The scientists who sniff water

Mel Suffet stood at a whiteboard, dry-erase marker in hand, and asked me sarcastically, “Did you have fish for dinner?” He wasn’t commenting on my breath. For the past hour, I’d been sitting in a mini-cubicle, taking part in a six-person odor-sniffing panel in his lab at the University of California, Los Angeles. I’d just told him that a water sample I’d sniffed had a musty and fishy odor.

“I’m not going to put it down because it’s wrong,” he said. “We don’t have a fishy odor.”

He paused from recording the other panelists’ descriptions, went to a fume hood, and pulled out the Erlenmeyer flask containing the sample in question. He told me to smell it again. After a few sniffs, I realized he was right: It was musty and pine scented. Clearly, smelling water is challenging.

The exercise was an example of how water scientists analyze funky-smelling water. The research of Suffet and other water scientists over the past 30-plus years has helped water utilities around the world solve problems with off-odors and off-tastes in drinking water. These scientists have used analytical chemistry and panels like the one I participated in to identify a spectrum of off-putting scent molecules in drinking water, allowing utilities to design treatments to keep their water taste- and odor-free.

In the U.S., off-odors and off-tastes are dealt with through so-called secondary maximum contaminant levels—recommendations from the U.S. Environmental Protection Agency that don’t require compliance. Some states enforce them more than others do. But in general, water with an unpleasant smell is seen as a nuisance, not a health problem. In fact, Suffet points out that almost all the odor compounds that he and others have studied create a stink at nanogram-per-liter concentrations or less—below levels that would typically raise concerns about toxicity through ingestion.

That doesn’t stop people from worrying that a weird smell coming from their taps is a sign something harmful is in the water. In fact, before modern sanitation, smelly water often indicated contamination by microbes or chemicals. “The public doesn’t want to smell anything in their water,” says Djanette Khiari, research manager at the Water Research Foundation. “They expect it to have no odor.”

In 2012, the American Water Works Association surveyed water utilities across the world and found that on average, a utility had 11 problematic taste- or odor-related episodes per year over the past five years. And 77% of the utilities said handling such issues was “quite” or “extremely” important.

The water industry got serious about sniffing its wares around the late ’70s. Earlier that decade, beverage companies started bottling water in plastic instead of glass, making it cheaper and easier to deliver their water to consumers, says Christy C. Spackman, who studies the history of drinking water aesthetics at Harvey Mudd College.

The increased popularity of bottled water added to utilities’ worries about consumer confidence in their water. “In the U.S., there was a real need to deal with negative public attention to municipal drinking water,” Spackman says.

To help address odor issues, water scientists from Southern California to Philadelphia to Paris started to borrow approaches from the food and beverage industry. Specifically, the water scientists adopted a technique called flavor profile analysis (FPA). In the middle of the 20th century, the food and beverage industry wanted to find ways to analyze the tastes and smells of their products to increase consumer acceptance. One of the triggers for this research was determining why soldiers during World War II rejected rations as unpalatable, Spackman says.

The consulting firm Arthur D. Little developed FPA in the 1950s as a way to help the industry translate people’s subjective experiences of a food’s or drink’s flavor into data that scientists could compile and analyze, Spackman explains. Water scientists took the process a step further and connected people’s sensory experiences to possible molecular culprits that water utilities could then look for and remove.

The FPA process for water involves a panel of at least four people who smell water samples and describe their odors—similar to what Suffet asked me and the other panelists to do in his lab. The samples sit in covered Erlenmeyer flasks and get heated to about 45 °C to help volatilize any odor molecules. Panelists swirl the flask, remove the cover, and then take short sniffs, or what Suffet calls “bunny sniffs.” They then write down the odors they smell and rate their intensities.

The intensity ratings follow a seven-point scale (1, 2, 4, 6, 8, 10, and 12) from barely detectable to very strong. Panelists sip 5%, 10%, and 15% sugar solutions to relate the solutions’ taste intensities to the 4, 8, and 12 odor-intensity levels, respectively. These levels correspond to weak, moderate, and very strong odors. The sweetness of the sugar solutions conveys that: The 5% solution has a hint of sweetness, whereas the 10% solution tastes like a sugary drink such as Kool-Aid, and the 15% is sickeningly sweet.

After each panelist’s descriptions and intensities for each sample have been recorded, the group discusses the odors they detected. Sometimes panelists go back and smell a sample again to see if they missed an odor. A sample of water gets assigned an official odor intensity only if 50% or more of the panelists detect it. The researchers then average the intensities for that odor. Water scientists have found that this intensity increases with the logarithm of the odorant’s concentration in the water.

One thing I learned during the UCLA panel is that it’s hard to come up with words to describe the experience of certain odors. One sample spiked with ethyl mercaptan, for example, had a pungent sulfur smell that created what felt like a garlic coating at the back of my throat. Other panelists described it as marshy or oniony.

“If I presented you with an odor and I didn’t give you any clues or words about what it might smell like, you’d have a hard time naming the odor,” says Gary Burlingame, director of the Bureau of Laboratory Services at the Philadelphia Water Department. “But if I gave you a list of names, you’d say, ‘Oh yeah, that’s what that smells like.’ ”

So as the water scientists developed their version of FPA decades ago, they looked for a way to give panelists odor clues. To do so, they again borrowed from the food and beverage industry. Producers of wine, beer, and coffee all use graphical tools called taste and odor wheels to break down the common flavors in their beverages. A person can refer to the wheel while sipping the drink and then find the flavor that best describes the experience.

