Enhanced methods for surface current estimation from X-band radar data

Li, Yi (2025) Enhanced methods for surface current estimation from X-band radar data. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (12MB)

Abstract

Estimating ocean surface currents accurately is crucial for a wide range of applications, including marine navigation, environmental monitoring, and coastal management. Traditional methods for measuring surface currents face challenges such as limited spatial coverage and sensitivity to environmental noise, making the development of more accurate and robust techniques a pressing need in oceanography. This thesis focuses on improving the accuracy and robustness of ocean surface current estimation using X-band radar image sequences by introducing two novel algorithms. In the first part of this thesis, a Symmetry of Doppler Shifts (SDS) method is introduced for retrieving surface current information from radar images. The method focuses on extracting the wave angular frequencies and corresponding wavenumbers from the radar image sequences. Then, Doppler shifts are calculated based on wavevectors that exhibit symmetry with respect to the origin in the wavenumber plane. These Doppler shifts are used to estimate both the speed and direction of surface currents. Simulations with synthetic data show that the SDS method achieves a root mean square error (RMSE) of 0.13 m/s for current speed and 1.4° for direction. The results indicate that the method performs with accuracy comparable to existing techniques under simulated conditions. The second part of this research builds on the SDS method by integrating it with an enhanced Polar Current Shell (PCS) algorithm. The improvements include the application of Kernel Density Estimation (KDE) for noise filtering, interquartile range filtering to remove outliers, and symmetry-based noise reduction. The modified PCS method also employs a single curve-fitting process, analyzing all wavenumbers in the PCS domain collectively rather than individually. The improved algorithm was validated with both simulated data and real-world radar data from a Decca radar (2008) and a Koden radar (2017). Results show that the modified PCS method reduces the RMSE for speed by 0.06 m/s and direction by 3.8° for the Decca radar, and by 0.02 m/s for speed and 4.6° for direction for the Koden radar, compared to the original PCS method.

Actions (login required)

View Item