NSS Convention 2026

Sinkholes are a defining feature of karst landscapes that play a critical role in hazard management and groundwater quality. In the United States, sinkhole damages are estimated to cost at least $300 million annually. They can also act as recharge points and contaminant pathways into karst aquifers, allowing for surface contaminants to enter groundwater. Recent advancements in geospatial analysis have improved the ability to detect and map sinkholes across large areas. High-resolution digital elevation models (DEM) derived from LiDAR data are used to calculate the morphometry and spatial distribution of detected sinkhole features. Sinkhole morphometry measurements can be used to understand the evolution and dissolution processes that shape karst landscapes.

This study evaluates how different mapping approaches influence sinkhole morphometry across diverse environmental settings. We compared two nationwide sinkhole datasets for the United States that were developed using fundamentally different methods. The USGS dataset uses a flow routing algorithm, while the Mihevc dataset uses a machine learning algorithm. These approaches also differ in DEM resolution, processing tools, parameter thresholds, and artifact removal procedures. We used QGIS and Python to quantify and compare morphometric parameters within a range of karst settings. Sinkhole polygons were smoothed to reduce pixelation effects associated with DEM resolution when calculating parameters like circularity. Differences in sinkhole size, shape, and spatial distribution help identify strengths and limitations related to each method.