A sub-10cm pass over a stand gives you something a cruise plot never will: every crown in the stand, not a sample of it. No expansion factor, no plot variance to defend in a basis call meeting. But a crown delineation is built from what a sensor sees looking down, and two things in particular don't show up the way a cruiser walking the ground would record them.
Suppressed stems under canopy
A crown model segments the canopy surface. It draws a polygon around the visible top of each tree and estimates size from that polygon. A suppressed stem riding under a dominant tree's branches isn't a separate feature on that surface. Its crown is folded into the shade of whatever's growing over it, so the algorithm never gets a boundary to draw.
A cruiser on a plot doesn't have that problem. They're standing inside the stand, and a 6-inch stem waiting its turn in the understory gets tallied whether or not it has daylight on its crown. That gap matters most in stands with real understory recruitment, or where an old thinning left pockets of suppressed stock that have spent the last decade creeping toward merchantable size. A full-stand crown count will show you the overstory accurately. It will undercount what's coming up beneath it, because there's nothing up top to count it by.
Multi-stem clumps in aerial counts
The second gap is the opposite problem: one crown, multiple stems. Coppice regen, low-forking hardwoods, anything that sent up two or three leaders from a single root collar reads as one crown from above, sometimes a slightly lobed one if the leaders have split enough to separate visually. The segmentation counts a crown. It doesn't know how many stems are feeding it from the ground.
A cruiser working that same clump with a diameter tape counts leaders, not canopy. If two of the three stems meet merchantable dbh, both go in the tally, both get a volume estimate. An aerial count, no matter how clean the imagery, is going to log that clump once. In stands with a history of coppice management or storm-damaged stock that resprouted from the stump, this can leave a real difference between the crown count and a stem-level cruise, especially in the smaller diameter classes where multi-leader growth is more common.
Where the ground cruise still wins, and where it doesn't need to
Neither of these gaps is a reason to throw out the drone pass. They're a reason to know what it's for. A full-stand count gives you per-tree crown size across the entire stand instead of extrapolating from a handful of plots, which is a real improvement for the dominant and codominant stock that makes up most of your merchantable volume and most of your cut estimate. What it won't replace on its own is species ID in a mixed stand, defect calls, or a diameter distribution pulled stem by stem in the places where the canopy is doing the counting instead of a person.
For a straightforward even-aged stand without heavy multi-stem history or thick understory recruitment, a full-stand crown pass gets you closer to the real stocking than a sampled cruise ever could, at a fraction of the field time. For stands where you know suppressed stems or coppice clumps are part of the story, the honest move is to pair the crown map with spot plots in those pockets rather than treat the aerial count as the whole answer.
If you're putting together a cut estimate and want a full-count crown map to start from, that's the offer behind Tree Crown Mapping.