Why PFAS Levels Don’t Add Up: Exploring Variability in Water Testing Results
Advisor: Ryan Grove
Per- and polyfluoroalkyl substances (PFAS) are often referred to as “forever chemicals” because of their extreme persistence in the environment and the human body, and are increasingly linked to cancer, endocrine disruption, and immune suppression. Although regulatory standards are emerging, reported PFAS concentrations often vary widely across datasets, raising questions about how sampling conditions influence measured values. This project investigates how time, depth, and environmental factors affect PFAS concentrations within a single water source: the Occoquan Reservoir in Northern Virginia. I expect to find that PFAS concentrations fluctuate significantly depending on sampling context, even when overall contamination levels remain stable. By identifying how methodological and environmental factors shape reported values, this project contributes to improving sampling protocols and strengthening confidence inPFAS testing. Because regulatory decisions and public trust rely on accurate water quality data, understanding the variability behind PFAS measurements is critical for environmental transparency and public health protection.
Project Blog
Post 1: Intro to My Project!
When you turn on your tap, you probably assume the water has been tested, regulated, and confirmed safe. I used to think the same. But over the past few years, I have learned that measuring contaminants is not always as straightforward as it sounds.
My project focuses on PFAS, a large class of synthetic chemicals often referred to as “forever chemicals” because of their resistance to heat, water, and degradation. Their persistence makes them useful in industrial and consumer products, but it also allows them to accumulate in ecosystems and the human body. Research increasingly links PFAS exposure to cancer, endocrine disruption, immune suppression, and developmental effects. Despite growing concern, testing frameworks typically measure only a small fraction of the thousands of known PFAS compounds.
My interest in this topic began when I studied how PFAS interact with proteins, which introduced me to their unusual chemical behavior and biological impact. Later, while conducting independent water sampling in Northern Virginia, I found significant discrepancies between my measured PFAS concentrations and those reported by a local utility. That experience raised an important question: are PFAS levels inconsistent because contamination changes, or because the way we sample and measure them affects the results?
Through systematic sampling across different depths, times, and environmental conditions, I hope to better understand how variable PFAS measurements really are. Ultimately, I want to contribute to more transparent and scientifically reliable water testing practices.
Week 1: Preparation!
The first week of my PFAS research project was mostly about building the scientific and logistical foundation for the months ahead. Before collecting any samples, I wanted to understand how and why PFAS concentrations vary in aquatic systems on a deeper level.
I spent most of the week reviewing literature on PFAS variability in surface water. Several studies show that PFAS concentrations can fluctuate significantly over time and across locations within the same water system due to environmental factors such as precipitation, temperature, and hydrological mixing. Some research has documented large temporal and spatial variability in PFAS concentrations, even within a single site. For example, a study by Tokranov et al. found that changing the location of testing in the same lake shifted findings up to 85%.
Alongside the literature review, I finalized my testing schedule. Because PFAS concentrations may shift after rainfall events or other environmental changes, I structured my sampling plan around weather patterns and different depths within the water column. I also obtained a schedule for local firefighting training exercises, since aqueous film-forming foam used in firefighting can introduce PFAS into nearby water systems and potentially influence short-term variability.
One of the biggest logistical challenges I anticipated was collecting deep-water samples. This week I solved that problem by acquiring a Van Dorn water sampler, a device specifically designed to collect water at precise depths. This will allow me to test whether PFAS concentrations differ between surface and deeper water layers.
Finally, I gathered additional PFAS testing kits and sampling materials to ensure consistent measurements throughout the project. With the planning stage complete, I am ready to begin the first round of sampling next week.
Week 2: Scouting
This week marked the transition from planning to actually preparing the field site for sampling. After spending last week reviewing literature and organizing my testing schedule, my main goal was to find a practical location where I could consistently collect water samples from the Occoquan Reservoir.
I visited several areas around the reservoir to evaluate possible sampling sites, looking for a location that would allow me to reach water that is less influenced by shoreline sediment and runoff. Eventually, I selected a bridge that extends over the reservoir. Sampling from the bridge allows me to collect water farther from the shore, which should provide more representative measurements of the reservoir itself. The bridge also offers a stable and accessible location that I can return to repeatedly throughout the project, ensuring consistency across sampling dates.
