Users of Geographical Information Systems (GIS) and Spatial Modelling Systems today have access to an unprecedented amount of high resolution and high-quality data through scanners, satellites, range finders, medical devices and other devices. Also their demands regarding speed and functionality have increased over the years.

To keep up with the resulting computational demands without sacrifice (i.e., reduction in resolution or scope of model), parallel computing appears to be the only solution. Parallel hardware is readily available at a good price-to-performance ratio (in the small to medium range).

Our primary research and development objective is to enable users, researchers and developers within these application communities to use parallel computers without paying the high price of having to deal with the complex issues inherent to it (i.e. transparent parallelism.)

Here we report on our first major milestone towards achieving this objective: the design and prototype implementation of an environment for parallel raster-based NEighbourhood MOdelling (NEMO). NEMO is primarily intended for (but not necessarily restricted to) coarse-grained parallelism with a limited number of processors. Neighbourhood modelling applications fall into three main categories: neighbourhood analysis, cellular automata and propagation. NEMO is not tailored to any specific application. Rather, it is designed to support applications falling under the umbrella of these three raster neighbourhood modelling categories. NEMO encompasses these categories through corresponding neighbourhood modelling drivers. Two additional client-server components Display Manager and Raster Database Manager handle all data I/O and visualization issues.

We demonstrate the validity of our approach and the efficiency of the system through the implementation of a variety of applications: forest fire modelling applications (in collaboration with Forestry Canada and K. Clarke, UC Santa Barbara), earth quake modelling, ice-tracking for the Canadian shipping (Canada Remote Sensing), Image processing, Cartography (implementation of nearly all of Tomlin's Map Algebra functions). (NEMO has been delivered to our industrial partner ALMERCO Inc. for commercialization.)