WEST LAFAYETTE, Ind. — Through the integration of aerial and ground mobile sensors and mapping systems, a team of digital forestry researchers from Purdue used state-of-the-art technology to locate, count and measure more than a thousand trees in a matter of hours .
“The machines count and measure each tree – it’s not an estimate using modeling, it’s a real forest inventory,” Songlin Fei said.the Dean’s Chair in Remote Sensing and professor of forestry and natural resources and head of Purdue University Digital Forestry Initiative. “This is a groundbreaking development on our path to using technology for rapid and accurate inventory of the global forest ecosystem, which will improve our ability to prevent forest fires, detect disease, to perform an accurate carbon count and make informed forest management decisions.”
The technology uses manned aircraft, unmanned drones and backpack-mounted systems. The systems integrate cameras with light detection and ranging, or LiDAR, units as well as navigation sensors, including global navigation satellite systems (GNSS) and inertial navigation systems (INS). A Purdue team led by Ayman Habib, Thomas A. Page professor of civil engineering and head of Purdue Digital Photogrammetry Research Groupwho co-led the project with Fei, designed and created the systems.
“The different parts of the systems take advantage of the synergistic characteristics of the acquired data to determine which component has the most accurate information for a given data point,” Habib said. “This is how we can integrate small and large scale information. A single platform cannot do it. We needed to find a way for multiple platforms and sensors – providing different types of information – to work together. This gives a full picture at an extremely high resolution. Small details are not lost.
A machine learning algorithm developed by the team to analyze the data is as important as the custom autonomous vehicles they created. The results of a study using their technology are detailed in an article published in the journal Remote Sensing.
“This system gathers a variety of information about each tree, including height, trunk diameter, and branching information,” Habib said. “In addition to this information, we maintain accurate location and time tags of acquired features.”
The result is like giving a person some much-needed glasses. What was once hazy and uncertain becomes clear. Their vision is improved and therefore their understanding of what they see.
LiDAR works like radar, but uses light from a laser as a signal. LiDAR sensors estimate the distance between the scanning system and objects using the time it takes for the signal to travel to objects and back to the sensor. On drones, planes or satellites, it takes measurements from above the tree canopy, and on roving vehicles or backpacks, it takes measurements from below the canopy. Airborne systems have continuous access to the GNSS signal to determine sensor location and orientation after GNSS/INS integration and provide reasonable resolution. Ground-based systems, on the other hand, provide more detail and finer resolution, while suffering from potential GNSS signal outages, Habib said.
“This cross-platform system and processing framework takes the best of each to deliver both fine detail and high positioning accuracy,” he said.
For example, if the backpack is in an area with limited access to the GNSS signal, a drone can step in and put that data in the right place, he said.
“It’s a breakthrough in applying new geomatics tools to forestry,” Fei said. “It solves a real and pressing challenge in areas such as agriculture and transportation, but it’s also incredible engineering and science that will be applied beyond one area.”
As the different platforms work together, the system also identifies data points from each that correspond to the same tree characteristic. Eventually, this could correlate and find out what the above-canopy data means in terms of what’s happening below the canopy, Habib said. That would be a giant leap in speed and the area of forest that could be covered.
LiDAR can be used to create 3D digital maps of trees and forests, making it possible to virtually assess tree growth, land cover and forest condition. A map created by the team is available here.
The Digital Forestry Initiative is part of Purdue Next moves. The team continues to work on scaling the technology and improving machine learning.
The Hardwood Tree Improvement and Regeneration Center and the US Department of Agriculture’s (Hatch Project No. IND10004973) fund this work.
Writer: Elizabeth K. Gardner; 765-441-2024; [email protected]
Sources: Songlin Fei; 765-496-2199; [email protected]
Ayman Habib; [email protected]
Comparative analysis of multi-platform, multi-resolution and multi-temporal LiDAR data for forest inventory
Yi-Chun Lin, Jinyuan Shao, Sang-Yeop Shin, Zainab Saka, Mina Joseph, Raja Manish, Songlin Fei and Ayman Habib
LiDAR technology is rapidly evolving as various new systems emerge, providing unprecedented data to characterize the vertical structure of the forest. Data from different LiDAR systems exhibit distinct characteristics due to a combined effect of sensor specifications, data acquisition strategies, as well as forest conditions such as tree density and canopy cover. Comparative analysis of cross-platform, multi-resolution, and multi-temporal LiDAR data provides guidelines for selecting appropriate LiDAR systems and data processing tools for different research questions, and is therefore of crucial importance. This study presents a comprehensive point cloud comparison of four systems, Linear and Geiger Mode LiDAR on a manned aircraft and Multibeam LiDAR on an Unmanned Aerial Vehicle (UAV), and backpack developed in-house , taking into account different forest covers. scenarios. The results suggest that proximal dorsal LiDAR can provide the finest level of information, followed by UAV LiDAR, Geiger mode LiDAR and linear LiDAR. The new Geiger mode LiDAR can capture a significantly higher level of detail while operating at a higher altitude compared to traditional linear LiDAR. The results also show: (1) that the percentage of forest cover has a critical impact on the ability of aerial and ground systems to acquire information corresponding to the lower and upper parts of the forest cover respectively; (2) all systems can obtain adequate ground points for the generation of digital terrain models, regardless of canopy cover conditions; and (3) point clouds of different systems are in agreement within ±3 cm and ±7 cm along the vertical and planimetric directions, respectively.
Agricultural communications: 765-494-8415;
Maureen Manier, Head of Department, [email protected]
Agricultural news page