Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Spatiotemporal variability of tetrachloroethylene in residential indoor air due to vapor intrusion: a longitudinal, community-based study


The migration of volatile contaminants from groundwater and soil into indoor air is a potential health threat at thousands of contaminated sites across the country. This phenomenon, known as vapor intrusion, is characterized by spatial and temporal heterogeneity. This study examined short-term fluctuations in concentrations of tetrachloroethylene (PCE) in the indoor air of residential homes due to vapor intrusion in a community in San Antonio, Texas, that sits atop an extensive, shallow plume of contaminated groundwater. Using a community-based design, we removed potential indoor sources of PCE and then collected twelve 3-day passive indoor air samples in each of the 20 homes. Results demonstrated a one-order-of-magnitude variability in concentration across both space and time among the study homes, although all measured concentrations were below risk-based screening levels. We found that within any given home, indoor concentrations increase with the magnitude of the barometric pressure drop (P=0.048) and humidity (P<0.001), while concentrations decrease as wind speed increases (P<0.001) and also during winter (P=0.001). In a second analysis to examine sources of spatial variability, we found that indoor air PCE concentrations between homes increase with groundwater concentration (P=0.030) and a slab-on-grade (as compared with a crawl space) foundation (P=0.028), whereas concentrations decrease in homes without air conditioners (P=0.015). This study offers insights into the drivers of temporal and spatial variability in vapor intrusion that can inform decisions regarding monitoring and exposure assessment at affected sites.


Volatile organic compounds (VOCs) are often found at higher concentrations indoors compared with the outdoor environment.1, 2 VOCs are capable of migrating from contaminated groundwater through overlying soil and building foundations, resulting in vapor-phase contaminant intrusion into indoor air.3, 4 Tetrachloroethylene (PCE), a VOC, is among the most frequently detected groundwater contaminants at hazardous waste sites in the United States.5, 6 The inhalation of vapors inside homes is an understudied field, but prior research suggests it may be an important pathway by which communities at hazardous waste sites are exposed to chlorinated VOCs (CVOCs) in groundwater.7, 8, 9, 10 Long-term exposure to CVOCs has been linked to cancer, kidney and liver disease and reproductive problems, such as pregnancy loss, developmental abnormalities and low birth weights.11, 12, 13, 14 Elevated rates of cancers, low birth weights, fetal growth restrictions, and cardiac defects have been reported at sites with CVOC vapor intrusion, although causality has not been established.15, 16, 17 Owing to these potential health risks and the frequency of PCE detection in contaminated groundwater, the potential for PCE exposure via vapor intrusion is an important consideration when making decisions regarding groundwater remediation.

Spatial and temporal variability has been observed in subslab and indoor air concentrations of CVOCs above contaminated groundwater plumes (e.g., Folkes et al.18, McDonald and Wertz19, Schreuder20, McHugh et al.21, Luo et al.22). Variability across space and time has also been observed in indoor radon concentrations, which also result from vapor intrusion (albeit from natural geologic sources rather than from anthropogenic contamination).23, 24, 25 Hence, an indoor air sample from a single point in space and time is unlikely to reflect community-scale exposure to vapor intrusion risks. Furthermore, previous work suggests that groundwater concentrations are not adequate surrogates for measuring vapor intrusion exposure potential, because variability in soil and household characteristics can lead to houses above relatively low groundwater PCE concentrations having higher PCE levels in indoor air than homes overlying higher concentrations, and vice versa.18, 26 In site assessments, often, only a single 24-h indoor air sample is taken from a small number of homes in an affected community, although Environmental Protection Agency (EPA) and state guidance recommend that multiple samples be collected from a single home following a multi-tiered approach to vapor intrusion investigations.27, 28 For example, in the EPA’s National Vapor Intrusion Database, the sampling frequency was as follows: single-point-in-time sample in 84% of buildings, 2 samples collected in 10% of buildings, 3–5 samples in 5% of building and more than 5 samples in 1% of cases. Collecting one or two samples, as is the current common practice, will not account for the potential spatial and temporal variability and may under or over estimate the true exposure risk. An inaccurate characterization of exposure may result in inaccurate human health risk assessments.

Previous work has helped describe the mechanisms governing vapor intrusion and potential causes of variability. Pressure-driven flow is an important mechanism for gas entry into homes.26, 29 Building underpressurization, changes in barometric pressure, wind, and diurnal fluctuations in temperature all can influence indoor–outdoor pressure differentials and hence vapor flow into homes.30, 31, 32 When these processes lead to negative building pressure (i.e., outdoor pressure greater than indoor pressure), the rate of vapor intrusion increases. However, the inter-relationships among these factors are complex, and the net effects on vapor intrusion are difficult to predict. For example, in some cases, higher wind speeds have been associated with lower indoor radon concentrations, while in others no relationship between wind speed and indoor radon concentrations has been observed.29, 33, 34, 35

Further understanding of the spatiotemporal drivers of vapor intrusion is needed to inform decisions about the extent of indoor air monitoring necessary to adequately estimate exposure risks in communities overlying contaminated groundwater. Yet, indoor air monitoring is intrusive, and residents can be resistant to allowing researchers or government personnel into their homes.36 Due in part to this challenge, other studies of temporal variability have focused on a single home rather than multiple homes, and studies of spatial variability have been able to collect only one or two 24-h samples in each home.

