Natural Pool Filtration: Airlift Pump vs Wetland Filter vs ?

Most people who write to me have already decided how their biofilter is going to work. They have watched a video, usually a good one, and they arrive asking how to build that particular thing. What they have not realised is that they made a choice, and that there were other choices sitting alongside it that would have suited their site better.

There are three broad approaches in common use for DIY natural pools. The airlift pump method, popularised by David Pagan Butler. The pumped upflow gravel bed, which covers bog filters, the OzPonds method and the American Aquascape systems. And the two-pump design I have settled on after ten years of building these things and, more usefully, ten years of being called back to look at ones that stopped working.

I am not neutral, and I will tell you plainly at the end where I landed. But everything here has its use case, and anyone who tells you a method is simply bad is selling you something. What follows is what each one actually does, what it costs you in money and in attention, and where it stops working.

One note on words before we start. The planted gravel area is called the regeneration zone, the wetland, the reed bed or the bog depending on whose book you read. They are the same thing. I will call it the regeneration zone throughout, because "wetland" makes people picture the plants doing the work, and the plants are not doing most of the work. More on that later.

Three different standards, and you must know which one you are building to

This is the distinction that makes the rest of the article make sense, and it is the one most often missed. Three quite different things get built with the same components and talked about in the same forums.

A garden or fish pond is built to be alive. The water quality target is health, not clarity. A green tinge through the warm months is normal, visibility of a metre is perfectly acceptable, and a mat of filamentous algae in spring is part of the year rather than a failure. If there are fish in it, the fish themselves are a substantial and permanent nutrient input, and the system is largely there to keep them alive rather than to keep the water looking like glass.

A swim pond, or recreational pond, is built to be swum in but designed to look like a pond. Informal planted edges, shelved or beach entry, an organic shape that reads as landscape rather than as pool. The water quality target is real but forgiving: clear enough to see the bottom, safe, pleasant to get into, with some seasonal variation accepted as the price of the look. Nobody is upset if it is a little softer in high summer.

A natural swimming pool is a different brief. Usually a more formal shape, defined edges, coping, and an expectation set by the chlorine pool the owner had before. The target is water that is clear right to the bottom for the whole season, no green cast, no sludge on the floor, and very little variation month to month. It is much closer to what people expect from a conventional pool, delivered without a single chemical.

All three are legitimate and I build all three. The mistake, and it is an expensive one, is building to pond standard and expecting pool results. A method that is excellent at the first standard can be an outright failure at the third, using identical parts, on an identical site. This is exactly why I rate the OzPonds approach very highly for a garden ecosystem pond and would not choose it for a family swimming pool, and it is why the same argument about skimming that a pond builder can reasonably shrug at is decisive for me.

One honesty note that applies at every level, including the top one. A natural pool of any kind will develop a very thin biofilm on its walls and floor. That is normal, healthy and brushable. If what you actually want is a surface with nothing biological on it anywhere, no natural system will give you that, and you should build a chlorine pool.

Which of the three you are building is the first decision, and it deserves an article of its own rather than a paragraph. For now, hold your answer in mind while you read the rest of this, because it changes which method is right for you.

What a biofilter has to do

Before comparing anything, it is worth being precise about the job, because every argument that follows comes back to this.

A natural pool is a nutrient management problem, not a cleaning problem. Leaves, pollen, dust, sunscreen, skin, insects and rain all put nitrogen and phosphorus into the water. Algae are simply the fastest organism at turning those nutrients into green. If the nutrients are there and the light is there, you get algae. A chlorine pool solves this by killing everything. A natural pool solves it by making sure something else has already taken the nutrients up, or by getting them out of the water before anything can.

So a biofilter has five jobs, and every design is a different compromise between them.

