The most precise line on a very old map

A company in this sector recently published an article about the tools it uses. It's worth reading, because it makes a claim a lot of people in this market seem to believe — and I think the claim is wrong.

The claim is that this is, at bottom, an engineering problem: that defensible mitigation depends on “state-of-the-art GIS technology that less technically equipped operators are not currently using,” and that firms without the same kit are, by implication, producing something weaker. The company is proud — reasonably — of rejecting “pre-defined polygons” and “desk-based assumptions” in favour of precise, high-resolution catchment delineation. Certainty about the quality of mitigation, it says, matters as much as certainty about its availability.

Let me grant the honest part first, because it is real. Catchment delineation is skilled work. Getting the boundary right — knowing which sub-catchment a site actually drains into, tracing flow paths across flat ground where the water's route is not obvious — genuinely matters, and doing it well takes proper hydrological work rather than a coloured-in map. Anyone who tells you otherwise is selling something. This is not an argument against good GIS. We use it too, and we take it seriously.

It's an argument about where the uncertainty actually lives, and about the difference between being precise and being certain.

Because here is the thing about a nutrient budget: precision at the catchment boundary is the last precise step in a chain that is, of necessity, broad-brush almost everywhere else. You can draw the line to the metre. It does not make the number certain — because the number was never settled at the boundary. It is settled upstream, in the national datasets the boundary is drawn from, and downstream, in the calculator the boundary feeds. And in both directions, the people who built the regime have been open that it deals in approximation, not precision. That candour is to their credit — and it's the whole of my point.

The boundary is a sensible working line, not a physical constant

Start with the catchment layer itself — the polygon a GIS platform delineates so carefully. Where does it come from?

Natural England explains, in its own metadata for the nutrient neutrality catchments, that the boundaries were “extracted from N2K Diffuse Water Pollution Plans Catchments data alongside WFD catchment data, River Basin Management Plans and Surface Water Catchment mapping.” So far, so technical. But it goes further, and — to its credit — it tells you plainly what happened next:

…these datasets were also used by Councils to generate their own mapping using local knowledge and their in house tools. Datasets were occasionally extended to estuary boundaries or to exclude parcels which local knowledge of hydrology proved to be inaccurate.

That is a sensible, pragmatic way to build a national boundary — blend the best available datasets, then let people who know the ground correct them where the hydrology on paper doesn't match the hydrology in the field. Natural England deserves credit for doing it that way, and more credit for saying so openly rather than presenting the line as a purely mechanical output.

But notice what it means. The object that state-of-the-art GIS delineates so precisely is, by its own authors' account, a working line — a reasonable, locally-informed, honestly-caveated boundary, not a razor-edged physical constant waiting to be traced to the nearest metre. Delineating it at ever-finer resolution does not make its origins any more exact. You can draw a very sharp line around a boundary that was, sensibly and admittedly, drawn with judgement. And it is still being refined: of the catchment polygons in the published layer, several carry no amendment date at all, and the most recent revisions are dated 2026. It is a live, maintained dataset — which is a good thing — but a live thing is not a fixed thing to be precise about.

The calculator is a pragmatic tool, and says so

Now go downstream, to the calculation the boundary feeds.

The competitor's article is proud of avoiding “desk-based assumptions.” Yet the method everyone in this market must work within is, by design, a desk-based lookup. And there's nothing wrong with that. The nutrient budget calculators the councils distribute (I have opened two, for the Broads and the Wensum) are, sensibly, built for use by every planning authority and applicant in the catchment. They are not bespoke models run afresh for each site; they look up pre-computed values against a handful of categorical keys. The column headings say so — “Farmscoper Farm Term,” “Farmscoper Soil Drainage Term” — because they draw on Farmscoper, an established, Defra-recognised agricultural model. Using a standard tool consistently is the right call for a national regime. It is what makes one applicant's answer comparable with the next.

What follows from that design is a set of deliberate, reasonable simplifications — and, again, they are disclosed, not hidden. On the wastewater side, the Broads calculator lists 82 named treatment works; the nitrogen figure is a single standard value for 80 of them, with phosphorus at a common default for most. On the land side, the inputs are grouped into workable classes: several soil-drainage descriptions map onto a smaller number of categories, and rainfall is grouped into bands you read off a map by matching a colour to a legend. None of this is sloppiness. It's what a usable national tool has to do: trade fine granularity for consistency, so one applicant's answer means the same thing as the next one's.

The only thing worth flagging is what happens to the impression of precision at the end. The coefficients come out of the spreadsheet carrying full floating-point tails — a phosphorus figure stored to seventeen significant figures. That is an artefact of the software, not a claim about the world, and Natural England would be the first to agree it does not mean seventeen figures of real-world certainty. But it is a neat illustration of the gap between looking precise and being certain: seventeen decimal places, resting on a farm type the methodology tells the applicant to “ascertain from the current farmer.”

Seventeen decimal places, resting on an input the method tells you to get by asking the farmer. Precise notation; approximate world. The regime knows this — and does not pretend otherwise.

The evidence base is necessarily not live — and no one should expect it to be

The trigger for the whole regime is a protected site being in unfavourable condition because of nutrients, a status recorded in Natural England's condition assessments. Those assessments, across the country, are mostly not recent: a large majority of SSSI units were last formally assessed some years ago, and several nutrient neutrality sites were last assessed around 2009 or 2010.

