SEPARATING STRATIGRAPHIC LAYERS AND LOCAL FACIES IN BOREHOLE RECORDS FROM GLACIAL SEDIMENTS USING STATISTICAL ANALYSIS

Maps and three dimensional models of glacial sediments at depth are essential to hydrologic and engineering applications. However, mapping even at site scales is made challenging due to the complexity of the sedimentary deposits and the paucity of data at depth. Typically, maps and models are based on a large number of lithology logs from public water well data, supplemented with a few borings with detailed records. The essential task in developing a geologic layer model is to determine the lateral connectivity between lithologic layers in the well records. This is often difficult due to the sparse sampling of borings relative to the horizontal scale of sedimentary variability, the highly variable quality of water well records, and the typical absence of objective stratigraphic information (i.e. paleosols, clast data, biostratigraphy, ash layers etc.). A potential solution to the problem is to calculate a composite boring record through statistical combination of all the individual borings within a specific area, typically a two dimensional geologic mapping unit. The composite records show the proportion of each lithology and the variability in lithology (entropy) with depth. While spatial specificity is sacrificed, statistical analysis of multiple wells reduces the noise due to error and inconsistent methodologies, and exposes patterns that maybe used to identify stratigraphic layers. In addition, percolation theory can be used to help estimate which layers may transmit water. Several case studies from Ohio-based glacial mapping projects are presented illustrating the utility of this modeling and interpretation approach.