• Surface area. The bacteria that process nitrogen live as a film on solid surfaces, not floating in the water. Gravel is cheap surface area. A cubic metre of 6 to 12 mm gravel gives you an enormous colonisable area for very little money.
• Oxygen. That processing is aerobic and it is oxygen-hungry. A bed that goes anaerobic stops working, starts producing hydrogen sulphide, and smells like it.
• Contact time. Water has to move slowly enough through the media for the biofilm to actually work on it. Push water through gravel too fast and you have built an expensive pipe.
• Nutrient export. Bacteria convert ammonia to nitrite and then to nitrate. Nothing in that chain removes nitrogen from the pool. It has to physically leave, by plant uptake and pruning, by removal of solids, by offgassing, or by dilution.
• Solids capture, and where it happens. This is the one that gets missed, and it is the one that decides whether your pool is still clear in year five.

Hold those five in mind and the differences between the methods stop being a matter of taste.

Method one: the airlift pump

David Pagan Butler did more than anyone alive to put natural pools within reach of ordinary people, and the airlift is the centre of his approach. It deserves a proper explanation rather than a caricature.

An airlift is not a pump in the usual sense. Nothing spins. You run an air line from a blower down to the bottom of a vertical pipe, often 110 mm, that stands submerged in the water. The bubbles rising inside that pipe make the column of water inside it lighter than the water outside it, so the column rises. Water enters at the bottom, is carried up with the bubbles, and spills out of the top. The bubbles are doing the lifting.

What it does well. It moves a large volume of water for very little electricity, provided you only need to lift it a few centimetres. It oxygenates the water while it moves it. There is no impeller to jam and nothing expensive submerged that will fail. It is genuinely buildable from parts, which matters enormously if you live somewhere pool equipment is costly or hard to get. And on a large, wildlife-oriented swimming pond in a cool climate with a light swimmer load, it is an extremely elegant solution. Huge regeneration zone, very low energy, a great deal of ecological resilience. I would not talk anyone out of it on that site.

Now the objections, which are bigger than most people realise.

It short-circuits. This is my main gripe and it is a layout problem, not a component problem. The airlift stands at the edge of the swim zone next to the regeneration zone. It takes water from right there and delivers it a metre away into the regeneration zone. That water then has to go somewhere, so it displaces water back out of the regeneration zone, into the pool, immediately next to where the airlift is drawing from.

You have built a loop about two metres across. Water goes round and round between the edge of the swim zone and the near end of the regeneration zone, and the rest of the pool barely participates. The flow meter, if you had one, would look excellent. The far end of your swim zone is effectively still water.

Stagnant water in a swimming pool is not a cosmetic problem. It is where the temperature stratifies, where debris settles instead of being carried, and where the oxygen goes first.

There is no skimming, and skimming is not decoration. This is the point I most want people to take away from this article, whichever method they choose.

A skimmer is not there to make the surface look tidy. The large majority of the nutrient load entering a natural pool arrives through the surface. Pollen. Dust. Dead insects. Dead plant material blowing off your own regeneration zone. Leaves from whatever is growing nearby. Every one of those things floats, at first.

While it floats, you can remove it whole, in one piece, with almost no energy, and it takes its entire nitrogen and phosphorus content out of the pool with it. Once it waterlogs and sinks, it is on the floor of your swim zone, it starts breaking down, and every gram of nutrient in it is released into the water for your regeneration zone to deal with. You have turned a debris problem into a nutrient problem, and nutrients are the thing you built the whole pool to control.

That is how sludge forms on the bottom of a natural pool. It is not bad luck. It is the arithmetic of not skimming, and you can see it in a great many airlift pools if you look at the floor rather than the surface.

You cannot set the flow rate. Getting the flow through the regeneration zone right matters more than almost anything else in the design, and I will do the arithmetic on that below. It is a sweet spot: too fast and there is no contact time, too slow and the pool does not turn over enough and the bed goes short of oxygen. With a pump you calculate it, set it and know it. With an airlift you get whatever that pipe diameter, that submergence and that blower happen to produce on the day. It is a guess, and it drifts as the pipe fouls.