In addition to scouting the location, I gathered most of the materials I will need for testing. The PFAS sampling kits and collection supplies have arrived, which means I now have the necessary containers and equipment to begin collecting samples while minimizing contamination. At the moment, I am still waiting for the Van Dorn water sampler to arrive. Once it does, it will allow me to collect water at specific depths rather than only at the surface, which is essential for studying how PFAS concentrations vary throughout the water column. Here is a picture of the Van Dorn bottle:
(Image from Novatech International, https://www.novatech-usa.com/USS-UNWTRSMPL01)
For now, the project is still in preparation mode, but with the site selected and materials nearly complete, I am getting very close to beginning the first round of sampling.
Week 3: The First Sample!
Though I was hoping to start deep water sampling this week, UPS seems to have different plans for me. Currently, I am still waiting on the Van Dorn water sampler to arrive. However, rather than pause data collection entirely, I decided to begin with a site I am already familiar with in Alexandria, where I conducted sampling over the summer.
I collected and sent out my first water sample from that same location to establish a point of comparison. My goal is to determine whether PFAS concentrations have changed significantly since my last measurements. This was because I came across a presentation by AlexRenew from February of this year during my research, which talked about additional measures they planned to implement to reduce PFAS contamination. By comparing current data to my previous results, I hope to get a sense of whether those efforts have had any measurable impact.
In addition to fieldwork, I have also started focusing on the data analysis component of my project. Because my research involves comparing PFAS concentrations across multiple variables such as time, depth, and environmental conditions, choosing an appropriate analytical model is very important. I have not yet finalized which statistical approach I will use, but I am currently exploring different methods and beginning to learn more about how artificial intelligence can support data analysis. Moving forward, I will continue refining both my sampling strategy and my analytical approach as more data becomes available.
Week 4: UPS Came Through!
This week, there was a huge advancement in my project! After several weeks of preparation, I was finally able to begin sampling. The Van Dorn water sampler finally arrived, which allowed me to begin to collect water at specific depths rather than being limited to surface samples. Armed with this new technology, I conducted my first full set of samples at the Occoquan Reservoir. I collected three samples: one at the surface, one near the bottom, and one at a depth right in the middle. By keeping the string length consistent, I will be able to use these exact depths in future trials, which is very important for maintaining consistency across datasets.
One of the biggest challenges I encountered was preventing contamination between samples. Because I am using the same equipment for each collection, there is a risk that PFAS compounds could accumulate on the sampler and affect future measurements. This was especially concerning for me, given how persistent and adhesive PFAS are.
To fix this problem, I researched proper decontamination methods and developed a cleaning protocol. Before each use of the sampler, I will now thoroughly wash it with Liquinox detergent and follow it with a final rinse using deionized water before collecting each sample. This process should help minimize cross-contamination and improve the reliability of my data.
With my first set of samples collected, I am now waiting for results from both the Occoquan and Alexandria tests. Once those come in, they will serve as a benchmark for future sampling and analysis moving forward.
Week 5: Unexpected Discoveries
This week, there was a major advancement in my project, as I received my first set of testing results. The first result came from the benchmark sample I collected in Alexandria, at the same site I tested over the summer. Based on the additional PFAS mitigation measures outlined in the recent AlexRenew presentation, I expected concentrations to decrease slightly. Instead, the result was the highest I have recorded so far at 23.8 ppt (way above the original EPA regulations). This shows that PFAS contamination at this site is still persisting and may even be increasing. This reinforces the importance of my project, as it highlights the need for greater transparency and accountability in how local water quality is monitored and communicated to Alexandria residents.
At the same time, I made an unexpected discovery in my home. I tested the water in my own house and found that the PFAS concentration was only 2.7 ppt, the lowest value I have measured so far. I had a hunch that it was because we have recently installed a water filter intended for hard water treatment. After further investigation, I found that the system uses reverse osmosis technology, which is known to be highly effective at removing PFAS.
This is significant because it suggests that effective PFAS filtration may already be accessible to consumers through systems not explicitly marketed for PFAS removal, potentially making these solutions more affordable.
This week, I also conducted a second round of testing in the Occoquan. I made sure to keep all confounding variables consistent, even down to the clothes I was wearing. Hopefully, this consistency will pay off when I get the results back.
While I am still waiting on the first set of Occoquan results, this week has already shown how both environmental data and everyday household systems can reveal important insights about PFAS exposure.
Week 6: First Results
This week, things started to feel more like real analysis rather than just data collection. While I continued collecting data by completing another round of sampling, this time under much hotter conditions (this week was very hot), and I also received the results from my first full set of Occoquan samples.
