Assessing risk in groundwater contaminated sites: The importance of preserving subsurface connectivity and integrating multiple data sources
Abstract
Industrial activities often produce accidental leaks of hazardous substances that enter the groundwater and risk human health and the environment. The fate of these contaminants is strongly controlled by Subsurface Architecture (SA), that is, the multiscale heterogeneity of the geologic formations and their connectivity features (e.g., karstic cavities). Even though the influence of SA on the physics of contaminant transport has been widely studied, its role on risk assessment has received less attention and deserves a fresh perspective in light of modern modeling tools. With this in mind, we performed a comparative study of different techniques for representing SA in groundwater models and their implications for the outcome of health risk assessments. We focus on a reactive chlorinated solvent moving through a highly heterogeneous aquifer with well-defined connectivity patterns discharging to a river. In this scenario, we perform a statistical analysis of risk due to human exposure by direct ingestion considering two key metrics: (a) the peak concentration and (b) the maximum exposure duration. Specifically, we used the synthetic system to generate synthetic observations of permeability (K), hydraulic head (H), and solute concentration (C). We then used different geostatistical techniques to create plausible SA configurations via: (i) homogeneous zonation of K, (ii) Gaussian sequential simulation, (iii) single normal equation simulation, and (iv) normal score ensemble Kalman filter. Our results indicate that preserving the connectivity of the aquifer is the main factor influencing the environmental metrics. The selection of the geostatistical method to approximate the original SA is a key factor that could negatively affect the representation of the aquifer's connectivity and, consequently, also negatively affects the value of the risk assessment. In this regard, we show that reliable risk assessment analyses require an accurate simulations of SA connectivity by integrating K, H, and C data, such as the normal score ensemble Kalman filter. We also show that the configuration of the sampling network conditions our ability to capture SA connectivity because it influences the data conditioning of models. This work has important implications for remediation of contaminated sites throughout the Nation.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.H13B..07A
- Keywords:
-
- 1829 Groundwater hydrology;
- HYDROLOGY;
- 1831 Groundwater quality;
- HYDROLOGY;
- 1832 Groundwater transport;
- HYDROLOGY;
- 1847 Modeling;
- HYDROLOGY