研究目的
To detect icebergs/growlers using satellite SAR scenes and compare the results with near coincident ground-based radar measurements and in-situ observations for validation.
研究成果
The study demonstrates the potential of using ground-based radar and GPS measurements to cross-validate iceberg detection from high-resolution C-band polarimetric SAR data, improving reliability and accuracy in polar regions.
研究不足
Time delays between GPS localizations and radar acquisitions may affect validation, especially for drifting icebergs. Ground-based radar has higher spatial resolution in near range, which might detect small ice features missed by SAR. The study is limited to a specific region and time period.
1：Experimental Design and Method Selection:
The study involved a fieldwork campaign in Kongsfjorden, Svalbard, using satellite SAR (RADARSAT-2 quad-pol images), ground-based radar (GPRI system), and in-situ GPS measurements for cross-validation of iceberg detection. The methodology included automatic detection algorithms for SAR data and adaptive thresholding for ground-based radar data.
2：Sample Selection and Data Sources:
Data were collected from April 15-19, 2016, covering icebergs, sea ice, and open water in Kongsfjorden. Satellite data included four RS-2 scenes, ground-based radar provided high temporal resolution images, and GPS positions of 17 icebergs served as ground truth.
3：List of Experimental Equipment and Materials:
RADARSAT-2 satellite, Gamma Portable Radar Interferometer (GPRI) ground-based radar, GPS devices for in-situ measurements.
4：Experimental Procedures and Operational Workflow:
The GPRI was installed on a hill, scanning the fjord every 2 minutes. Satellite images were acquired with specific incidence angles. GPS data were collected by circum-navigating icebergs. Data processing included calibration, multilooking, geocoding, segmentation, and thresholding for detection.
5：Data Analysis Methods:
Used unsupervised segmentation (EM algorithm) for SAR data, adaptive thresholding for radar data, and connected component labeling. Detection accuracy was assessed by comparing results from different sensors.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