The results were not what I expected. I expected the deep water to have the highest concentration, as I thought the PFAS might sink to the river floor. However, both the surface sample and the deep-water sample came back at around 36 ppt, which suggests that PFAS may be more evenly distributed throughout the water column than I initially thought. However, the middle-depth sample was noticeably lower at 31 ppt. While the difference might not seem huge, it is still significant enough to raise questions, especially since I expected a more consistent gradient. I will double check this by seeing if the pattern continues in my other samples.
Right now, one of my hypotheses is that this could be related to density differences or how PFAS compounds interact with water layers, but I am still researching this and trying to better understand the chemistry behind it. It could also be influenced by mixing patterns or environmental conditions that I have not fully accounted for yet.
In addition to analyzing these results, I have decided to expand my sampling locations. I plan to test streams near semiconductor plants and around the Dulles area to see whether proximity to potential industrial sources affects PFAS concentrations. Up to this point, I have mainly focused on variability within a single water body, but this will allow me to start exploring how location itself might play a role.
Week 8: More Results and a Slight Bump in the Road
Last week, I got back my second full set of sampling results. The data, which is attached to this post, shows a pattern that is actually not that consistent with what I saw in my first round of testing, as it contains a few details that make me want to look into more factors that are influencing the results.
For this set, the total PFAS concentrations were 33.7 ppt (middle), 29.8 ppt (bottom), and 39.2 ppt (surface). The surface sample is the highest, the bottom is the lowest, and the middle falls somewhere in between, whereas my first sample set had the bottom as the highest and the middle as the lowest. This shows that the general structure may not be as set in stone as I thought, which does not fully align with what I originally expected.
Though I explored the density hypothesis in my last post, since this new discovery, I’ve been looking at other possible reasons. This pattern could be explained by a combination of stratification, mixing, and sediment interaction. Surface water may have higher concentrations due to atmospheric deposition or runoff, and the middle layer might act more like a mixing zone, which could explain why the results varied so much. As for the bottom, I will keep looking into why the results jumped so much.
This week also came with an unexpected challenge. Part of my Van Dorn sampler actually broke off during use, which was definitely not ideal. I ended up having to go to Home Depot and buy some PVC cement to repair it. Before doing that, I also made sure to check the chemical composition to confirm that it wouldn’t introduce any additional PFAS contamination into my samples. Fortunately, everything is now fixed and functioning properly, and I’ve continued sampling without further issues. Aditionally, it finally rained this week, meaning I was able to get a sample influenced by precipitation, which was something I was worried about not being able to do!
On the analysis side, I’ve started applying computational methods to make sense of all this data. I’ve been organizing everything into structured datasets and using graphs to visualize trends, especially how PFAS concentrations change with depth. I’m also beginning to explore regression models and simulations to understand how environmental factors might influence these values. Even with a relatively small dataset, these methods are already helping me see patterns that aren’t immediately obvious from the raw numbers.
Hopefully, as the dataset keeps growing, the patterns become clearer, and I’ll start to move from just collecting data to actually interpreting it.
Week 9: From Collection to Analysis
This week, I ran into another limitation. Only, this time it was not equipment, it was materials. After my most recent round of sampling, I realized that I only have three testing kits left, which is enough for just one final full sample set. I decided to save that last set for next Wednesday, right after a few days of rain, since precipitation is one of the environmental factors most likely to influence PFAS concentrations. That means my data collection phase is starting to come to a close, and my focus is shifting much more heavily toward analysis.
Looking back at my results so far, some patterns are starting to feel consistent rather than random. Across multiple trials, surface water tends to have the highest PFAS concentrations, while middle-depth samples are often slightly lower. For example, in one dataset the surface reached around 39 ppt while the middle depth dropped closer to 31–33 ppt. That repeated “dip” in the middle is something I did not expect at the beginning of this project, and it suggests that PFAS distribution is being influenced by more than just simple vertical mixing.
Now, I am working on organizing all of my data into a structured dataset so I can analyze it computationally. I have started using graphs to visualize trends, especially depth profiles that show how concentrations change from surface to bottom. I am also beginning to explore statistical methods like correlation and regression to see which variables, such as depth, temperature, or weather conditions, actually have the strongest relationship with PFAS levels.
With sampling almost complete, the project is starting to shift from collecting numbers to understanding what those numbers actually mean.