A Computational Geometric and Graph Theoretic Approach to Reducing Dimensionality on Raster Data Problems

Large scale mathematical models often involve a trade off between computational length and detail. In general, the more detailed the data, the more time it takes for the model to process. Models that use geographic scale data are particularly susceptible to this inflation; fine resolution data (on the order of m2 [meters squared]) brings great benefits, but demolishes the computation time. This thesis presents a method for reducing the dimensionality of large scale data in a systematic manner to maximize the benefits of fine resolution data while minimizing the computational time increase, then applying the method to a simulated invasive species problem using geographic data.