Date of Award
Fall 12-20-2024
Level of Access Assigned by Author
Open-Access Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Spatial Information Science and Engineering
Advisor
Torsten Hahmann
Second Committee Member
Kate Beard Tisdale
Third Committee Member
Silvia Nittel
Additional Committee Members
Aaron Weiskittel
Marshall (Xiaogang) Ma
Abstract
The Digital Forest project represents a stride towards sustainable forest management and the preservation of global forest resources. It achieves this by identifying and developing essential ontologies for forestry data integration, creating a comprehensive knowledge graph of forestry data, and providing a semantic interface that enables users without semantic expertise to interact with the data. The Digital Forest Knowledge Graph seamlessly integrates diverse datasets, encompassing environmental, spatiotemporal, and ecological dimensions. This initiative introduces three primary ontologies: the Spatial Temporal Aggregate Data (STAD) Ontology Design Pattern, the Forest Inventory Ontology (FIO), and the Preferences for Environments Ontology (PrefEnvO).
The STAD ontology addresses the ambiguity of aggregate data and enhances the interoperability of spatio-temporal statistical data by offering a structured framework for specifying the semantics of aggregations. With advances in data collection techniques, spatio-temporal data has become increasingly available across many scientific disciplines. For easier processing, analysis, storage, and often due to security and privacy concerns, data is frequently distributed in aggregated forms rather than as individual data points. For instance, understanding and predicting weather conditions is more practical when
examining climate aggregates like climate normals. Thus, statistically aggregated spatial and temporal data often suffice to concisely describe, analyze, and predict events and their interrelations. However, most aggregate data available on the web lack the rich semantics needed to guide correct data retrieval, reuse, and integration. The STAD ontology design pattern developed in this project helps specify the semantics of aggregations, thereby guiding the semantically correct integration of spatial-temporal statistically aggregated data and knowledge.
Forest inventory involves systematically gathering data and information about forests for assessment or analysis. These inventories are crucial for providing data that enhances understanding and management of forest ecosystems by offering essential insights into the composition, structure, and health of forested landscapes. However, the heterogeneity and lack of standardization in representing forest inventory data pose significant challenges for data integration, sharing, and interoperability across various research projects and management initiatives. The Forest Inventory Ontology (FIO) addresses these challenges by standardizing the representation and integration of forest inventory data. This ontology ensures consistency and interoperability across different forest inventory datasets by providing a standardized vocabulary for describing various attributes of forest inventories. It encompasses two core modules: the Location Pattern and the Inventory Pattern. The Location Pattern represents geospatial attributes associated with inventory locations, including relationships among inventory plots and administrative boundaries. The Organism Inventory Pattern focuses on modeling the attributes of organisms inventoried within inventory events, such as species identification, diameter measurements, health status, and growth stage. This ontology aims to serve as a foundational framework for developing interoperable forest inventory systems.
The Preferences for Environments Ontology (PrefEnvO) provides a framework for capturing and analyzing species-environment relationships. Environmental preferences greatly influence species distributions and consequently ecosystem dynamics. Existing
ontologies often lack the granularity needed to represent the preferences of diverse organisms across various environmental variables. The PrefEnvO offers a structured vocabulary for delineating species’ preferences for environmental variables, considering spatial and temporal dimensions. By formalizing these preferences, the ontology supports habitat suitability assessments and ecosystem dynamics studies.
These ontologies, together with datasets spanning climate, soil, topography, geology, and tree inventories in Maine, have been integrated into the Digital Forest Knowledge Graph. By leveraging the three primary ontologies and other existing ontologies such as GeoSPARQL, QUDT, and NCBITaxon, it provides a robust framework for accurately representing spatio-temporal aggregations, forest inventory data, and species-environment relationships. This integration allows users to query and explore environmental data, tree inventories, and species-specific preferences, offering valuable insights for informed decision-making, sustainable forest management, and biodiversity conservation.
To facilitate broader user interaction with the Knowledge Graph, a user-friendly interface with streamlined query capabilities has been developed. This interface allows users to navigate environmental data, forest inventory records, and species-specific environmental preferences without needing to know SPARQL and GeoSPARQL. This enhanced accessibility promotes interdisciplinary collaboration and extends the benefits of the Digital Forest to a wider audience.
Recommended Citation
Wiafe-Kwakye, Kingsley, "Development of an Ontology-Based Approach to Spatio-Temporal Data Analysis for Forest-Environment Interactions" (2024). Electronic Theses and Dissertations. 4112.
https://digitalcommons.library.umaine.edu/etd/4112
Files over 10MB may be slow to open. For best results, right-click and select "save as..."
