Spatial & Temporal Modeling Using Artificial Intelligence/Pattern Recognition Technique

Dr. Debasmita Misra
College of Engineering and Mines, University of Alaska, Fairbanks

Physics-based models for spatial and temporal simulations are quite complex and involve inclusion of several parameters to attain desired accuracy. Recently, pattern-learning type algorithms such as the artificial neural networks (ANN) and Support Vector Machines (SVM) have gained popularity in simulating similar processes yielding comparable accuracy. Through two distinct case studies, one spatial and the other temporal, I will be presenting the outcome of application of ANN, SVM, and a new approach developed in UAF called the Multiple Regressive Pattern Learning Technique (MRPRT) (Oommen, 2007). From these studies, one could conclude that pattern recognition methods could be used even in sparse data scenarios to simulate complex hydrological processes.

For more information contact: Jeff Derry (907) 322- 3026