Overview

This project will develop an unmanned aircraft system (UAS) that can fly under a forest canopy, collect lidar and multispectral data of surface and ladder fuels, and convert these data into usable fuels products using machine learning. An autonomous fleet of these drones could assess three-dimensional subcanopy fuels, bulk characteristics (living vs. dead), and moisture content, providing near-real-time data at various scales to pre- and active-fire management teams.

Science Area

Wildfires in temperate forests have been increasing in size and severity, and wildfire management has shifted from a primarily suppression-based approach to a more proactive fuels management approach that combines fuels reduction and ecosystem restoration. Monitoring pre- and post-fire landscapes at scales matching management action is challenged by an inability to characterize subcanopy structure, namely surface and ladder fuels. Currently, spaceborne and most airborne active and passive sensors are unlikely to fully capture the fuel bed composition at landscape scales relevant to fire risk and spread modeling.

Technology

This project will develop, test, and demonstrate a UAS fleet each equipped with lidar, RGB cameras, and a high-performance GNSS-aided inertial navigation system. Multispectral cameras may also be included pending further evaluation. The sensor payload will be designed to weigh less than ten pounds, enabling each drone to conduct mapping operations lasting 20 to 30 minutes. Sensor data will be used to simultaneously navigate safely beneath the forest canopy and to construct dense point cloud representations of the subcanopy forest environment, with point cloud layers colorized with available sensor data.

Advancements

• Lidar data, GNSS/INS, and camera data will be fused to construct high-density multi-channel 3D point cloud maps of the subcanopy forest environment.
• Global and local trajectory planning will ensure that the entire region of interest is mapped and that the UAS follows a dynamically-safe trajectory through the subcanopy environment.Machine learning techniques will be developed to convert the colorized point-cloud data into fuels metrics relevant to forest management with a goal of rapid turnaround of the products following a flight.

Principal Investigator

Jonathan Greenberg is an associate professor in Remote Sensing at the University of Nevada, Reno. His research focuses on understanding the impacts of climate change and land use change on vegetated ecosystems through the use of advanced remote sensing technologies. Greenberg’s work spans scales ranging from individual plants to global ecosystems and includes both terrestrial and aquatic environments. He utilizes a wide range of state-of-the-art remote sensing data, including hyperspectral, hyperspatial, multi-temporal, thermal, and lidar imagery. His expertise supports the development of innovative approaches for ecosystem monitoring, wildfire science, and environmental analysis. Greenberg earned his B.S. from Boston University in 1996 and his Ph.D. from University of California, Davis in 2004.