3D Reconstruction of deciduous Trees using low-cost UAV- and Crane-based Photogrammetry for Monitoring Shoot Elongation across entire Canopies
This study demonstrates that low-cost UAV and CraneCam-based photogrammetry can accurately reconstruct 3D models of entire deciduous trees with sub-centimeter precision and high completeness, providing a viable method for monitoring shoot elongation across whole canopies to better understand climate change impacts on primary tree growth.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine trying to measure how much a tree is stretching its legs in the spring. You can't just grab a ruler and measure the whole thing because the branches are too high, too twisty, and covered in leaves that haven't even fully opened yet. Scientists have been great at measuring how thick a tree gets (like checking its waistline), but measuring how long its new shoots grow (its height growth) has been a total mystery.
This paper is about a team of researchers who decided to solve this puzzle by turning trees into giant 3D video games. They wanted to see if they could use cheap drones and a special camera crane to build a perfect digital twin of a tree, down to the very tip of its newest, thinnest twig.
The Main Discovery: Drones Can See the Tiny Stuff
The big news is that yes, you can build a super-accurate 3D model of a whole tree using a drone that weighs less than a bag of flour (under 250 grams). When they flew these drones around trees in a Swiss park, they managed to create a digital cloud of points that matched the real tree with an accuracy of 5 to 6 mm. That's about the thickness of a pencil eraser!
They tested two types of drones: a tiny, super-light one (the DJI Mini 3 Pro) and a slightly heavier, more professional one (the DJI Phantom 4 Pro V2). Both worked, but the heavier one was like a sharp-eyed hawk compared to a slightly blurry owl. The professional drone gave a 3D point accuracy of 5.1 mm, while the tiny one was at 6.5 mm. The professional drone also managed to reconstruct 98% of the tree's branches, while the tiny one still got a very impressive 92%.
The "Fake Twig" Test
How do you know the drone isn't just making things up? The team got creative. They 3D-printed a fake branch (they called it a "Ground-Truth Branch") with specific, known thicknesses: 3 mm, 5 mm, 7 mm, and 10 mm. They taped this fake twig onto a real tree and then flew the drones over it.
The result? The drones could see the thinnest 3 mm part of the fake twig. However, the drones had a little trouble with the exact width. The professional drone guessed the thickness was about 3 mm too wide, and the tiny drone guessed it was 7 mm too wide. So, while they can see the tiny shoots, they aren't quite perfect at measuring their exact thickness yet.
What They Ruled Out (The "No-Go" Zone)
The researchers didn't just try one thing; they tried a few things that didn't work.
- The "Blurry" Drone: They tested a third drone (the DJI Mavic 3T) but quickly kicked it out of the game. Its camera sensor was too small, making the images too blurry to see the tiny branches.
- The Flying Laser: They also tried using a drone with a laser scanner (LiDAR) attached. This was a bust. The laser scanner was too "noisy" and couldn't see thin branches clearly; it made a 1 to 2 cm thick branch look like it was twice as wide. They concluded that for measuring tiny shoot lengths, this method is a dead end.
- The "Crown Surface" Trick: They noted that past studies often just measured the "skin" of the tree (the outer shape of the leaves). But that doesn't tell you how much the individual branches are growing inside. This paper insists you need the full 3D skeleton, not just the skin.
The Crane and the "Construction Site" Camera
Not all the data came from drones. In a second location, they used a camera system mounted on a giant construction crane (called a CraneCam). This is like having a camera that sits on a crane arm and takes photos as the crane spins around. Even though the camera wasn't as close as the drones (meaning the "pixels" were bigger, about 5.2 mm at the top of the tree), it still managed to create a detailed 3D model. This suggests that if you have a forest with a giant crane nearby (like the Swiss Canopy Crane II), you could monitor trees without needing a pilot to fly a drone in the wind.
The "Skeleton" Problem
The ultimate goal is to turn these point clouds into a "skeleton"—a digital stick-figure version of the tree where you can measure exactly how much each branch grew. The team tried using a computer program called TreeQSM to do this.
Here is the catch: The skeletons they got so far are a bit messy. They look like a tree, but they have "glitches"—some branches are disconnected, and some fake branches appear where there are none. The authors are careful to say that right now, these skeletons are not yet ready to be used for precise growth measurements. It's a promising start, but the computer still needs to learn how to untangle the mess of leaves and overlapping branches.
The Challenges
It wasn't easy. The researchers had to fly the drones manually because no software exists yet that can handle flying so close to a tree in such a tight pattern. They also had to fight the wind, which made the branches shake and the photos blurry. They had to fly 26 times over a few months to get a full time-lapse of the tree growing.
The Bottom Line
This paper proves that you can build a highly detailed 3D map of a whole tree using cheap, consumer-grade drones, capturing details as small as 3 mm. It rules out laser scanners for this specific job and shows that while the 3D models are amazing, turning them into a perfect "skeleton" for measuring growth is still a work in progress. The authors suggest that with better AI and more time, this could become a standard way to watch trees grow in real-time, helping us understand how climate change affects their development.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.