This study addresses the need for community-wide assessment of spatiotemporal variability in vapor intrusion risks. The study, the first of its kind in the southern United States, integrated longitudinal and cross-sectional data collection at a contaminated site adjacent to the former Kelly Air Force Base in southwest San Antonio, Texas. We examined the effects of household characteristics and meteorological conditions on observed fluctuations in indoor air PCE concentrations to determine whether changes in (a) meteorological conditions, (b) soil type, (c) groundwater concentration and (d) household characteristics significantly explain spatiotemporal variability in indoor PCE concentrations attributable to vapor intrusion. A better understanding of the drivers of temporal and spatial variability in vapor intrusion can inform decisions regarding monitoring and exposure assessment in affected communities.

The case study site is a low-income neighborhood overlying extensive plumes of CVOCs in groundwater emanating from the former Kelly Air Force Base. These plumes extend five miles to the southeast of the base and underlie 30,000 homes. The shallow groundwater lies 1–12 m below the homes. PCE concentrations in the groundwater range from 1 μg/l to 200 μg/l in the residential areas. Off-base groundwater remediation began in 2004 and is ongoing.

The EPA evaluated a cohort of 24 houses for vapor intrusion in May 2008 and February 2009. During the May sampling event, the EPA collected one or two samples beneath each home’s foundation, outdoor air samples in selected locations, and a single indoor sample in a subset of homes. The sampling protocol followed EPA method TO-15, in which 6-liter collection devices known as summa canisters (in this case, with a PCE detection limit of 0.14 μg/m3) are deployed to capture an air sample later analyzed in a laboratory.37 For homes in which indoor air was tested, the EPA verified that all indoor sources had been removed by scanning each home with a real-time trace atmospheric gas analyzer. Figure 1 shows the results for PCE for the sampling events. The indoor air concentrations ranged from non-detectable to 1.83 μg/m3. Ambient air sampled for PCE averaged 0.055 μg/m3. The elevated subslab concentrations of PCE (4–600 μg/m3), along with the very low outdoor PCE concentrations, provide one line of evidence suggesting that PCE vapors are migrating from the groundwater into homes.

Figure 1

Summary of EPA’s previous subslab, crawl space and indoor air measurements for PCE.

As this previous sampling was carried out only on a single day, the study design and results were not sufficient to evaluate the temporal variability in PCE concentration across the community. We previously modeled the scope of indoor air contamination in the community by employing a stochastic house-by-house approach based on the Johnson–Ettinger algorithm to account for variability and uncertainty in the parameters that influence vapor intrusion potential.38 This modeling study estimated that PCE concentrations may exceed screening levels (0.41 μg/m3 at the time of the analysis) in up to 72% of the homes, demonstrating potential vapor intrusion risk and highlighting specific neighborhoods that may be at higher risk. This present study was conducted to obtain field data to further explore the spatiotemporal variability suggested by our previous stochastic model.

Materials and methods

We sampled indoor air for PCE in 20 homes over a 12-day period during summer (July–August) 2011 (Figure 2). We resampled nine of the homes over another 12-day period in winter (February–March) of 2012. For the winter period, we divided the homes into those with evidence of vapor intrusion (at least one detection above 0.25 μg/m3) and those without. We randomly selected six homes from those that showed evidence of vapor intrusion and an additional three from the homes with no detectable PCE.

Figure 2

Map showing the location of the 20 testing sites (ovals, black ovals for homes sampled in winter) and the PCE concentrations in the underlying groundwater plume.

Indoor Air Sources Identification

PCE is a commonly used solvent that is contained in many common consumer products. A survey of indoor homes unaffected by vapor intrusion found a median PCE level of 0.9 μg/m3,39 and previous work has identified common consumer products containing PCE.40 To identify and remove potential confounding indoor PCE sources, our study team sampled each home in real time with the Hazardous Air Pollutants on Site (HAPSITE) field portable GC/MS (Inficon, Syracuse, NY, USA) before the deployment of the passive samplers. Previous work has used the HAPSITE as an effective tool to identify and remove indoor CVOC sources.41 Before deploying the HAPSITE, we asked residents about the presence of common household products that could contain PCE and explained the importance of removing these sources for the duration of the study. We then asked to examine storage spaces for automotive, cleaning, and home repair supplies. We removed liquid/spray spot cleaners (15 homes), automotive lubricants (8 homes), and strong advesives or shoe glue (4 homes). No participant was known to use dry cleaning services or work in the dry cleaning industry. We also removed certain types of air fresheners because we found they interfered with the HAPSITE analysis. Next, within each home, we conducted an area-by-area investigation with the HAPSITE device to identify any additional potential vapor sources. After cleaning the concentrator, we collected a 5-min air sample that the HAPSITE automatically analyzed for PCE (detection limit, 0.18 μg/m3). Any detected household PCE sources were removed for the duration of the study period, and the air was resampled 3 h after removal to assure that no confounding sources remained. Sampling for vapor intrusion began 24 h later. In a randomly selected subset of 10 homes, the indoor air was reanalyzed with the HAPSITE on the fifth day of the study to evaluate whether confounding sources had been reintroduced into the home. We found no evidence of additional PCE sources during the mid-study resampling. In 18 of the homes, residents had a detached storage shed or garage, whereas two homes had no garage. These detached structures were not evaluated or included in the analysis.

