Frontiers in Environmental Science (Sep 2020)
Feasibility Study for the Application of Synthetic Aperture Radar for Coastal Erosion Rate Quantification Across the Arctic
Abstract
The applicability of optical satellite data to quantify coastal erosion across the Arctic is limited due to frequent cloud cover. Synthetic Aperture Radar (SAR) may provide an alternative. The interpretation of SAR data for coastal erosion monitoring in Arctic regions is, however, challenging due to issues of viewing geometry, ambiguities in scattering behavior and inconsistencies in acquisition strategies. In order to assess SAR applicability, we have investigated data acquired at three different wavelengths (X-, C-, L-band; TerraSAR-X, Sentinel-1, ALOS PALSAR 1/2). In a first step we developed a pre-processing workflow which considers viewing geometry issues (shoreline orientation, incidence angle relationships with respect to different landcover types). We distinguish between areas with foreshortening along cliffs facing the sensor, radar shadow along cliffs facing away and traditional land-water boundary discrimination. Results are compared to retrievals from Landsat trends. Four regions which feature high erosion rates have been selected. All three wavelengths have been investigated for Kay Point (Canadian Beaufort Sea Coast). C- and L-band have been studied at all sites, including also Herschel Island (Canadian Beaufort Sea Coast), Varandai (Barents Sea Coast, Russia), and Bykovsky Peninsula (Laptev Sea coast, Russia). Erosion rates have been derived for a 1-year period (2017–2018) and in case of L-band also over 11 years (2007–2018). Results indicate applicability of all wavelengths, but acquisitions need to be selected with care to deal with potential ambiguities in scattering behavior. Furthermore, incidence angle dependencies need to be considered for discrimination of the land-water boundary in case of L- and C-band. However, L-band has the lowest sensitivity to wave action and relevant future missions are expected to be of value for coastal erosion monitoring. The utilization of trends derived from Landsat is also promising for efficient long-term trend retrieval. The high spatial resolution of TerraSAR-X staring spot light mode (<1 m) also allows the use of radar shadow for cliff-top monitoring in all seasons. Derived retreat rates agree with rates available from other data sources, but the applicability for automatic retrieval is partially limited. The derived rates suggest an increase of erosion at all four sites in recent years, but uncertainties are also high.
Keywords