It is not quiet, and it is fussy to build. Air blowers are noisy, so the blower has to be housed somewhere, and housing it properly is a job in itself. And for a DIY builder the airlift is more finicky than it looks in a video: weighted air lines, a vertical 110 mm pipe held true, submergence depth right, air distribution even. It is all doable. I just do not see what you get back for the trouble.

What about the aeration? The usual defence is that the airlift oxygenates the whole system, and that this is worth the compromises. I do not think it is, on a properly designed bed. A regeneration zone taking a slow, continuous flow of already-oxygenated water does not go short of oxygen, and it certainly does not need bubbles pumped into it.

I will be honest that one part of this argument gets overstated, including by people arguing my side. Plant roots do release oxygen, but locally and modestly, and the claim that root oxygenation alone keeps a bed aerobic is doing more work than the evidence supports. The real reasons a well-built bed stays aerobic are duller than that. The water arriving is already carrying dissolved oxygen. The flow is slow enough that it is never stripped faster than it is replaced. And, in a bed where water enters at the top and works downward, the most oxygenated water meets the highest oxygen demand first, at the surface where the fresh organic material is sitting. That is a design outcome, not a plant miracle.

Who it is right for. A large regeneration zone, a cool climate, low bather load, a flat site, a builder who wants as little machinery as possible, and somewhere spares are hard to buy. On that pool it is a fine choice.

Method two: the pumped upflow gravel bed

This family covers a lot of ground: the classic bog filter out of the koi pond world, the OzPonds method, and the American Aquascape systems, which are close cousins. The principle is the same in all of them. A pump pushes water in underneath a gravel bed, through a perforated pipe or a distribution void, and the whole flow is forced upward through the media and the plant roots before returning to the pool.

Upflow has a real advantage and it is worth naming: pushing water up through a bed spreads it more evenly than pulling it down, with less design effort. Every litre has to pass the biofilm, and the plants sit exactly where the nutrients are. If you want a bed that distributes acceptably without much thought about the pipework, upflow is the forgiving option.

I do not build upflow, for reasons I will come to. But the advantage is genuine and I am not going to pretend otherwise.

The basic bog filter. At its simplest this is one pump, a pipe under the gravel, and a bed. It is the cheapest and simplest thing in this article, and on a wildlife pond with light loading it is close to unimprovable. The nutrient stripping is genuinely good, better than most people expect, and the planting establishes fast because the roots are in moving, nutrient-rich water rather than in soil.

Its weakness is that one pump does everything, so every solid in the water is delivered into the gravel, and gravel does not digest a leaf. Organic solids collect in the lower part of the bed, they mineralise, they block the pore spaces, flow finds the easy channels and stops visiting the rest of the bed, and the parts that no longer see flow go anaerobic. The gravel becomes the thing that fouls rather than the thing that filters. It happens slowly enough that nobody can name the day it went wrong, and the fix is to dig the bed out and wash it.

The OzPonds and Aquascape refinement, which is a real improvement. These designs add two things, and both of them matter.

The first is an intake bay. Instead of a pipe stub somewhere, you build a bay that water is drawn into, positioned and shaped so that it also skims the surface. Debris is pulled in off the top rather than being left to sink. I rate that concept highly. It is the cheapest way anyone has found to get the surface-nutrient problem partly under control with a single pump, and it is the thing I would steal from these designs if I were only allowed to steal one.

It is an improvement rather than a solution, though, and it is worth being clear about why. Everything the intake bay pulls in is still in your pool system, sitting in the bay. The leaves have to be physically fished out, and somebody has to actually do that. Worse, the fine material, the pollen, the dust and the broken-down organic matter that carries most of the nutrient load, does not sit conveniently on top waiting to be lifted. It works its way down into the gravel and rocks of the bay and stays there, releasing nutrients back into the water that is about to be drawn through the filter. You have collected the debris but you have not exported it.