Indoor Air Sampling

During the summer sampling event, we collected a total of eight duplicate samples (16 total samples) over a period of 12 days in each of the 20 study homes. For the winter sampling period, four sample pairs per home were collected sequentially over a 12-day period in February and March. A total of 392 samples were collected (186 paired measurements). Passive monitoring devices were shipped to the field site, and duplicate field blanks were included in each sampling season. In each case, samplers were left in place for three days to ensure sufficient detection sensitivity. Duplicate samples were taken to help assure the quality of the collected data and avoid losing information if a device was mishandled. Figure 3 shows an example sampling schedule.

Figure 3

Example data collection schedule per house.

We deployed the indoor air monitoring devices on a simple, freestanding apparatus (constructed for this study) that enabled the sampling tubes to be hung in the breathing zone, 1.5 m (4.5 ft) above the floor. We located the samplers on the ground floor in an unused room, if available, or otherwise in a location where the monitors were less likely to be disturbed.

The sampling protocol followed ISO 16017-2:2003 (Indoor, Ambient and Workplace Air-Sampling and Analysis of Volatile Organic Compounds by Sorbent Tube/Thermal Desorption/Capillary Gas Chromatography—Part 2: Diffusive Sampling). Consistent with this protocol, the sampling devices were small (6.35 mm diameter × 89 mm height), stainless steel tubes packed with an engineered adsorbent, Chromosorb-106, with a demonstrated affinity for chlorinated solvents. Beacon Environmental Services (Bel Air, MD, USA) thermally conditioned the samplers and shipped them to the study site. Previous studies have shown that these devices provide results comparable to those of summa canisters.42

Consistent with ISO 16017-2:2003 and also with US EPA Method TO-17 guidelines for sorbent samplers, Beacon Environmental analyzed the resulting samples with a Markes International thermal desorption system coupled with an Agilent 7890 Gas Chromatograph/ 5975 Mass Spectrometer (TD-GC/MS).43 The concentration was calculated from the measured mass, exposure duration, and sorbent tube uptake rate for PCE (0.46 ml/min). The Beacon Environmental laboratory’s reporting limit is 0.25 ng per sampling tube, yielding a detection limit of 0.13 μg/m3. All field sample measurements were below the analytical system’s upper calibration limit of 5.0 ng; therefore, no sample dilutions were required. The continuing calibration verification values for the system check compounds were all within ±20% of the true values. Laboratory method blanks were run with each sample batch to identify contamination present in the laboratory. In addition, laboratory control samples were included with each of the analytical batch samples and included the PCE compound. The average recovery rate of PCE for these samples was 95%.

In total, 392 total samplers were collected, with two lost due to leaks and sample handling errors. Five additional samplers were not used because the second internal standard was outside of the control limit, resulting in 385 observations. In these cases, we used only a single measurement to assign concentration. For all other cases, we averaged the two duplicate samples to estimate the measured concentration for a total of 186 distinct observations. On an average, the relative percentage difference was 8.1% among duplicate samples that exceeded the detection limit.

Model Covariates

Meteorological data were acquired from the weather station at the former Kelly Air Force Base, which is within a 1.0–4.5-km radius of the study homes. Hourly data for temperature, wind speed and humidity were averaged for the appropriate time period (based on the start and stop time for each sampler). Barometric pressure generally follows a diurnal cycle. The daily pressure drop was calculated as the difference between the crest and subsequent trough of the curve for each cycle (determined from hourly pressure measurements), with the first cycle commencing at the time of sampler deployment. These daily pressure drops were then averaged over the 3-day exposure time for each sample tube. Information on chemical groundwater concentrations was acquired for April 2011 through the Kelly Air Force Base Semi-Annual Compliance Plans (1998–2011) from the Air Force Real Property Agency. Concentrations were interpolated from 900 monitoring wells using a Bayesian Maximum Entropy approach (see refs 38, 44). All homes were located within 480 m of a monitoring well, with the majority of homes within 100 m of a well. While groundwater concentrations exhibit temporal changes, for this analysis the value was assumed to be constant for the study period. Groundwater depth was not included because the temporal resolution of the data was insufficient to allow such an analysis. The soil type beneath each home was determined from the Bexar County Soil Survey. For all homes, the identified soil type was either Houston black clay or Lewisville silty clay.45

Information was collected daily from participants about use of air conditioners, fans, and windows. These data were consolidated into a binary variable based on whether the home used any air conditioning. In all homes that used air conditioning, the windows were kept closed. As open windows and cooling systems were strongly collinear, we only included the air conditioning variable in the model. Only one of the 20 homes used a dryer inside the home (others used dryers located in detached structures), so this variable was not considered in the model. Information on the age and square footage of each home was acquired from the Bexar County Appraisal District.

