Cross-Scale Spatiotemporal Modeling Using an Integrated Data Framework

点击翻译按钮获取中文摘要

This research project will address the challenges associated with managing scales in space and time within a single, unified analytic framework. The choice of scale is an important question in the analysis of geospatial data. For example, the spatial analysis of socioeconomic variables at the state level may mask local processes taking place at the county or community level. Historically, spatial and temporal analysis has proceeded either separately or in a loosely coupled research design. This project will develop and extend a multi-scale framework for the visualization and analysis of geospatial data. The framework will resolve fundamental issues of scale handling in data analytics, advance knowledge about cross-scale spatio-temporal phenomena, and aid scientists in looking more deeply into the interplay among various environmental and social processes. New tools will be developed and made publicly available. The utility of these tools will be tested in case studies, including an analysis of wetland habitats in coastal Louisiana and Hawaiian rainfall patterns. The project will support graduate students whose participation will advance their own professional development. The collaboration between the University of Hawaii and the University of Colorado at Boulder will increase geographic diversity and the presence of women and underrepresented minorities in computer science, earth science, and spatial data science.This research project will develop a theoretical framework for multi-scale data representation, modeling, and analysis. Multi-scale analysis of spatio-temporal data is a longstanding concern for analytic systems in many disciplines. The problem of handling scale is epitomized in the well-known modifiable areal unit problem and its temporal equivalent. These issues are partially due to the traditionally held views of time as a linear sequence and space as a flat layer. This project extends research on the Triangular Model (TM), a 2D representation of time, into higher dimensional models. The project will test the utility of TM in analyzing linear spatial data, refine conceptual and computational aspects of a 3D Pyramid Model (PM) for multi-scale spatial analysis, and integrate the TM and PM into an 5D analytical framework for multi-scale spatio-temporal analyses. Topologies, statistics, and machine learning methods will be developed on the models and framework to support multi-scale queries, visualization, and quantitative modeling. The questions to be answered in the project include: 1) what additional knowledge can be gained by analyzing spatio-temporal variations, patterns, and relationships of phenomena across in the TM and PM frameworks? and 2) in what ways can multi-scale representations of spatio-temporal data facilitate the modeling of human-environment interactions?This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.