Compare that with a mesh basket. My skimmer runs a 300 micron stainless steel mesh that lifts out in one movement. You can pull it daily if you want to, or every few days, and every time you do, a genuinely significant quantity of organic matter and the nutrients locked up in it leaves the property. Nothing about it is clever. It is simply the difference between a place where debris collects and a place where debris leaves.

The second is a settlement void underneath the bed. Water is delivered into a large open space, often a big-diameter pipe or chamber, before it rises into the gravel. Because that space is large, the water slows right down when it gets there, and anything heavy enough drops out. The fines settle as sludge at the bottom of the void, where they are out of the media and can be pumped out later rather than dug out.

That is a properly thought-out piece of design. It means the system has somewhere for solids to go that is not the gravel, and it means the sludge and its nutrients can actually be exported from the pool. I much prefer this to any downflow arrangement, because at least here the debris, the sludge and the nutrients have a designed route out.

I have built these. They work, and they work very well.

What they still do not do. There is no strong circulation in the swim zone. The intake bay skims near itself, and the return trickles back in, but nothing is forcefully moving the body of water in the pool where people actually swim. On a garden pond that does not matter much. On a pool with swimmers, sun and a real bather load, it does.

A pump buys you options an airlift cannot. Worth saying plainly, because people underestimate this. Even with a single pump you can put the return into the bottom of the swim zone as a jet, drive circulation across the whole pool, and push water from the far end back towards the regeneration zone. You can do the reverse and deliberately draw water out of the bottom of the regeneration zone. You can put a pre-filter in the line before the bed. None of that is available to you with an airlift, because an airlift has no head to spare and no line to put anything in.

Who they are right for. For a garden ecosystem pond I rate OzPonds very highly: cheap, understandable, forgiving and genuinely DIY-friendly. For a swim pond with a modest bather load, either version is a reasonable answer, especially if you take the intake bay seriously and add a second pump for skimming and circulation, which you can do.

Upflow versus downflow is the wrong argument

Before I get to my own method, the thing I most want to correct.

The internet argument about these systems is almost always upflow versus downflow, and it is not the important question. Upflow is better, we can agree on that and move on. The questions that actually determine whether your pool is clear in five years are these four.

1. What goes into the regeneration zone? If solids go in, the bed is a settlement trap with plants on top, and its working life is measured in a few years.
2. How fast does water move through it? This sets contact time, and contact time is the whole point of the media.
3. How uniformly does it move through it? A bed where 30% of the area takes 90% of the flow is a bed three times smaller than the one you paid for.
4. Where do the captured nutrients ultimately go? Not where are they captured. Where do they physically leave the pool. If the answer is "they stay in the gravel", you have built a nutrient store, not a filter.

Every design decision I make follows from those four, not from the direction of flow.

For the record, and it surprises people who have read the forums: I build downflow. I draw water down through the bed and collect it underneath. I have just told you that upflow distributes more evenly, so that needs explaining, and I will do it properly in the next section. The short version is that the distribution problem is solvable with design, and pulling downward buys me two things that I am not willing to give up.

The arithmetic, because this is where people guess

Two calculations decide the size and speed of the whole system, and neither is difficult.

Flow rate is set per square metre of regeneration zone, not per litre of pool. Suppose you are running 3,000 litres an hour through a regeneration zone with a surface area of 30 m². Each square metre is receiving 100 litres an hour. That loading rate is what you are designing to, and it is what tells you whether a bed is being asked to do too much.

Contact time falls straight out of the bed dimensions. Take that same bed at 30 m² of surface and 800 mm deep. That is 24 m³ of gravel. Washed gravel is roughly 35% void space, so the water actually inside the bed at any moment is about 8,400 litres. At 3,000 litres an hour, water spends nearly three hours in the media.

