Quantitative Assessment of Salt Coverage in Seismic Images Using Connected Component Labeling

Abstract

Quantifying subsurface salt structures plays a critical role in petroleum exploration and seismic interpretation, as salt bodies influence hydrocarbon migration, trap formation, and seismic wave propagation. While most existing research has focused on semantic segmentation of salt regions using deep learning architectures such as U-Net, Attention U-Net, and PINN-U-Net, limited work has addressed the quantitative measurement of salt coverage and distribution across large seismic datasets. In this work, we present a detailed quantitative salt coverage analysis using Connected Component Labeling (CCL) applied to ground-truth masks from the TGS Salt Identification Challenge dataset. The algorithm systematically identifies distinct salt regions and measures their individual and cumulative areas. From seismic masks, an average salt coverage of 24.79% was observed, with a standard deviation of 31.83%, reflecting high variability in salt distribution. This approach not only quantifies salt extent but also establishes a framework for linking salt morphology with depth, paving the way for data-driven geophysical interpretation.