A Complete High-Resolution Coastline of Antarctica Extracted from Orthorectified Radarsat SAR Imagery

A complete, high-resolution coastline of Antarctica, extracted from an orthorectified mosaic of Radarsat-1 synthetic aperture radar (SAR) images through a sequence of automated image processing algorithms, is presented. A locally adaptive thresholding method is used to segment the orthorectified SAR images, while image-object formation and labeling, and edgetracing techniques are used to process the segmented images into vector-based cartographic products of coastline, defined here as the boundary between continental ice or rock exposures and sea ice covered ocean. The absolute accuracy of planimetric positioning of the resultant coastline is estimated to better than 130 m, and its spatial resolution (25 m) is adequate for supporting cartographic and scientific applications at 1:50,000 scale. This radar-image-derived coastline gives an accurate description of geometric shape and glaciological characteristics of the Antarctic coasts and also provides a precise benchmark for future change-detection studies. Introduction Coastline delineation is a prerequisite for the geographic description of the physical form of a continent. This seemingly simple task was still unfinished for the Antarctic continent until recently (Fox and Cooper, 1994). Some portions of the Antarctic coast have never been accurately mapped, and other portions have experienced dramatic changes in positions and shapes since they were mapped. Information about geographic position, orientation, and geometric shape of the Antarctic coastline is essential for geographic exploration, and nautical and aerial navigation. In addition, the area and volume of the Antarctic Ice Sheet are intricately linked to global climate, and changes in the shape and extent of the ice sheet may be diagnostic indicators of changing climate (Mercer, 1978; Williams and Hall, 1993). Constructing an accurate coastline map is an important step toward establishing a baseline for future change detection studies in order to understand the response of the Antarctic Ice Sheet to climate change. In this paper, we present a complete high-resolution Antarctic coastline, which we define as the boundary between the ocean and the continental ice sheet, including ice shelves and coastal rock outcrops. The coastline is extracted from 1997 Radarsat-1 synthetic aperture radar (SAR) image mosaic. The mosaic is compiled from individual SAR image frames that have undergone rigorous terrain correction, geocoding, and radiometric correction and balancing, and therefore A Complete High-Resolution Coastline of Antarctica Extracted from Orthorectified Radarsat SAR Imagery Hongxing Liu and Kenneth C. Jezek furnishes a planimetrically accurate image base for interpreting and delineating coastline features. A chain of image processing algorithms has been applied to the orthorectified images to automate the coastline extraction. The final coastline products in ArcInfo vector line coverage format will be provided to the National Snow and Ice Data Center and to the British Antarctic Survey for eventual distribution to the science community. We believe that this highly accurate coastline will be useful for a wide range of cartographic and scientific applications. In the following sections, we will first give an overview of the image data and summarize the techniques used to create the orthorectified SAR image mosaic. Then we describe our automated coastline extraction algorithms. Next, the relative and absolute accuracy of the extracted coastline is evaluated, and the shape property of the coastline is also analyzed. In the final section, we draw some conclusions. Image Data Source and Methodology Overview Antarctica is the coldest, windiest, and, on average, highest continent on Earth. The size of Antarctica approximates the areas of the United States and Mexico combined. The continent was first sighted in 1819, and the first landing was made in 1821 (Larsgaard, 1993). The harsh climate, inaccessibility, and long dark winters imposed serious difficulties and challenges on the conventional surveying and mapping techniques. By the end of the Second World War, more than half of the Antarctic coastline was still uncharted, and the interior was largely blank (Larsgaard, 1993). At present, the largerscale maps (1:250,000 or larger) only cover portions of Antarctica, including the Antarctic Peninsula, Transantarctic Mountains, and parts of Dronning Maud Land and Marie Byrd Land, while most parts of the continent remain poorly mapped at 1:1,000,000 or smaller scale. Severe geolocation errors of up to 30 km have been observed and reported in the existing maps (Larsgaard, 1993; Williams et al., 1995; Jezek et al., 2000). Satellite remote sensing technology overcomes the obstacles of surface travel and provides an efficient means for mapping the Antarctic continent. Optical satellite sensors, especially AVHRR and Landsat, have been used extensively in mapping and studying Antarctic continent since the late 1970s (Swithinbank, 1988; Williams et al., 1995; Ferrigno et al., 1996; Mullins et al., 1999). However, optical sensors are affected by clouds, and images are often saturated due to the P H OTO G R A M M E T R I C E N G I N E E R I N G & R E M OT E S E N S I N G May 2004 6 0 5 H. Liu is with the Department of Geography, Texas A&M University, College Station, TX 77843 ([email protected]). K.C. Jezek is with the Byrd Polar Research Center, The Ohio State University, Columbus, OH 43210 ([email protected]. ohio-state.edu). Photogrammetric Engineering & Remote Sensing Vol. 70, No. 5, May 2004, pp. 605–616. 0099-1112/04/7005–0605/$3.00/0 © 2004 American Society for Photogrammetry and Remote Sensing 03-003.qxd 4/5/04 10:48 PM Page 605