Computer Analysis of Coastal Ocean Features in Synthetic Aperture Radar Imagery (Japan, Gotoh Islands).

Abstract

Listen

Translate article

A wide variety of oceanographic phenomena is revealed in synthetic aperture radar (SAR) ocean imagery. Four basic classes of SAR oceanographic phenomena, surface waves, fronts, wind patterns and internal waves, are of oceanographic interest. These classes are defined in light of their generation mechanisms and as a function of Bragg scattering, which accounts for their visualization in SAR ocean imagery. Digital exploitation of SAR imagery offers image enhancement and parameter extraction capabilities and is currently an area of extensive research. The state of the art in digital exploitation of SAR imagery is reviewed, with special attention to SAR speckle reduction and ocean parameter extraction. Digital processing of SAR optically processed imagery gathered over the Gotoh Islands of Japan is the focus of the dissertation research. Preprocessing is done to digitize the imagery, correct for digitization shading nonlinearities, convert the imagery from the radar slant-plane to the ground-plane, and correct for antenna range dependent power attenuation that may be present. Power spectral techniques are employed to extract wind direction and wavelength, and to estimate wave period, phase speed, wave height, and aerodynamic parameters associated with sea surface roughness through proven empirical models. Imagery with spectra and associated ocean parameters superimposed is generated digitally. Texture analysis techinques are studied to determine their applicability to SAR ocean image parameter extraction. The inertia texture measure is shown theoretically and experimentally to extract wavelength. Other texture measures are analyzed for their ability to further characterize image periodicities.