r.terracost is a scalable approach for computing least-cost-path surfaces on massive grid terrains. The module outperforms standard solutions as dataset size increases relative to available memory.


r.terracost computes a least-cost surface for a given cost grid and set of start points using an approach that scales to large grids. For details, see paper "TerraCost: A Versatile and Scalable Approach for Path Computations on Massive Grid-Based Terrains" by Hazel, Toma, Vahrenhold and Wickremesinghe (2006). The basic idea is to split the grid in tiles. numtiles is the number of tiles. Run with -i to see the recommended value for numtiles. When numtiles=1 it runs Dijkstra's algorithm in memory. When numtiles > 1 it runs an external SP algorithm that consists of 5 steps. For debugging purposes, it is possible to run the five steps separately. When running in separate steps, an intermediate config file stores temporary information, and other streams (see below) contain intermediate data. IOLibrary temporary streams will be in STREAM_DIR.



r.cost, r.drain, r.walk


Laura I. Toma,

Last changed: $Date$