Resolving and Analyzing Landfast Ice Deformation by InSAR Technology Combined with Sentinel-1A Ascending and Descending Orbits Data

Zhiyong Wang, Jian Liu, Jinning Wang, Lihua Wang, Meng Luo, Zihao Wang, Ping Ni, Hao LiSensors 2020 · DOI

TL;DR

This paper presents a novel method using Sentinel-1A satellite data to precisely map the horizontal and vertical deformation of landfast ice in the Baltic Sea. By combining ascending and descending orbit data, the researchers can overcome the limitations of single-orbit InSAR, providing a more comprehensive understanding of ice dynamics crucial for sea ice modeling and safety.

What Problem Does This Paper Address?

Accurately monitoring and understanding the deformation of landfast ice is critical for several reasons. Landfast ice, which is attached to the coast, islands, or reefs, plays a significant role in the marine environment's thermal balance and influences ocean salinity. It's also vital for human activities like constructing ice roads for transportation in polar and high-latitude regions. However, traditional methods for measuring sea ice deformation are often difficult to implement in remote, harsh environments, and existing remote sensing techniques have limitations. Specifically, single-orbit InSAR (Interferometric Synthetic Aperture Radar) can only measure deformation along the satellite's line-of-sight (LOS), making it challenging to disentangle horizontal and vertical movements. This ambiguity can lead to inaccurate assessments of ice stability and potential hazards.

Methods

The core innovation of this study lies in its approach to resolving 3D ice deformation using multi-orbit InSAR data. The methodology involves several key steps:

  • Fast Ice Boundary Extraction: The researchers developed an automated method to identify the precise boundaries of landfast ice. This was achieved by combining interferometric coherence images (which indicate the stability of the signal) and SAR amplitude images (which show surface features). Texture features from SAR amplitude images, analyzed using the Gray-Level Co-occurrence Matrix (GLCM), were crucial for this segmentation.
  • Deformation Transformation Model: A novel model was developed to convert the one-dimensional LOS deformation measurements from ascending and descending Sentinel-1A orbits into independent horizontal (east-west) and vertical (up-down) deformation components. This model leverages the geometric relationship between the satellite's viewing angles and the deformation directions. Notably, the north-south deformation was assumed to be negligible due to the near-polar orbit of the satellite.
  • InSAR Data Processing: Standard InSAR processing techniques were employed, including baseline estimation, co-registration, speckle filtering (using the Lee algorithm), interferogram generation, phase filtering (Goldstein algorithm), and phase unwrapping (minimum cost flow algorithm).

Datasets

  • Sentinel-1A SLC (Single Look Complex) Images: Four images acquired in February 2018 were used, comprising two ascending and two descending orbit pairs. These were chosen for their VV polarization, which is considered more sensitive to sea ice properties than HH polarization.
  • ACE2 (Altimeter Corrected Elevations, v2) DEM: A Digital Elevation Model from NASA was used for InSAR processing.
  • NOAA GSHHG Database: Coastline data was used as a land mask for accurate fast ice boundary delineation.
  • Swedish Meteorological and Hydrological Institute (SMHI) Ice Charts: These were used as reference data for sea ice type classification.
  • Meteorological Data: Daily data on temperature, precipitation, and wind speed/direction from the Kemi Ajos meteorological station were collected as auxiliary information to understand deformation drivers.

Key Insights and Ideas

  • 3D Deformation Resolution: The combination of ascending and descending Sentinel-1A data is a powerful technique for disentangling horizontal and vertical landfast ice deformation, overcoming a significant limitation of single-orbit InSAR.
  • Wind as a Primary Driver: Strong southwest winds were identified as the principal cause of landfast ice deformation, driving ice floes and causing shear stress.
  • Protective Role of Consolidated Ice: The inner fast ice remained relatively stable, shielded by the presence of consolidated ice at its periphery, which acted as a buffer against external forces.
  • Automated Fast Ice Boundary Detection: The proposed method for automatically extracting fast ice boundaries using coherence and SAR amplitude texture features is a valuable contribution for efficient and accurate ice mapping.

Key Results

  • Maximum Deformation: Within a 960 km² fast ice region in the Baltic Sea between February 2nd and 16th, 2018, the study found a maximum horizontal deformation of -44 cm (east to west) and a maximum vertical deformation of 16 cm (uplift).
  • Deformation Direction: The dominant horizontal deformation was observed to be from east to west, directly correlating with the prevailing southwest winds.
  • Coherence Values: Mean coherence values were around 0.43 for one image pair and 0.39 for another, indicating moderate stability of the ice surface during the observation period. Differences in coherence were observed between sea ice and open water, and also near the shore due to grounding and ice ridges.
  • Comparison with Previous Studies: The results showed good consistency with previous InSAR studies in the Baltic Sea, with minor differences attributed to variations in study area, time span, and the presence of icebreakers or large ice cracks in prior research.

Limitations and Future Work

  • Lack of Field Measurements: The authors acknowledge the absence of ground-truth measurements (e.g., GPS, leveling) to directly validate their deformation results. This is a common challenge in remote sea ice research.
  • Accumulated Deformation: InSAR measures deformation over the time baseline of the interferograms (12 days in this study). This means it captures accumulated or counteracted movements and may not precisely reflect sudden, short-term events like ice cracking.
  • North-South Deformation Assumption: The assumption of negligible north-south deformation, while standard for near-polar orbits, could be a limitation if significant ice movement occurs in this direction.
  • Future Work: The paper suggests that future research should focus on incorporating more advanced radar technologies like TerraSAR-X/TanDEM-X and exploring three-dimensional imaging radar altimeters for more accurate and efficient sea ice parameter retrieval and deformation monitoring.

Summary updated

Abstract

Detailed mapping of landfast ice deformation can be used to characterize the rheological behavior of landfast ice effectively and to improve sea ice modeling subsequently. In order to analyze the characteristics, trends and causes of deformation comprehensively and accurately, the Sentinel-1A ascending and descending orbits data were used to detect the horizontal and vertical deformation of the fast ice in the Baltic Sea. Firstly, the fast ice edge lines were acquired through feature extraction with interferometric coherence images and SAR amplitude images. Then, the deformation transformed model was constructed according to the geometric relationship of multi-orbits deformation measurements. Finally, the landfast ice deformations were resolved and the horizontal and vertical deformations were obtained. The results showed that the maximum deformation was-44 cm in horizontal direction and 16 cm in vertical direction within the fast ice region of 960 km<sup>2</sup> during the time from 2 to 16 February 2018. The southwest wind was the principal reason for the deformation, which made the deformation mainly occur in the horizontal direction from east to west. Moreover, the inner fast ice kept stable due to the protection of outer consolidated ice. The results showed that the deformation trend and characteristics can be better understood by using InSAR technology that was combined with multi-orbits SAR data to resolve and analyze the landfast ice deformation.
BibTeX
@inproceedings{wang2020resolving,
  author = {Zhiyong Wang and Jian Liu and Jinning Wang and Lihua Wang and Meng Luo and Zihao Wang and Ping Ni and Hao Li and Ping Ni and Hao Li},
  title = {{Resolving and Analyzing Landfast Ice Deformation by InSAR Technology Combined with Sentinel-1A Ascending and Descending Orbits Data}},
  booktitle = {Sensors},
  year = {2020},
  doi = {10.3390/s20226561},
}
Figure