Custom Engineering · Scenario Customization

Custom Forestry Survey Systems

Forestry systems have to cope with a canopy that hides most of what a survey normally measures, and with access that is difficult in every direction.

10 of 10Method note

Overview

Design for the ground control network first: canopy blocks both line of sight and GNSS quality.
Decide whether the deliverable is canopy metrics, terrain, or both, and state what cannot be delivered.
Season determines visibility; build the survey calendar around leaf-on and leaf-off periods.
Terrain and canopy models should be versioned together, since a change in one often explains a change in the other.

Forestry customisation starts with the sampling design

Forestry survey is the clearest case of a discipline where the aircraft is the smaller half of the problem. What determines whether a survey produces an inventory or merely a picture is the sampling design: what is measured everywhere, what is measured on plots, and what is inferred from the plots. Get that right and a moderate aircraft produces an inventory; get it wrong and an excellent aircraft produces images.

The three-tier structure is standard. Canopy extent and height are measured wall-to-wall at moderate resolution. Species composition, health and understorey condition are measured on representative plots at higher resolution or with a different sensor. Volume and yield are then estimated by extrapolating plot measurements across the area using the wall-to-wall canopy layer.

The custom work is designing that tiering for a specific forest: which metrics, at what plot density, with which sensor, at what season, and with what field plot programme to calibrate against. That design is specific to the stand type, the terrain and the client's inventory requirements, and it is the deliverable we scope first.

Sensor selection against the deliverable

A visible-light camera delivers canopy extent, colour, texture, gaps and — through photogrammetric reconstruction — a canopy height model. That is sufficient for a surprising amount of forestry work and it is the cheapest data per hectare. Its limit is structural: it sees the top of the canopy and infers what is below.

Where the deliverable needs a ground surface under canopy or stem-level structure, a LiDAR payload is required, because it returns multiple elevations per pulse as it passes through gaps. The trade is mass, power, cost per hectare and a heavier calibration and processing workload — but under closed canopy, no camera resolution substitutes.

Where the deliverable includes vegetation condition or species discrimination, multispectral data adds information that neither of the other two provides. It is not a substitute for either; it answers a different question and it brings its own calibration and illumination discipline.

Field plots, and why they are not optional

Ground truth is the part most often omitted from forestry programmes and most often regretted. A set of measured field plots, tied to the airborne data, is what turns a canopy model into a calibrated estimate rather than a relative one. Without them, the airborne data is internally consistent and externally unanchored.

Field plots are a ground task with real cost, and they should be scoped as part of the programme rather than discovered as a gap at the analysis stage. The number, distribution and measurement protocol depend on the stand variability and the accuracy the inventory requires, and designing them is part of the engagement.

Seasonality is tied to the same decision. For terrain and stem structure, leaf-off acquisition is usually better; for species and health, leaf-on usually is. Whichever is chosen must be held across years, because an inventory is a time series and a season change breaks comparability in the same way a sensor change does.

Access, connectivity and the remote-terrain reality

Forestry terrain is often remote, which brings specific constraints: limited launch sites, poor or absent connectivity, long transit and weather that changes with elevation. Launch sites need to be surveyed and recorded, and on large blocks a staged site may be the difference between one sortie a day and several.

Connectivity matters more than teams expect, because it determines whether data can be verified on site or only after return. Plan for offline verification: sufficient onboard storage, a field check procedure that does not need a network, and a clear rule about re-flying uncertain sorties.

Where data must be processed within the client's own infrastructure — common where forest inventory data is commercially or regulatorily sensitive — our on-premise deployment and data management work covers that model.

Starting a forestry engagement

The path is: define the metrics and the inventory schema; design the tiered sampling and the field plot programme; select the sensor against the deliverable rather than against capability; fix the season and the acquisition geometry; then run a pilot block and verify the metrics before committing to the full area.

The pilot block matters because forestry deliverables are derived rather than direct. Verifying that the chain produces the metric the client needs — on a small area, with the field plots in place — is much cheaper than discovering the gap after flying the whole estate.

Send your area, stand type, required metrics, inventory schema, accessible season and access constraints to our engineering team. We will propose the sampling design and sensor set, and be explicit about what field work the inventory requires.