A cruiser on the ground settles a forked-stem call in about two seconds. Walk up, look at the base, see one root collar feeding two leaders, done, it's a fork. From the air, that same tree can look like two separate crowns standing shoulder to shoulder, or one lumpy crown hiding a split you'd never catch without walking it. Getting this wrong at scale either inflates your stem count or buries cull volume you needed to flag before the sale.
What a forked crown actually looks like from above
At sub-10cm resolution, a true fork usually shows a single, continuous crown outline with an uneven top. Instead of one clean apex, you get two bulges of roughly similar height pushing up against each other, often with a faint seam or shadow line running between them where the leaders haven't fully closed canopy. The outer edge of the crown stays smooth and unbroken all the way around. That's the tell: one boundary, two high points.
Two adjacent single-stem trees read differently. Each crown has its own closed, roughly circular or oval outline, and there's a visible gap or pinch point between them, even if it's a narrow one. Shadow helps here more than color does. On a midday pass, a true gap between crowns throws a thin shadow line clear down to the understory. A fork's internal seam rarely does, because the two leaders are close enough together that the shadow stays shallow and doesn't separate the canopy underneath.
NIR response can back this up. A genuine split in the canopy (two separate trees) tends to show a sharper reflectance drop at the boundary than the softer gradient you get across a fork's internal seam, since there's real sky and lower-story vegetation in that gap rather than just two lobes of the same crown pressed together.
Low codominant forks, where the split happens near breast height rather than up in the canopy, are the hardest call from the air regardless of GSD. If the fork point is below where the crowns start separating visually, you're relying on crown shape and stem spacing patterns in the stand rather than a clean visual boundary. That's where a full-count pass earns its keep over a sampled plot. A full-count pass lets you look at every crown individually and flag the ambiguous ones for a ground check, instead of extrapolating one cruiser's fork rate across the compartment and assuming the plot average holds.
Why the call matters for the cut estimate
A forked stem isn't automatically a cull, but it's a flag. Low forks, especially with included bark at the crotch, carry higher odds of stem failure, decay columns running up from the split, and reduced sawlog recovery even when the tree stands straight otherwise. A buyer pricing the tract wants to know how many stems in the stand carry that risk before they bid, not after the mill starts rejecting logs.
Miscounting forks as two separate stems inflates your per-acre stem count and skews your volume estimate high. Miscounting two close-growing singles as one forked tree does the opposite, and it can mask a real cull indicator if one of those two stems is split and the other isn't. Either error compounds across a full stand faster than it would across a sampled plot, simply because the plot never claimed to catch every tree in the first place.
A stem-level crown map built from a full-stand pass lets you review every ambiguous crown on screen, cross-reference it against NIR and shadow, and decide which ones warrant a ground check before the cruise crew ever walks in. That's the difference between a full-count crown map for the stand and a sampled estimate that averages the fork rate out and hopes the average holds.
If you're ready to stop guessing fork rates from a plot and start counting every crown in the stand, that's exactly what Tree Crown Mapping was built to do.