That is the number to hold on to, and it is why bed depth is not a detail. Double the flow rate to 6,000 litres an hour and contact time halves to under an hour and a half. Build the same bed at 300 mm instead of 800 mm and you have cut it to about an hour, on the same footprint, with the same pump, for the sake of some excavation. Go the other way and halve the flow to 1,500 litres an hour and the contact time looks wonderful, but now your 40,000 litre pool turns over only 0.9 times a day and the far end of the swim zone has gone quiet.

There is a floor on depth for a second reason, which I will come back to under root intrusion.

Turnover, for the same example. 3,000 litres an hour is 72,000 litres a day. On a 40,000 litre pool that is 1.8 turnovers a day, which is where I like to be for a pool people swim in.

This is exactly what you cannot do with an airlift. Not because airlifts are bad, but because you cannot set the number, so you cannot check the number.

What it takes to move 3,000 litres an hour. Roughly, and check these against the actual equipment in front of you rather than trusting my catalogue memory:

• A small pond pump rated at 3,000 litres an hour against low head draws somewhere around 40 W. Running continuously that is about 0.96 kWh a day.
• An air blower big enough to drive an airlift delivering a comparable flow at a few centimetres of lift is typically a linear pump around 60 litres of air a minute, drawing something like 50 to 60 W.

So on energy the two are close, and the airlift is not obviously winning. It gets you aeration for that money, and it costs you flow control, skimming, layout freedom and quiet. That is the trade, stated fairly, and it is why I do not see the payoff.

Method three: the two-pump design

This is what I build, and it comes from one decision: skimming and biological filtration want opposite things, so they get their own pumps.

Skimming wants high flow, rarely. Skimming is a surface-tension effect. A short, fast draw pulls a real surface current across the entire pool and sweeps the film into the skimmer. A slow continuous trickle makes a small dimple near the weir and lets everything else drift about until it waterlogs and sinks. So the skimming pump is large and runs on a timer. On my pools it runs five to ten minutes in every hour at high speed, and its discharge also feeds the pool jets, so the same burst drives circulation through the whole body of water. The debris it collects goes into a 300 micron stainless steel mesh basket, which lifts out in one movement and tips into the garden.

Biological filtration wants low flow, always. The second pump runs a slow continuous flow through the regeneration zone at whatever rate the arithmetic above says. It never sees floating debris, because the other circuit already took it. So the bed is not being asked to be a rubbish trap. It is being asked to be a biofilm reactor, which is what gravel is genuinely good at.

The power cost of running two pumps is smaller than it sounds, because the big one is barely on.

Read the last row twice. The two-pump system uses less energy per day than a continuously running air blower, moves more water, and does aggressive whole-pool skimming that the airlift does not do at all. Applying a tariff of R2.80 or about 20 US cents a unit, that is roughly R98 or seven dollars a month.

The reason is that the expensive pump is off 55 minutes in every hour, and the pump that runs all day is a small one.

Why I pull water down through the bed rather than up. I promised to come back to this, because I have already conceded that upflow spreads water more evenly for less design effort.

The distribution problem is real, and it is solvable. Even spread through a bed is a function of how large the collection orifices are, how far apart they sit, and how the underdrain beneath the gravel is laid out and balanced. Get those three right for your design flow and a downflow bed distributes perfectly well. It is arithmetic and careful setting out rather than luck, which is precisely why it is the part of a plan set people tell me they could not have produced themselves. So I am not accepting worse distribution. I am accepting more design work in exchange for two things.

The first is that debris ends up where I want it. Whatever survives the skimmer is drawn downward onto the top surface of the regeneration zone, spread thinly across a large area, sitting in the open air and light. That is the best possible place for it. An upflow bed pushes water out of that surface, so anything settling there is being lifted and pushed back into the pool.

The second is that I get the water in a pipe, and that changes what I can do next. An upflow bed delivers its treated water as a slow, diffuse seep across the whole top surface, which then spills back into the pool. That water is going straight home and there is nowhere to intercept it. Pull downward and the same water arrives in a collection pipe under pressure, where I can send it somewhere before it returns. What I send it to is a submerged aquatic zone, and that is worth explaining properly, because it is the least understood part of natural pool planting.

