We examined three publicly available datasets, including EnviroStor, GeoTracker, and the US EPA’s Superfund Data and Reports from DTSC, RWQCB, and the EPA, respectively, to identify contaminants of concern at each location. These datasets provide detailed records of site contamination, monitoring activities, and remediation efforts. In addition, guidance, input, and insight from our community partners (Greenaction for Health and Environmental Justice), academic researchers, and state government agency experts helped us refine the study area and more accurately model the potential movement of VOCs from contaminated sites to nearby buildings. Most data included in this repository is derived from publicly available sources. Parcel-specific data was not included in this repository as it is publicly available elsewhere. Publicly available data is linked to the original source website and can be downloaded there. 

Particle Flow Model

The San Francisco Bay Region’s groundwater basins are comprised of aquifer materials ranging from unconsolidated fill to fractured metamorphic rock complexes and surficial geologic features (e.g., paleochannels, alluvial fans) that influence the movement of groundwater across a gradient. To model the hydrologic fate of dissolved VOCs in groundwater originating within site parcel boundaries, we used MODPATH 7 (Pollock, 2016). The particle tracking was based on high-resolution (10 m x 10 m) one-layer, steady-state groundwater flow models conducted previously to quantify unconfined groundwater responses to sea-level rise (Befus et al., 2020). These models used a homogeneous and isotropic hydraulic conductivity of 1 m/day and a Bay constant head set to the mean higher high water (MHHW) tidal datum. No groundwater pumping or other remediation activities (e.g., enhanced drainage or impermeable barriers) were included in these models. For particle tracking, each site parcel identified within the San Francisco Bay Area was seeded with one particle per model grid cell, entering the flow model via recharge at the top of the model. All particles were allowed to flow until either a strong sink or a discharge location led to the particle leaving the model, and the San Francisco Bay was set as a secondary stop condition for particles. MODPATH 7 calculates sub-grid scale particle trajectories, such that the computed particle trajectories can include multiple vertices within a single groundwater cell (Pollock, 2016).

Sewer Connectivity Model

Using particle flow models developed for 21 sites in MODPATH, we created a spatial model that identifies the potential transport of VOCs through sewer systems and into buildings. This spatial model was built in GIS software (ESRI ArcGIS Pro 3.2.2). Since sewer line spatial data is not publicly available, we used roads as proxies (OpenStreetMap). Potential VOC pathways through the sewer line were drawn outwards from the point of intersection of each road and the modeled flow path from the parcel of origin.

Beckley and McHugh (2020) identified the distance contaminants may travel through sewer systems based on elevation gradients, providing a maximum uphill distance of 228.6 meters (750 feet) and a downhill distance of 685.8 meters (2250 feet) to represent the farthest distance that VOCs might travel through a sewer line (Beckley & McHugh, 2020). Z-values for elevation were identified at 76.2-meter (250 feet) intervals, starting from 0 and extending up to 685.8 meters from the intersection of flowlines with streets. Elevation data was obtained from USGS 10-meter resolution DEM topographic layers. If the difference in the Z-value between 0 and 228.6m was negative, indicating a downhill slope, our model extended the potential VOC travel distance 685.8m. If the difference in the Z-value between 0 and 228.6 meters was positive, indicating an uphill slope, the model limited the potential VOC travel distance to 228.6 meters.

Identifying Potentially Exposed Buildings and Their Uses

To identify buildings potentially impacted by vapor intrusion from VOC-containing sewer systems, we used 2020 parcel data to map buildings in proximity to sewer lines. Since sewer laterals typically connect buildings to main sewer lines, we created a 30m buffer around the potentially exposed sewer lines to account for the typical distance between sewers and buildings.

Using the 30-meter buffer, we identified the structures located within the zone of potential VOC exposure from the sewer line. The buildings were classified according to their primary use as defined in 2020 parcel data acquired from LandVision. The building uses of primary interest were residential buildings, schools, and daycares. In addition, the total number of potentially impacted structures was determined for each of the 17 individual and 4 combined sites. Residential buildings, schools, and daycares were of particular interest as they may house populations especially vulnerable to the health effects of VOC inhalation (Kuang et al., 2021; Madaniyazi et al., 2022).

Given the ongoing redevelopment of residential areas in the San Francisco Bay Area, we included priority development areas and previously permitted developments as defined by the Association of Bay Area Governments (ABAG). Lastly, we included the CalEnviroScreen percentile score for each site.

Data Management

Data was managed using three primary software platforms: RStudio, Python, and ArcGIS Pro. RStudio (R 4.3.1) was used to quantify the number of potentially exposed buildings. Python was used to automate the process of modeling contaminant movement through groundwater. ArcGIS Pro was used to model sewer intersections, identify buildings occupied by susceptible populations, and identify the social vulnerability of parcels intersecting those locations.