Expert system tools for describing and predicting the coastal ocean environment

The complex and variable nature of the littoral environment poses a challenge to individuals interested in describing and predicting the coastal ocean environment for commercial, civil, or military applications. Inadequate oceanographic insight and data in many littoral regions provides further complication. Focusing on the intended use of the coastal scene description, however, reveals that not all aspects of the coastal scene are relevant (and need to be described) for every application. The Johns Hopkins University Applied Physics Laboratory is investigating the suitability of expert system technologies for developing and refining a process-oriented description of the coastal environmental scene to support decisions relating to naval tactics and operations in the littoral. Emulating the reasoning of human analysts, the authors strive to capture the physical cause-and-effect relationships of the dominant coastal processes. Two formalisms which are being investigated for reasoning about coastal processes based on sparse data, Dempster-Shafer theory and Bayesian belief networks, are introduced