Characterizing Attribute Distributions in Water Sediments by Geostatistical Downscaling

Information about attributes such as contaminant concentrations or hydraulic properties in benthic sediments is typically obtained in core sections of varying lengths, and only the average value is measured in each section. However, an estimate of the attribute distribution at a uniform spatial resolution is often required for site characterization and the design of appropriate risk-based remediation alternatives. Because attributes exhibit spatial autocorrelation, geostatistical methods have become an essential tool for estimating their spatial distribution. The purpose of this paper is to optimally infer the spatial distribution of sampled attributes at a uniform resolution from fluvial core sampling data, using a downscaling technique formulated as a geostatistical inverse problem. We compare geostatistical downscaling to the more traditional approach of point-to-point ordinary kriging for a hypothetical case study, and for total organic carbon observations from the Passaic River, New Jersey. Although frequently used to interpolate measurements, ordinary kriging is shown not to be able to estimate the spatial distribution of attributes accurately, because this approach assumes that data are sampled at a uniform resolution. Geostatistical downscaling, on the other hand, is shown to resolve this problem by explicitly accounting for the relationship between the known average measurements and the unknown fine-resolution attribute distribution to be estimated.