Diffraction loss from partial Fresnel zone obstruction, explained

You ran the path profile, confirmed 0.6 F1 clearance at every obstacle, and signed off on the hop. Eighteen months later the link is fading on wet mornings and nobody touched the antennas. Partial obstruction is almost always the culprit, and it's the kind that gets missed because the path still technically clears.

What's actually happening at the obstruction

A Fresnel zone is a volume: an ellipse in cross-section that bulges widest at mid-path. Line-of-sight clearance only tells you the direct ray is unobstructed. It says nothing about how much of that surrounding ellipse is clear.

When something intrudes into the zone without touching the direct ray, like a growing canopy that reaches up to 0.4 F1 or 0.3 F1 of clearance instead of the 0.6 F1 your design called for, the path doesn't go dark. It diffracts. Some of the signal energy bends around the obstruction and some gets absorbed or scattered by it, and what arrives at the far end is weaker than the free-space calculation assumed. That shortfall is diffraction loss, and it's additive to whatever fade margin you built in.

Roughly how much loss, and why it isn't linear

Knife-edge diffraction geometry gives you the shape of the curve even without running the Fresnel-Kirchhoff integral by hand. At 0.6 F1 clearance, added loss is close to zero, which is exactly why 0.6 F1 became the standard design target. Clearance right at the direct line-of-sight edge, 0.0 F1, costs you around 6 dB. Push the obstruction further in, so it's blocking the direct ray by a quarter wavelength or more, and loss climbs past 10 dB fast, then keeps climbing the deeper it goes.

The part that matters for planning: the curve is steep near the LOS line and shallow out near 0.6 F1. That means a canopy sitting at 0.5 F1 clearance is doing you almost no harm yet, but the next couple of meters of growth cost you disproportionately more than the couple of meters before it did. Once vegetation crosses that knee, a link degrades slowly for years, then falls off a shelf in one wet season.

This is also why a signal-quality complaint is a lagging indicator. By the time BER climbs enough for someone to file a ticket, the tree has usually been eating into the zone for several seasons already. The loss was there at 2 dB, then 4 dB, then 6, quietly consuming margin that was budgeted for rain fade, not foliage.

Where this bites on real hops

Low hops over regrowth, agricultural buffer strips, or anything crossing a creek line with a windbreak are the usual candidates, since those are exactly the spots where canopy height isn't static and nobody's walking the right-of-way every season to check it. A hop that cleared fine at commissioning with young trees at 8 m can be sitting at 14 m five years later, and the same obstacle that cost you nothing at 0.6 F1 clearance is now costing you 8 to 10 dB, with no update to the link budget to reflect it.

The fix is knowing which specific spans have canopy creeping toward the 0.6 F1 line before the dB loss starts compounding, so trimming gets scheduled where it's needed instead of guessed at across every hop. Link Vegetation Monitor runs that check once a year, pulling canopy height along each path and flagging it against your Fresnel clearance envelope so a span crossing the line shows up as a line item, not a fault ticket.

If you'd rather see which hops are already eating into that margin than wait for the next outage report, that's the problem Link Vegetation Monitor was built to flag first.