HR: 0830h
AN: OS11A-07    [Abstracts]
TI: Lava Morphology Classification of a Fast-Spreading Ridge Using Deep-Towed Sonar Data: East Pacific Rise
AU: * Meyer, J
EM: jmeyer@geol.sc.edu
AF: Dept. of Geological Sciences, Univ. South Carolina , Columbia, SC 29208 United States
AU: White, S
EM: swhite@geol.sc.edu
AF: Dept. of Geological Sciences, Univ. South Carolina , Columbia, SC 29208 United States
AB: Classification of lava morphology on a regional scale contributes to the understanding of the distribution and extent of lava flows at a mid-ocean ridge. Seafloor classification is essential to understand the regional undersea environment at midocean ridges. In this study, the development of a classification scheme is found to identify and extract textural patterns of different lava morphologies along the East Pacific Rise using DSL-120 side-scan and ARGO camera imagery. Application of an accurate image classification technique to side-scan sonar allows us to expand upon the locally available visual ground reference data to make the first comprehensive regional maps of small-scale lava morphology present at a mid-ocean ridge. The submarine lava morphologies focused upon in this study; sheet flows, lobate flows, and pillow flows; have unique textures. Several algorithms were applied to the sonar backscatter intensity images to produce multiple textural image layers useful in distinguishing the different lava morphologies. The intensity and spatially enhanced images were then combined and applied to a hybrid classification technique. The hybrid classification involves two integrated classifiers, a rule-based expert system classifier and a machine learning classifier. The complementary capabilities of the two integrated classifiers provided a higher accuracy of regional seafloor classification compared to using either classifier alone. Once trained, the hybrid classifier can then be applied to classify neighboring images with relative ease. This classification technique has been used to map the lava morphology distribution and infer spatial variability of lava effusion rates along two segments of the East Pacific Rise, 17 deg S and 9 deg N. Future use of this technique may also be useful for attaining temporal information. Repeated documentation of morphology classification in this dynamic environment can be compared to detect regional seafloor change.
DE: 3035 Midocean ridge processes
DE: 3045 Seafloor morphology and bottom photography
DE: 3094 Instruments and techniques
SC: Ocean Sciences [OS]
MN: 2005 Joint Assembly