Pond water quality
The water change is the one piece of pond maintenance nothing else substitutes for. A biofilter converts waste; it does not remove it. Everything the filter finishes with, plus everything it never touches, stays in the pond until you carry it out in a bucket of old water. Here is what a working schedule looks like, and the three ways a well-intentioned water change goes wrong.
Ask five keepers why they change water and you will get five answers, most of them partly right. It helps to separate what the biofilter handles from what only dilution handles.
Nitrifying bacteria take ammonia to nitrite and nitrite to nitrate. That is the end of the line in most ponds. Nitrate accumulates. So do dissolved organic compounds — the tea-colored, protein-rich residue of food, mucus, leaf litter and fish waste that a mechanical filter never captures because it is in solution, not suspension. Neither leaves on its own.
Two further effects are less obvious. First, the carbonate buffer is consumed continuously: nitrification is acid-producing, and every day the biofilter works it draws down alkalinity. If your fill water is harder than your pond, a water change puts buffer back. That is the quiet reason ponds on regular water changes rarely see a pH crash, and why carbonate hardness is worth understanding before you set a schedule. Second, in heavily stocked ponds, koi excrete growth-inhibiting hormones that suppress their own development; dilution is the only practical way to reduce them.
| What is in the water | Does the filter remove it? | Does a water change remove it? |
|---|---|---|
| Ammonia / ammonium | Yes — converted by nitrifiers | Yes, by dilution |
| Nitrite | Yes — converted by nitrifiers | Yes, by dilution |
| Nitrate | No — it is the end product | Yes — this is the main route out |
| Dissolved organics (DOC) | Largely no | Yes, by dilution |
| Growth-inhibiting hormones | No | Yes, by dilution |
| Carbonate buffer (KH) | No — the filter consumes it | Replenished only if the fill water is harder |
The consistent answer from keepers who run stable ponds year after year: modest and frequent beats rare and large. A 10–20% change every week is the common working range for a stocked koi pond. A small daily trickle change — fresh water in, overflow out, running continuously — achieves the same weekly volume with almost none of the disturbance, and is what most serious systems eventually move toward.
The two are not equivalent in how the fish experience them. A 20% weekly batch is one noticeable step in water chemistry every seven days. The same 20% spread across seven days is a change the fish essentially never register. Both work. The trickle is gentler.
What does not work is saving it up. A 60% change once a season moves the same volume of water on paper and puts the pond through a chemistry swing it has no buffer for.
Feeding rate is the real driver. Waste production scales with what goes in, and what goes in scales with water temperature. In midsummer, on heavy feeding, a densely stocked pond justifies the upper end of the range. In winter, with dormant fish that are not being fed at all, the pond is barely producing waste and needs far less.
| Water temperature | Fish activity and feeding | Water change |
|---|---|---|
| 75°F+ (24°C+) | Peak metabolism, heaviest feeding | Upper end — toward 20% weekly, or the daily trickle equivalent |
| 65–75°F (18–24°C) | Active, feeding well | 10–20% weekly |
| 55–65°F (13–18°C) | Appetite tapering, lighter feeds | Lower end of the range |
| 45–55°F (7–13°C) | Feeding stops | Small and infrequent, temperature-matched |
| Below 45°F (7°C) | Dormant | Much less — only enough to keep the system from stagnating |
Stocking density and filter capacity shift the whole table. A lightly stocked pond with generous filtration sits at the bottom of every band. A show pond carrying more fish than the volume really wants sits at the top of every band and still benefits from more.
This is the step most often skipped, and it is the one that turns a routine into a problem. A water change is only a buffer top-up if the fill water carries more buffer than the pond does. Test the water coming out of the hose the same way you test the pond: total alkalinity (KH) and pH at minimum. US kits report alkalinity in ppm as CaCO3; if you prefer dKH, roughly 17.9 ppm is 1 dKH.
| Fill water vs. pond | What happens on every change | What to do |
|---|---|---|
| Fill KH higher than pond KH | Buffer is replenished; pH stability improves | Nothing — this is the ideal case |
| Fill KH about the same | Buffer holds steady; dilution still works | Watch KH over the season as the filter draws it down |
| Fill KH lower than pond KH (soft municipal or well water) | Every change dilutes the buffer further | Buffer the pond separately; smaller, more frequent changes |
| Fill water is RO, softened or rainwater | Essentially no buffer at all | Do not use it unbuffered as the primary source |
Soft fill water can cause the instability you were trying to prevent. If your municipal supply is softer than the pond, a diligent weekly water change quietly lowers alkalinity week after week. The pond reads fine until the buffer runs out, and then pH becomes unstable within a single day. If your fill water is soft, the water change still needs to happen — but the buffer has to be maintained separately, not assumed.
Done carelessly, the maintenance task meant to protect the fish is the thing that harms them. Three failure modes account for almost all of it.
