Item

Abstract

Increasing needs for real-time flood forecasting and adaptation to flood risks associated with climate change require more detailed and accurate inundation information. Although river levees are an important element that defines the extent and depth of flood waters, data on the height and the location of levees that can be introduced into global river models are not well developed. Therefore, we developed an automatic levee detection algorithm for the Kinu River in Japan (basin area: 1,760 km2) and confirmed its applicability to the Mississippi River (basin area: 2,981,076 km2), a large river with a different climate. This algorithm automatically determines the presence of levees when multiple levee conditions are met from high-resolution Digital Elevation Models (DEMs). In the case study of the Mississippi River in the USA, 3DEP which is a 10-meter resolution DEM data can detect the height and the location of levees, especially in the middle and lower basin and urban areas. As in the Japanese case, detection has some weaknesses, such as bridge junctions and areas with wide top edges, but the detection capability is considered sufficient for obtaining levee information for the global-scale modeling. However, unlike Japan, where assets are concentrated in flood plains, the USA has vast tracts of land, and levee locations are determined based on economic rationality. Thus, areas where flood control is planned by cities, such as the upper basin, may be overlooked by levees.