Comparative Evaluation of Folk-Based Seabed Classification Schemes Using MBES Data in the East Sea of Korea

다국어 초록 (Multilingual Abstract)

As coastal and marine management and marine spatial planning (MSP) gain increasing importance, the demand for detailed seabed information continues to grow. Particularly, seabed sediment classification provides essential baseline data for marine ecosystem conservation, seabed habitat mapping, and coastal development. Although most seabed sediment classifications are derived from the Folk system, region-specific and application-oriented variations have been adopted in different studies, making the selection of an appropriate classification scheme critical for a given area. The objectives of this study were to evaluate four proposed seabed classification schemes to determine the most suitable scheme for the geological and environmental conditions of the East Sea of Korea; generate a predictive seabed sediment map using the selected scheme; and qualitatively assess its spatial patterns for practical applicability in this region. In this study, the applicability of four Folk-based classification schemes was assessed in the East Sea of Korea. A total of 575 grab samples and multi-beam echo sounder (MBES) data were analyzed. A total of 23 variables were extracted from the MBES data, including bathymetry, backscatter intensity, and texture metrics derived from the grey level co-occurrence matrix (GLCM). The grab samples were classified using four Folk-based classification schemes, referred to as Scheme 1 through Scheme 4. AutoML was applied to identify and evaluate the optimal algorithm for each classification scheme. The results showed that the Random Forest model under Scheme 1 achieved the highest overall accuracy (OA) of 89.6%, while Scheme 4 demonstrated the best performance among the four schemes with an OA of 76.74%. Notably, the mud class consistently exhibited high accuracy across all schemes, with a producer’s accuracy (PA) of 76.94% and a user’s accuracy (UA) of 89.96%. This study confirms the suitability of multiple Folk-based classification schemes for the East Sea of Korea and provides valuable insights for future applications, including seabed cover mapping, MSP, and marine ecosystem management.