hikes.guideLake District

Every “The View” panorama is generated from a terrain model, not photographed. Here is exactly how — and what it does, and does not, claim.

Home › Features › How we compute the view

How we compute the view

By Lee McCoy — the method, the sources, and the honest limits, stated in the open. Updated 14 July 2026.

On the summit of a Wainwright you look out and think: what’s that fell over there — have I climbed it? Our panoramas answer that. Each is a computed 360° skyline with every visible named top labelled by distance and bearing. Wainwright drew view diagrams from every top by hand; this is the computed, updatable, uncertainty-honest successor.

Every panorama is computed, not photographed. It shows what the terrain model says you could see in perfectly clear air — not a picture of a particular day. Real-world haze routinely limits the view to well under the computed maximum.
On this page
  1. The terrain model
  2. Casting the horizon
  3. Two questions, not one
  4. Skyline vs mid-ground
  5. What we don’t claim
  6. How we check it
  7. Sources

The terrain model

Everything starts from one elevation model: Copernicus GLO-30 (Copernicus / ESA), a global digital elevation model at 30 metre resolution, free to use with attribution. We reproject it onto the British National Grid (OSGB36 / EPSG:27700, Ordnance Survey) so that distances and bearings are computed in metres on the same grid the rest of the site uses. We use one model for the whole run — a seam between two models with different error profiles is exactly the kind of artefact a careful reader would find, so we don’t create one. GLO-30 reaches Snowdonia, the Galloway hills, the Isle of Man and the Pennines, which is as far as any real Lakeland sightline goes.

Two honest properties of GLO-30 matter. First, it is a surface model: it includes tree canopy and buildings, so a sightline grazing woodland can carry a few metres of noise. Second, its stated vertical accuracy is 4 metres (the LE90 figure — 90% of points fall within it) in the product specification (Copernicus / ESA); on the steep, rugged ground typical of the Lakes the error runs larger still. We therefore treat anything inside a conservative 10 metre band as uncertain, and never call it a confident sighting (see below). Ten metres is deliberately more than the 4 metre spec — it absorbs the steeper-terrain error, the surface-model canopy noise, and the fact that a 30 metre grid places a thin ridge only approximately.

Casting the horizon

From each of the 214 Wainwright summits we cast rays outward every 0.25° — 1,440 rays for the full circle — and walk each ray across the terrain to find the highest point on the skyline in that direction. Along the way we apply the curvature of the Earth and atmospheric refraction (the slight bending of light that lets you see a little further than a straight line would). The observer’s eye is set 1.6 metres above the summit — a representative standing height across adults and children; we don’t ask for yours, because at these distances it changes nothing you’d notice. Distances are straight-line, summit to summit.

Two questions, not one

A naive panorama answers one question — is it visible? — with a yes/no that hides how much the answer depends on the day. We split it into two honest questions, and label every top on both.

1. Geometry: does it clear the horizon?

Refraction is not a constant, so we cast every sightline three times — at a poor, a standard and an exceptionally clear-air value — and look at the spread. A top earns Confident only if it clears the skyline even on the poor day and by more than the terrain model’s error. Anything that clears only in kinder air, or by less than that error, is Marginal — honestly uncertain. This is what stops a 7-metre clearance masquerading as a solid sighting.

2. Haze: how often is the air clear enough?

Here is the subtlety that surprised us. We expected the far horizon to be geometrically marginal — on the edge of visibility. It isn’t. Snowdon, 164 km from Scafell Pike, clears the Welsh ridge in front of it by a robust 62 metres at every refraction value: geometrically, it is rock-solid. The reason you see it only a handful of days a year isn’t geometry — it is haze, the aerosol and moisture in 164 km of air, which our terrain model knows nothing about.

So the rarity rides on a second axis, keyed to distance: near tops are visible whenever it’s not raining on you; the far four-nations horizon needs a rare, cold, clear day. That is why a distant label reads “Snowdon · 164 km · a handful of days a year”: geometrically confident, atmospherically rare. The caveat becomes the headline.

Skyline vs mid-ground

A top can be visible but sit in front of higher ground, rather than breaking the skyline itself. We distinguish the two: a top that forms the horizon at its bearing is marked on the ridge; a visible-but-lower top is a mid-ground label. It matters for how the panorama reads — and it is in the data, not just the picture.

What we don’t claim

How we check it

Because every one of the 214 fells gets an automatic, honest self-label, the checks exist to prove the confidence model is calibrated — not to hand-verify 214 individual views. We spot-check a handful of fells in both directions: the Confident tier should reproduce published summit view-indicator plates, and the rare tier should match the known long-sighting lore. On our canary sightlines it does: Snowdon from Scafell Pike lands confident-but-rare; Criffel from Skiddaw — a 7-metre clearance, inside the model’s error — is correctly held at marginal, while from Scafell Pike, where it clears by 15 metres, it is confident. And the celebrated four-kingdoms view from Scafell Pike comes out true: on the clearest days the far horizon holds Snowdonia (Wales), the Isle of Man, the Galloway hills (Scotland) and the Mourne Mountains of Ireland — each a real, named summit. The model earns that by rejecting the impostors: an obscure 214-metre hill near Belfast that briefly mis-flagged “Ireland” was dropped once we required a distant nation-sighting to be a substantial peak. Anything a tier gets wrong is now a visible, systematic bug rather than a needle in a haystack. Cross-checks against specific toposcopes are ongoing and recorded as we do them.

The companion summit intervisibility report is the same computed dataset seen a different way — which Wainwrights can see which.

Sources

  1. Copernicus / ESA — Copernicus DEM Product Handbook (v5.0) — GLO-30 at 30 m; absolute vertical accuracy 4 m (LE90)
  2. Database of British and Irish Hills — DoBIH v18.4 (CC BY 4.0) — hill names, heights and prominence
  3. Ordnance Survey — OSGB36 / British National Grid (EPSG:27700) — the projection we reproject onto

Fell names, heights and prominence for our own tops come from the hikes.guide fell database; distant tops from the Database of British and Irish Hills (CC BY 4.0). Terrain: Copernicus GLO-30. Generated 14 July 2026.