Estimating the position and variability of buried bedrock surfaces in the St. Louis metro area

Abstract

The precise position of bedrock surface is the most important variable for evaluating seismic site response. The subsurface data suggest that the bedrock elevations in the St. Louis area are proportional to ground surface elevation, and bedrock depths tend to thin in dissected, loess-covered uplands. Automated contour maps of the estimated bedrock depth or elevation are commonly constructed using geographical information systems (GIS) that employ various interpolation algorithms. In deeply incised terrain interpolation techniques often make erroneous predictions because they tend to under- or overestimate surfaces influenced by paleolandscapes, such as incised channels that are subsequently filled, or by employing single contour models across areas of differing geomorphic settings. This paper compares two different models for estimating dissected and/or eroded bedrock surfaces beneath the greater St. Louis area, which varies with the geomorphic setting. These models include: 1) the depth-to-bedrock derived from ordinary kriging, and, 2) bedrock elevations derived from cokriging. Cross-sections derived from the GIS programs suggest that the estimated depth-to-bedrock tend to simplify the actual situations because they assume near-constant thickness between data points, ignoring natural undulations caused by previous erosion. These simplified surfaces of depth-to-bedrock were adjusted by considering data, which does not pierce the bedrock interface. The estimated bedrock elevation does not conform to local topographic variations in rugged, hilly terrain, and tends to over-smooth the natural undulations caused by stream incision. In rugged, deeply incised terrain the interpolation of bedrock depths yields more realistic estimates of the buried bedrock surface than those derived from bedrock elevations. In the major Holocene floodplains, we developed a technique employing complex curve-fitting of channel cross-sections to produce more realistic estimations of the spatial variability of depth-to-bedrock within the deeply-incised channels.