Exploratory data analysis of synthetic aperture radar (SAR) measurements to distinguish the sea surface expressions of naturally-occurring oil seeps from human-related oil spills in campeche bay (gulf of Mexico)

We are thankful to Pemex and to MDA Geospatial Services for making available the high-quality RADARSAT dataset, as well as to the three anonymous reviewers that greatly contributed to improve the manuscript with thoughtful criticism. The Human Resource Program (PRH) of the National Agency of Petroleum, Natural Gas, and Biofuels (ANP) provided the financial support for this research.

An Exploratory Data Analysis (EDA) aims to use Synthetic Aperture Radar (SAR) measurements for discriminating between two oil slick types observed on the sea surface: naturally-occurring oil seeps versus human-related oil spills—the use of satellite sensors for this task is poorly documented in scientific literature. A long-term RADARSAT dataset (2008–2012) is exploited to investigate oil slicks in Campeche Bay (Gulf of Mexico). Simple Classification Algorithms to distinguish the oil slick type are designed based on standard multivariate data analysis techniques. Various attributes of geometry, shape, and dimension that describe the oil slick Size Information are combined with SAR-derived backscatter coefficients—sigma-(σ_o), beta-(β_o), and gamma-(γ_o) naught. The combination of several of these characteristics is capable of distinguishing the oil slick type with ~70% of overall accuracy, however, the sole and simple use of two specific oil slick’s Size Information (i.e., area and perimeter) is equally capable of distinguishing seeps from spills. The data mining exercise of our EDA promotes a novel idea bridging petroleum pollution and remote sensing research, thus paving the way to further investigate the satellite synoptic view to express geophysical differences between seeped and spilled oil observed on the sea surface for systematic use.