The drinking water wheel has four tastes, eight smells, and one mouthfeel/nosefeel category, which includes sensations such as cooling and astringent. To serve on a drinking water FPA panel, a person must get trained to recognize the odors and tastes on the wheel. The key is to relate the sensory experience created by each odor and the descriptor that’s been assigned to it. For example, my untrained nose initially referred to that pine scent as fishy. After smelling it again, the word “pine” clicked in my head—the odor I struggled to explain until I had access to a more concrete description.

Through such a training process, FPA panelists can be confident that they’re all talking about the same experience when they refer to a given odor description. And that is important for the success of the method, Spackman says. “Sensory experience is no longer ineffable—something that cannot be described, that is distinctly individualistic,” she says. “It’s become something shared. And since it’s shared we can potentially quantify it and make our sensory experience into something that can be manipulated scientifically.”

For drinking water, that shared experience translates into potential molecules to hunt down and remove through treatment processes. The taste and odor wheel contains not only descriptions of sensory experiences but also molecules that scientists have determined to be responsible for those tastes and odors. These odor molecules can have natural or synthetic origins. When a panel identifies a fishy odor, for example, scientists can then go looking for molecules, such as 2-trans-4-cis-7-cis-decatrienal, produced by some algae.

That’s just one of the odorants that originate in the drinking water source. For example, two common odor culprits are geosmin and 2-methylisoborneol. Cyanobacteria that grow in the rivers, lakes, and reservoirs used for drinking water produce these earthy, musty metabolites.

Other odorants arise during the treatment process. The most common odor complaint from consumers, Burlingame says, is a chlorine smell created by the disinfectant that treatment plants add to water to prevent microbial growth. This disinfection process can also convert phenolic compounds already in the water into chloro- and bromophenols that have a medicinal smell.

To lower the concentrations of odor molecules below levels that cause consumer complaints, utilities can dilute the smelly water with nonsmelly sources. But there are also options to treat the water to break down or remove the odorants. The solution is often one of two choices: oxidation with chlorine, ozone, or permanganate, or adsorption by activated carbon. The choice will depend on the specific odorant. The earthy compound geosmin isn’t degraded by chlorine, so utilities have to use ozone or activated carbon to treat it. Meanwhile, grassy and fishy odors involving aldehydes are best dealt with by oxidation.

When treating smelly water, two concentration benchmarks matter, Suffet says: the threshold concentration at which people can first start detecting the odor, and the recognition concentration at which people can name it. “You want to treat the chemical so that its concentration is between the odor threshold and odor recognition concentrations, where the customers won’t complain,” Suffet explains.

But what happens when an FPA panel smells something that isn’t on the wheel? For example, in the 1980s, Burlingame worked on a problem in Philadelphia involving water that panels described as smelling like cucumber. Having never come across a water compound with that odor, he searched the food and beverage literature and found a nonadienal molecule reported to have a cucumber scent. The panels agreed that it and the water smelled the same. Then, using gas chromatography/mass spectrometry, researchers confirmed the presence of the compound in a reservoir more than 480 km upstream. Now trans-2-cis-6-nonadienal, which is produced by algae, is on the wheel.

A more recent case of a novel odor in drinking water involved a chemical spill. On the morning of Jan. 9, 2014, a chemical storage tank on the Elk River in West Virginia started to leak. About 37,800 L of a mixture of chemicals, mainly 4-methylcyclohexanemethanol (4-MCHM), spilled into the river, which serves as a drinking water source for Charleston, W.Va. Residents started to notice licorice odors in their water.

Andrea M. Dietrich, an environmental engineer at Virginia Tech, and colleagues analyzed the odors of the cis and trans isomers of 4-MCHM using a technique called GC-olfactometry (GC-O). The technique takes the volatile effluent from a GC column and directs it to a port where people can sniff the molecules as they come out. Coupling GC/MS with GC-O allows scientists to connect a scent with molecular mass and structural information to identify an odorant. On the basis of descriptions from odor panels, Dietrich’s team determined that trans-4-MCHM was responsible for the strong licorice odor (Environ. Sci. Technol. 2015, DOI: 10.1021/es5049418).

At the same time, Suffet was studying 4-MCHM with water quality consultant Michael J. McGuire, who was one of Suffet’s former graduate students. They used odor panels to estimate the threshold and recognition concentrations of the molecule, which were 0.15 µg/L and 2.2 µg/L, respectively. The data helped explain why residents still smelled the chemical even when some analytical methods, which had a detection limit of about 10 µg/L for 4-MCHM, couldn’t find it.

After 30-plus years of research, water scientists feel like the field has a good handle on many drinking water odor problems. “We’ve probably cataloged and identified more than 90% of the odors and tastes that anybody around the world would experience,” Burlingame says. But he acknowledges that scientists still have a lot to learn about our senses of smell and taste and how they affect our experience with drinking water.

And recently, Burlingame, Suffet, and others have started to apply the FPA and odor wheel process to air pollution scents, such as those from landfills and wastewater treatment plants. The scientists think the water field’s expertise could help clean up odors wafting from these sites into neighborhoods.

After the odor panel exercise at UCLA ended, Suffet told me about a project he’ll be working on to analyze the odors from a landfill in the Los Angeles area. He plans to train a panel to evaluate odors downwind of the site. After spending an afternoon sniffing smells described as rancid or rotten vegetables, I don’t know if I’d volunteer for such a panel.

The work may stink, but fortunately for our noses, somebody does it.