A Unified Framework for Hybrid GNSS-SSP Spatial Data Quality Theory

Thesis Abstract

This study addresses the fragmented understanding of spatial data quality within Hybrid GNSS-SSP (Satellite-based Positioning) frameworks by proposing a unified theory that integrates data quality constructs across positioning, sensing, and processing stages. The aim is to develop a coherent framework that reconciles positional accuracy, temporal integrity, sensor fusion reliability, and semantic quality indicators to support robust decision-making in surveying and geoinformatics. Specific objectives are (1) to conceptualize a composite data quality model that unifies GNSS-derived accuracy, SSP-based semantic reliability, and data provenance; (2) to operationalize the model into measurable indicators and a diagnostic toolkit; (3) to evaluate the framework's predictive capability for data quality under varying environmental conditions and sensor configurations; (4) to test the framework's applicability across urban, suburban, and rural contexts; and (5) to formulate governance guidelines for data quality assurance in hybrid GNSS-SSP workflows. A mixed-methods research design is employed. The quantitative component uses a multi-site observational study in three metropolitan regions with 120 GNSS-SSP data captures per site, totaling 360 samples, to assess positional error, time synchronization discrepancy, and SSP semantic confidence scores. Data collection instruments include a standardized GNSS receiver log, high-resolution SSP sensor streams, and a provenance-tracking instrument integrated into the data fusion pipeline. The qualitative component involves semi-structured interviews with 18 professionals (6 per site) and a thematic analysis of system logs to capture contextual factors influencing data quality, such as multipath exposure, atmospheric anomalies, and data fusion assumptions. Validity and reliability are ensured through triangulation, pilot testing of instruments, inter-rater reliability checks for coding (Cohen’s kappa > 0.8), and calibration of the SSP ontologies against established geospatial semantics. Analytical methods comprise regression-based modelling to quantify the relationship between GNSS error metrics, SSP semantic confidence, and provenance indicators, employing hierarchical linear modelling to account for site-level variance. Structural equation modelling (SEM) will test the hypothesized causal pathways within the unified data quality framework, while time-series analysis evaluates temporal stability of quality indicators during data fusion. A Bayesian updating mechanism is embedded to adapt quality assessments as new data streams are integrated. The model specification includes latent variables for positional quality, temporal integrity, semantic reliability, and data lineage, with manifest indicators derived from error statistics, clock drift measurements, confidence scores, and provenance attributes. The study also applies information-theoretic measures (AIC, BIC) for model comparison and validation against independent holdout datasets. Key expected findings include (1) evidence that a unified data quality index outperforms isolated GNSS or SSP quality measures in predicting downstream decision accuracy; (2) identification of critical thresholds where environmental conditions (e.g., urban canyon effects) or sensor fusion configurations precipitate quality degradation; (3) validation of the SEM model demonstrating robust links among accuracy, semantic reliability, and data lineage; and (4) practical guidelines for configuring data fusion pipelines to maintain target quality levels under resource constraints. The study anticipates that the unified framework will reveal synergistic effects between GNSS geometry quality and SSP semantic confidence, moderated by provenance quality and processing latency. Contributions to knowledge include a theoretically grounded, empirically validated framework that integrates spatial data quality across GNSS and SSP domains, offering a transferable blueprint for quality assurance in hybrid geospatial systems. It extends existing theories of quality in geoinformatics by introducing a holistic, multi-criteria, provenance-aware model that supports adaptive quality management in real-time data fusion contexts. The study concludes that a unified spatial data quality framework enhances reliability in surveying outcomes, enables more trustworthy geographic information products, and informs policy on data governance, standards development, and system design. Recommendations include implementing the diagnostic toolkit in operational workflows, adopting provenance-aware quality assurance protocols, and extending the framework to incorporate emerging sensor modalities such as LiDAR and descriptor-based semantic enrichments.