Eco-informatics Tools for Coral Reef Ecology

Campbell Allen (2010). Eco-informatics Tools for Coral Reef Ecology MPhil Thesis, School of Geography, Planning & Environmental Mgt, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
s4152672_MPhil_abstract.pdf Final Thesis Lodgement - Abstract application/pdf 87.30KB 0
s4152672_MPhil_totalthesis.pdf Final Thesis Lodgement application/pdf 6.88MB 7
Author Campbell Allen
Thesis Title Eco-informatics Tools for Coral Reef Ecology
School, Centre or Institute School of Geography, Planning & Environmental Mgt
Institution The University of Queensland
Publication date 2010-02
Thesis type MPhil Thesis
Supervisor Professor Jane Hunter
Total pages 143
Total colour pages 9
Total black and white pages 134
Subjects 05 Environmental Sciences
Abstract/Summary The Great Barrier Reef is the world's largest continuous coral reef ecosystem. It is internationally recognised through world heritage status, and is an iconic, cultural and economically significant resource. Undertaking research is an important part of anticipating problems and providing the best protection for the Great Barrier Reef. As result, a number of independent and diverse research organisations undertake research aiming to understand and help conserve this unique and vital ecosystem. To obtain a thorough understanding of the complex interactions and relationships between entities and processes in an ecosystem context, research organisations in today’s highly multi-disciplinary ecological research environment require greater integration of disparate, independently collected research data. The emergent field of ecological informatics (eco-informatics) endeavours to improve the scale and scope of ecological science by facilitating the capacity to query integrated datasets across a range of variables and spatio-temporal scales. The specific aims of this thesis were as follows: (a) To define a common data model for representing varied and disparate coral reef research data and associated metadata (drawing on existing relevant data models and standards). (b) To map the various datasets into the common model. (c) To identify alternative data storage architectures for integrated data that could potentially provide a viable long-term storage and data management solution that is extensible and sufficiently robust to evolve with changing research objectives and data volumes. (d) To identify a range of exemplary cross-dataset queries that are typically required by users and that can be used to analyse trends and processes in coral reef ecosystems across scales of variables, space and time. (e) To evaluate and compare the capabilities of the alternative architectures to respond to the set of specified queries in a timely manner. (f) To create an online Web tool with a search interface that supports the queries identified in (d). To achieve these aims, data models and data storage technologies that potentially provide the most robust and efficient methodologies to integrate ecological datasets into a unified data store were investigated. A flexible, extensible and robust data model (based upon observations and measurements) was created in order to capture the scientific data at a common generic level, whilst also capturing the specific variables contained in the data, through extensions. Following the development of the model, data translation scripts were written to convert the available scientific datasets into the unified observations and measurements data model for the various data store implementations. Alternative data storage implementations were investigated to provide a long-term storage and data management solution that would be extensible and robust enough to evolve with changing research objectives and data volumes. A set of ten test queries was identified for the benchmarking framework. These queries were designed to (a) specifically test the capabilities of the system; and (b) reflect the types of queries used by ecological researchers. These ranged from simple data retrieval without complex variable filtering to complex multivariate ecological queries with filtering on ecological variables in space and time. The data storage implementations were then evaluated by comparing the data loading performance times and the time taken to resolve the test queries for each data store at varying data volumes. The derived metrics were used in a time-to-event statistical analysis to determine the data storage implementation that performed the set of test queries in the shortest time. Following data storage evaluation, online Web-based data access tools were then designed and implemented. The Web interface uses the Google Web Toolkit (GWT) API’s to create an AJAX Web application and server as well as GWT remote procedure calls (RPC) to enable client-server communication over the Internet. This particular tool utilises the unified data store schema and data to enable the researcher to efficiently extract synergistic information about the Capricorn Bunker Region in the southern Great Barrier Reef across broad temporal and spatial scales. The data model developed in this thesis proved that the generic logical separation of scientific observation from measurement is extensible and robust enough to capture all datasets that were used in this project. Both an OWL ontological and a relational schema data model were implemented. The data conversion resulted in the original heterogeneous data being mapped into both the RDF triple store and the relational data formats, which were then used as the source data for the data storage implementations that were benchmarked. Semantic Web RDF triple stores and relational data storage systems were identified as the only two choices that met the stated data storage system aims of this project. Consequently, a set of implementations were benchmarked to determine which data storage solution performed best. A rigorous evaluation of different data storage technologies indicated that traditional relational data storage systems outperformed the new semantic data stores in efficient query performance. This was especially true for complex, multivariate ecological queries that included filtering in space and time. The Semantic Web RDF triple stores were found to be particularly suited to the data integration task at hand due the ability of ontologies to semantically define and link the data. Relational data stores with GIS spatial extensions, however, proved to be the optimal approach for high performance query resolution and large volume data handling. Though Semantic Web data storage systems offer data integration benefits, their poor query resolution performance ruled out their use as an underlying eco-informatics data storage solution in their current implementations. The Web-based tool created in this project enables the creation of complex multivariate queries, which can be used to investigate the intricate ecological processes in a spatially and temporally convergent context. The use of Web services, mapping interfaces and a Web-based online data access tool greatly enhances access to the unified data store by allowing the researcher to quickly and easily query, browse and export the data. My research demonstrates the feasibility of using a generic data model to integrate disparate, homogeneous scientific data in order to create a unified ecological data store. In addition, my research highlights how the use of an integrated multi-disciplinary research data store in combination with Web services and a Web-based tool can enable efficient, online data discovery and access to highly diverse ecological data over a broad range of spatial and temporal scales. In doing so, my thesis outlines a blueprint for an eco-informatics framework, which can deliver access to a robust unified data store for ecological observation data.
Keyword Eco-informatics
coral reef
ecology
ontology
OWL
RDF
semantic Web
Relational database
ajax
Query Performance
Additional Notes Colour: 44, 74, 90-91, 99-103 Landscape: 48-49, 52, 76-77, 81-85 A3 landscape: 143

 
Citation counts: Google Scholar Search Google Scholar
Created: Thu, 21 Oct 2010, 20:27:14 EST by Mr Campbell Allen on behalf of Library - Information Access Service