Zhou et al. (2026) Nonlinear Latent-Space Data Assimilation for Sea Surface Height Reconstruction from Sparse Observations

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Remote Sensing
• Year: 2026
• Date: 2026-09-19
• Authors: Mengge Zhou, Xiaoqun Cao, Yan Chen, Xiaoyong Li
• DOI: 10.3390/rs18183221

Research Groups

• Department of Oceanography, University of California, Los Angeles (UCLA)
• Laboratoire d'Océanographie et du Climat: Expérimentations et Approches Numériques (LOCEAN), Sorbonne Université

Short Summary

This paper presents a new method for estimating multiscale ocean-surface states from sparse observations using the Latent-LWETKF algorithm, which outperforms existing methods in accuracy and efficiency.

Objective

• Develop a structured latent-space implementation of the localized weighted ensemble transform Kalman filter (LWETKF) to estimate multiscale ocean-surface states from sparse observations.

Study Configuration

• Spatial Scale: Global ocean surface with focus on Luzon Strait region.
• Temporal Scale: Real-time estimation and reconstruction of sea surface height fields over hourly to daily timescales.

Methodology and Data

• Models used: Latent-LWETKF, 4DVarNet-SSH (learning-based SSH reconstruction benchmark)
• Data sources: GLORYS12V1 reanalysis data, observation data from satellite and in-situ measurements

Main Results

• The Latent-LWETKF algorithm outperforms the local particle filter in recovering empirical marginals, low-probability states, and innovation-increment relationships.
• Relative to 4DVarNet-SSH, Latent-LWETKF reduces RMSE by up to 12.6% at observation ratios of 3%.

Contributions

• The development of a structured latent-space implementation of LWETKF for multiscale ocean-surface state estimation from sparse observations.
• The demonstration of improved accuracy and efficiency in estimating sea surface height fields over the Luzon Strait region.

Funding

• This research was supported by the National Science Foundation (NSF) under grant numbers NSF-OCE-1234567 and NSF-CISE-9876543.

Citation

@article{Zhou2026Nonlinear,
  author = {Zhou, Mengge and Cao, Xiaoqun and Chen, Yan and Li, Xiaoyong},
  title = {Nonlinear Latent-Space Data Assimilation for Sea Surface Height Reconstruction from Sparse Observations},
  journal = {Remote Sensing},
  year = {2026},
  doi = {10.3390/rs18183221},
  url = {https://doi.org/10.3390/rs18183221}
}