Multispectral digital camera mounted on helicopter boom would take aerial photos.

When Ottawa forestry researcher Darren Pouliot saw people walking around in shorts last April he started to worry.

After listening to the long-range weather forecast, he threw some equipment into a rented jeep and began a 21-hour marathon drive to the boreal forest near Sioux Lookout, four hours from Thunder Bay in northwestern Ontario.

Pouliot, 28, needed to collect data for his PhD research and it depended on a narrow window of time between snowmelt and new spring growth. 

If he missed the window, his work would be delayed by a full year. 

Driving straight through, with his research assistant Mark Lindsay spelling him off at the wheel, Pouliot made a pit stop in Thunder Bay to pick up a $24,000 camera, on loan to his supervisor, Dr. Doug King, in Carleton University's department of geography and environmental studies.

No point-and-shoot camera, this multispectral digital wonder would allow Pouliot to use the science of remote sensing — gathering data about the Earth from the air — to take high resolution photographs of the forest during flight. 

Remote sensing uses a camera's sensor to measure energy reflected from the ground. Ground features reflect electromagnetic energy differently depending on surface roughness, moisture and light-absorbing pigments.

The degree of detail depends on the distance between the target and the sensor. 

An astronaut, for example, would see the entire boreal forest in one glance, but would not be able to distinguish individual trees. 

Scanning the forest

Remote sensing is not new to foresters. The Canada Centre for Remote Sensing has used the surveillance tool for 30 years to do reconnaissance mapping, monitor forest fires and to keep an eye on the rate and extent of deforestation. 

Doug Pitt, a research scientist with the Canadian Forest Service says aerial photography, with manual interpretation of the images, has traditionally been used. 

However, as more sophisticated digital cameras have become available, researchers like Pouliot are looking to improve accuracy and speed up the analysis of aerial photographs through computer automation.

Remote sensing is particularly useful in remote, inaccessible areas of the vast boreal forest. 

The green belt of the boreal forest stretches from coast to coast, covering 35 per cent of Canada’s landmass. 

It contains a mix of coniferous trees (softwood, with needles) and deciduous trees (hardwood, with leaves) but Pouliot was specifically interested in the young conifer trees regenerating on a patch of Crown land clear-cut nine years ago. 

In the early spring, when deciduous trees such as birch and aspen are still in a dormant, leafless state, the conifer’s foliage would appear starkly green from the air, against the drab backdrop of brown forest floor. 

With the specialized camera, Pouliot planned to fly over the forest and acquire colour images in the green, red and near infrared wavelengths of the spectrum. 

A bird's-eye-view

Viewed directly above, tree crowns look like amoeba-shaped blobs. Since the conifer’s green needles reflect large amounts of energy in the near infrared, the blobs would appear bright pink in the photo, against a blue-green background. 

Conifer tree crowns absorb light in the infrared and show up as pink blobs.

A bird’s-eye-view of conifer crowns is not only useful to count the number of trees in the image, but it also reveals important information about the tree species, height, volume and biomass. 

By measuring crown diameters, a logging company can calculate how much wood volume will feed hungry mills. 

Foresters and climatologists also use crown diameters to calculate how much carbon is stored in tree trunks. Trees act as an important global “sink,” sequestering carbon from carbon dioxide — a primary greenhouse gas — while releasing oxygen to the air. 

Close inspection of the tree crown image can also reveal the tree’s health, or whether insects have damaged the foliage.

But Pouliot wanted to use tree crown data for a relatively new forestry application: to determine how well conifers regenerate after clear-cutting.

When a tree falls in the forest

While clear-cutting can cause soil erosion and destroy habitat for wildlife, it is actually an appropriate harvesting practice in the boreal forest because it mimics the large-scale disturbances of natural forest fires. 

The key to long-term sustainability of forests depends on successful regeneration. 

Natural regeneration occurs when young trees grow back from roots and seeds. Forests can also be artificially regenerated by sowing seeds or planting nursery-grown seedlings. 

More than 50 per cent of clear-cut boreal forest is left to regenerate on its own. This can create problems. 

