Mapping hazardous mining-induced sinkhole subsidence using unmanned aerial vehicle (drone) photogrammetry

Abstract

Accurate subsidence inventory data, based on an understanding of local topography, are a crucial first step toward reliable subsidence prediction and mapping future subsidence hazards. However, conventional, human-based methods of surveying and mapping subsidence suffer from data omissions and errors due to problems regarding accessibility, safety, and manual digitization. This study employed unmanned aerial vehicle photogrammetry to compile an accurate subsidence inventory map of abandoned mine areas. A Phantom 2 Vision+ drone was used, which is inexpensive yet appropriate for detailed topographic surveying of small-sized mine sites with a history of subsidence. An autonomous flight plan was designed, taking into account the extent of target mapping areas. A series of 29 aerial photographs were obtained within 2 min; digitally georeferenced orthoimage and digital terrain model (DTM) with 5 cm resolution could be obtained by processing with coordinate information of pre-installed ground control points (GCPs) within 30 min. sinkhole-type subsidence, including locational information, was identified from the geocoded high-resolution orthoimage and the DTM, and its area and volume were calculated to be 427 m2 and 2323 m3 (length 25 m, width 23 m, depth 9.1 m), respectively, from its modeled shape. Contour lines (10 cm interval), slope, and curvature were produced using the DTM. Validation using the GCP locations showed an error of approximately 14 cm in the generated DTM, which was considered acceptable for subsidence mapping purposes. The proposed approach enables accurate, rapid, low-cost, and safe surveying and mapping, which complements conventional surveying methods at sites of mining subsidence.