TerraCost addresses the problem of computing
multiple-source weighted least-cost-path surfaces for
grid terrains. Currently, this functionality is 
provided by the GRASS module r.cost. Our approach, 
implemented in GRASS as r.terracost, expands this
functionality such as to allow massive terrains 
to be processed efficiently. We obtain this 
efficiency by combine memory- and disk-based 
techniques, and, as a by-product of the algorithm's 
modular design, we can actually benefit from
cluster-connected computing resources (if available).
Experiments show that TerraCost’s algorithms perform 
well in practice: Our implementation outperforms 
standard solutions as dataset size increases relative 
to available memory and our distributed solver obtains 
near-linear speedup when preprocessing large terrains 
for iterated computations with varying parameters.