Zhou et al. (2025) Snow effects on altimeter waveforms over sea ice in the Weddell Sea — Part I: Radar waveform decomposition

Identification

• Journal: Remote Sensing of Environment
• Year: 2025
• Date: 2025-11-08
• Authors: Lu Zhou, Henriette Skourup, Julienne Stroeve, Sahra Kacimi, Stefanie Arndt, Weixin Zhu, Alek Petty, Lanqing Huang, Shiming Xu
• DOI: 10.1016/j.rse.2025.115112

Research Groups

• Institute for Marine and Atmospheric Research, Utrecht University, Netherlands
• Department of Geodesy and Earth Observation, National Space Institute, DTU Space, The Technical University of Denmark, Denmark
• Clayton H. Riddell Faculty of Environment, Earth, and Resources at the University of Manitoba, Canada
• Jet Propulsion Laboratory, California Institute of Technology, USA
• Alfred-Wegener-Institut, Helmholtz-Zentrum für Polar- und Meeresforschung, Germany
• Department of Earth System Science, Tsinghua University, China
• ESSIC, University of Maryland, USA
• Department of Earth Sciences, University College London, United Kingdom

Short Summary

This study analyzes Ku-band CryoSat-2 and Ka-band KAREN altimeter waveforms over the Weddell Sea to decompose scattering contributions from snow surface, snow volume, and ice surface. It finds that snow-volume scattering significantly contributes to Ku-band returns, often as much as or more than the snow-ice interface, while Ka-band is dominated by surface/near-surface snow scattering.

Objective

• To quantify the respective contributions of snow surface, snow volume, and ice surface scattering to Ku-band (CryoSat-2) and Ka-band (KAREN) radar altimeter signals over Antarctic sea ice.
• To examine how snow layering and wetness modulate backscattered energy across these interfaces.

Study Configuration

• Spatial Scale: Weddell Sea, Antarctica. CryoSat-2 footprint: approximately 1.65 km across-track, 305 m along-track. KAREN footprint: approximately 5 m along-track, 12 m across-track. Sentinel-1B spatial resolution: 90 m with 40 m pixel spacing.
• Temporal Scale: CryoVEx Antarctic campaign: December 2017 to January 2018. Specific KAREN flights: 2 January and 15 January 2018. Sentinel-1B acquisitions: 1, 3, and 15 January 2018.

Methodology and Data

• Models used:
  • Forward Backscatter Emulation Model (FBEM)
  • Convolutional Neural Network (CNN)
  • Levenberg–Marquardt (LM) algorithm
  • k-nearest neighbors (KNN) classification algorithm
• Data sources:
  • Satellite radar altimeter: CryoSat-2 (CS-2) L1b waveforms and L2 data (Ku-band, ~13.6 GHz).
  • Airborne radar altimeter: MetaSensing Ka-band radar altimeter (KAREN) Level 1c data (34.525 GHz) from CryoVEx campaign.
  • Satellite Synthetic Aperture Radar (SAR): Sentinel-1B (S1B) EW Level-1 GRD MR Product (C-band, 5.405 GHz).

Main Results

• Under typical Antarctic conditions, snow-volume scattering contributes as much as, or more than, the snow–ice interface to Ku-band CryoSat-2 returns.
• Ka-band KAREN signals are dominated by surface/near-surface snow scattering with minimal penetration to the ice surface.
• Wet snow amplifies upper-layer backscatter and increases signal attenuation, reducing effective penetration depth. Wet snow conditions result in higher skewness, kurtosis, and pulse peakiness, and reduced stack standard deviation, suggesting more specular-like waveforms.
• Sensitivity tests identify volume scattering and ice-surface roughness as primary controls on waveform shape.
• The CNN regression model achieved a mean Normalized Root Mean Square Error (NRMSE) of approximately 0.30 for retrieving scattering parameters and snow depth.
• For CS-2, mean snow-ice interface backscatter (𝜎0 𝑠−𝑖) is around -2.6 dB, comparable to snow volume backscatter (𝜎0 𝑣−𝑠) at approximately -3.8 dB, and both are stronger than snow surface backscatter (𝜎0 𝑠−𝑠) at approximately -5.7 dB from CNN decomposition.
• For KAREN, snow surface backscatter (𝜎0 𝑠−𝑠) at approximately -2.5 dB is slightly stronger than snow volume backscatter (𝜎0 𝑣−𝑠) at approximately -4.1 dB, while ice surface backscatter (𝜎0 𝑠−𝑖) at -8.3 dB is comparatively weak.

Contributions

• Provides a physics-based decomposition of Ku-band and Ka-band altimeter waveforms over Antarctic sea ice, explicitly quantifying contributions from snow surface, snow volume, and ice surface scattering.
• Highlights the significant and often underestimated role of snow volume scattering in Ku-band altimetry over Antarctic sea ice, challenging the common assumption of dominant snow-ice interface returns.
• Demonstrates the utility of a Convolutional Neural Network (CNN) for efficient and robust decomposition of radar waveforms, reducing dependency on iterative fitting and initial assumptions.
• Offers insights into the impact of snow wetness and surface roughness on radar backscatter, informing future altimeter retrieval schemes.
• Supports the development of dual-frequency strategies for upcoming missions like ESA’s CRISTAL by providing a diagnostic framework for interpreting waveform variability.

Funding

• INTERAAC project (co-funded by the National Key Research and Development Program of China, project no. 2022YFE010670, and the Research Council of Norway, grant no. 328957).
• National Natural Science Foundation of China (project no. 42030602).
• International Partnership Program of Chinese Academy of Sciences (grant no.: 183311KYSB20200015).

Citation

@article{Zhou2025Snow,
  author = {Zhou, Lu and Skourup, Henriette and Stroeve, Julienne and Kacimi, Sahra and Arndt, Stefanie and Zhu, Weixin and Petty, Alek and Huang, Lanqing and Xu, Shiming},
  title = {Snow effects on altimeter waveforms over sea ice in the Weddell Sea — Part I: Radar waveform decomposition},
  journal = {Remote Sensing of Environment},
  year = {2025},
  doi = {10.1016/j.rse.2025.115112},
  url = {https://doi.org/10.1016/j.rse.2025.115112}
}