Using Cross-Mission SAR Data for a Multidecadal Coastline Change Monitoring and Assessing the Influences of SAR-Related Factors
Journal
IEEE Journal of Oceanic Engineering
Start Page
1
End Page
13
ISSN
0364-9059
1558-1691
2373-7786
Date Issued
2024
Author(s)
DOI
10.1109/JOE.2024.3425968
Abstract
Conservation of estuarine blue carbon (BC) environments has been a vital issue due to their carbon sequestration capacity and ecosystem services. Thus, these habitats are included in many biodiversity goals, e.g., the sustainable development goal (SDG) 14 and Ocean Decade Challenge 2. Since the shoreline is a critical indicator for assessing the management of BC systems, it is necessary to accurately extract shoreline positions and continuously monitor their change. The utilization of synthetic aperture radar (SAR) has been favored because of its cloud-penetrating ability and weather independence. Nonetheless, most studies rely on backscattered intensity and pixelwise analysis, which is greatly affected by the speckle of SAR. Also, a coarse-to-fine strategy is commonly used, which is computationally intensive and requires fine-tuning of parameters. Therefore, there is no long-term SAR-based shoreline change tracking available. The goals of this article include 1) proposing an automatic, robust, and texture-considered SAR-based shoreline extraction approach; and 2) investigating the influences of oceanic and SAR-related factors on shoreline delineation reliability. A 19-year (2003-2021) cross-mission SAR dataset is used, which is synchronized with tidal records to control the tidal effect. Different SAR-derived information, e.g., normalized radar cross section (NRCS), interferometric coherence, and polarimetric parameters, are integrated with two region-based methods to establish the optimized approach. The results indicate that the copolarization NRCS together with the morphological active contour model yields the most reasonable shorelines. By applying this strategy to a BC conservation site in southern Taiwan, a quantitative assessment of shoreline dynamics is achieved and reveals an extreme erosive rate of -40.8 m/yr at the river mouth sand banks. Lastly, by analyzing the usability of the SAR images, it is found that although there is a positive correlation between the NRCS at the ocean surface and wind speed, the radiometric quality of the SAR sensors, i.e., the noise-equivalent sigma zero and the radiometric resolution, is more decisive.
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Type
journal article