Run a tree inventory¶

Turn a segmented forest plot into a tree list: one row per tree with its stem position, DBH, stem curve, lean, height, crown base height, crown size, crown volume and how crowded it is by its neighbors. Then summarize the plot (stems/ha, basal area, QMD, Lorey's height, canopy cover, biomass), and build a QSM for every tree in one run. Each DBH comes with the evidence behind it: how well the circle fit and how much of the stem was actually seen. Browse and sort the trees in the Tree Table, click a tree to fly to it, record species and status, and export the list and the stem curves as CSV.

How every value is measured, and the published method it follows, is on Tree inventory measurements and Stand metrics.

Before you start¶

The inventory reads the labels the segmentation tools write, so run them first, on the same cloud:

Step	Writes	Why the inventory needs it
Segment ground points	`ground_class`	Keeps ground out of the trees; the fallback ground source.
Generate a DEM with Height above ground ticked	`height_above_ground`	The terrain under each tree. Every height, and breast height itself, is measured from it.
Segment individual trees	`tree_instance`	One row per tree id.

Without a DEM, heights are less reliable

Without the height-above-ground column, the ground is estimated from the ground-labeled points. Without those either, each tree's own lowest point is used, which is biased on slopes. Those trees are flagged ground_from_tree_min, and the panel says which ground source it will use before you run.

Sky/miss points and deleted points are never measured.

Run it¶

Select the cloud, then open Tools → Tree Inventory (or press Cmd/Ctrl+K and search "inventory").

Check the settings:

Setting	Default	Meaning
Breast height	1.3 m	Where DBH is taken, above the uphill ground and along the stem. Or 4.5 ft (1.37 m), the US convention.
Circle search	RANSAC	The robust search that finds the stem's points in a slice: RANSAC, or the randomized Hough transform. Both are refined by the same precise fit.
Crown voxel	0.10 m	Only affects crown volume: the number of cubes of this size that hold a point. Smaller cubes report less volume, larger ones approach the crown envelope. Use roughly 2–3× the crown's point spacing, and one value across trees you compare.
Min points / tree	50	Trees with fewer points are skipped (and counted in a warning).
Competition	6 m	Search radius of each tree's competition index (the CI column). Every neighbor with a DBH within the radius adds its DBH ÷ this tree's DBH ÷ their distance, so bigger, closer neighbors count more. Trees closer than this to the plot edge are marked in the CSV's `edge` column, since their outside neighbors are missing.

Click Run inventory. The bar shows which tree is being measured. Cancel stops it between trees.

The trees are measured one at a time, straight from the cloud's store, so a large plot doesn't need to fit in memory at once.

The Tree Table¶

Each row is one tree:

Column	Meaning
Tree	The `tree_instance` id. A ⚠ marks a tree with a quality flag. Hover the row to read it.
DBH cm	Diameter at breast height.
H m	Height above the ground at the stem base.
CBH m	Crown base height.
Crown ⌀ m	Mean crown diameter.
CPA m²	Crown projected area.
Lean °	Stem lean from vertical.
BA m²	Basal area.
H/D	Slenderness.
CI	Hegyi's competition index: higher means larger, nearer neighbors.
Overlap %	Share of the crown projection covered by other crowns.
QSM m³	Woody volume of the tree's QSM (after Build QSMs).
AGB kg	Above-ground biomass by the Stand tab's method (when one is chosen).
Species, Status, Label	Your own entries.

Sort by clicking a column header. Click it again to reverse. Trees with no value in that column always sort last.
Click a row to select the tree and frame it in the viewer. Its overlay turns pink.
Species, Status (live, dead, damaged, uncertain) and Label are yours to fill in. They are exported with the tree list. When you re-run the inventory on the same cloud, an entry is kept only for the same tree: the same id, with its stem base within 0.5 m of where it was. Re-running Segment Trees renumbers trees, so an entry whose id now names a different stem is dropped rather than moved onto the wrong tree, and a warning says how many were dropped. Running the inventory on another cloud does not lose them: each cloud keeps its own entries and entered plot area for this session, and they come back when you re-run that cloud.
The table always belongs to the last run. If you have since selected another cloud, the panel says whose inventory it is showing. If you have since changed that cloud (re-segmented it, edited it, or moved it), the panel says the result is out of date.

