INTERPRETING LINEAMENTS USING REMOTE SENSING AND GIS ANALYSIS FOR PREDICTING CAVE DIRECTIONS: A CASE STUDY OF THAM SAI THONG CAVE AREA, CHIANG RAI, NORTHERN THAILAND

Abstract

This study presents a semi-automated geospatial approach for predicting subterranean pathway orientations in the Tham Sai Thong cave area, Chiang Rai, Northern Thailand. High-resolution Sentinel-2 imagery and a Digital Elevation Model (DEM) derived from the ALOS satellite were utilized to objectively extract geological lineaments. Their spatial correlation with karst depressions and surveyed cave entrances was analyzed to infer probable subsurface pathways. Preprocessing involved band stacking and Topographic Position Index (TPI) enhancement, while Canny edge detection followed by the Hough Transform facilitated efficient, semi-automated lineament extraction. Although the core extraction and overlay procedures are parameterized and reproducible, parameter calibration and final depression validation involve limited expert-guided refinement. Karst depressions were identified via DEM differential analysis (fill-sinks method) and validated against topographic and satellite data. Cave entrances, accurately surveyed by GNSS, were integrated with high-density lineament zones using spatial buffering and a Multiplicative Boolean Overlay (Map Algebra). Two dominant lineament orientations—NW-SE and NNE-SSW—were observed, consistent with the regional tectonic framework and interpreted principal stress orientation associated with the Mae Chan Fault system. The predicted cave directions demonstrated high accuracy, with spatial overlaps of 93.67% with Shepton Mallet Caving Club data and 81.47% with Department of Mineral Resources maps, demonstrating the practical utility of integrated GIS and remote sensing for structured cave direction prediction in structurally controlled karst terrains.