Community-Based Design

In partnership with a local community organization, the Committee for Environmental Justice Action (CEJA), we designed the research question, chose appropriate methods, recruited participants, and collected data. In this study, community cooperation was especially important, because the sampling protocol necessitated accessing participating homes on a daily basis over the sampling period and that participants adhere to the removal of products that may confound the results. To recruit participants, CEJA representatives circulated flyers describing the study. If a community member responded, CEJA arranged a meeting with a member of our study team, who then offered additional information about the data collection process. One participant in each household helped collect information on heating, cooling and mechanical ventilation type for each home, and completed an activity diary for each day of the study. The activity diary asked about use of products that might affect indoor air PCE concentrations and/or transport of vapors from the subsurface into the home (e.g., use of cleaning products, mechanical cooling devices, windows, and clothes dryers).

Statistical Analysis

We employed a longitudinal multivariate regression modeling approach to examine the temporal associations between the observed indoor PCE concentrations (dependent variable) in each home and barometric pressure drop, wind speed, and other meteorological characteristics. The form of the regression model was chosen to account for the detection limit (0.13 μg/m3) of the sampling device as well as the longitudinal nature of the data collection. Typical techniques, such as exclusion of data, the assignment of one-half the detection limit to non-detects, or the substitution of a value randomly selected from an appropriate distribution have been shown to bias parameter estimates and, in the case of the latter approach, bias the variance.46, 47 To avoid such biases, we used the Tobit model, an extension of the probit analysis developed by Tobin48, which has been proven to provide an unbiased maximum likelihood approach for analyzing measurement data with detection limits.48, 49 As the distribution of observed concentrations was right-skewed, we used a log-transformed dependent variable. Further details and equations are provided in the Supplementary Information.

In this analysis, we employed clustered robust estimates of SE. To account for repeat and overlapping observations in each home and for the relatively small sample size to estimate the SE on each β coefficient, we employed a method that is robust to serial autocorrelation and performance is good across a variety sample sizes (see Arellano50, Kezdi51, Hansen52). Stata IC (Version 12) was used for statistical analyses, with an a priori significance level of 0.05.

To analyze the influence of changes in meteorological conditions on the within-home variation over time, a distinct intercept was modeled for each home in the regression, so that time-invariant characteristics would not bias the model. To investigate the variation between homes, we evaluated a pooled population average that examined both time-varying meteorological variables and time-invariant household characteristics.


Table 1 shows the characteristics and minimum and maximum PCE concentrations measured in the 20 homes (all of which completed the study in its entirety). Figure 4 shows the detailed results for each house. PCE was detected in 12 out of the 20 homes. The average PCE concentration across all samples above the detection limit was 0.28 μg/m3 (Table 2); however, concentrations fluctuated as much as one order of magnitude (Figure 5).

Table 1 Characteristics of the 20 homes included in this study.
Figure 4

Indoor air concentration of PCE by study home.

Table 2 Summary statistics for the key continuous variables included in the regression model.
Figure 5

Temporal variation in indoor PCE concentrations in homes with at least one sample above the detection limit.

In general, the measured concentrations were low, although about half exceeded the EPA Region 6 risk-based screening level for resident air of 0.33 μg/m3 that was in place at the time of sampling. In April 2012, the EPA revised its PCE screening level to 9.4 μg/m3, which is higher than all of the concentrations observed in this study. Nonetheless, the previous EPA analyses showing elevated subslab PCE concentrations and extremely low ambient concentrations (Figure 1) suggests that vapor intrusion may be an important source of the PCE observed in these homes, particularly because we removed indoor sources before sampling. (As an additional check on potential indoor sources, we also examined the correlations between measured PCE concentrations and self-reported days when cleaning products were used, but none of these correlations were significant.) Neither field blanks nor laboratory method blanks samples had any measurable concentrations of PCE. Despite the low PCE concentrations observed in this study, the results nonetheless provide valuable new information on factors both within and between homes that influence variability in indoor PCE concentrations at sites affected by vapor intrusion.

Within-Home Temporal Variability

Table 3 shows the results of the regression model examining the effects of weather variables on within-home temporal variability in PCE concentrations. As shown, barometric pressure drop, wind speed, relative humidity and season all significantly predict the observed temporal variations. Specifically, indoor concentrations increase with magnitude of the pressure drop (P=0.048) and humidity (average marginal effect, P<0.001), while concentrations decrease as wind speed increases (P<0.001) and during winter (P=0.001). As noted above, a similar relationship between wind speed and indoor concentration has been observed in previous studies of radon.29, 33, 34 A recent, detailed study of a single house at a vapor intrusion site did not observe strong correlations between seasonal winds and vapor intrusion but suggested that wind may contribute to short-term vapor intrusion changes.53 In the model, humidity may be capturing some of the short-term effects of rainfall, such as groundwater rise. During rain events the humidity levels exceed 90%, whereas the average during the study period was 62%. In summary, for this community based on the regression analysis, it is expected that PCE concentrations, in homes that are prone to vapor intrusion, will be higher during summer, during low-wind events, or when large barometric pressure drops occur.

