•  
  •  
 

Abstract

The completeness and accuracy of collecting and investigating geotechnical and marine engineering geological conditions are key to the success or failure of engineering development. Traditional marine engineering and geotechnical surveys often require the assistance of geophysical methods to fully present the geological conditions. Electrical Resistivity Tomography (ERT) is an advanced geophysical method that produces accurate cross-sections of underground resistivity. Although 2D ERT has been widely used in geotechnical engineering and marine engineering, many uncertainties remain hidden behind the vivid and colorful resistivity profiles. To apply 2D ERT profiles effectively in geotechnical and marine geological surveys and monitoring, it is essential to address these uncertainties and avoid over-interpretation of the test results. Interpreting color ERT profile images is often a significant challenge for engineers. Therefore, this study aims to reprocess 2D ERT images using a pattern recognition method of cluster analysis and explores the use of image recognition technology to assist in interpreting stratigraphic boundaries or characteristic areas on the ERT profiles, thereby enhancing image analysis. To simulate real formations, this study developed seven numerical models. The results show that the ERT profiles can clearly indicate layer positions and characteristic areas through the cluster analysis method. In the field test case, this study applied ERT detection for underground concrete tunnels. The results show that the cluster analysis method can accurately define characteristic areas on the ERT profile, which significantly reduces the error of visual interpretation and avoids over-interpretation; at the same time, it eliminates many uncertainties in ERT profiles and effectively improves the spatial analysis capabilities of 2D ERT data.

Download
COinS