Marginal plants and submerged plants are not doing the same job.

Marginal aquatics are the plants everybody pictures: the reeds, sedges and irises standing up out of the gravel at the edges. Their roots are in the saturated media, their leaves are in the air. They photosynthesise using carbon dioxide from the atmosphere, and they draw nutrients mostly through their root systems from the substrate and the water passing through it. They put on a lot of biomass, that biomass is largely above the water, and pruning it is a genuine nutrient export. They are also what makes the pool look like the picture that made you want one.

Submerged aquatics live their whole lives underwater. Hornwort, elodea, milfoil, pondweed and their relatives. Every part of the plant is in contact with the water, so they do not have to take nutrients up through roots and move them anywhere. They absorb dissolved nutrients directly across their entire surface area, out of the water column, continuously.

That difference matters enormously for phosphate. Phosphate is usually the nutrient that decides whether you get algae, and the phosphate you care about is dissolved in the water, not sitting in the substrate. A submerged planting takes that dissolved phosphate up far more effectively than a marginal planting does, per square metre, because it is bathed in the water rather than merely rooted beside it. It is also competing with algae for exactly the same nutrient in exactly the same place, which is the whole game. As a bonus it releases oxygen straight into the water during daylight rather than into the air.

Submerged plants are not a free lunch. They need light, so the zone has to be shallow and the water has to be clear, which means it works downstream of the filtration rather than instead of it. They die back seasonally, and if you leave the dead material in there it hands every nutrient it took up straight back. Harvesting is not optional.

So my regeneration zone is two phases, not one. Water goes down through the gravel bed with its marginal planting, where the biofilm does the bacterial work and the solids stay on top. It is collected underneath and passed through a submerged aquatic zone, which strips dissolved phosphate out of the water column. Then it returns to the pool. The two zones are doing different chemistry on the same water, in the order that suits each of them, and pulling downward is what makes that sequence possible at all.

Where the nutrients actually go, which is question four. Debris that does make it past the skimmer settles on the top surface of the regeneration zone rather than on the floor of the swim zone, which is exactly where you want it. Sitting on an open, aerobic gravel surface across a large area, most of it simply breaks down, and a good deal of the nitrogen leaves the system as gas. The phosphorus that gets released is taken up by the planting, both the marginals and the submerged zone, and it leaves the pool when you cut them back. Every six months to a year I vacuum that surface, or simply disturb it by hand so the loosened material drifts off and is picked up by the skimming circuit and lifted out in the mesh basket. That is a twenty minute job, not an excavation.

The honest problems with my own method. Two pumps is two pumps: more capital cost, more plumbing, more fittings, and more that can eventually fail. If your budget is genuinely tight, that is a fair objection and I am not going to argue you out of it.

For that reason I also have a design that runs on a single pump, for anyone willing to compromise a little on clarity in exchange for fewer failure points, less plumbing and lower capital cost. It is a real trade rather than a lesser version pretending not to be, and on some projects it is the right call. What you give up is the ability to skim hard and filter slowly at the same time, and you will see that in the water at the margins rather than in any dramatic way.

The other watch item is root intrusion. Roots go looking for water and nutrients, and an underdrain is full of both, so given long enough they will find the collection orifices and start closing them. Two things deal with this.

Depth is the main defence. I build my beds deep, 800 mm as an absolute minimum. That puts a substantial thickness of media between the root mass, which stays in the upper part of the bed where the oxygen and the nutrients are, and the collection system at the bottom. A shallow bed puts the roots and the pipework in the same place and then asks you to be lucky.

Species selection is the other. No Typha, no Phragmites, however good they look in photographs. Those are the ones that will find your underdrain and mean it. Better-behaved plants, Cyperus textilis among them, do the same job without the aggression, and even then I am choosing with the pipework in mind.