Table 3 Average marginal effects for the within-home variability of log PCE indoor air concentration (log-μg/m3) due to vapor intrusion.

Spatial (between-home) variability

Table 4 shows the regression model exploring the effects of household, environmental, and meteorological characteristics on between-home variability. The model results include all samples, including those from homes in which no indoor PCE was detected. As shown, indoor air PCE concentrations increase with groundwater concentration (expressed as a logarithmic term, P=0.030), a slab-on-grade foundation (P=0.028), magnitude of the barometric pressure drop, (P=0.036) and humidity (expressed as a quadratic term, P=0.04). On the other hand, concentrations decrease in the absence of an air-conditioning unit (presumably because windows are opened, P=0.015) and with wind speed (P=0.004). Although not statistically significant, larger homes tended to have higher indoor PCE concentrations, whereas lower PCE concentrations were measured in older (and presumably leakier) homes. Together, all of these included covariates are highly significant (P<0.001).

Table 4 Population-averaged effects of model covariates on between-home (spatial) variability of log PCE indoor air concentration (log-μg/m3) due to vapor intrusion.


The indoor air concentrations observed in this study were similar to the results previously found in the EPA investigation (Figure 1). The highest concentration measured by the EPA summa canister (1.83 μg/m3) was on par with the highest observations in this study (1.50 μg/m3). We do however observe a short-term temporal variability in indoor air concentrations that cannot be captured with a single-point-in-time sampling event. The relationship between meteorological conditions and indoor PCE observed here could help indicate the potential range of concentrations when only a single measurement is possible.

Several previous studies have found relationships between meteorological variables and vapor intrusion similar to those observed here. Radon studies have shown that atmospheric pressure drops contribute to the total radon entry rate into a building and can increase indoor concentrations by a factor of two over a daily timescale.54, 55 Although atmospheric pressure fluctuations do not produce a net positive flow rate into the homes over longer time intervals, they cause short-term changes in radon entry because of increases in the pressure differential between the subslab and indoor air.54 That is, indoor air responds more quickly to an ambient pressure change than subslab air, hence leading to short temporal variations in subslab-indoor air pressure differentials that, in turn, affect advective flow of contaminant vapors into buildings. Similarly, a previous investigation of the intrusion of (unchlorinated) hydrocarbon vapors into a building in Australia found that semidiurnal decreases in barometric pressure caused a negative pressure differential between the building interior and subslab, increasing the rate of advective mass transfer of hydrocarbons into the indoor air.56 Our analysis also suggests that an ambient pressure drop may increase the mass of PCE flowing into the home.

In studies in northern climates, in homes with basements, higher concentrations of CVOCs have been observed in the winter compared with other seasons.26, 53 The inverse relationship observed here, in the hot and arid San Antonio climate, may be partly explained by the tighter sealing of homes during the summer months (to keep out the heat), higher subsurface and groundwater temperatures, dessication of the shallow soils, or any combination of these. It should be noted that the temperatures in February and March 2012 were mild, ranging from 12–21 °C. A detailed survey of homes in Houston, Texas, with characteristics similar to households in this study observed the lowest air exchange rates in the summer.57 Seasonal data on vapor intrusion in southern climates is limited; however, summertime increases in radon concentrations were observed in Alabama homes with crawlspaces and were attributed to preferential flow pathways associated with the site’s karst geomorphology.58 Using air conditioning has also been associated with higher indoor radon concentrations.59 This observation is worth further investigation, because it suggests that seasonal effects on vapor intrusion in southern climates may differ from those in northern climates. However, the small sample size (necessitated by funding limitations) and selection criteria for resampled homes may have biased our observations.

The results of this study are limited by the small sample size and few homes studied compared with the size of the potentially affected population. Measurements were not taken during a rainstorm or during freezing conditions. The relationships identified here may not be generalizable to other sites, especially those with a different climate, hydrogeology and housing stock. The homes included in the study were a convenience sample and not a random sample. Therefore, selection bias among the households that chose to participate in the study may have influenced the results of the analysis. The passive sampling devices showed relative good precision, although in some cases, the difference in the duplicates was high and the use of the average of the measurements may bias the results. We did not collect summa canister samples (to compare the accuracy of the passive sampling devices) nor were ambient air samples collected. The sampling of the EPA and previous sampling by the local health department gave no indication of elevated ambient levels of PCE, but possible intrusion from outdoor sources may have affected our measurements.

Also worth noting are the potential advantages of the community-based research design used in this study. The study required daily access to each home and the motivation of residents to complete the entire protocol (including foregoing use of products that may contain PCE). Experience at other sites has suggested that access to homes is a barrier to data collection in vapor intrusion studies, particularly because at such sites animosity may exist between the community, those responsible for the pollution, and the involved government agencies.36 In this case, participants appeared to adhere to the study protocol and participated for the duration of the study.