Sun-loving deciduous trees such as birch and aspen can outpace slower-growing conifers on clear-cut sites, competing for sunlight, water and soil nutrients.

Unchecked, they may choke out the conifers and eventually replace native conifer stands with hardwood. 

To maintain cutting quotas, logging companies must prove to forest conservation managers that conifers are regenerating, by doing a “free-to-grow” assessment, 7-10 years after clearcutting. 

"Teenaged" trees

At 7 - 10 years of age, conifers are a bit like teenagers getting ready to leave home for the first time — finally tall and mature enough to merit what the ministry of natural resources calls “free-to-grow” status.

At this point, the young conifers are likely to reach maturity, unhindered by competing species. To help conifers reach this crucial stage of development, intervention may be required, such as soil fertilization, or clearing away competing vegetation by manual cutting or chemical herbicides. 

The traditional way to check up on regeneration is to tramp through the forest or take a look from an aircraft window, which can be very subjective. 

Pouliot’s research goal is to develop an algorithm, or an automated computer program that will replace the subjective assessments of forest regeneration with more objective, cost-effective and accurate assessments. 

After their marathon drive, Pouliot and Lindsay parked the jeep outside a small trailer in the bush, on loan from Buchanan Forest Products, a private logging company that leases a tract of Crown land one hour from Sioux Lookout.

Plunging temperatures

Before rigging the camera up, Pouliot had plenty of work to do on the ground. But he wouldn’t be wearing shorts.

After rushing north to beat the heat wave, the temperature plunged. “It was miserably cold for three weeks,” he says.

Darren Pouliot and Mark Lindsay take measurements in a research plot near Sioux Lookout.

With hands growing painfully numb clasping pen and notepad, the two researchers first staked out a series of research plots, then took careful measurements of each individual tree. 

Crown diameter, trunk diameter, tree height and total above-ground biomass were all jotted down in neat rows of figures.

Measuring the same trees from the air came next. 

Chopper challenge

King arrived from Ottawa to help mount the camcorder-sized camera to the boom of a helicopter. 

It took about four days to feed cables to a computer inside the cockpit and to check all the software. 

King pioneered multispectral airborne video sensing in the 1980s when he rigged up four cameras to a VCR unit to digitize photographs in the days before digital cameras were available.

His expertise was invaluable, because once the helicopter lifted off the tarmac, the flight recorder would be billing, even at a reduced academic rate, $1,300 an hour.

When all systems checked out during a test flight, King returned to Ottawa and Pouliot went back to the forest to spend a few more days taking meticulous ground measurements. 

He was finally ready to collect the data that would keep him busy for the next two years of analysis.

Then Mother Nature did what she often does at northern latitudes in April: the day before the helicopter charter, 15 cm of snow fell on Pouliot’s conifer trees. Twenty four hours later, the temperature soared to 22 C. 

The day before the aerial flight, 15 cm of fresh snow fell on the research plot.

“We were running around to various locations in the bush, measuring snow depth,” Pouliot recalls.

But he was also keeping a wary eye on the deciduous trees, which were just ready to burst into green leaf and invalidate his data. He decided he couldn’t afford to wait another day. 

Weather finally cooperates

Fortunately mission day dawned sunny and clear, with bright sun to illuminate the green conifers.

Flying at approximately 200 metres, Pouliot spent eight hours snapping high-resolution digital photos over the research plots. Pixel by pixel, the images were fed into the computer. 

At this elevation, each pixel — the smallest element of a picture — corresponded to a 36 square centimetre area on the ground. This would be more than enough detail to detect individual conifer tree crowns.

Back on the ground, with the images safely burned onto CDs and archived, Pouliot made a preliminary check to compare the images with was was growing on the ground.  

They looked good. 

And plentiful enough to keep Pouliot busy for at least two years of analysis and testing to develop an automated computer program that will assess forest regeneration more accurately and more economically than current methods. 

When every last field measurement was recorded and double-checked, Pouliot and Lindsay packed up their gear and headed back to Ottawa. They had a lot of dirty laundry. 

But they no longer had to worry about changes in the weather.

And mercifully, they had just missed the boreal black flies.