Data and code for: "Seasonal preconditioning and triggering of alpine mass movements at an active rock slope from machine-learning seismic monitoring" by Chmiel et al.

This repository contains data and code supporting the manuscript: "Seasonal preconditioning and triggering of mass movements at an active rock slope in the Swiss Alps from machine-learning seismic monitoring" by Małgorzata Chmiel (1,2), Fabian Walter (2), Lena Husmann (2), GiacomoBelli (3), Emanuele Marchetti (3), Clément Hibert (4), Nils Hählen (5), Christian Kienholz (6), submitted to Journal of Geophysical Research: Earth Surface (1) Observatoire de la Côte d'Azur, Université Côte d'Azur, CNRS, IRD, Géoazur, 250 rue Albert Einstein, Sophia Antipolis 06560 Valbonne, France(2) Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Birmensdorf, Switzerland(3) Department of Earth Sciences, University of Firenze, Firenze, Italy(4) Institut de Physique du Globe de Strasbourg, University of Strasbourg/EOST, Strasbourg, France(5) Natural Hazard Division of Canton Bern, Interlaken, Switzerland(6) GEOTEST AG, Zollikofen, Switzerland The repository includes seismic-derived event catalogs, infrasound training data, auxiliary environmental monitoring datasets, the trained Random Forest model, and the analysis notebooks used to reproduce the results and figurespresented in the manuscript. ## Repository Structure ### 1. Ancillary_data/Auxiliary monitoring datasets used for event validation and environmentalinterpretation (01 October 2021 – 31 December 2024).- **Borehole data** — ground temperature time series from thermistor strings (borehole TB2).- **GNSS data** — displacement time series (1-hour sampling; stations GPS12 and GPS5).- **Precipitation data** — Doldenstock meteorological station (10-minute sampling).- **Air temperature data** — Doldenstock meteorological station (10-minute sampling). ### 2. Infrasound_data_2021/Infrasound dataset (miniSEED) used to construct the labeled training catalog(October–December 2021).- Air-pressure time series- State-of-health (SOH) data Channels CH00–CH03 correspond to array elements SPT00–SPT03 (Table S1 of the manuscript), SPT00 is the central element. | Name (element) | Easting | Northing ||---|---|---|| CH00 (SPT00) | 620714 | 149949 || CH01 (SPT01) | 620703 | 149869 || CH02 (SPT02) | 620674 | 150011 || CH03 (SPT03) | 620785 | 149961 | ### 3. Model_original/Machine-learning outputs and seismic event catalogs.- **catalog_complete/** — final processed seismic catalog (12,962 events, 8 January 2022 – 13 December 2024) after post-processing and removal of 33 fieldwork-related detections (see fieldwork_exclusion_windows.csv). Columns: event_id, start_time_utc, end_time_utc, duration_s, seismic_energy_proxi (the seismic energy proxy Es of the manuscript, arbitrary units), classifier_probability.- **detected_events_2022/ 2023/ 2024/** — raw detection outputs prior to final post-processing. Same column structure.- **fieldwork_exclusion_windows.csv** — 30-minute exclusion windows around on-site fieldwork activities on four days; detections fully contained within a window (start and end inside it) were removed from the catalog (33 events).- **RF_model.pkl** — trained Random Forest classifier used for mass-movement detection. ### 4. Jupyter_notebook/Notebooks used to train the model and reproduce the manuscript figures.Figure numbers refer to the revised manuscript:- Figure 5: frequency-energy analysis (maximum-likelihood b-values with bootstrap confidence intervals, SeismoStats)- Figures 6–8: multi-year catalog, seasonal patterns, and triggering statistics- Figure 9: coupling between deep-seated creep (GNSS) and surface mass-movement activity ## Time Coverage01 October 2021 – 31 December 2024. All timestamps in UTC. ## Additional NotesContinuous seismic waveform data are archived at the Swiss Seismological Service (network 9S) and are subject to a five-year embargo period, after which they become publicly available. Infrasound data used for bulding the training catalog (October–December 2021) are included in this repository.