Analysis of high-resolution satellite images for iceberg detection in the Barents Seas

Iceberg observations in the Northern and Eastern Barents Sea have been studied by use of high-resolution satellite images, supported by in situ observations from field expeditions by Arctic and Antarctic Research Institute in 2005 and 2006. Satellite images used in the studies include optical images from Landsat and Terra ASTER and Synthetic Aperture Radar (SAR) images from ENVISAT and RADARSAT. Images with resolution of about 15 m have been used, including alternating polarisation SAR images from ENVISAT. Since most of the icebergs in the Barents Sea are small, less than 100 m in horizontal extent, it was important to use relatively high resolution. Optical images give more reliable identification of icebergs of size 50 m and more compared to SAR, because SAR has high-frequency speckle noise disturbing the iceberg detection. On the other hand, optical images are limited by cloud and darkness, and the Barents Sea region is cloud cover most of the time. SAR can provide good data independent of cloud and light conditions, but iceberg observations are ambiguous and need to be confirmed by other observations. Iceberg detection depends also on the background conditions. Icebergs occur in (1) open water, (2) drifting sea ice, and (3) in fast ice near coasts and in archipelagos. In fast ice, studies showed that icebergs of 50 m or more could be well observed in optical and SAR images.
In open water, SAR images can identify icebergs under moderate to low wind conditions, while optical images can be used in cloud-free conditions. The size of icebergs observable in open water was not studied in this project, but previous studies suggest that 100 m large icebergs are observable by the present satellites. Icebergs located in drifting ice are the most difficult to observe, because the background sea ice can have similar signature as the iceberg itself. The detection capability in all three situations can be improved by repeated use of satellite images. The project has reviewed data from previous Russian aircraft surveys to get an overview of the temporal and spatial distribution of icebergs. The interannual variability in iceberg distribution can be very large, and there is need for a monitoring system to control the extent and drift of icebergs. Scenarios for monitoring of icebergs are recommended where frequent coverage by satellite images is supplemented by aircraft and ship observations. Also tagging of selected icebergs by GPS/ARGOS transmitters will be an important part of a monitoring system. New high-resolution satellite images, both SAR and optical, will be available in near future. These data will improve the possibility to monitor icebergs from space.