Counting Trees from the Sky
When most people look at a wooded parcel listing, they see “40 acres, wooded.” Maybe the listing mentions hardwoods. Maybe it says “mature timber.” There’s usually an aerial photo showing a green canopy. That’s all you get.
What you actually need to know — how many merchantable trees are on the property, how big they are, what they might be worth, whether the stand has been selectively harvested or clearcut in the past — requires hiring a forester to walk the property and do a timber cruise. That costs money and takes time, and nobody does it during the browsing phase of a land search.
So buyers make one of the most expensive decisions about a wooded property — whether the timber is a significant asset — based on essentially zero information.
What’s in the Point Cloud
USGS 3DEP LiDAR covers most of the rural land in the eastern United States. A plane flies over, fires millions of laser pulses at the ground, and records every surface each pulse hits — ground, understory, mid-canopy, upper canopy. The result is a three-dimensional point cloud: a spatial record of every surface in the forest, measured at roughly 2-5 points per square meter.
From that point cloud, you can derive a canopy height model — a map showing how tall the trees are at every point across the property. That’s the starting point for most remote sensing forestry. It tells you the height of the canopy, which is useful but limited.
TALON goes further. Instead of just measuring canopy height, the trunk detection pipeline attempts to identify individual trees and estimate their characteristics.
How the Detection Works
The algorithm works from the inside of the canopy outward. In a LiDAR point cloud, tree trunks don’t show up as clean cylinders — the laser pulses that make it through the canopy hit branches, bark, and understory vegetation at various heights. But there’s a pattern: in the bole zone — the section of trunk between the first major branches and the lower canopy — returns tend to cluster in tight vertical columns.
The pipeline processes the point cloud in horizontal slabs, looking for clusters of returns at trunk-like spacing. A real trunk produces clusters that appear at consistent positions across multiple slabs — a vertical stack of hits that traces the bole. Random canopy hits don’t do this.
For each detected trunk, the pipeline estimates height from the canopy height model directly above it, then uses regional allometric equations to estimate diameter at breast height (DBH) from height. The relationship between height and diameter varies by species group — hardwoods grow differently than pines — so the model accounts for forest type using national land cover classification.
The output is a tree inventory: location, estimated height, estimated DBH, and size class (sapling, pole timber, small sawtimber, large sawtimber) for every detected tree on the parcel.
What It Gets Right and What It Doesn’t
This is not a timber cruise. It’s important to be clear about that.
The detection works well for dominant and co-dominant trees — the ones that form the upper canopy and have clear vertical structure in the LiDAR data. These are also, conveniently, the trees that matter most for timber valuation. A 70-foot oak with 18 inches of DBH is detectable because it’s big enough to produce consistent returns through the canopy. A 30-foot suppressed understory tree isn’t reliably detectable at public LiDAR density.
The allometric models — height to diameter — are calibrated from regional forestry data, but they carry uncertainty. The same height tree could be a different diameter depending on site conditions, species, competition, and genetics. TALON reports estimates as ranges (low, mid, high) rather than point values, reflecting this uncertainty.
Species identification from LiDAR alone isn’t reliable at public data density. The pipeline classifies by broad forest type (deciduous, conifer, mixed) using NLCD land cover, not by individual species. A buyer looking at a parcel will know approximately how many large hardwoods are present, not whether they’re red oak or white oak.
What It Means for a Buyer
Even with these limitations, the inventory changes the conversation.
Before TALON: “40 acres, wooded.” Could mean anything from a recently clearcut parcel with 10-year-old regrowth to a mature hardwood stand with six figures of stumpage value. No way to tell from the listing.
After TALON: “40 acres, approximately 850 detected trees, 340 in the sawtimber size class, estimated canopy height averaging 72 feet, estimated stumpage value range $35,000-$65,000.” That’s enough to know whether the timber is a footnote or a primary asset — and whether the property is worth a forester’s visit for a proper cruise.
The data to do this has been sitting in publicly available LiDAR archives, covering millions of acres of rural land. The trees have been standing there, measured by lasers, waiting for someone to count them.