Disentangling Static and Dynamic Landslide Drivers: A Dual-Phase Bayesian Framework for a Paraglacial Tyrolean Basin - Supplementary Data

Disentangling Static and Dynamic Landslide Drivers - Code Repository

This repository contains the R and JavaScript code required to reproduce the analysis presented in: "Disentangling Static and Dynamic Landslide Drivers: A Dual-Phase Bayesian Framework for a Paraglacial Tyrolean Basin" (Hunyadfalvi, 2025).

Script Execution Order

To fully reproduce the results and figures, execute the scripts in the following order:

GEE_CHIRPS_Processing.js: (Google Earth Engine) Run this in the GEE Code Editor to extract the raw satellite precipitation metrics.

00 Spatial Join Landslides to WLV.R: Links landslide polygons to WLV catchment IDs and historical event dates.

01 Process Gauge Data.R: (NEW) Automates the calculation of gauge statistics and spatially filters them to the Study Area ROI (Replacing the manual QGIS step).

02 Rainfall Interpolation (Ordinary Kriging).R: Performs Ordinary Kriging to generate the downscaled MAP map (Figure 1).

03 Phase 1 Susceptibility Model.R: Fits the INLA-SPDE Hurdle model (Occurrence + Magnitude).

04 Visualize Latent Spatial Field.R: Creates the publication-quality Rainfall Map.

05 Visualize Mean Annual Precipitation.R: Creates the publication-quality Latent Spatial Field Map.

06 Phase 2 Dynamic Trigger Analysis.R: Performs Firth Penalized Regression to identify the convective triggering threshold.

Software Requirements

R (>= 4.0.0)

INLA (R-INLA), fmesher, sf, terra, gstat, ggplot2, logistf, tidyverse