Computational Oil-Slick Hub for Offshore Petroleum Studies

The paper introduces the Oil-Slick Hub (OSH), a computational platform to facilitate data visualization of large database petroleum signatures observed on surface ocean with synthetic aperture radar (SAR) measurements. This Internet offers an information search and retrieval system resulting from >20 years scientific projects that interpreted ~15 thousand offshore mineral oil “slicks”: natural “seeps” versus operational “spills”. Such Digital Mega-Collection Database consists satellite images oil-slick polygons identified in Gulf Mexico (GMex) Brazilian Continental Margin (BCM). A series attributes describing slicks are also included, along technical reports papers. Two experiments illustrate use OSH selection subsets mega collection (GMex variables BCM samples), which artificial intelligence techniques—machine learning (ML)—classify into seeps or spills. GMex variable dataset was analyzed simple linear discriminant analyses (LDAs), three-fold accuracy performance pattern observed: (i) least accurate subset (~65%) solely used acquisition aspects (e.g., beam mode, date, time, name, etc.); (ii) best results (>90%) were achieved inclusion location (i.e., latitude, longitude, bathymetry); (iii) moderate performances (~70%) reached using only morphological area, perimeter, perimeter area ratio, etc.). sample six traditional ML methods, namely naive Bayes (NB), K-nearest neighbors (KNN), decision trees (DT), random forests (RF), support vector machines (SVM), neural networks (ANN), most effective algorithms per were: RF (86.8%) for Campos, Santos, Ceará Basins; NB (87.2%) Campos Santos SVM (86.9%) (iv) (87.8%) Basin. can assist different concerns (general public, social, economic, political, ecological, scientific) related exploration production activities, serving as important aid discovering new exploratory frontiers, avoiding legal penalties oil-seep events, supporting oceanic monitoring systems, providing valuable environmental studies.