Toward an Architecture for Subalpine Forest Monitoring Using Commercial Off-the-Shelf Unmanned Aircraft Systems and Sensors

摘要

Forests of the western United States have, in recent years, been subjected to increasingly severe disturbance regimes. To study interactions between these disturbances and their impact on forested systems, ecologists have traditionally relied on a combination of fine-scale ground surveys and coarse-scale satellite imagery. Unmanned aircraft systems have the potential to bridge a gap in resolution between these sources of data. The work presented here takes an exploratory step in bridging this gap. A prototype unmanned aircraft system consisting of commercial off-the-shelf components, namely a quadrotor drone and a compact multispectral camera, was developed. An accompanying concept of operations for flying the system was also designed, and was carried out to collect imagery over plots of forest in the Colorado Front Range. The images were processed into three dimensional point cloud reconstructions and multispectral reflectance maps. These are to be used to explore development of segmentation and classification algorithms to derive forest metrics from the data, as well as spectral unmixing models on moderate resolution satellite imagery from Landsat, which estimate the percent of each pixel occupied by land cover types. It was found that the concept of operations enables sufficiently safe and robust operation of the prototype system in the rugged subalpine environment. However, it was determined that a more stable aircraft platform was needed in order to collect imagery of sufficient quality for scientific analysis. This work represents a first step in enabling large-scale, regular, and detailed observations of forest metrics from unmanned aircraft systems in subalpine forest ecosystems, which may inform efforts to understand the effects of disturbances on these systems across multiple scales.