RIASSUNTO
To monitor dynamic changes in sea ice using SAR (Synthetic Aperture Radar) imagery in polar regions, we propose an object-oriented regional segmentation and dynamic object identification method that discriminates between sea ice and open water, classifies sea ice into different types, identifies sea-ice objects, and detects and represents dynamic changes. The results demonstrate the validity and feasibility of the proposed method.
INTRODUCTION
As a useful remote sensing technology, Synthetic Aperture Radar (SAR) has become the most important method of observing sea ice on the regional scale, independent of cloud cover or nightlight conditions (Sandven and Johannessen, 2006; Piotrovskaya, 2007; Barale and Gade, 2008), especially in the polar regions where few ground stations are located (Pettersson et al., 1999; Johannessen et al., 2008). In the past decade, several Northern Hemisphere countries, such as the United States, Canada and Russia, have used high-resolution SAR from ERS-1/2, RADARSAT-1 and ENVISAT imagery as the main data sources for sea-ice monitoring for ship traffic navigation, fisheries and offshore operations (Jackson and Apel, 2004; Scheuchl et al., 2004). In addition, several processing platforms based on high-resolution SAR imagery—such as the Radarsat Geophysical Processing System (RGPS) (Kwok, 1998), the MAp-Guided Sea Ice Classification System (MAGSIC) (Clausi et al., 2010), and Advanced Reasoning using Knowledge for Typing of Sea Ice (ARKTOS) (Soh and Tsatsoulis, 2002; Soh et al., 2004)—have been established for scientific research and operational applications. These platforms have been applied in studies of sea-ice segmentation and classification, changes in sea-ice extent and thickness distribution, and the identification and motion of sea ice, with the aim of understanding changes in sea-ice dynamics and the contribution of such changes to global environmental change (Jackson and Apel, 2004; Stroeve et al., 2007; Holland et al., 2006).