View all

Forestry Keeps Buying Software When the Problem Is Data

Canadian forestry has digitized almost everything except the forest.

We run GIS platforms, digital cruising apps, harvest scheduling software, detailed cost models and mill systems that track fibre through production. We can calculate area to a fraction of a hectare and forecast product value to the dollar.

But the inventory data feeding those systems still comes from a sample, collected much the way it was a generation ago. We have become very good at processing inventory data. We have not become much better at collecting it. That gap is starting to cost us.

A standard built for a different margin

In British Columbia, the Cruising Manual sets a sampling error objective of under 15 per cent at two standard errors for scale-based cutting authorities, with tighter targets, typically 10 per cent, for cruise-based ones.

That standard is statistically sound. It was also set for an industry with more room in its margins. When a block was clearly profitable, a swing in volume rarely changed the decision to harvest, and paying for better inventory data didn't make sense.

That room is gone. As costs rise faster than timber values, knowing exactly what's in the block matters more. A block that once absorbed a weak estimate now breaks even on it, or loses money.

So the question is no longer whether our inventory data meets the standard. It's whether meeting the standard is good enough to make the call.

Precise calculations, uncertain inputs

Digital cruising was a real step forward. It cut compilation time, removed transcription errors and made calculations consistent. But a faster way to compile a sample doesn't make the sample any more precise.

Take a cruise estimate of 50,000 cubic metres at 15 per cent sampling error. The sampling confidence interval runs from 42,500 to 57,500 cubic metres. That is a 15,000-cubic-metre range, yet most downstream plans carry forward only the single number.

And total volume is the easy part. A cruise can meet its target on total net merchantable volume and still tell you little about species mix, diameter distribution or where the merchantable timber sits in the block. Those are what drive harvesting productivity, mill recovery and, in the end, whether the block makes money.

We calculate profitability to the dollar on inputs that carry far more uncertainty than the spreadsheet admits. A precise calculation is not a certain result.

Asking more than the inventory data can answer

The questions we ask of inventory data have changed too. Commercial thinning, partial cutting and selection systems all depend on how trees are distributed within a stand, not just how much volume it averages. Two stands with the same estimated volume can carry very different harvesting costs and product values.

Our inventory systems were answering the questions they were built for. But we are asking them new ones.

The tools have caught up

Airborne lidar has already moved from pilot to standard practice, and Natural Resources Canada reports that coverage now exceeds half of Canada's managed forest. Drone-mounted lidar and multispectral sensors go a step further, measuring individual trees across an entire block: location, height, crown size and spacing. Combined with field plots and machine-learning models, that inventory data also supports estimates of species, diameter and volume.

These methods have limits. Dense, multi-layered stands make tree detection harder. Species and volume estimates are only as good as the models behind them. Field validation still matters, and any forester should insist on it.

But the shift is real. Instead of a stand average, you can see what is actually in the block, tree by tree, before you commit to harvesting it.

Where the next dollar should go

None of this is an argument against software. Our planning, cruising and mill systems are essential. But software cannot create certainty that isn't in its inputs. If we keep investing in systems while leaving inventory data behind, we will build ever more precise models of conditions we don't actually know.

The case for better inventory data shouldn't be measured in cost per hectare or trees counted. It should be measured in decisions. Can you spot a marginal block before committing to it? Predict species and piece size well enough to improve product allocation? Know where thinning pays?

We've spent decades getting better at calculating what a forest is worth. It's time to invest as seriously in knowing what's actually there.

That is the gap Flash Forest's Forest Intelligence Service was built to close. We survey whole blocks with drone-mounted lidar and multispectral sensors, measure every tree rather than a sample, and are clear about which attributes are measured and which are modelled. The inventory data feeds the planning and appraisal systems you already run, so the software you've invested in finally gets the source data it deserves.

Sources

‍

Interested to help on our mission?
Get in Touch