This study provides evidence of spatial variability as well as short-term and seasonal variability in PCE concentrations due to vapor intrusion. These results suggest that a single-point-in-time sample of indoor air in homes at risk of vapor intrusion is not adequate for characterizing the temporal variability in exposures. This study contributes to the body of evidence suggesting that vapor intrusion potential fluctuates on short and seasonal timescales and suggests that evaluating temporal variability is needed to adequately characterize the occurrence of vapor intrusion in a home. While PCE concentrations detected at this site did not exceed the new PCE risk-based standards, acknowledging spatial and temporal variability as well as understanding the drivers of these processes may be significant in designing and conducting vapor intrusion investigations at other sites, where concentrations may exceed EPA’s standards.


  1. 1

    Dodson RE, Levy JI, Houseman EA, Spengler JD, Bennett DH . Evaluating methods for predicting indoor residential volatile organic compound concentration distributions. J Expo Sci Environ Epidemiol 2009; 19: 682–693.

    CAS  Article  Google Scholar 

  2. 2

    Adgate JL, Eberly LE, Stroebel C, Pellizzari ED, Sexton K . Personal, indoor, and outdoor VOC exposures in a probability sample of children. J Expo Sci Environ Epidemiol 2004; 14: S4–S13.

    CAS  Article  Google Scholar 

  3. 3

    Environmental Quality Management I User’s Guide for Evaluating Subsurface Vapor Intrusion into Buildings. 2004.

  4. 4

    Johnson PC, Ettinger RA . Heuristic model for predicting the intrusion rate of contaminant vapors into buildings. Environ Sci Technol 1991; 25: 1445–1452.

    CAS  Article  Google Scholar 

  5. 5

    McCarty PPL . Groundwater Contamination by Chlorinated Solvents: History, Remediation Technologies and Strategies. In Situ Remediation of Chlorinated Solvent Plumes. Springer New York, NY. 2010 pp 1–28.

    Google Scholar 

  6. 6

    ATSDR 2007 CERCLA Priority List of Hazardous Substances that will be the Subject of Toxiciological Profiles and Support Document. 2007.

  7. 7

    Provoost J, Reijnders L, Swartjes F, Bronders J, Carlon C, D’Alessandro M et al. Parameters causing variation between soil screening values and the effect of harmonization. J Soils Sediments 2008; 8: 298–311.

    CAS  Article  Google Scholar 

  8. 8

    Little JC, Daisey JM, Nazaroff WW . Transport of subsurface contaminants into buildings. Environ Sci Technol 1992; 26: 2058–2066.

    CAS  Article  Google Scholar 

  9. 9

    Fischer ML, Bentley AJ, Dunkin KA, Hodgson AT, Nazaroff WW, Sextro RG et al. Factors affecting indoor air concentrations of volatile organic compounds at a site of subsurface gasoline contamination. Environ Sci Technol 1996; 30: 2948–2957.

    CAS  Article  Google Scholar 

  10. 10

    Ferguson C, Krylov V, McGrath P . Contamination of indoor air by toxic soil vapours: a screening risk assessment model. Build Environ 1995; 30: 375–383.

    Article  Google Scholar 

  11. 11

    Doyle P, Roman E, Beral V, Brookes M . Spontaneous abortion in dry cleaning workers potentially exposed to perchloroethylene. Occup Environ Med 1997; 54: 848–853.

    CAS  Article  Google Scholar 

  12. 12

    Aschengrau A, Weinberg JM, Janulewicz PA, Gallagher LG, Winter MR, Vieira VM et al. Prenatal exposure to tetrachloroethylene-contaminated drinking water and the risk of congenital anomalies: a retrospective cohort study. Environ Health 2009; 8: 44.

    Article  Google Scholar 

  13. 13

    Beliles RP . Concordance across species in the reproductive and developmental toxicity of tetrachloroethylene. Toxicol Ind Health 2002; 18: 91–106.

    CAS  Article  Google Scholar 

  14. 14

    ATSDR Toxicological profile for Tetrachloroethylene (PERC) 1997.

  15. 15

    CDPHE Analysis of Diagnosed vs. Expected Cancer Cases in the Vicinity of the Redfield Plume Area in Southeast Denver County, 1979-1999 2002.

  16. 16

    ATSDR. Health Consultation. Endicott Area Investigation Health Statistics Review Endicott, Broome County, New York. 2006.

  17. 17

    Forand SP, Lewis-Michl EL, Gomez MI . Maternal exposure to tetrachloroethylene and trichloroethylene through soil vapor intrusion and adverse birth outcomes in New York State. Environ Health Perspect 2012; 120: 616–621.

    CAS  Article  Google Scholar 

  18. 18

    Folkes D, Wertz W, Kurtz J, Kuehster T . Observed spatial and temporal distributions of CVOCs at Colorado and New York vapor intrusion sites. Ground Water Monitor Remediat 2009; 29: 70–80.

    CAS  Article  Google Scholar 

  19. 19

    McDonald GJ, Wertz WE . PCE, TCE, and TCA vapors in subslab soil gas and indoor air: a case study in upstate New York. Ground Water Monitor Remediat 2007; 27: 86–92.

