Look at a wooded hill.
Choose a grove of trees the size of two basketball courts. Get up there and count every tree in that space, measure its height, check its health, and record all results.
Repeat 2.8 billion times.
Or call the Forest Service Rocky Mountain Research Station and check out their new tree map. In a 21stLast century’s version of seeing the forest for the trees, a team of fire researchers in Missoula analyzed all those pixels of satellite data to inventory the woods covering nearly a third of the continental United States.
“A lot of other people have tried to do this with limited success,” said Karin Riley, one of the researchers who helped develop the Tree Map. “When we first started getting into it, a lot of them told us we would fail. It’s a really tough problem.
What Riley and his co-designers Mark Finney and Isaac Grenfell did was start with a database called Forest Inventory Analysis – a catalog of field sample plots where a forester walked around and recorded each type and tree condition within a 30 by 30 meter square. However, this inventory includes only one sample per square mile or more.
People also read…
Then they took another database called LANDFIRE, developed by firefighters from satellite images. It reveals large-scale forest conditions like burned areas or disease infestations, also on a 30 by 30 meter grid.
Combine these two gigantic wads of information with topographical details like elevation, slope, and sun exposure, apply machine learning algorithms, and the result is a computer program that can examine any of these 2.8 billion pixels and impute their type, condition and characteristics. of all the trees inside.
“Click on any pixel on the map,” Riley said, “and it will bring up a list of trees — say, 40 trees, 30 of which are ponderosa pine, those sizes, all derived from the other maps.”
With this information, a wildfire incident manager can tell a ground crew to avoid building a line on a hill covered in dangerous dead snags. Planners looking for locations to set prescribed fires can target these snag areas or locations with low amounts of scenic or harvestable older trees. Loggers can more accurately estimate the amount of marketable timber they can expect from a timber sale. Biologists can identify key wildlife habitats.
“The snag maps produced by this are already available,” Finney said, adding that they were field-tested in real wildfires in 2020. The maps can be downloaded to a firefighter’s cellphone.
The map can also look into the future. As it is updated each year with new satellite and ground data, it can predict how landscapes might react to a thinning project versus a prescribed burn, or what a stand of trees would look like after 50 years of growth under different climatic conditions.
“It’s never been done before on a national scale,” Finney said. “Anyone, anywhere who needs access can have it. You don’t have to be a genius. People are already knocking on the door to gain access.