Using SAR images to delineate ocean oil slicks with a texture-classifying neural network algorithm (TCNNA)

Abstract

Satellite-borne synthetic aperture radar (SAR) data are widely used for detection of hydrocarbon resources, pollution, and oil spills. These applications require recognition of particular spatial patterns in SAR data. We developed a texture-classifying neural network algorithm (TCNNA), which processes SAR data from a wide selection of beam modes, to extract these patterns from SAR imagery in a semisupervised procedure. Our approach uses a combination of edge-detection filters, descriptors of texture, collection information (e.g., beam mode), and environmental data, which are processed with a neural network. Examples of pattern extraction for detecting natural oil seeps in the Gulf of Mexico are provided. The TCNNA was successful at extracting targets and rapidly interpreting images collected under a wide range of environmental conditions. The results allowed us to evaluate the effects of different environmental conditions on the expressions of oil slicks detected by the SAR data. By processing hundreds of SAR images, we have also found that the optimum wind speed range to study surfactant films is from 3.5 to 7.0 m·s–1, and the best incidence angle range for surfactant detection in C-band is from 22° to 40°. Minor postprocessing supervision is required to check TCNNA output. Interpreted images produce binary arrays with imbedded georeference data that are easily stored and manipulated in geographic information system (GIS) data layers.