I'm currently busy with my Masters Thesis "Analyse the Impact of DEM Uncertainty within Hydrological Modelling using Geostatistics". I'm developing a python geostatistics model that runs a monte carlo simulation (conditional sequential guassian simulation) 1000 iterations where the DEM is altered simulating error and a stream network is derived. The 1000 simulated river networks are used to determine where the biggest uncertainty lies within the derived river network. The following requires that I run Flow Accumulation a 1000 times for my study area.
I'm looking for advice from the community that have sucessfully multiprocessed grid processes such as Flow Accumulation or Flow Direction, where each cell is visited to determine the accumulation or direction. How can I subprocess th following with python to improve the performance of my model.