The Threefold Force
Base satellite, high resolution, LiDAR, drone — we have every layer. The skill is putting each one where it does its best work.
A vegetation management programme covers the whole network, but the money is decided on a few kilometres. Trees growing into the line are not spread evenly along the route — they appear in patches, and those patches change from year to year.
A clearance, a planting, a change of land use — creates a new patch, or removes an old one.
A storm, a dry year — or simply a stand reaching the height at which it starts to matter.
No single layer answers every question
Detailed imagery over every kilometre — even where nothing is changing. The detail lands where nobody needs it, and the rest of the toolkit sits unused.
It watches the whole network and shows where the bands are shifting — but it does not measure the individual tree, so the final question stays open.
The money is not in the survey, it is in the intervention
In pruning, felling, forestry work, call-outs and outages. A monitoring system earns its place when it works continuously, across the whole network, and sends each data layer only where its question is.
Good monitoring, moreover, does not add to the cost of the work — it cuts into it: it sends the crew where the work actually is, and skips the sections that need nothing this year.
A monthly fee that keeps working
A single large survey project strains the annual budget, then goes stale within eighteen months. A monthly fee spreads out, stays predictable and — more importantly — shows the current state, not last year’s.
One measurement — Most of the network does not even run through woodland
So that this is not merely a claim: we selected three network areas and measured them end to end. In total, more than 159,000 kilometres of medium-voltage line, with a ±10 metre band on either side, compared against public 30 metre land-cover data.
Take the first. On this 22,580 kilometre network, only 5.2 percent runs through woodland. The rest passes through built-up areas, farmland and low vegetation — where the growth either never reaches the conductor, or is mown and managed every year.
The other two areas told a very different story. Same method, same band width — and the share of wooded sections ranges from 5.2 to 39.1 percent . On one network every twentieth kilometre needs vegetation attention; on another, almost every third.
There is no single right number. There is no industry average to plan from — which is exactly why it must be measured before anyone writes a budget into a spreadsheet.
The first force — The base system: the whole network, continuously
The customer does not buy data; they get a framework with satellite processing already inside it. This layer covers the whole network, not occasionally but continuously: the same section gets a fresh reading continuously.
Which span is fine, which one has started growing, where the stand’s condition has declined, where drought stress shows.
It does measure — but its purpose is not to give the height of every single tree to the centimetre. The purpose is watching the bands: where the situation is changing, and in which direction.
It is which thirty kilometres to inspect this year. The classification answers exactly that, and it refreshes every month.
What this yields is more than any single image could give. Readings have been accumulating over the same place for years, across seasons and growth stages — and from that time series the pace of leafing, year-on-year change and declining condition can all be read. Our own processing is what brings that out.
The second force — A global model with local ground truth
A model trained on one country works in that country. We train the same height and species-composition model across biomes on several continents — because networks also run through several climates, and because diversity is what makes the model stable even where it has never been.
The same species grows differently in another climate, and behaves differently beside a conductor. A single-country model cannot know this.
We compare what the model says against measured, tree-level truth. Where it diverges, we refine it with local data.
The refinement feeds back into the global model, so the system gets more accurate with every deployment — not only where it happens to be working.
So the customer is not buying a boxed product but a model calibrated to their own area, with every earlier validation standing behind it.
The third force — High resolution — but only where it is needed
Once the first two layers single out a section, the targeted layer takes over — high-resolution imagery, LiDAR or drone. At that point we are not buying an image — we are choosing a tool for a purpose: one kind of image tells you how tall the tree is, another tells you whether something has been built on the route.
Nobody has to work out which tool fits each section: our system knows every available data product and proposes the right one for the purpose. We recommend; the customer decides.
Is the resolution fine enough for the question; is it from a single day or stitched from several years; what is the supplier’s minimum order and minimum swath width; does the corridor-shaped area or the bounding box work out cheaper.
