The tallest hill is rarely the problem

Both things that eat clearance peak at mid-path and vanish at the towers, so a lower obstacle in the middle beats a higher one near a tower.

Looking for the highest ground answers the wrong question

Asked to check a microwave hop, the natural move is to find the tallest thing between the two towers and see whether the beam clears it. That instinct is wrong often enough to be worth unlearning, because the two things that actually consume clearance do not care much about height on its own — they care about where along the path the height is.

Both penalties peak in the middle

The first Fresnel zone is the ellipsoid around the direct ray within which obstacles cause meaningful diffraction loss. Its radius at any point is proportional to √(d₁·d₂/D), where d₁ and d₂ are the distances to the two ends. At either tower, one of those is zero and the zone has no width at all. Halfway along it is at its widest.

The Earth's bulge — the amount the ground rises into the path because the planet is curved — is proportional to d₁·d₂. Same story: nothing at the ends, most in the middle.

On a 20 km hop at 7 GHz under standard atmosphere, that works out at:

• At 2 km from the tower: zone radius 8.8 m, bulge 2.1 m — about 11 m of penalty.
• At mid-path: zone radius 14.6 m, bulge 5.9 m — about 20 m of penalty.

So a hill in the middle is charged roughly nine metres more than the same hill near a tower. Which is why this is entirely possible:

• A 46 m ridge at 2 km leaves 78% of the first zone clear. Fine.
• A 42 m hill at 10 km leaves 49%. Not fine.

The taller obstacle passes and the shorter one fails. A survey that reports "highest point on route: 46 m at 2 km" has told you about the wrong hill.

Sixty per cent, and where it comes from

The usual planning rule is to keep obstacles clear of at least 60% of the first Fresnel zone radius. Below that, diffraction loss climbs quickly; above it, the loss is small enough to ignore. It is a convention rather than a law of physics, and different operators sit at different points — 100% of F1 for critical links, less than 60% on short hops where the zone is narrow anyway.

What matters is stating which convention is in use. "The path is clear" means nothing on its own; "the path holds 60% of F1 at its worst point, which is at 10 km" is a statement somebody can check.

The K factor is the number to stress-test

Radio bends slightly downward in the atmosphere, which is modelled by pretending the Earth is bigger and flatter than it is. K = 4/3 — an effective radius one third larger than reality — is the standard temperate-climate planning value.

It is a typical condition, not a permanent one. Sub-refractive conditions push K below 1, making the effective Earth rounder than the real one and lifting the apparent ground into the beam. These are the conditions under which marginal paths fail, and they are not rare: they arrive with particular temperature and humidity gradients, often in the early morning.

Re-running a profile at K = 0.6 takes a second and is the single most useful thing you can do to a design that looks fine. A hop with comfortable clearance at 1.33 and nothing left at 0.6 is a hop that will produce intermittent, weather-correlated outages that nobody can reproduce.

Bare earth is not what is actually there

Elevation models describe the ground. Mature trees are 15 to 30 m of obstacle sitting on top of that ground, they are not in the model, and they grow — a path signed off against bare earth with 5 m to spare has a decade, at most.

Buildings are the same problem with a shorter fuse and no seasonality. Anything on top of the ground is part of the obstacle, whether or not the dataset knows about it, and the honest way to handle it is to add the height explicitly and see what it costs.

Raising an antenna moves the problem

Solving for the antenna height that clears a path has a trap in it. Raise one end and the line of sight tilts, which changes the clearance at every point — and can change which point has the least. Solving against the worst point of the original profile can produce a height that clears that point and fails somewhere else.

The reliable approach is to solve against the whole profile: pick a height, check every point, and search for the lowest height at which all of them pass. It is slower and it is right.

Where terrain data comes from, and how it lies

If you have survey data, use it. If you have a licensed high-resolution model, use that. Failing both, a public elevation service will give you something usable — provided you know how it is wrong.

An elevation model reports an average over each cell. Averaging takes the tops off peaks and fills in valleys, so summits come back lower than they are. Checked against published heights, one common public source returned Ben Nevis 12 m low and Snowdon 52 m low.

That bias runs in the dangerous direction. A path checked against shaved-down terrain looks clearer than it is. Public elevation data is a good way to find where the problem on a route is; it is not a basis for signing one off.

It is also the one thing on this site that involves sending anything anywhere, which is why the tool shows you the exact payload — coordinates and nothing else — and does nothing until you press the button. Typed and pasted profiles do the same arithmetic and disclose nothing.

Frequently asked questions

Why is the tallest obstacle usually not the controlling one?

Because the Fresnel zone radius and the Earth's bulge are both proportional to terms that vanish at the towers and peak at mid-path. On a 20 km hop at 7 GHz the penalty is about 11 m near a tower and about 20 m in the middle, so a 42 m hill at mid-path can fail a link that a 46 m ridge at 2 km clears comfortably.

What does 60% of the first Fresnel zone mean?

Keep obstacles at least 60% of the first zone radius away from the direct ray and diffraction loss is negligible. It is a widely used planning convention rather than a law — some links are planned to 100% of F1, and short hops sometimes accept less. What matters is saying which one you used.

What is the K factor and why check 0.6?

K expresses how much the atmosphere bends radio downward, modelled as an enlarged effective Earth. K = 4/3 is the standard temperate value. Sub-refractive conditions push K below 1, making the effective Earth rounder and lifting ground into the beam — which is precisely when marginal paths fail. A hop that clears at 1.33 and has nothing left at 0.6 will produce intermittent weather-correlated outages.

Should I add height for trees?

Yes. Elevation models are bare earth. Mature trees are 15 to 30 m of obstacle that no terrain dataset contains, and they grow. A path signed off against bare earth with 5 m to spare has a decade at most.

Why solve antenna height against the whole profile?

Because raising one antenna tilts the line of sight and changes clearance everywhere, which can change which point is worst. Solving against the original worst point can produce a height that clears that point and fails at another.

How accurate is public elevation data?

Coarse, and biased in the dangerous direction. Models average over each cell, which shaves peaks — one common public source returned Ben Nevis 12 m low and Snowdon 52 m low. A path checked against shaved terrain looks clearer than it is. Use it to find where the problem is, not to sign a path off.

Does the path profile tool send my data anywhere?

Only if you choose to fetch terrain and press the button, and then only coordinates — no site names, frequencies, antenna heights or identifiers. You are shown the exact payload first. Typing or pasting a profile does the same arithmetic and sends nothing.

What is the difference between blocked and short of clearance?

Blocked means the ground is physically above the line of sight — there is no path at those antenna heights. Short of clearance means the beam passes over the obstacle but too close to it, so diffraction loss eats into the fade margin. The second is a design problem; the first is a different route.