, 205
Surface expression analysis of LiDAR elevation data has emerged as a powerful prospecting tool for cavers, offering the ability to identify candidate entrances and karst features before committing to fieldwork. However, the accuracy of automated detection methods has remained inconsistent, and the technical barrier to performing this analysis has kept it out of reach for most grottos and individual explorers.
We present two advances addressing both challenges. First, a detection pipeline calibrated against over 11,000 verified cave entrances across Tennessee and Alabama, with scoring weights derived empirically from how well each method ranks known caves rather than set by intuition. Validated results show 75% of known entrances detected within 50m of a generated candidate, with approximately 59% appearing in the top 10 ranked results for their search area. We discuss where the pipeline succeeds, where it fails, and what the failure modes reveal about the limits of terrain-based detection across different geologies and entrance types.
Second, these methods have been packaged into a web-based platform that requires no GIS software, no programming knowledge, and no data acquisition skills. A caver draws a search area on a map and receives a ranked list of GPS coordinates to investigate. This removes the primary barrier that has kept LiDAR prospecting a niche capability, and opens it to the broader caving community.
Together, these advances represent a step toward making LiDAR-based cave prospecting both more reliable and more democratic.
Zach Englebert