Data Summary

Bedrock Surface Topography Digital Elevation Raster of Pennsylvania

2023 - Pennsylvania Department of Conservation and Natural Resources

API
REST: https://imagery.pasda.psu.edu/arcgis/rest/services/pasda/DCNR_BedrockElevation/MapServer
WMS: https://imagery.pasda.psu.edu/arcgis/services/pasda/DCNR_BedrockElevation/MapServer/WMSServer?SERVICE=WMS&request=getcapabilities

ADDITIONAL RESOURCES
n/a

ABSTRACT
PaGS assembled 214,851 relevant well records from Pennsylvania, GroundWater Information System (PaGWIS) and other unpublished PaGS reports. Each well used in the analysis contains a measurement of the depth to bedrock (in feet) and/or a notation indicating if bedrock was encountered during well drilling, these attributes allow well records to be separated into two datasets – bedrock wells (wells that penetrate bedrock) and drift wells (wells that did not encounter bedrock).Topographic Position Index (TPI) is a quantitative landform analysis that uses land surface elevation data to determine landforms such as ridge, upper slope, middle/flat slope, lower slope, and valley. A composite TPI raster for each of Pennsylvania’s 23 physiographic sections was generated. Each well data point was attributed to a physiographic section and assigned a TPI value based on its location. The square root of depth-to-bedrock was calculated for each well. A linear regression relationship between the TPI and the square root of sediment thickness was established for five TPI classes (ridge, upper slope, middle/flat slope, lower slope, and valley) in each of the 23 physiographic sections. This statistical relationship was used to create a surrogate model for depth to bedrock to predict sediment thickness across the state. Synthetic data points were generated from the surrogate model to fill in areas of low well data density. A combination of bedrock well data points and synthetic data points were used to generate the first-iteration sediment thickness model through a natural neighbor interpolation technique. Iterative refinements to the sediment thickness model were made by comparing model predictions to drift well data points. If the total depth of the drift well was less than the predicted thickness of sediment at that location, then the drift well data point was ignored. If the total depth of a drift well was greater than the predicted thickness of sediment at that location, then the drift well data point was added to dataset and a new sediment thickness model was generated. In total, 413, 474 data points were used in the modeling process – 207,130 empirically derived well points and 206, 344 synthetic points derived from the surrogate model.The final sediment thickness model was resampled to a 100-meter grid digital raster conforming to a similar resolution surface topography digital elevation raster. The surface topography grid was smoothed to remove detail before subtracting the sediment thickness to create a bedrock elevation map. The degree of smoothing was applied proportionally to the magnitude of sediment thickness. Portions of the surface topography grid that correspond to sediment thickness greater than 365 feet received the maximum amount of smoothing; likewise, portions of the surface topography grid that correspond to zero sediment thickness received no smoothing. The remaining portions of the surface topography grid that correspond to sediment thickness between 0 and 365 feet received gradational smoothing proportional to the sediment thickness.This 100-meter grid bedrock elevation raster was calculated by subtracting the sediment thickness model from the conditionally-smoothed surface topography digital elevation raster.