Exploring the Potential of Machine Learning-Assisted Spaceborne GNSS-R Measurements for Ocean Eddy Detection

This paper explores the potential of Global Navigation Satellite System Reflectometry (GNSS-R) technology for detecting ocean eddies through the retrieval of Sea Surface Height (SSH) data. The study utilizes a Specular Point (SP)-based Convolutional Neural Network (CNN) tailored for processing and analyzing spaceborne GNSS-R data to retrieve SSH. Additionally, it calculates the global Absolute Dynamic Topography (ADT) and determines the Root Mean Square Error (RMSE). By employing a gridded global map, the ADT achieves an RMSE of 0.46 meters for a day. Theoretically, with the Nyquist-Shannon sampling theorem, this demonstrates GNSS-R's potential for detecting ocean eddies with amplitudes exceeding approximately 1.0 meter. Results indicate that GNSS-R could effectively contribute to global oceanographic studies and provide valuable insights into ocean eddy detections.