Topology-based terrain segmentation using Apache Spark

Terrain topology plays an important role in simulations and segmentation. A widely used terrain representation is the Triangulated Irregular Network (TIN). However, topological analysis on TINs is challenging due to high time and memory requirements, which limit the size of the terrain that can be analyzed.

We address this problem by introducing a novel framework for efficient and scalable analysis of large TINs based on Morse theory using Apache Spark. The proposed framework, named Morse-Spark, is based on a novel data structure for encoding the minimal information of a TIN, and integrates efficient algorithms for computing terrain morphology. To evaluate the effectiveness and scalability of such a framework, we compare Morse-Spark against a vanilla Spark implementation and three well-established software libraries for the topological analysis of TINs through a comprehensive set of experiments. Our experimental evaluation with real-world TINs shows that Morse-Spark can effectively handle datasets at least 20 times larger than existing approaches.