On the Use of Parallel Processing for Interactive Analysis of Large GIS Datasets
The computational intensity of analytical operations provided in GIS software can introduce disruptive computationally-induced latencies into decision-making processes. Though parallel processing can be used to improve the performance of GIS operations, the geographical configuration of input datasets can degrade performance when particular data decomposition strategies are used. We outline this problem and demonstrate its effects in a set of computational experiments. These experiments use a spatial interpolation algorithm to process datasets that contain three levels of control point density that are arranged in different geographical orientations. Finally, we suggest strategies to overcome the problem that are based on a preliminary assessment of input datasets.