Incorporating uncertainty, objective, and subjective data in geologic site characterization

In current geotechnical engineering practice, soil or rock stratigraphy is usually obtained from borehole data. Interpolation between boreholes is performed by projecting borehole data on a cross-sectional plane, either by hand drawings or using CAD or GIS software, followed by linear or non-linear interpolation which is based on subjective interpretation. This methodology for obtaining the geology of a site does not truly honor the three-dimensional nature of the data, and it does not capture the uncertainties present in the interpolation. This paper describes NOMAD, a three-dimensional ground profiler, which has been expanded with KRIBS. KRIBS can create ground profiles using a combination of Indicator CoKriging and Bayesian updating to capture both subjective and objective information in geologic site characterization. Details are discussed followed by an application of KRIBS to a case study in which parts of the geology of the Aberjona Valley are represented. The Aberjona watershed area, in Woburn, Massachusetts, consists primarily of glacial outwash and glacial till, overlying granodioritic bedrock; it is the location of several Superfund sites. The ground profiles created with NOMAD and KRIBS play an important part in the subsurface characterization.