Mapping karst depressions and sinkholes in anthropogenically influenced areas

Abstract

Abstract A comprehensive sinkhole inventory is crucial for accurately assessing sinkhole susceptibility. The presented study introduces a semi-automatic approach for mapping potential sinkholes through a geomorphometric analysis of digital elevation models complemented by statistical analysis. This approach aims to augment current sinkhole inventories in quantity and quality. We chose the distinct karst terrain along the southwestern edge of the Harz Mountains in Lower Saxony as our focal area. The genesis of most sinkholes in this area is linked to subrosion processes within the evaporitic Zechstein Group, predominantly driven by the hydration of anhydrite to gypsum followed by its steady dissolution. The region exhibits diverse karst formations, ranging from bare and slightly covered karst to a fold-block landscape where resilient strata overlay subrosion zones. Human influence has also significantly reshaped this landscape over the past millennium, introducing numerous artificial structures and posing challenges to fully automated sinkhole detection. We employed a statistically refined landform classification based on geomorphons to identify localities of potential sinkhole formations, followed by a masking process to refine our results. In analyzing the distribution of these potential sinkholes, we took into account various geological and environmental factors, such as different types of karst, fault lines, and land cover, while purposefully excluding anthropogenic features from our analysis. The results revealed previously undocumented sinkholes in areas where the overburden layers, particularly in the Buntsandstein formation, are increasingly thick. The proposed approach proves to be particularly adept at detecting sinkholes in regions where they are isolated or have not been extensively surveyed yet.