Not km², not scenes, not credits. The unit a network operator already thinks in.
| What you want to know | What we pull in for it | Billed as |
|---|---|---|
| How tall are the trees beside the line? | Surface and terrain model; canopy height computed from the difference, broken down by span | € / km / year |
| What has changed since the last survey? | Comparison of repeat images with known acquisition dates — new growth, clearance, construction | € / km |
| Has anything been built on the route? | High-resolution imagery on which a building or a new road is unmistakable | € / km |
| Where did trees fall on the line after the storm? | Priority acquisition across the whole network within 24–72 hours — here the question is time, not price | case by case |
What else it answers — The same chain, different questions
Vegetation is our foundation — our oldest and best-validated field. But the technology is the same: satellite time series, height model, change detection and object recognition. Whatever can be answered this way, we can do — the only difference is what we look for.
Vegetation along the network
Tree height by span, band classification, growth tracking. Trained and validated against LiDAR.
Rapid post-storm assessment
Where trees fell on the line, where the route broke, where vehicles cannot get through — hours before the first patrol, across the whole network at once.
Here the question is time, not price.
Unpermitted construction
A new building, extension or outbuilding on the route or in the safety strip — before anyone moves in, and before it becomes a legal case.
Encroachment into the safety strip
Not just vegetation: stockpiled material, containers, parked vehicles, staging areas under the line.
Railway overhead line
The same task on a different track: vegetation along the rails and the surroundings of the overhead line.
Changes to assets
Substation, tower, access road, fence — what was built, what disappeared, what deteriorated since the last image.
Land-use change
Clear-cutting, a new road, illegal dumping, earthworks around the route — what becomes a risk tomorrow.
Terrain and elevation model
0.5 m surface and terrain model for design, route selection, line-of-sight and sag calculations.
The real question — It is not whether we can measure height
That question is settled here. We can give height, band classification and growth trend — and we can measure how accurately we do so. The question is at what cost we do it, and how often.
Because there is a trap here that every vegetation programme runs into: no single layer gives both the detail and the rhythm. LiDAR gives the tree-level truth, but it is flown at long intervals — between two flights it misses the change, and in five years a tree grows by exactly the amount the whole question is about. High-resolution imagery and drone give detail where they are sent; the continuous base gives the whole network every month. On its own, each leaves the same gap — dynamics or detail.
The threefold force resolves precisely this dilemma. Not by compromise — it is not that you get a slightly worse version of everything. It is that each layer does the job it is best at: the continuous base monthly, the model always, and LiDAR, high resolution or drone exactly where and when the other two have said they are needed.
What we do not replace — We do not take the human out of the process
This is worth saying plainly, because many promises today claim the opposite. The actual intervention — what to cut, how, under what conditions, on whose land, under what permit, with what machinery you can get there — is always decided on site. This cannot be determined from orbit, and we do not claim otherwise.
What we do is two things. We shorten the path: the crew does not spend the week finding the problem but solving it. And we optimise the risk: we direct attention where the network is genuinely vulnerable, so that the chance of failure converges, step by step, towards zero.
The decision stays with the specialist. All we want is that when they go out, they already know what they will find.
Which span is at risk — from the height model and the risk matrix. That is where the machine’s job ends.
The specialist records on site: species, condition, access, protected species. Offline-first — it works without signal, then syncs.
Pruning, felling, stump removal — with local knowledge, permits and an understanding of the conditions. No image can tell you this.
What we ultimately deliver
Not satellite imagery and not a data file. A processed data catalogue: tree height by span, band classification, growth trend, risk zones — loadable into GIS, as a report, and as a work package the contractor can pick up the next morning.
The customer does not have to choose a satellite, sign contracts with suppliers or compare minimum order sizes. The three forces work in the background, and only one thing surfaces: the answer to what was asked.
And something is coming. Within a few weeks we will publish a tool on our website where anyone can…
… and from here we say nothing.
We think every network operator has wanted something like this — and simply never got it from anyone. Worth checking back.
Every layer in its own role — the high-resolution image, the LiDAR flight or the drone goes only where the other two layers have already said it is needed.