Avinash Rude

Date of Award


Level of Access

Open-Access Dissertation

Degree Name

Doctor of Philosophy (PhD)


Spatial Information Science and Engineering


Kate Beard-Tisdale

Second Committee Member

Silvia Nittel

Third Committee Member

Neal R. Pettigrew


Recent advancement in sensor technology has enabled the deployment of wireless sensors for surveillance and monitoring of phenomenon in diverse domains such as environment and health. Data generated by these sensors are typically high-dimensional and therefore difficult to analyze and comprehend. Additionally, high level phenomenon that humans commonly recognize, such as storms, fire, traffic jams are often complex and multivariate which individual univariate sensors are incapable of detecting. This thesis describes the Event Oriented approach, which addresses these challenges by providing a way to reduce dimensionality of space-time series and a way to integrate multivariate data over space and/or time for the purpose of detecting and exploring high level events. The proposed Event Oriented approach is implemented using space-time series data from the Gulf of Maine Ocean Observation System (GOMOOS). GOMOOS is a long standing network of wireless sensors in the Gulf of Maine monitoring the high energy ocean environment. As a case study, high level storm events are detected and classified using the Event Oriented approach. A domain-independent ontology for detecting high level xvi composite events called a General Composite Event Ontology is presented and used as a basis of the Storm Event Ontology. Primitive events are detected from univariate sensors and assembled into Composite Storm Events using the Storm Event Ontology. To evaluate the effectiveness of the Event Oriented approach, the resulting candidate storm events are compared with an independent historic Storm Events Database from the National Climatic Data Center (NCDC) indicating that the Event Oriented approach detected about 92% of the storms recorded by the NCDC. The Event Oriented approach facilitates classification of high level composite event. In the case study, candidate storms were classified based on their spatial progression and profile. Since ontological knowledge is used for constructing high level event ontology, detection of candidate high level events could help refine existing ontological knowledge about them. In summary, this thesis demonstrates the Event Oriented approach to reduce dimensionality in complex space-time series sensor data and the facility to integrate ime series data over space for detecting high level phenomenon.