Formatted Title:

Integrating Geostatistical Interpolation Methods and Site Conditions to Define Remedial Action Areas for the Lower Duwamish Waterway Upper Reach

Background/Objectives:

The Lower Duwamish Waterway (LDW) Superfund Site is a 5-mile-long channelized waterway and the main marine industrial corridor in Seattle, Washington. The sediment remedy is being designed and constructed in three separately managed reaches (upper, middle, and lower), with remediation construction of the 2-mile upper reach (river miles 3.0 to 5.0) scheduled to begin in fall 2024. Polychlorinated biphenyls (PCBs) are the primary chemicals of concern (COCs) in the waterway, accounting for nearly 90% of the remedial action level (RAL) exceedance area footprint in the upper reach; therefore, PCBs were the focus of advanced geostatistical interpolation. Further, the U.S. Environmental Protection Agency’s RALs apply only to near-surface sediments (e.g., 0 to 10 and 0 to 45 centimeters) and, thus, defining RAL exceedance areas required 2D interpolation. This presentation describes the interpolation phasing and integration of engineering considerations to develop remedial action areas (RAAs), which represent the actual areas for remedial construction.

Approach/Activities:

Following evaluation of various geostatistical methods, indicator kriging (IK) was selected as the interpolation method for delineating RAL exceedance areas for PCBs. IK has the advantage of mapping probabilities of RAL exceedances so that boundary uncertainties can be addressed, either with additional data or engineering designs. Thiessen polygons were used to interpolate areas of other contaminants with RAL exceedances. Data were collected in three separate pre-design investigation (PDI) phases to refine geostatistical interpolation after each PDI phase and provide better data gap identification to inform each subsequent data phase. After all geostatistical interpolation was completed, engineering, site physical conditions, and interpolation uncertainty considerations were then applied to develop RAAs. This approach resulted in a greater level of confidence in capturing RAL exceedances in the engineering design. The design factors varied for each remedial technology (i.e., dredging, capping, and enhanced natural recovery) and included depth of contamination, constructability, site physical conditions (e.g., geotechnical stability and offsets from structures and utilities), U.S. Army Corps of Engineers federal navigation channel maintenance needs, adjacent site conditions (including remediated early action areas, upland site cleanups, site uses and infrastructure, and habitat areas), and review of other available information related to the conceptual site model.

Results/Lessons Learned:

The uses of a phased PDI data collection approach and IK as the primary interpolation method to identify areas of RAL exceedances addressed regulatory concerns about interpolation uncertainty by providing a greater degree of confidence through several interpolation refinements. The interpolation process, including refinements, resulted in RAL exceedance areas comprising 7.6 acres of the upper reach. Following application of engineering considerations, site physical condition, interpolation uncertainty, and constructability factors, the resulting total RAA acreage was 15.1 acres and reflects a conservative remedial design approach that further increases the degree of confidence that the remedial action will achieve its goals. Geostatistical interpolation is a powerful tool that can be integrated with traditional engineering design considerations to manage uncertainties and address regulator concerns during remedial design.

Menu

Katy Gross

Biography