A contour tree based spatio-temporal data model for oceanographic applications

Abstract

To present the spatio/temporal data from oceanographic modeling in GIS has been a challenging task due to the highly dynamic characteristic and complex pattern of variables, in relation to time and space. This dissertation focuses the research on spatio-temporal GIS data model applied to oceanographic model data, especially to homogeneous iso-surface data. The available spatio-temporal data models are carefully reviewed and characteristics in spatial and temporal issues from oceanographic model data are discussed in detail. As an important tool for data modeling, ontology is introduced to categorize oceanographic model data and further set up fundamental software components in the new data model. The proposed data model is based on the concept of contour tree. By adding temporal information to each node and arc of the contour tree, and using multiple contour trees to represent different time steps in the temporal domain, the changes can be stored and tracked by the data model. In order to reduce the data volume and increase the data quality, the new data model integrates spatial and temporal interpolation methods within it. The spatial interpolation calculates the data that fall between neighboring contours at a single time step. The Inverse Distance Weighting (IDW) is applied as the main algorithm and the Minimum Bounding Rectangle (MBR) is used to enhance the spatial interpolation performance. The temporal interpolation calculates the data that are not recorded, which fall between neighboring contour trees for adjacent time steps. The “linear interpolation” algorithm is preferred to the “nearest neighbor’s value” and “spline” interpolation methods, for its modest accuracy and the simple implementation scheme. In order to evaluate the support functions of the new data model, a case study is presented with the motivation to show how this data model supports complicated spatio-temporal queries in forecasting applications. This dissertation also showcases some work in contour tree simplification. A new simplification algorithm is introduced to reduce the data complexity. This algorithm is based on the branch decomposition method and supports temporal information integrated into contour trees. Three types of criteria parameters are introduced to run different simplification methods for various applications.