Wang, Wei (2011) Signal Processing Methods to Improve Ocean Surface Wave Estimation from a High Frequency Surface Wave Radar. PhD thesis, University of Sheffield.
|PDF (The thesis text) |
Available under License Creative Commons Attribution-Noncommercial-No Derivative Works 2.0 UK: England & Wales.
High frequency surface wave radars are operated as a remote sensor to measure ocean surface parameters to ranges exceeding 200-300 km from the coastline. The Bragg peaks in the power spectrum of backscattered radar electromagnetic signals from ocean waves reveal the Bragg resonant effect and the second order continuum reflect a hydrodynamic and electromagnetic modulation on the radio waves from the ocean waves. The power spectrum is thus utilized to invert the ocean wave directional spectrum by a non-linear integral equation. Further integrations of the ocean wave directional spectrum yield the estimates of wave parameters: significant waveheight, mean wave period, mean direction, directional spread, etc. Beside sea echoes, non-sea echoes or interferences (collectively termed ‘clutter’), are also received by radar antenna receivers and included in the power spectrum. Clutters which occupy the second order continuum are treated as sea echoes in the inverse algorithm and cause inaccurate estimation of the ocean wave directional spectrum. Thus clutter mitigation is the main purpose of this thesis and is intensively investigated. A three-step image processing approach is proposed in this thesis which could mitigate visible clutters, e.g. radio frequency interferences and ionospheric interferences, and also invisible clutters. The kernel implements a decomposition of the mixture space and then a projection of the mixture space into a desired subspace. In addition to this main approach, various signal processing methods are also investigated for improving the wave estimates, e.g. wavelet analysis, AR modeling, adaptive filtering algorithm. The clutter mitigation scheme is validated by operational use on a whole month of Pisces data and exhibits some improvements in the accuracy of wave estimates. To aid the operational use, a statistical pattern recognition method is also developed. Finally, the best schemes are chained together for a sequential operational use in terms of providing better wave estimation.
|Item Type:||Thesis (PhD)|
|Department:||The University of Sheffield > Faculty of Science (Sheffield) > School of Mathematics and Statistics (Sheffield)|
|Deposited By:||Miss Wei Wang|
|Deposited On:||13 Jan 2012 16:17|
|Last Modified:||13 Jan 2012 16:17|
Repository Staff Only: item control page