A Risk-Based Approach for Karst Geohazard Prediction in Central Luconia using the DBSCAN Algorithm

Summary This study presents a risk-based approach for karst geohazard prediction in Central Luconia, focusing on carbonate reservoirs in Cycle IV-V limestone. Karst formations pose significant challenges in subsurface exploration due to their complex geometry and impact on reservoir properties. To address this, we integrate seismic interpretation, deep learning technique, and clustering analysis (DBSCAN) to assess karst-related risks. The methodology involves processing 3D seismic data, extracting key seismic attributes, and applying DBSCAN clustering to identify high-risk zone based on the spatial distribution of karst features. A risk cube visualization is utilized to enhance the assessment, incorporating voxel-based volume calculations to quantify risk levels. The results demonstrate that DBSCAN effectively delineates karst features with varying risk intensities, providing valuable insights into their spatial distribution. This approach offers a systematic framework for karst risk assessment, which can be further refined for carbon storage evaluations and geohazard mitigation strategies.