- Geosmin and 2-methylisoborneol (MIB) are byproducts of cyanobacteria and actinomycetes living in your reservoir — the human nose detects them at single-digit parts-per-trillion, far below any health concern and far below what conventional treatment removes well.
- Powdered activated carbon and advanced oxidation work, but they are expensive, slow to dial in, and reactive by nature — you are paying to remove a compound after the reservoir has already made it.
- The compound is made in a place and a season you can predict. Suppressing the biomass that produces it, where it grows, is the lower-cost control point — which is why the durable fix lives upstream of the intake.
The complaint that is not about safety
The water is safe and your customers are angry anyway. The phones light up with reports of an earthy, musty, dirty-lake taste, the complaints cluster over a few days, and every finished-water compliance number is well inside its limit. Nothing is wrong with the water in any regulatory sense. What is wrong is that people can taste their reservoir, and to a customer that is indistinguishable from the water being bad.
The reason this problem feels so disproportionate is a quirk of human biology. The two compounds responsible — geosmin and 2-methylisoborneol (MIB) — are detectable by the human nose at concentrations around 5 to 10 nanograms per liter, single-digit parts-per-trillion. That is roughly a thimble of compound in a large reservoir. No treatment plant was designed to chase a contaminant at that detection floor, and no health standard requires it to. You are removing a compound purely because people can taste it at a level three or more orders of magnitude below where it would matter chemically.
Where the smell actually comes from
Geosmin and MIB are secondary metabolites — molecules produced by cyanobacteria and, in some systems, by actinomycetes, as part of normal metabolism. They are made in the reservoir, by living biomass, in specific places and at specific times. Understanding that geography is the whole game.
Planktonic cyanobacteria in the photic zone are one source, and those events often track the same nutrient-driven bloom dynamics that produce toxin risk. But a large share of stubborn taste-and-odor problems come from benthic mats — cyanobacteria and actinomycetes growing on the sediment and on shallow littoral surfaces in low-light, nutrient-rich zones. These producers are invisible to a surface grab sample and to a sonde profiling the open water. The compound they release diffuses upward and outward, and depending on where your intake draws from and how the reservoir is stratified, it arrives at the plant on a schedule the operators experience as random. It is not random. It is the product of a specific community growing in a specific place under specific conditions. The same population can also release a pulse of geosmin or MIB when it collapses — so an event can be triggered not by a bloom but by the end of one, when the cells lyse and dump their stored metabolites, and a turnover or mixing event can carry a deep-water signal to the intake within days.
Two consequences follow. First, the release is often decoupled from the most visible biomass — a clear-looking reservoir can deliver a severe taste-and-odor event from benthic production while the open water looks fine. Second, the timing is tied to stratification, temperature, and the nutrient supply feeding those mats, which means it is forecastable if you understand the system and bewildering if you do not.
Why the plant is the expensive place to win
The plant-side tools are real and they work. Powdered activated carbon adsorbs geosmin and MIB; advanced oxidation will destroy them. But both are reactive controls applied to a compound the reservoir has already manufactured, and both carry structural disadvantages.
- Carbon dose has to be tuned to a concentration that swings by the day, often with little warning, so you either overdose continuously as insurance or you lag the event and field complaints anyway.
- The compounds are hard to adsorb efficiently in the presence of competing natural organic matter, so the carbon you pay for is partly spent on background organics rather than on the molecule you are targeting.
- Oxidation that is strong enough to destroy these stable molecules raises its own questions about disinfection-byproduct formation and oxidant cost.
- None of it changes next week's load. You re-buy the entire response every time the reservoir produces another pulse.
Fixing it upstream inverts the economics. If the compound is being made by a biomass you can locate, in a season you can anticipate, driven by a nutrient supply you can characterize, then suppressing that production is a control applied once to the cause rather than continuously to the symptom. The diagnostic questions are about the reservoir, not the plant: which organisms are producing, where in the reservoir they grow, what nutrient supply sustains them, and how the stratification and intake geometry deliver the compound to the withdrawal point. We map bloom and production hotspots, characterize the nutrient drivers — including the internal phosphorus loading that fuels late-summer biomass — and identify the conditions under which the next event will arrive. The output is a handle on the cause, not a better guess at the dose.
What the complaints actually cost
Taste-and-odor has no MCL, which is exactly why it is dangerous to a utility. Because it is not a compliance violation, it does not show up in the regulatory ledger — it shows up in the political and reputational one. A bad season generates complaint volume, consumes staff time, drives some fraction of customers to bottled water, and erodes the public confidence that takes years to rebuild and is the foundation for every rate case and bond measure that follows. The carbon and oxidant spend is the visible cost; the trust cost is the larger and quieter one.
Set against that, a reservoir-scale diagnosis that identifies the production source and its drivers is the cheaper instrument, because it attacks the recurring cause instead of re-buying the response each season. If your plant is dosing carbon to mask a smell that is manufactured upstream in your reservoir, that is a conversation worth having.
Stop masking the smell. Start managing the source that makes it.