COMPARING LIDAR AND LEGACY DIGITAL ELEVATION MODELS TO QUANTIFY TOPOGRAPHIC CHANGE IN AREAS OF MOUNTAINTOP REMOVAL COAL MINING, MCDOWELL AND PIKEVILLE QUADRANGLES, KENTUCKY

A number of studies have linked mountaintop removal coal mining (MRCM) with adverse human health effects, water quality deterioration, and habitat loss. Quantifying the extent of mountaintop removal and associated valley filling is thus an important step in assessing current effects and predicting possible future consequences of MRCM. Remotely sensed spectral imagery has been used to document changes in the spatial extent of MRCM over time, but does not provide topographic or volumetric information. Comparison of the current 1.5 m statewide LiDAR-based DEM to a re-projected and interpolated legacy DEM from the 1/3 arc-sec (approximately 10 m) National Elevation Dataset for an area covering the McDowell and Pikeville 7.5-minute quadrangles in eastern Kentucky shows that simple subtraction produces a noisy elevation difference map reflecting known MRCM footprints in addition to many smaller areas of apparent but likely non-existent topographic change. A sample of 1000 random points from an inferred no-change area within the difference map yielded a symmetric and near-normal distribution of errors with a mean of +0.99 m, a standard deviation of ±6.35 m, a minimum of -43.9 m, and a maximum of +59.7 m. Ideally, all of those values would be zero. After adjustment for the non-zero mean, the 95% confidence threshold for real topographic change between the two DEMs is at least ±18 m. Further investigation of patches of calculated change in areas without obvious signs of MRCM, but in which many small underground mine entrances are shown on historical topographic maps, shows that the calculated DEM differences are more likely a consequence of localized horizontal errors in the NED DEM rather than mining. Prominent topographic features such as ridge crests and slope toes differed by 30 m to 45 m horizontally between the (presumably more accurate) LiDAR and NED DEMs in non-MRCM areas of calculated change, but showed virtually no horizontal difference in non-MRCM areas with little or no calculated change. Ongoing research includes development of methods to correct non-stationary horizontal errors in the legacy DEM to reduce the amount of erroneously calculated topographic change.