A Python Program for Extracting Vertical Features from a LiDAR-DEM with Shallow Water Physics

Abstract:

A critical challenge for flood predictive modeling is describing the geometric complexities of the terrain, particularly in the unstructured mesh generation process. For example, the elevation of natural barriers (e.g., beach dunes) and anthropogenic features (e.g., levees, and raised roadbeds) are likely to be smoothed out due restraints in horizontal mesh resolution. This can have a dramatic impact on the propagation of flood waters across the landscape. In order to accurately and automatically represent these vertical terrain features, an open-source program PyVF been developed that can extract significant vertical features from a high-resolution LiDAR-derived DEM. The algorithm can be applied in coastal and upland regions.

Vertical features are also employed to guide the mesh generation process for flood models. The required length of each extracted vertical feature needs to vary with the local mesh element size; however, (in the mesh generation process) it is not clear what the local element size should be. PyLTEA is a toolbox to aid in the development of a mesh sizing function based on shallow water physics using localized truncation error analysis (LTEA). The mesh node placement and density created by PyLTEA provide a criterion for determining the varied length of vertical features and the appropriate spatial distribution of the identified vertical features for unstructured mesh generation. The combination of PyVF with PyLTEA for vertical feature delineation improves the unstructured element mesh model across the coastal land margin. In addition, the program provides varying parameters to satisfy various landforms and research objectives.