    CAS  Article  Google Scholar 

  20. 20

    Schreuder WA . Uncertainty approach to the Johnson and Ettinger vapor intrusion model. Principia Mathematica Lakewood, CO. 2006.

    Google Scholar 

  21. 21

    McHugh TE, Nickles T, Brock S . Evaluation of spatial and temporal variability in VOC concentrations at vapor intrusion investigation sites. Proceedings of Vapor Intrusion: Learning from the Challenges, Providence, RI. 2007.

  22. 22

    Luo H, Holton C, Dahlen P, Johnson PC . Field and modeling studies of temporal variability of subslab soil gas and indoor air concentations at a house overlying a chlorinated compound-impacted groundwater plume. Bioremed Sustain Environ Tech 2011 Reno, NV, 2010.

  23. 23

    Davies B, Forward J . Measurement of atmospheric radon in and out door. Health Phys 1970; 19: 136.

    Google Scholar 

  24. 24

    Steck DJ, Capistrant JA, Dumm JP, Patton ES . Indoor radon exposure uncertainties caused by temporal variation. Proceedings of the 11th Conference of International Radiation Protection Association, Madrid, Spain. 2004.

  25. 25

    Groves-Kirkby CJ, Denman AR, Phillips PS, Crockett RGM, Sinclair JM . Comparison of seasonal variability in European domestic radon measurements. Natural Hazards Earth Syst Sci 2010; 10: 565–569.

    Article  Google Scholar 

  26. 26

    Fitzpatrick NA, Fitzgerald JJ . An evaluation of vapor intrusion into buildings through a study of field data. Soil and Sediment Contamination 2002; 11: 603–623.

    CAS  Article  Google Scholar 

  27. 27

    US EPA OSWER Draft Guidance for Evaluating the Vapor Intrusion to Indoor Air Pathway from Groundwater and Soils (Subsurface Vapor Intrusion Guidance) 2002 EPA530-D-02-004.

  28. 28

    Eklund B, Beckley L, Yates V, McHugh TE . Overview of state approaches to vapor intrusion. Remediat J 2012; 22: 7–20.

    Article  Google Scholar 

  29. 29

    Nazaroff W, Feustel H, Nero A, Revzan K, Grimsrud D, Essling M et al. Radon transport into a detached one-story house with a basement. Atmos Environ 1985; 19: 31–46.

    CAS  Article  Google Scholar 

  30. 30

    Garbesi K, Sextro RG . Modeling and field evidence of pressure-driven entry of soil gas into a house through permeable below-grade walls. Environ Sci Technol 1989; 23: 1481–1487.

    CAS  Article  Google Scholar 

  31. 31

    Adomait M, Fugler D . A Method to Evaluate Soil Gas VOC Influx into Houses. Proceedings of Air & Waste Management Association’s 90th Annual Meeting and Exhibition 1997.

  32. 32

    McHugh TE, Beckley L, Bailey D, Gorder K, Dettenmaier E, Rivera-Duarte I et al. Evaluation of vapor intrusion using controlled building pressure. Environ Sci Technol 2012; 46: 4792–4799.

    CAS  Article  Google Scholar 

  33. 33

    Nazaroff W, Doyle S . Radon entry into houses having a crawl space. Health Phys 1985; 48: 265.

    CAS  Article  Google Scholar 

  34. 34

    Turk BH, Prill RJ, Grimsrud DT, Moed BA, Sextro RG . Characterizing the occurrence, sources, and variability of radon in Pacific Northwest homes. J Air Waste Manage Assoc 1990; 40: 498–506.

    CAS  Article  Google Scholar 

  35. 35

    Luo H . Field and modeling studies of soil vapor migration into buildings at petroleum hydrocarbon impacted sites. ProQuest. 2009.

    Google Scholar 

  36. 36

    Siegel L . Stakeholders' views on vapor intrusion. Ground Water Monit Remediat 2009; 29: 53–57.

    Article  Google Scholar 

  37. 37

    US EPA Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air: Compendium Method TO-15 2nd edn US EPA. 1999 EPA/625/R-96/010b). Cincinatti, OH.

  38. 38

    Johnston JE, Gibson JM . Probabilistic approach to estimating indoor air concentrations of chlorinated volatile organic compounds from contaminated groundwater: a case study in San Antonio, Texas. Environ Sci Technol 2011; 45: 231–252.

    Google Scholar 

  39. 39

    Dawson HE, McAlary T . A compilation of statistics for VOCs from post-1990 indoor air concentration studies in North American residences unaffected by subsurface vapor intrusion. Ground Water Monitor Remediat 2009; 29: 60–69.

    CAS  Article  Google Scholar 

  40. 40

    Gorder KA, Dettenmaier EM . Detailed indoor air characterization and interior source identification by portable GC/MS. Air and Waste Management Association Vapor Intrusion Specialty Conference, Chicago, IL. 2010.

