Data Mining of Satellite-Based Measurements to Distinguish Natural From Man-Made Oil Slicks at the Sea Surface in Campeche Bay (Mexico)

Campeche Bay, located in the Mexican portion of the Gulf of Mexico, has a well-established activity engaged with numerous oil rigs exploring and producing natural gas and oil. The associated risk of oil slicks in this region – that include oil spills (i.e. oil floating at the sea surface solely attributed to man-made activities) and oil seeps (i.e. surface footprint of the oil that naturally comes out of the seafloor reaching the surface of the ocean) – leads Pemex to be in a continuous state of alert for reducing possible negative influence on marine and coastal ecosystems. Focusing on a monitoring strategy, a multi-year dataset (2008-2012) of synthetic aperture radar (SAR) measurements from the RADARSAT-2 satellite is used to investigate the spatio-temporal distribution of the oil slicks observed at the surface of the ocean in the Campeche Bay region. The present study is an exploratory data analysis that seeks to discriminate between these two possible oil slick types: oil seeps and oil spills. Multivariate data analysis techniques (e.g. Principal Components Analysis, Clustering Analysis, Discriminant Function, etc.) are explored to design a data-learning classification algorithm to distinguish natural from man-made oil slicks. This analysis promotes a novel idea bridging geochemistry and remote sensing research to express geophysical differences between seeped and spilled oil. Here, SAR backscatter coefficients – i.e. sigma-naught (σ^o), beta-naught (β^o), and gamma-naught (γ^o) – are combined with attributes referring to the geometry, shape, and dimension that describe the oil slicks. Results indicate that the synergy of combining these various characteristics is capable of distinguishing oil seeps from oil spills observed on the sea surface to a useful accuracy.