Fully Automated Breakline Identification in HEC-RAS 2D Flood Propagation Modeling

Abstract:

Land surface topography plays a crucial role in flood risk analysis as it significantly affects 2D flow propagation and flow interactions with human life, economy, and environment. In this study, we introduce fully automated terrain analysis methods for the identification of breaklines and the generation of suitable unstructured meshes in HEC-RAS 2D hydrodynamic models. Breaklines are a way to force cell faces to follow a terrain ridge or crest hinge line. Terrain analysis is used to recognize automatically the ridge network and to identify breaklines representing relevant natural and man-made topographic structures such as embankments and roads (see the AGU FM abstract ID# 678439). Breaklines are used to indicate where detail in the description of land surface topography is needed so that accuracy and computational efficiency can be addressed simultaneously. Our universally applicable methods are illustrated by reconstructing the flood event occurred in Modena, northern Italy, in January 2014 after a levee failure along the Secchia River. Three 2D hydrodynamic models are evaluated. These models share the same 1-m grid digital elevation model obtained from a lidar survey as underlying topographic data, the same boundary and initial conditions, roughness coefficients, and computational time steps. They only differ in the computational mesh. In model 1 (left figure), a 10-m regular grid is used, whereas breaklines are not used. In model 2 (middle figure), an unstructured mesh is enforced along breaklines traced manually by a skilled operator. In model 3 (right figure), an unstructured mesh is enforced along breaklines extracted automatically by using the developed terrain analysis methods. Computational efficiency increases by a factor of 300 when unstructured meshes are used in preference to structured meshes. Breaklines extracted automatically are found to be consistent with those traced manually by a skilled operator. Levees and roads are found to affect significantly the water movement and, in this respect, only models 2 and 3 are found to be reliable. The new automated methods for breakline identification were found to be essential for developing reliable and computationally efficient 2D flood propagation models such are those needed to solve flood related societal challenges over large geographical areas.