Baek, W.-K.; Jung, H.-S.; and Kim, D., 2020. Oil spill detection of Kerch strait in November 2007 from dual-polarized TerraSAR-X image using artificial and convolutional neural network regression models. In: Jung, H.-S.; Lee, S.; Ryu, J.-H., and Cui, T. (eds.), Advances in Geospatial Research of Coastal Environments. Journal of Coastal Research, Special Issue No. 102, pp. 137-144. Coconut Creek (Florida), ISSN 0749-0208.

The oil spill is the main marine disaster. It is known that the data mining based method performs better in detecting oil than the traditional SAR based method to distinguish from lookalikes. Recently, artificial neural networks were employed to detect the oil spill. For that synthetic aperture radar intensity map, intensity texture map, co-polarized coherence map, co-polarized phase difference texture map, and digital elevation model was employed as input. The detection performance based on data mining largely depends on the used data, the characteristics of the study area, and the used model. Improving detection performance through various existing data mining models is a very important task since minor improvements in detection can have a significant impact on disaster damage mitigation. Artificial and convolutional neural network regression models were applied to detect the oil spill of Kerch strait in November 2007. The two models showed F1-scores of 0.832 and 0.823 respectively. The results of this study would contribute to monitor oil spill dynamic and mitigate damages caused by the oil spill.