Correlation analysis of high-resolution ocean microwave radiometric images

Recent remote sensing studies have shown that a high-resolution passive microwave imaging technique enables detailed observations of ocean variables. Multi-channel sets of ocean microwave radiometric images (OMIs) were obtained using the NOAA airborne Polarimetric Scanning Radiometer (PSR) that operated at five frequencies: 10.7, 18.7, 21.5, 37, and 89 GHz and two polarizations. Radiometric features such as mosaic spot-type structures were extracted digitally from the OMIs using enhanced image processing methods. In this paper, we explore their statistical characteristics including power 2-D Fourier spectra and both covariance and correlation matrices. Our analysis has shown that the correlations can be observed at low and mid-spatial frequencies where the ocean radiometric features are revealed in the form of quasi-periodic brightness variables (hot and cold spots). Geophysical interpretation based on the random field models and a hydrodynamic-electromagnetic approach allows us to explain the nature of stochastic ocean microwave image textures. We suppose that the low-contrast radiometric features are associated mostly with the spatial variability of the ocean surface roughness and the near-surface wind. Results of the statistical image processing and related hypotheses are presented and discussed