I want to be careful here, because this is where a lazy version of the argument would reach for a cheap shot, and it would be wrong to. Assessing the condition of tens of thousands of SSSI units across England is an enormous, resource-hungry task, and no agency on any realistic budget could keep every unit under continuous, real-time review. The dates being old is not evidence of neglect; it is the arithmetic of a vast estate and finite people. And it is not the only evidence in play — the Environment Agency monitors and classifies water quality separately, and that record is more recent, though its most recent national classification is itself a few years old.

So the fair statement is a modest one, and it is not a criticism of anybody. The formal evidence base for these decisions is, of necessity, not a live feed. It cannot be. The condition record is often several years old; the water-quality record is a little fresher but not real-time either. That is simply the nature of monitoring the natural world at national scale — and it sets a ceiling on how much certainty is available to anyone, however good their software, before a single catchment boundary is drawn.

Certainty is the word that matters

Which brings us to the word the company used, and the word the law uses. They are the same word, and that is the whole point.

The legal test behind every one of these decisions — the Habitats Regulations test under every Appropriate Assessment — is certainty. An appropriate assessment, in the standard formulation, must contain:

…complete, precise and definitive findings which are capable of removing all reasonable scientific doubt as to the absence of adverse effects on the integrity of the site.

Complete, precise, definitive. All reasonable scientific doubt, removed. Now, here's the part that matters — and it reflects well on Natural England, not badly. NE does not pretend its method delivers that on its own. It describes the nutrient calculation, in its own published methodology, as having “been developed as a pragmatic tool,” adding that “for each input there is a degree of uncertainty.” It handles that uncertainty exactly as an honest regulator should — by building in a precautionary margin and advising authorities to err on the safe side, rather than by claiming a false precision. The people who built this regime have never sold certainty. They have been candid that they are managing uncertainty carefully, which is a different and more honest thing.

That candour is precisely why the marketing rings false. When a firm sells “certainty about the quality of your mitigation,” it is offering something the regime's own authors have expressly, and rightly, declined to claim. You cannot buy, with proprietary GIS, a certainty that the underlying framework — honestly — does not assert. A sharper catchment boundary is a genuine improvement to one link in the chain. It cannot manufacture a certainty that was never available at the boundary in the first place, because the boundary was never where the doubt lived.

This is the difference between precision and accuracy that every first-year engineer is taught and this market keeps forgetting. Precision is how many decimal places you can quote. Accuracy is whether the number is right.

You can be extraordinarily precise about a figure built from grouped categories and a several-year-old assessment — and the precision does not make it any truer. It just makes it look truer.

The rigour that actually counts

None of this means technical skill is beside the point. It matters enormously — but not in the way the article implies. The question isn't whether you own the most expensive tool. It's whether your working can be checked.

Take the one genuinely advanced capability the piece points to: the proprietary terrain software that traces where a discharge really drains across flat ground, where the nearest blue line on a map tells you nothing. That is worth having. Knowing the actual flow path, rather than assuming the closest watercourse, is exactly the sort of thing this assessment ought to rest on — and we think so strongly enough that we do it ourselves. We just do it differently: we reproduce that same analysis, of where the water actually goes, from open and published data, by methods that can be re-run by anyone who cares to check them.

The difference is not capability. It is trust. A proprietary flow path asks a regulator to accept a precise-looking line because of the software that drew it. A reproducible one lets them redraw it. When the legal test is the removal of all reasonable scientific doubt, a number a regulator can independently reproduce is worth more than a number they have to take on trust — and it makes their job easier, not harder.

That principle runs through the rest of it, and it points the opposite way to precision for its own sake. We measure to the water's edge, not merely to the centreline of the nearest channel. Where a criterion cannot be positively evidenced, we resolve it toward less credit, never more — the cautious reading, not the flattering one. We show each contested judgement on a map for the regulator to examine, rather than asserting it and asking them to take our word. And we would rather a report never issued at all than issue one that claimed more than the evidence supports. That is a different idea of rigour from “we own tools you don't, so trust our numbers.” It is: here is the working, on data you can obtain, by methods you can repeat, rounded down where the evidence runs thin — check it.

The honest version

We have a commercial interest in this, and I'm not going to pretend otherwise. We install package treatment plants; so, as it happens, does the company whose article prompted this one. We're competitors, not neutral observers, and you should weigh everything above accordingly.

But the argument underneath is not an attack on anyone's data, least of all Natural England's. NE has built a pragmatic, transparent regime under real constraints, and has been honest at every turn about the uncertainty it manages — that is the behaviour of a good regulator, and this piece should not be read as a complaint about it. The argument is narrower, and it is aimed squarely at a single overclaim: that proprietary precision at the catchment boundary amounts to certainty. It doesn't — and the regime's own authors are the first to say the certainty on offer is bounded. Precision you can't check isn't certainty. It's confidence with good production values.

So by all means invest in the best technical tools going — we are glad the sector is getting more capable, not less. But be honest about what precision at the boundary can and cannot buy. It can draw a beautiful line. It cannot, by itself, remove a single reasonable doubt — because the doubt was never at the line. For that you need working someone else can pick up and repeat, and the discipline to round down when the data runs out.

The most precise line in the world, drawn around a working boundary and a farmer's best estimate, is still just a very neat drawing of something the rest of us should be allowed to check.

Quotations and figures are from Natural England's own published material — the Nutrient Neutrality Generic Methodology (NECR459) and the nutrient neutrality catchments dataset metadata — and from POSTnote 755 (Parliamentary Office of Science and Technology, December 2025). The certainty standard is the Habitats Regulations test applied at Appropriate Assessment. Short quotations from the third-party article are reproduced for fair comment.