Story originally published on Txchnologist
Trees don’t make work easy for the scientists who want to study them and
the landscapes they create, since the leaves, fruits and flowers of a
forest’s canopy generally stretch far above human reach. So, a team of
researchers is turning to drones, programming these pilot-less flying
machines to buzz over forests while snapping pictures.
“I flip the switch on the controller and away it goes on its own,” says
Jonathan Dandois, a doctoral candidate at the University of Maryland,
Baltimore County, as he launched a drone over a patch of forest on
campus on an overcast October day. “It will climb up to about 100 meters
above and will continue on its route.”
The spiderlike machine carries a point-and-shoot camera aimed at the
ground as it travels along a preprogrammed path defined by GPS
coordinates. But Dandois and fellow researchers are trying to do more
than just get a bird’s-eye view of the treetops.
Dandois is working on Ecosynth,
a suite of tools his team is developing with National Science
Foundation funding. Their goal is to allow anyone from a professional
ecologist to a citizen scientist to generate an interactive,
three-dimensional map of any landscape. He says that a confluence of
ever-cheaper drones — Dandois prefers the less ominous term autocopters —
with more accessible and user-friendly 3-D mapping and visualization
software has made this project possible.
Insights from above
Dandois envisions the project doing more than making forest surveys
easier. “It might inspire ecologists to rethink the ways they collect
data,” He says.
This could potentially include developing algorithms to find particular
features, such as flowers, in images collected from above.
In June, Dandois went to a Smithsonian Tropical Research Institute study site in Panamato
test the system. Trees there are surveyed at regular intervals by
people on the ground, so the information collected until now has been
limited to things like trunk diameter that could easily be measured on
foot, says Helene Muller-Landau, an ecologist with the institute. “But
(these measurements) are not necessarily the most interesting or useful
things you want to know.”
Ecologists’ other options, which include satellite data or
aircraft-borne scans, have drawbacks. These can have low spatial
resolution, are collected infrequently, or cost a lot, Muller-Landau
says.
Muller-Landau is interested in year-to-year variation in the cycles by
which different species of trees grow and shed their leaves, fruit and
flower. Fluctuations in these cycles reflect trees’ responses to annual
variability in climate, and hint at how they will fare in the face of
long-term climate change. To document these cycles, she wants to gather
images every week over the course of years, something not currently
feasible.
(A view of agricultural land in Costa Rica captured by a drone. Courtesy Jonathan Dandois, Ecosynth.org)
The Panamanian test run encountered a few issues: an autocopter crash
and the realization that the images they collected needed to overlap
more. However, the system proved largely successful, she says. Viewing
the images in 3-D makes tree height, as well as gaps in the forest,
leafless trees and fruits easier to identify.
“This technique is so new I think as people start to explore it, we will
figure out a lot more things we can do with it,” she says.
Maturing tech opens opportunities
Ecosynth’s approach became feasible when the cost of hobbyist-grade
multi-rotator autonomous copters fell to about $5,000 around three years
ago. At the same time, the 3-D reconstruction software necessary to
compile all of the images was becoming easier to use. Ecosynth computer
scientists are working to make the software more accessible by
developing their own open-source 3-D mapping program.
Scanning the Earth’s surface from above, a technique known as remote
sensing, requires a great deal of technical sophistication. But Ecosynth
intends to open this field to just about anyone.
Because 3-D data often comes in prohibitively large files, engaging
newcomers means making this information easy to visualize and access
remotely, Dandois says. This is where Ecosynth’s social dimension comes
in. Its collaborators are developing a system to transform the 3-D map
into a virtual world users can access with nothing more than a standard Internet browser. In it, users can interact with each other in real time by, for example, highlighting trees and chatting. A demonstration is available on the Ecosynth website.
Ultimately, making this system as accessible as possible means ditching
the copters, since some training is necessary to use them, Dandois
says.
NSF funding is also supporting Ecosynth’s attempt to develop a system
that allows users to collect images with a smartphone or tablet in lieu
of a copter. This could allow a user to walk through a forest, a park or
a backyard and collect images that could be employed to generate a
virtual version of anyone’s ecosystem, he says.