Hydrology-Aware Constrained Delaunay Triangulation
Basic Idea: To preserve the hydrology using a constrained triangulation in conjunction with a modified TIN. The cost metric used by TIN is changed so a point is considered to have the highiest cost if it has the largest amount of water times the gradient flowing uphill. This differs from TIN where a point is selected based on its absolute distance from the terrain reconstruction.
Properties:
- ODETLAP is not used (cuts down on compute time and improves scalability)
- Storing the edges does not require a lot of overhead since we already know that the ridges and rivers are connected. We simply have to store the sizes of each ridge-river line, along with the points
- We eliminate small rivers/ridges. This is done because I believe these points are mostly redundent and are more expensive to store then they are worth
- TIN is used after the Ridge-River constrained triangulation is constructed. The purpose of this is to add points that have a high amount of error that were not included in the intial construction. Error is based on the potential energy metric described in the GIScience paper.
Implementation:
We use a constrained triangulation method developed by Jonathan Richard Shewchuk which can be found here. triangle ZIP (159K)