Multi-Contextual Learning in Spatio-temporal Neighborhoods

This dissertation presents a multi-contextual learning framework within spatiotemporalneighborhoods to tackle analytical challenges in Earth observation data. The rapid rise in the amount of environmental data —as demonstrated by the European Space Agency’s Copernicus program expanding from 2 to 20 petabytes per year and NASA’s Earth Observing System producing 16 terabytes per day—poses difficulties to conventional analytical methods for managing multi-source data fusion, cross-scale pattern recognition, and spatial autocorrelation. The work develops a context-aware spatio-temporal data analysis approachwith neighborhood-based spatio-temporal framework at its foundation. The framework employs Voronoi tessellation for micro-neighborhood generation and attributebased grouping for macro-neighborhood generation. By incorporating contextual information from both spatial proximity and attribute similarity, the approach captures nuanced patterns that traditional methods tend to ignore. This multi-contextual learning framework is validated through two complementaryapplication domains that serve as case studies. The Greenland Ice Sheet case demonstrates how the application of neighborhood analysis successfully encapsulates intricate melt behavior, accounting for local variability, seasonality, and couplings between temperature, albedo, and other variables. Importantly, the framework aids in comparing surface and subsurface processes influencing change in the ice mass in marine-terminating glaciers in Southeast Greenland. Digital twin simulations form a second test case by demonstrating that the same neighborhood-based approach is capable of defining areas with analogous variance structures within high-resolution atmospheric data. The methodological contributions of the dissertation are: (1) multi-contextuallearning for spatio-temporal neighborhood formation; (2) Graph Deviation Networks for multivariate anomaly detection in such neighborhoods; (3) a technique for differentiating surface versus subsurface process dominance in analyzing ice mass change through hotspot analysis; and (4) spatial clustering for variance analysis. Each of these components tackles intrinsic challenges in spatio-temporal data analysis while offering practical solutions for environmental monitoring application scenarios. Results indicate that multi-contextual learning in spatio-temporal neighborhoods substantially enhances detection and interpretation capability for complex Earth observation data, with immediate implications for environmental monitoring, modeling, and satellite-based observational systems.