A Sea-Ice Lead Detection Algorithm for Use With High-Resolution Airborne Visible Imagery

The detection of leads, or cracks, in sea ice is critical for the derivation of sea-ice freeboard from altimetric measurements of sea-ice elevation. We present an approach for lead detection in sea ice using high-resolution visible imagery from airborne platforms. We develop a new algorithm, i.e., the sea-ice lead detection algorithm using minimal signal (SILDAMS), that detects clouds, extracts leads, and classifies ice types within leads from airborne visible imagery. Cloud detection is based on an assessment of local variances of pixel brightness across image scenes and where available coincident altimetric measurements are used to confirm suspected cloudy scenes. The lead extraction step computes affine time-frequency distributions (minimal signal) for the Red, Green, and Blue channels of each image. The transformed outputs are combined to take advantage of three channels simultaneously. Finally, lead pixel geolocations are extracted using a set of uniform thresholds for ice typing (including open water, thin ice, and gray ice) within leads along each flight line. SILDAMS was tested using data from the Digital Mapping System (DMS). DMS digital photographs represent the highest resolution (approximate to 10 cm) visible imagery available over sea ice and were collected during NASA Operation IceBridge sea-ice flights in the Antarctic and the Arctic in 2009 and 2010, respectively. We demonstrate that SILDAMS has a high lead detection capability of 99%.