Generating urban landscape meshes using CGAL

[giuliocrp]

Exa-MA WP1 - Terrain Mesh Generation Internship

Objective

During my internship, I focused on improving the terrain mesh generation process for urban energy simulations as part of the Exa-MA¹ project at Cemosis². The goal was to optimize the 3D geometric reconstruction of urban landscapes by generating constrained triangulated meshes aligned with contour lines, reducing unnecessary complexity in flat areas.

Key Components

1. Terrain Mesh Generation

• Lambda-Generated Meshes:
  • Half-Sphere Terrain:
    • This mesh was generated using a lambda function representing a half-sphere:

• Wave-Like Terrain:
  • Another lambda-generated mesh with a wave-like surface:

• GPS Data-Driven Meshes:
  • Grenoble, France (Zoom Level 16):
    • A real-world terrain mesh generated using elevation data from the Mapbox Terrain-RGB API³. This high-resolution mesh captures the detailed topography of Grenoble.

2. Contour Line Generation

• Implemented a method to generate and constrain terrain meshes along contour lines:

3. Re-Triangulation

• Using the CGAL⁴ library, the contour-constrained meshes were re-triangulated:

Impact and Future Work

The developed framework successfully reduces mesh complexity, making large-scale urban energy simulations more computationally efficient. Future work could involve merging multiple tiles for broader terrain coverage, optimizing performance for larger datasets, and integrating urban elements like buildings and vegetation to create comprehensive 3D urban models.

More information on the terrain mesh generation methods and visualizations can be found here.

References

Footnotes

1. Exa-MA. Methods and Algorithms for Exascale. 2024. Available at: Exa-MA ↩

2. Cemosis. Center for Modeling and Simulation in Strasbourg. 2024. Available at: Cemosis ↩

3. Mapbox. Mapbox Terrain-RGB v1. Mapbox Documentation. 2024. Available at Mapbox Terrain-RGB v1 ↩

4. CGAL: The Computational Geometry Algorithms Library. 2024. Available at: CGAL ↩