Recent advances in machine learning (ML)-based earthquake detection and location techniques combined with dense seismic networks have dramatically increased the quantity of low-magnitude earthquakes that can be detected and accurately located. We analyze the seismicity of the Northern Apennines (Italy) using data from the Alto Tiberina Near Fault Observatory (TABOO-NFO), an ideal site for applying modern detection techniques due to the presence of a dense seismic network, intense microseismic activity involving a complex fault system and deep fluid circulation. We use a ML-based workflow tailored to the TABOO area from 2010 to 2023 to construct an earthquake catalog with 420k events relocated with double difference. We leverage available data in the TABOO-NFO by using manually picked waveforms to train a new regional version of PhaseNet, a DL-phase picker designed for high accuracy P and S wave phase identification. The catalog provides new insights into geological and structural features in the area. We evaluate its quality by comparing the distribution of the hypocenters with the known active faults of the seismogenic region at TABOO. Our catalog illuminates structures and fault geometries with great detail, allowing detailed characterization of the spatiotemporal evolution of both the microseismicity occurring along the ATF and the shallow sequences. The focal mechanisms we obtain show more complex fault kinematics for deeper events compared minor, shallower fault segments. We illuminate a correlation between earthquake source properties and host rock lithology that includes the spatial pattern of the seismicity and the frequency-magnitude behavior.

Seismic waves contain information about the earthquake source, the geologic structure they traverse, and many forms of noise. Separating the noise from the earthquake is a difficult task because optimal parameters for filtering noise typically vary with time and, if chosen inappropriately, may strongly alter the original seismic waveform.  Diffusion models based on Deep Learning (DL) have demonstrated remarkable capabilities in restoring images and audio signals. However, those models assume a Gaussian distribution of noise, which is not the case for typical seismic noise. Motivated by the effectiveness of "cold" diffusion models in speech enhancement, medical anomaly detection, and image restoration, we present a cold variant for seismic data restoration. We describe the first Cold Diffusion Model for Seismic Denoising (CDiffSD), including key design aspects, model architecture, and noise handling. Using metrics to quantify the performance of CDiffSD models compared to previous works, we demonstrate that it provides a new standard in performance. CDiffSD significantly improved the Signal to Noise Ratio (SNR) by about 18% compared to previous models. It also enhanced Cross-correlation by 6%, showing a better match between denoised and original signals. Moreover, testing revealed a 50% increase in the recall of P-wave picks for seismic picking. Our work show that CDiffSD outperforms existing benchmarks, further underscoring its effectiveness in seismic data denoising and analysis. Additionally, the versatility of this model suggests its potential applicability across a range of tasks and domains, such as GNSS, Lab Acoustic Emission, and DAS data, offering promising avenues for further utilization.