Parallelizing drainage network algorithm using free software: Octave as a solution

Drainage network is a product, normally derived from a DEM (digital elevation model), widely used in environmental and civil engineering, and particularly in hydrology. The computation requirements increase exponentially as the size DEM increases, limiting the applications when a fast analysis is necessary. This is specially noticeably when working with multiple flow direction (MFD) drainage networks. Nowadays, some solutions have been explored, but focusing on Graphical Processing Units (GPU) technology. We propose a CPU-based approach which has the advantage that uses free software such as Octave and MPI wrap for it. Our parallelized algorithm not only improved the time computation but also allows adaptive behavior to different cluster settings.