Data-driven reconstruction of incomplete earthquake catalogs: Validation and application to the Yutian region, Xinjiang, China

The integration of geospatial data analytics and statistical methodologies offers significant potential for addressing data limitations in earthquake catalog analysis and statistical seismology. This study presents a data-driven framework for reconstructing incomplete earthquake catalogs in regions with sparse monitoring networks, a critical challenge for earthquake catalog analysis and statistical seismology. We systematically evaluate and apply a statistical replenishment (SR) method to address this incompleteness. Using a high-quality Japan catalog and controlled deletion experiments, we first quantify the method performance, demonstrating robust recovery (F1-scores > 0.81) in temporal and spatial dimensions. When applied to the Yutian region, the SR technique substantially enhances event counts and spatial coverage, particularly for small magnitude events. The replenished catalog yields lower and more stable magnitude of completeness ( M c) estimates and reveals clearer spatiotemporal patterns of the b -value. Notably, persistent low b- values ( b < 0.6) are identified preceding the major earthquakes, highlighting zones of sustained high differential stress that align with known active faults. While the method does not aim to recover exact individual earthquakes, it effectively restores the statistical integrity of catalogs, providing a more reliable basis for parameter estimation and statistical analysis of seismicity. This work demonstrates that data-driven reconstruction can improve earthquake catalog completeness, providing a practical framework for statistical seismological analyses in regions with sparse monitoring networks.