What I never do is wrap anything in geotextile. People suggest it constantly and it is a trap. It does not stop roots in the long term, it only delays them. And while it is failing to do that, it is quietly doing something worse: you have installed a large filter mat, buried under 800 mm of gravel, that will progressively blind with fines and that you have no possible way of cleaning. You cannot backwash it, you cannot lift it, and you cannot see it. When it finally chokes, the fix is to dig the entire bed out. It solves a manageable problem by creating an unrecoverable one.

And the last honest thing, which cuts against a lot of natural pool marketing including some of my own earlier material. The reeds are not the filter. The gravel and the biofilm growing on it are doing the bulk of the bacterial work, and they are doing it whether the marginals above them are thriving or sulking. The marginals earn their place: they take up nutrients, they are an export route every time you cut them back, they shade the bed and they are the reason you wanted a natural pool rather than a swimming pool. I would never build without them. But if you believe they are the treatment, you will size the planting and undersize the gravel, and you will end up with a green pool surrounded by beautiful reeds.

The plants that punch above their weight are the submerged ones, and almost nobody puts enough of them in. That is the part of the planting I would fight for.

Comparing them honestly

How to choose for your own site

Ignore the method for a moment and answer five questions.

Which of the three standards are you building to? Answer this before anything else, honestly, and answer it about the pool you will actually be looking at in February rather than the one in the photograph. If a garden ecosystem pond is genuinely what you want, you can build far more cheaply and simply than anything I would specify, and you should. If you want water that is clear to the bottom all season with nobody in the family complaining, the requirements tighten sharply and most of the cost difference between the methods is accounted for right here.

How much organic material lands on your water? This decides more than anything else. A pool under deciduous trees, or anywhere with a real autumn, needs the surface cleared mechanically and cleared early, or you are simply feeding the pool. If that describes you, a system with no skimming will disappoint you, and the second pump earns its cost.

Where can the regeneration zone physically go? If it must sit at the same level immediately beside the pool, everything here works. If it needs to be higher, further away, or behind a structure, the airlift is out on hydraulics alone.

How many people swim, and how often? Bather load is nutrient load. A family pool in daily summer use is a different animal from an ornamental pond someone occasionally gets into.

How warm does the water get? Warm water holds less dissolved oxygen and cycles nutrients faster, which shortens the margin for error in every design here. Hot climates are where undersized, low-oxygen beds fail first.

Where I settled, and why

Ten years of building these, and of being called out to look at ones that were not working, has taken me to one conclusion: for a similar amount of money and labour, you can have a pool that behaves like a natural swimming pool rather than like a pond you can swim in. That means water that is clear right to the bottom, low maintenance, and no sludge accumulating on the floor.

The design that gets there is a regeneration zone large enough to take the flow slowly, plus a second pump dedicated entirely to skimming and circulation. Almost every struggling natural pool I have been asked to rescue had the same underlying story: one flow rate doing two jobs, solids going into the media or onto the floor, and a bed that had quietly turned into a sludge store two or three summers before anyone noticed.

That does not make the other methods wrong. For a garden ecosystem pond I would happily build OzPonds and have. For a big wildlife swimming pond in a cool climate with a light bather load, Pagan Butler's approach is elegant and I understand exactly why people love it. Everything has its use case. Mine is the pool that a family swims in several times a week and expects to look immaculate, and that is a harder brief than it sounds.

If you have not committed to a method yet, the useful next step is to see a complete regeneration zone set out properly: laterals, header, orifice sizing to flow rate, bed depth, planting and both pump circuits, on a real drawing rather than in a diagram. That is what my natural pool plans contain, and the hydraulic calculations in them are the part buyers tell me they could not have worked out for themselves. If your site is unusual enough that none of these obviously fits, a one-on-one consultation is usually a faster way through it than another month of reading.