Convoy Protection Overwatch Sim

So how do you actually protect a patrol?

The uncomfortable arithmetic above doesn't change much as you move the sliders. You can buy a better interceptor — higher Pk, faster reaction, a second fire channel — and it helps at the margins. But the binding constraint isn't the quality of the kinetic system. It's the sensor horizon. In close country the treeline sets your warning time, and no procurement decision changes where the trees are.

A drone that crests the canopy 100 m from the road at 45 m/s gives the nearest vehicle barely two seconds. That is less than most realistic detect–decide–engage chains even before the first round leaves the launcher. Tighter convoy spacing buys mutual support — more launchers inside engagement range of each crossing point — but it also concentrates the targets, and a two-axis pincer deliberately splits that mutual support in half. Saturation is cheap for the attacker: every drone beyond your per-window shot capacity is a guaranteed leaker, and the sim's saturation counter shows how quickly that ceiling is reached.

What actually moves the outcome is elevation. A sensor-only overwatch asset pre-loads the reaction timer — the convoy's launchers are already cued when the drone breaks the treeline, which converts marginal windows into kills. But sensor-only cover does nothing about saturation: the geometry still limits how many shots fit in the window. An armed overwatch layer changes the geometry itself, engaging over the canopy before attackers ever reach their pop-up points.

The map makes the horizon effect literal. The grey wash is the column's dead ground — terrain where a drone at cruise height is invisible to every sensor in the column, because the sight line from a 3 m hull sensor cannot clear a 20 m canopy. Raise the sensor on a mast above the canopy and each wall collapses into a finite skirt; put the sensor at altitude and the wash all but vanishes — with overwatch up, the green tint marks the only terrain still hidden from it: the under-canopy interiors of the woods themselves. Same forests, three different worlds, purely from sensor height.

The conclusion we keep arriving at, scenario after scenario: organic hard-kill on the vehicles is necessary but not sufficient. It is the last layer, not the plan. Ground-based point defence handles leakers; the engagement that decides the fight has to happen above the trees. For a patrol of this size, that means some form of organic or on-call air-based protection — top cover — treated as a standing requirement rather than a luxury.

Abstract model — no real-world system parameters. Terrain line of sight is 2.5D: polygonal forest blocks with a uniform canopy height, ray-checked against sensor heights (3 m hull, 25 m mast, 400 m overwatch) and the drones' altitude profile (launch climb above the canopy, masked run beneath their own treeline, low cruise in the open). EW, acoustics, and soft-kill are deliberately out of scope for this scenario; they shift the detection curve but not the geometry argument. Headless engine: every run, plus the armed-overwatch counterfactual, is simulated in full — same seed, same dice — before the first frame is drawn.