PulseTrendingJournal ClubResearchersJournalsExplore
Instagram
HomeTrendingJournal ClubExplore
Synapse
⌘+K
Synapse
April 3, 2026Remote SensingOpen Access

Physics-Aware Hybrid CNN–Transformer Network for GNSS-R Sea Surface Wind Speed Estimation

View Full Paper
Ask AI
Bookmark
Share

Authors

BABaiwei AnWQWeiwei QinWKWeijie Kang

Discussion

Loading...

Member takes

Overview

Hybrid neural network estimates sea surface wind speed using GNSS-R data, suggesting improved accuracy and efficiency.

Key Points

  • The research aims to enhance the accuracy of wind speed retrieval from GNSS-R data by addressing scattering complexity and model limitations.
  • Developed a Physics-Aware Hybrid CNN–Transformer Network (PA-HCTN) for wind speed estimation.
  • Used CNN for local feature extraction from delay-Doppler maps (DDMs).
  • Integrated Transformer encoder for modeling global context and cross-attention module for fusing auxiliary physical parameters.
  • Incorporated geophysical model function-constrained loss to improve physical consistency.
  • Achieved an RMSE of 1.35 m/s and an R2 of 0.75 in wind speed retrieval.
  • Outperformed existing models in retrieving wind speeds, especially under high-wind conditions.
  • Validated effectiveness using independent NDBC buoy data from multiple sites.

Cite This Study

An et al. (2026) studied this question.

synapsesocial.com/papers/69cf5d4e5a333a821460b093https://doi.org/10.3390/rs18071053
View Full Paper
Ask AI
Bookmark
Share