publications
Wesche C, Dierking W, 2011. Iceberg detection in SAR images in two test regions of the Weddell Sea, Antarctica, Journal of Glaciology, in Review.
Wesche C, Dierking W, 2010. Separation of icebergs and sea ice in SAR images, Proceedings of ESA Living Planet Symposium, 28.6. – 2.7.2010, Bergen,
Norway / European Space Agency, In: Lacoste-Francis H (ed.), 1 CD-ROM. Noordwijk : ESA Communication Production Office (SP/European Space Agency; 686)
(ESA-SP; 686).
Talks
Wesche C, Dierking W, 2010. Eisberge und Meereis in SAR Bildern, DFG Berichtskolloquium SPP Antarktisforschung auf der 24. Internationalen Polartagung vom 6.9. - 10.9.2010 in Obergurgl, Österreich.
Wesche C, Dierking W, 2010. Separation of icebergs and sea ice in SAR images, ESA Living Planet Symposium 28.6. - 2.7.2010 Bergen, Norway.
Wesche C, Dierking W, 2010. Separation of icebergs and sea ice in SAR images, FRISP Workshop, 21. - 24.6.2010 in Bad Bederkesa, Germany.
Wesche C, Dierking W, 2009. Detection of icebergs using ERS SAR images, Cryospheric Science Workshop 4.11. - 6.11.2009 Sheffield.
Homepage
Homepage Meteorologie der Polargebiete - AWI
Research funding organisation
German Research Foundation
Project number: DI 909/3-1
Funding period: 01.12.2008 – 30.11.2008