Author(s): Md. Touhidul Islam; Keisuke Yoshida; Satoshi Nishiyama; Koichi Sakai

Linked Author(s): Md. Touhidul Islam, keisuke yoshida

Keywords: Irborne LiDAR; Floodplain vegetation; Hydrodynamic-numerical modeling; Submerged infrastructure; Topo-bathymetric green LiDAR

Abstract: For decades, remote sensing technologies based on light detection and ranging (LiDAR) have been used to survey stream channel and floodplain topography, as well as vegetation distributions. However, traditional platforms, such as airborne LiDAR bathymetry (ALB), have drawbacks, including flight altitude, point density, and scanning angle, which prevent the sensor from conducting a practical survey of vegetated rivers. More recently, unmanned aerial vehicle (UAV)-based LiDAR offers a way to remotely characterize vegetated river environments with several inherent benefits over ALB. This study as a first to compare the accuracy of a new advanced UAV-borne green LiDAR system (GLS) (i.e., TDOT GREEN, Amuse Oneself Inc., Japan) and an existing ALB in surveying the vegetated and gravel-bed lower Asahi River, Japan, based on correspondence between LiDAR-derived, depth-averaged numerically simulated, field- and sensor-based data. Findings on topo-bathymetry and reproducibility of low-water flow patterns revealed that both system-derived data were reasonably identical to benchmarked field observations, with outperformed accuracy of a few dozen centimeters, despite a few missing bathymetric data in deeper areas during the UAV campaign. In addition to proactively addressing a continuous topography of land and water area shallower than or equal to 2.08 m depth (0.8 FTU, turbidity near water surface), it was highlighted that recent GLS point clouds over the ALB were able to depict the river's submerged artificial infrastructure in detail, which could be helpful in maintenance and construction tasks. Furthermore, vegetation attribute monitoring results, including detailed mapping of vertical structures, types, and heights, revealed that the cost-effective GLS performed relatively well, providing almost identical performance to ground-truth observations and respective high spatial resolution aerial images with a ground sampling distance of 3 cm/pixel. To conclude, after further comprehensive studies in the same or different reaches, the novel technique could be regarded as promising for management measures in shallow streams with vegetated floodplains because the LiDAR-derived information can feasibly be used as input and validation data in the hydrodynamic-numerical model and floodplain vegetation characterization.

DOI: https://doi.org/10.3850/IAHR-39WC252171192022987

Year: 2022