Ship Signatures in RADARSAT-1 ScanSAR Narrow B Imagery: Analysis with AISLive Data
Abstract
RADARSAT-1 ScanSAR Narrow B (SCNB) mode imagery at 50 m (nominal) spatial resolution of high density shipping regions were acquired along with contemporaneous Automatic Identification System (AIS) data via an AISLive snapshot service. Within the images are 1805 signatures of isolated ships ranging from 40 m to 350 m in length, for which AIS data, including ship identification, size, and velocity, are available. The AIS data were verified through an authoritative ship database, which identified errors in the static AIS data. Dynamic errors were also characterized since pairs of location observations were available. The mean error between the AIS-predicted ship position and the ship signature position (124 m) is smaller than the average ship length in the overall data set (147 m). The data were used to model the radar cross section (RCS) of ships in SCNB data as a simple geometric function of the ship length. The variability about the normalized RCS, a consequence of ship type, local incidence angle, ship aspect angle, and sea state, could not be described by a single probability density function, but a Log-Normal distribution was appropriate for certain ship length ranges. Ship RCS variability observed in the SCNB data is consistent with that observed previously in Fine mode data, but there is an overall scaling differing between the SCNB and Fine mode data sets. This work demonstrates that the fusion of RADARSAT image signatures of ships with AIS data is straightforward and provides unprecedented insight to observed ship signatures. However, the AIS data, at least the static fields, should be routinely verified prior to fusion with other sources of ship information.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2007
- Accession Number
- ADA640083
Entities
People
- John Wolfe
- Paris W. Vachon
- Ryan A. English
Organizations
- Defence Research and Development Canada