Never do a large emergency water change to correct a reading. High nitrate, low KH and drifting pH are all corrected safely by more frequent modest changes. A single large change moves every parameter at once and does it faster than the fish can compensate. Warm water also holds less oxygen, so a big afternoon change in summer can pull dissolved oxygen down at the same time; keep aeration running throughout.
A weekly batch change is straightforward: draw down, refill through a dechlorinator, done. It costs half an hour and depends entirely on somebody remembering.
A continuous trickle change — a metered feed of treated water in, an overflow out — delivers the same weekly volume as a slow, invisible drift the fish never notice. It also means chemistry never sits still long enough to drift far. The catch is that a trickle running unattended has two ways to fail: it can stop, and the pond quietly declines with nobody noticing for weeks; or it can fail open, and an unbounded fill either overflows the pond or, with a stuck drain, empties it. Float valves and mechanical autofills fail this way more often than people expect.
Most keepers test weekly, often on the same day they change water. That is a reasonable discipline, and it is also a sample of a few minutes out of 10,080 in a week.
The things that actually go wrong between tests move faster than that. Oxygen falls overnight, bottoms out before dawn, and has recovered by the time anyone is holding a test vial. Buffer depletion is invisible until it is not: KH declines slowly for weeks and then pH loses its footing in an afternoon. A dechlorinator dosing pump that stops working leaves no trace in a Sunday-morning test unless the failure happened to occur on Saturday night. Weekly spot-checks confirm where the pond has settled. They do not tell you what it did in between, which is where the events live.
The routine described above — a small, regular, temperature-sane, dechlorinated water change — is exactly what H2Koi automates. The design constraints are the ones the failure modes dictate. The fill and drain are bounded by independent low and high level marks, so the pond can neither be drained nor overfilled if a valve sticks or a sensor misreads. The system draws only from a dechlorinated supply prepared upstream; it does not treat water itself, and it will not pull raw municipal water into the pond. And it runs continuously rather than in large batches, because that is the gentler pattern for the fish.
Automated water changes are built to the individual pond — plumbing, drain, supply and level marks are specific to the site. We have never found another system that does it for a koi pond, and neither have our clients.
What H2Koi does, and what it does not. The sonde continuously measures temperature, pH, dissolved oxygen (mg/L and % saturation), conductivity, turbidity, ammonium (NH4) and nitrate (NO3), with ORP available as an option. From those it computes KH (0–20 dKH), ammonia (NH3), nitrite, salinity, TDS. KH here is a derived value, not a titration — it will track the trend of your buffer, but your drop test remains the reference. General hardness (GH) is not reported at all. A self-cleaning wiper keeps the sensors clear so the readings stay comparable week to week.
It is continuous early warning from industrial-grade sensors. It does not detect disease or parasites. It tells you that something is moving, sooner than a weekly test would. What happens next is still your call.
For a stocked pond, 10–20% weekly is the usual working range, or a small daily trickle change that adds up to roughly the same volume. Shift toward the upper end in summer when feeding is heavy and stocking is dense, and much lower in winter when the fish are dormant and not being fed.
In almost every routine situation, yes. Large single changes move pH, temperature, hardness and dissolved gases all at once, faster than fish can compensate. Two or three modest changes over a week accomplish the same dilution without the swing. Reserve large changes for genuine emergencies, and even then match temperature carefully and keep aeration running.
Yes. Municipal water carries chlorine or chloramine at every volume, and the filter bacteria are more sensitive to it than the fish are. Small untreated top-offs done repeatedly are a common reason a biofilter underperforms with no obvious cause. Treat every drop of municipal water that enters the pond.
That works for free chlorine, given enough time, surface area and agitation. It does not work for chloramine, which is deliberately stable and will not gas off in a barrel overnight. Most US utilities use chloramine or switch to it seasonally, and you generally will not be told when. Use a product rated for chloramine rather than relying on standing time.
Far less, and far more carefully. Dormant fish are not being fed, so waste production is minimal and the pond needs little dilution. Any water added should be close to pond temperature — a cold hose into a cold pond can still drop it several degrees, and dormant koi have almost no capacity to cope with that. Aeration and leaving the fish undisturbed matter more than the schedule at this point in the year.
It will lower the number and create a new problem. Nitrate accumulates slowly, so it should be brought down slowly: increase change frequency rather than doing one large change. If nitrate keeps climbing back, the underlying causes are usually overfeeding, overstocking, or a change schedule that has quietly slipped — and those are what to address.
Yes, and it is worth doing properly. The requirements are a metered feed from a dechlorinated supply prepared upstream, a matching overflow or drain, and independent low and high level limits so the pond cannot be drained or overflowed if something sticks. That combination is what H2Koi builds. We have never found another system that does it for a koi pond, and neither have our clients.
Continuous monitoring of temperature, pH, dissolved oxygen, conductivity, turbidity, ammonium and nitrate, with ORP as an option, plus derived KH, ammonia, nitrite, salinity, TDS.
Explore H2Koi