  41. 41

    Gorder KA, Dettenmaier EM . Portable GC/MS methods to evaluate sources of cVOC contamination in indoor air. Ground Water Monitor Remediat 2011; 31: 113.

    CAS  Article  Google Scholar 

  42. 42

    Odencrantz JE, Thornley SC, O’Neill H . An evaluation of the performance of multiple passive diffusion devices for indoor air sampling of VOCs. Remediat J 2009; 19: 63–72.

    Article  Google Scholar 

  43. 43

    Woolfenden E, McClenny W Compendium Method TO-17. Determination of volatile organic compounds in ambient air using active sampling onto sorbent tubes. Compendium Methods for the Determination of Toxic Organic Compounds in Ambient Air, 2te Auflage US Environmental Protection Agency. 1999, 1–53.

  44. 44

    Christakos G, Bogaert P, Serre ML . Temporal GIS: Advanced Functions for Field-based Applications. Springer Berlin; New York. 2001.

    Google Scholar 

  45. 45

    Taylor FB, Hailey RB, Richmond DL Soil Survey, Bexar County, Texas: US Department of Agriculture, Soil Conservation Service, US Dept. of Agriculture Soil Conservation Service: Washington, DC. 1966.

  46. 46

    Lubin JH, Colt JS, Camann D, Davis S, Cerhan JR, Severson RK et al. Epidemiologic evaluation of measurement data in the presence of detection limits. Environ Health Perspect 2004; 112: 1691.

    CAS  Article  Google Scholar 

  47. 47

    Helsel DR . Less than obvious-statistical treatment of data below the detection limit. Environ Sci Technol 1990; 24: 1766–1774.

    CAS  Article  Google Scholar 

  48. 48

    Tobin J . Estimation of relationships for limited dependent variables. Econometrica 1958, 24–36.

  49. 49

    Slymen DJ, de Peyster A, Donohoe RR . Hypothesis testing with values below detection limit in environmental studies. Environ Sci Technol 1994; 28: 898–902.

    CAS  Article  Google Scholar 

  50. 50

    Arellano M . Computing robust standard errors for within-groups estimators. Oxford Bull Econ Stat 1987; 49: 431–434.

    Article  Google Scholar 

  51. 51

    Kezdi G . Robust standard error estimation in fixed-effects panel models. Hungarian Stat Rev 2004; 9: 95–116.

    Google Scholar 

  52. 52

    Hansen CB . Asymptotic properties of a robust variance matrix estimator for panel data when T is large. J Econ 2007; 141: 597–620.

    Article  Google Scholar 

  53. 53

    Holton C, Luo H, Guo Y, Johnson PC, Gorder K, Dettenmaier E . Long-term and short-term variation of indoor air concentration at a vapor intrusion study site. 22nd Annual International Conference on Soil, Water, Energy and Air, San Diego, CA. 2012.

  54. 54

    Robinson AL, Sextro RG, Riley WJ . Soil-gas entry into houses driven by atmospheric pressure fluctuations—the influence of soil properties. Atmos Environ 1997; 31: 1487–1495.

    CAS  Article  Google Scholar 

  55. 55

    Holford DJ, Schery SD, Wilson JL, Phillips FM . Modeling radon transport in dry, cracked soil. J Geophys Res 1993; 98: 567–580.

    CAS  Article  Google Scholar 

  56. 56

    Patterson BM, Davis GB . Quantification of vapor intrusion pathways into a slab-on-ground building under varying environmental conditions. Environ Sci Technol 2009; 43: 650–656.

    CAS  Article  Google Scholar 

  57. 57

    Yamamoto N, Shendell DG, Winer AM, Zhang JJ . Residential air exchange rates in three major US metropolitan areas: Results from the RIOPA Study 1999-2001. Indoor Air 2009; 20: 85–90.

    Article  Google Scholar 

  58. 58

    Wilson D, Gammage R, Dudney C, Saultz R . Summertime elevation of 222Rn levels in Huntsville, Alabama. Health Phys 1991; 60: 189–197.

    CAS  Article  Google Scholar 

  59. 59

    Radford EP . Potential health effects of indoor radon exposure. Environ Health Perspect 1985; 62: 281.

    CAS  Article  Google Scholar 

Download references


This project was supported in part by the National Science Foundation Graduate Research Fellowship Program and the Passport Foundation. We recognize the invaluable assistance of the Committee for Environmental Justice Action, Southwest Workers Union, Jessica Garcia, Sandra Garcia and Juan Rodriguez. We are grateful for the assistance of Harry O’Neill at Beacon Environmental Services, in addition to the assistance of Tyler Fitch, Dami Olagunju and Mandie Kramer.

Author information



Corresponding author

Correspondence to Jill E Johnston.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies the paper on the Journal of Exposure Science and Environmental Epidemiology website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Johnston, J., Gibson, J. Spatiotemporal variability of tetrachloroethylene in residential indoor air due to vapor intrusion: a longitudinal, community-based study. J Expo Sci Environ Epidemiol 24, 564–571 (2014).

Download citation


  • vapor intrusion
  • indoor air
  • tetrachloroethylene

Further reading


Quick links