ABSTRACT Here, we develop a supervised framework for identifying probabilistic earthquake-risk states in the central and southern Tan-Lu fault zone and adjacent areas of eastern China using multiple catalog-derived seismicity features and a deep learning approach. Using the China Earthquake Networks Center catalog from 1979 to 2025, seven seismological parameters (b, M, η, D, Mf, N, and Rm) were calculated for ML≥2.0 earthquakes. These parameters were transformed into probability-based time-series features and used as inputs to an enhanced long short-term memory model with dual-task learning and dropout regularization. The model was trained using data from 1979 to 2020 and evaluated on an independent testing period from 2021 to 2025. Training was repeated using five predefined random seeds, and same-date predicted probabilities were averaged with equal weights. In the broad study area, the seven-parameter probability-transformed ensemble achieved an area under the receiver operating characteristic curve (ROC-AUC) of 0.650 and average precision (AP) = 0.297, compared with 0.535 and 0.258 for the raw-parameter ensemble; the no-skill AP baseline was 0.209. Input-feature ablation showed that the full seven-parameter model outperformed models using N alone or N together with b. Comparison with an epidemic-type aftershock sequence model indicated that the relative performance of statistical and data-driven models depends strongly on the forecasting task. Although elevated scores were not restricted to labeled pre-earthquake windows and the separation between positive and negative score distributions weakened in the independent test period, the continuous scores retained some ranking information. Accordingly, the model output is interpreted as a continuous risk-ranking score rather than as a deterministic forecast of earthquake occurrence on any specific date. Probability-based representations of multiple catalog-derived seismicity parameters, combined with deep temporal modeling, may provide a useful complementary approach to regional identification of pre-earthquake risk windows.
ABSTRACT We present a spectral decomposition applied to nearly 10,000 earthquakes (ML 2.0–5.5) recorded along the East Anatolian Fault Zone (EAFZ) between January 2019 and February 2024, focusing on spatial and temporal variations in source parameters surrounding the 6 February 2023 seismic doublet. Results indicate that M0–fc scaling is consistent with approximate self-similarity, with nearly constant stress drop Δσ across the analyzed magnitude range. The observed linear relationship between M0 and radiated energy Er demonstrates the internal consistency of the estimated source parameters, although this relationship alone should not be interpreted as evidence of stable energy partitioning. Despite the structural segmentation of the EAFZ, Δσ remains statistically similar across the Nurdağı and Sürgü segments, indicating no clear segment-scale difference in Δσ within the analyzed dataset. To investigate spatial variations across fault segments, we examined normalized stress-drop and scaled-energy anomalies (Δσ* and E˜r*, respectively) as proxies for mechanical heterogeneity and scaled radiated energy. Δσ* exhibits relatively smooth spatial variations, whereas E˜r* shows more localized variability. These distinct spatial patterns may reflect differences in the sensitivity of the two parameters to stress conditions and local structural or mechanical heterogeneities. Our results indicate that small-magnitude earthquakes encode meaningful insights into fault segmentation and stress evolution, providing a multiscale perspective on rupture mechanics. These observations can enable detailed segment-level analyses and offer a transferable framework applicable to other tectonically complex fault systems, improving our understanding of earthquake physics and mechanical heterogeneity in strike-slip environments.
ABSTRACT Reliable measures of phase-pick confidence are essential for using neural-network phase pickers in automated seismic monitoring and catalog construction. Using the Pacific Northwest Seismic Network data set, we show that the spread of predicted probability mass functions (PMFs) tracks relative picking difficulty even when models are trained only with one-hot arrival-time labels, which assign probability one to the catalog pick time and zero to all other times, and without explicit quality information. Controlled label-perturbation experiments indicate that the predictive distributions inherit both systematic bias and label variability from the training targets, allowing probabilistic structure to emerge under deterministic supervision. We also find that disagreement across separately trained models increases for more difficult samples but can be substantially reduced by training on ensemble-averaged soft labels. PMF interquartile range follows the expected ordering of analyst-assigned quality tiers more consistently than either signal-to-noise ratio or a simple magnitude–distance metric, whereas its modest correlations with these measures indicate that it provides complementary information about picking difficulty. Together, these results show that PMF-derived measures such as interquartile range provide data-driven, reproducible indicators of picking difficulty, with practical value for both seismic monitoring and retrospective catalog construction.
The timing of past earthquakes is generally represented in the form of probability density functions (PDFs). The integration of this data from multiple sites is essential to investigate the size, spatial-temporal organization, and recurrence of large earthquakes along active fault systems. In this work, we introduce a new framework for integrating event-age PDFs from multiple sites into a unified earthquake chronology. Since a large rupture is defined by the overlap of PDFs from different trench sites, we implement the Bhattacharyya coefficient to compare earthquake PDFs and quantify the overlap degree for contemporaneous events. This approach enables a quantitative assessment of all possible rupture correlations. Second, we develop a weighted-mean method to combine overlapping PDFs into a single distribution representing the time of the rupture. These methodologies are implemented in multisite integrated chronologies (MUSIC), an open-source toolbox for correlation assessment and combination of paleoseismic data. MUSIC offers the potential to improve seismic hazard assessments along major fault systems. We demonstrate the use of MUSIC tools to better constrain the timing of large, clustered earthquakes along the central-eastern sections of the Altyn Tagh fault (ATF). The weighted-mean method yields narrower age uncertainties than the conventional mean approach and preserves contributions from all correlated paleoseismic datasets. We then compute earthquake recurrence intervals along the ATF using a Monte Carlo approach that accounts for event-age uncertainties. We found that the recurrence behavior shows a regular increase in the frequency of large earthquakes during clustering periods. Therefore, if the central-eastern ATF is currently in a cluster phase, the similar to 370-yr elapsed time implies elevated probabilities of an Mw 7.0+ earthquake occurring within the next decades.
The 1975 Ms 7.3 Haicheng earthquake, notable for its numerous short-term precursors and successful public warning, provides a unique case to test precursory scale increase (P) theories. Here, we systematically analyze the seismicity preceding the Haicheng earthquake using a composite catalog that combines pre-1970 events with post-1970 events. An automated P detection algorithm is applied to identify statistically significant P anomalies at both short and long timescales preceding the mainshock. We compare these patterns with cumulative Benioff strain and the load-unload response ratio. The results reveal coherent P anomalies before the Haicheng mainshock at both short and longer timescales, suggesting a potential linkage between seismic acceleration, stress loading-unloading behavior, and cumulative energy. These findings provide new insights into the multitimescale preparation of the Haicheng earthquake and highlight the potential value of P for medium-term earthquake forecasting.
