Authors: Ming Shen*, University of Tennessee, Yingkui Li, University of Tennessee, Nathan McKinney, University of Tennessee, Maofeng Tang, University of Tennessee
Topics: Remote Sensing
Keywords: Kudzu, LiDAR, Sentinel-2, Remote sensing classification
Session Type: Poster
Start / End Time: 9:55 AM / 11:35 AM
Room: Lincoln 2, Marriott, Exhibition Level
Presentation File: No File Uploaded
Kudzu, an invasive plant species, has been spreading rapidly in the southeastern U.S. in recent years. These epiphytic plants climb to the top of native trees and impede their access to necessary resources, causing death of many trees and great economic loss every year. It is of critical importance to identify and map Kudzu distribution. With high altimetric accuracy of georeferenced 3D point cloud, LiDAR (Light Detection and Ranging) technology has been proven very useful in tree species classification. The recently available 10-m resolution Sentinel-2 satellite imagery provides abundant spectral information at “red edge” bands that are also sensitive to vegetation. This research investigates the benefits of combining LiDAR point cloud with Sentinel-2 imagery in Kudzu classification. This synthetic approach integrates height and surface features from LiDAR point clouds, spectral features from Sentinel-2 imagery, and topographic features from LiDAR-derived digital elevation model. Support vector machine are used to identify Kudzu from surrounding trees based on these features. The results show that point cloud density, terrain roughness, green and near-infrared reflectance are more distinguishable than other features. The combination of multi-source data provides better classification results than using point cloud or imagery alone.