Your entries live in this session

Species, status and labels are not saved with the cloud. Export the tree list to keep them.

Quality flags¶

Flag	Meaning	What to do
`partial_arc`	Less than half the stem's circumference was seen at breast height.	Usual for a single scan position. The DBH is less certain; scanning from more positions fixes it.
`few_points`	Fewer than 20 points on the DBH circle.	Check the stem in the viewer; the scan may be sparse there.
`high_residual`	The circle fit is noisy: branches, ivy, or buttresses at breast height.	Inspect the DBH circle overlay.
`no_stem`	No circle could be fitted at breast height (often an occluded stem). DBH, basal area and slenderness are blank.	Check the segmentation; the tree may be merged with a neighbor or cut off.
`ground_from_tree_min`	No ground data under this tree.	Run Segment Ground and Generate DEM.

Stand summary¶

The Stand tab summarizes the plot from the tree list:

Figure	Meaning
Plot area	The convex hull of the ground points (measured), or the area you enter.
Trees	Trees counted: DBH at or above Min DBH (default 5 cm), and not marked dead unless you untick Exclude dead trees.
Stems / ha, Basal area	Per hectare of plot area.
QMD	Quadratic mean diameter.
Lorey's height	Mean height weighted by basal area.
Mean / max height	Of the counted trees.
Canopy cover	Share of the plot under at least one crown.

Below them are the DBH and height class histograms; set each class width in the box beside it.

Enter the plot area for small plots

The measured boundary is the scanned ground's outline, which is rarely a surveyed plot's. For small plots, where the edge matters most, enter the surveyed area.

Biomass (optional): pick a method and it is estimated per tree (the AGB kg column) and totalled for the stand, in Mg and Mg/ha.

Method	Needs	For
Chave 2014	DBH, height, wood density	Tropical trees.
Jenkins 2003	DBH, a species group	United States species groups.
QSM volume × density	a QSM per tree (QSM tab), wood density	Any tree whose woody structure is well scanned.

One species group and one wood density apply to the whole stand. Trees a method can't estimate are left out of the total, and the summary says how many. Stand metrics gives each method's range and source.

Stand summary CSV writes the figures, the settings that produced them, and the class tables.

QSM per tree¶

The QSM tab builds a QSM for every tree in the table in one run:

Set Max points per tree (default 60,000, at most 500,000). Larger trees are thinned evenly on a voxel grid.
Leave Wood points only ticked if the cloud has leaf/wood labels: a QSM models the woody structure.
Click Build QSMs. Each tree's model is added to the scene, named after the tree. Cancel stops between stages.

Each tree's woody volume joins the table (QSM m³) and can feed the QSM biomass method. A tree that can't be modeled (too few points, or a skeleton that doesn't connect) is counted as failed without stopping the others. QSM metrics CSV lists every tree, with the reason for each failure.

Overlays¶

With DBH & stem overlays ticked, the viewer draws, for every tree:

the DBH circle at breast height, perpendicular to the stem;
a cross at the stem base on the ground;
a line from the stem base up the stem axis to breast height.

The overlays stay on screen after you close the panel. Untick the box to hide them. They also hide while their cloud is hidden, and once the cloud has changed since the run (re-segmented, a DEM or edit that rebuilt it, or a transform in progress), because they would no longer sit on its stems. Re-run the inventory to bring them back.

Export¶

Tree list CSV — one row per tree, with every measurement, its quality evidence and your entries.
Stem curve CSV — one row per 0.5 m stem slice: position, diameter and fit quality.
Stand summary CSV — on the Stand tab.
QSM metrics CSV — on the QSM tab, one row per tree.

The columns are described in File formats.