Acoustic and Geomorphologic Characterization of Submarine Seeps - A Geoclassification Approach
Abstract
Seep morphologies related to the global phenomenon of submarine hydrocarbon expulsion remains poorly constrained despite growing interest in academia, governmental institutions, and industry. The Exploration Division of Fugro USA Marine, Inc. is a global leader in the identification and characterization of submarine hydrocarbon seeps using multibeam echosounder (MBES) data integrated with seismic data and geochemical coring. Using publicly available data, we qualitatively classified seeps based on observed morphologic differences. MBES derivatives such as water column acoustic anomalies, slope, backscatter intensity, seafloor rugosity, and regional geologic structure can be used to quantify the acoustic and morphologic signatures of different seep types. These indicators can then be evaluated to determine if any relationships to potential genetic controls such as underlying fluid migration style or the origins/geochemical composition of hydrocarbon sources exist.
Quantifying the acoustic and geomorphic signatures associated with hydrocarbon seepage will allow us to apply machine learning concepts to seep identification. Machine learning classifiers have recently gained traction in marine exploration applications [e.g. McClinton et al., 2012, and Suresh et al., 2015]. We utilize MBES bathymetric, backscatter, and water column datasets to develop deep learning classifiers that can statistically differentiate between seep types. An adaptive neuro-fuzzy inference system (ANFIS) combines a structural neural network for robust learning capability with the flexibility of fuzzy logic [Jang, 1993]. The structure of neural networks mimics the biological neural system with nodes connected by pathways to exchange information. Classification using fuzzy statistics rather than discrete statistics yields a probability that any given point falls within a certain class [Zadeh, 1965, 1973]. The probability generated by fuzzy classifications provides a user-defined uncertainty threshold that reduces risk in hydrocarbon exploration endeavors. The definition and quantification of acoustic and geomorphic parameters is essential to construct a robust ANFIS that can accurately differentiate between seep types.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFMOS33D1926M
- Keywords:
-
- 1910 Data assimilation;
- integration and fusion;
- INFORMATICSDE: 3045 Seafloor morphology;
- geology;
- and geophysics;
- MARINE GEOLOGY AND GEOPHYSICSDE: 3094 Instruments and techniques;
- MARINE GEOLOGY AND GEOPHYSICSDE: 3099 General or miscellaneous;
- MARINE GEOLOGY AND GEOPHYSICS