Mapping Levees Using LiDAR Data and Multispectral Orthoimages in the Nakdong River Basins, South Korea

Mapping levees is important for analyzing levee surfaces, assessing levee stability, etc. Historically, mapping levees has been carried out using ground surveying methods or only one type of remote sensing dataset. This research aims to map levees using airborne topographic LiDAR data and multispectral orthoimages taken in the Nakdong River Basins. Levee surfaces consist of multiple objects with different geometric and spectral patterns. This research investigates different methods for identifying multiple levee components, such as major objects and eroded areas. Multiple geometric analysis approaches such as the slope classification method, and elevation and area analysis are used to identify the levee crown, berm, slope surfaces, and the eroded area, with different geometric patterns using the LiDAR data. Next, a spectral analysis approach, such as the clustering algorithm, is used to identify the major objects with different spectral patterns on the identified components using multispectral orthoimages. Finally, multiple levee components, including major objects and eroded areas, are identified. The accuracy of the results shows that the various components on the levee surfaces are well identified using the proposed methodology. The obtained results are applied for evaluating the physical condition of the levees in the study area.