UAV-Based LiDAR and Optical Imagery Fusion for Fine-Scale Classification of Aquatic Plant Associations in Lakeshore Wetlands

Youwen Wang^1,2Gao Jian^3,4,5*Qingchun Guo^1,2Wei Wang^1,2Gan Liu^1,2Juhua Luo^6*

• ¹Nanjing Normal University Key Laboratory of Virtual Geographic Environment Ministry of Education, Nanjing, China
• ²Jiangsu Engineering Lab of Water and Soil Eco-remediation, Nanjing Normal University, Nanjing, China
• ³Key Laboratory of Intelligent Health Perception and Ecological Restoration of Rivers and Lakes, Ministry of Education, Hubei, China
• ⁴Hubei Key Laboratory of Environmental Geotechnology and Ecological Remediation for Lake & River, Hubei, China
• ⁵School of Civil and Environment, Hubei University of Technology, Hubei, China
• ⁶Nanjing Institute of Geography and Limnology Chinese Academy of Sciences, Nanjing, China

The innovation here is the new classification of aquatic vegetation based on the association level using unmanned aerial vehicle (UAV)-mounted sensing technology, and a light detection and ranging (LiDAR) method to acquire point cloud data and high-resolution red, green, and blue (RGB) imagery. This research focuses on aquatic vegetation in the littoral zone of East Lake Taihu. By innovatively introducing UAV and LiDAR provide clear single images of both exterior and atmospheric surfaces by using a point cloud canopy height model (PCHM), VDVI (visible-band difference vegetation index, spectral information) and a decision tree classification model for littoral aquatic vegetation at the association level. In terms of data processing, improving data reliability through point cloud gridding and alignment with field quadrats. After integrating point cloud and optical image data, we interpret canopy height and spectral information of aquatic associations by precisely identifying and mapping vegetation types to their individual vegetation associations. This is the first study to achieve fine-scale classification of aquatic vegetation at the association level in lakeshore wetlands based on UAV-LiDAR fusion technology. Results showed the classification accuracy for these associations ranging from 79.80% to 97.40%. The higher canopy associations have greater classification accuracy with an overall classification accuracy of 87.93% and a kappa coefficient of 0.855. The new association classification method can improve data results on scientific management of littoral aquatic ecosystems.