Semi-Quantitative Direct-Push Data Can Improve Contaminant Delineation and Mass Discharge in Groundwater

Reliable mapping and delineation of contaminant plumes and accurate estimation of contaminant mass discharge (CMD) are critical for groundwater risk assessment and planning of remedial actions at contaminated sites. However, traditional interpolation methods are often challenged by low-density sampling resulting in improper plume delineation. This study introduces a probabilistic censoring method that enhances geostatistical interpolation by incorporating comparably cheap, high-resolution, but semi-quantitative data collected from direct push-probes in the subsurface. The method converts halogen-specific detector signals into binary presence-absence indicators, which are interpolated using indicator kriging to generate a probability field of contaminant distribution. The probability field is then used to censor a spatial concentration field derived from traditional groundwater sampling, retaining interpolated concentration values only in areas where contamination is likely. We apply the method to a site contaminated with chlorinated solvents using two datasets with different sampling densities. Results show that, using our new method, plume fringes became more clearly defined and the total area with low concentrations (<10 μg L^-1) increased by 41-85%. CMD estimates were reduced by 13-18%, while relative uncertainty remained largely unchanged. The method integrates seamlessly with traditional interpolation methods and our censoring workflow can be applied to other forms of direct-push data (e.g., relative permeability). As such, the framework offers a useful method for incorporating semi-quantitative field measurements into concentration interpolation and CMD estimation at contaminated sites.