The file corresponds to a code written using the R software version 4.0.5 (R Core Team, 2021). We used a Bayesian robust regression to predict the posterior probability P(L) at which a given location yi in our study areas (north Patagonia, Chile) is classified as part of a landslide source, transport, or deposition area.
We used the NUTS sampling scheme implemented in the STAN probabilistic programming language (Carpenter et al., 2017) to draw samples from the joint posterior distribution via the R package brms (Bürkner, 2017). We ran four independent Hamiltonian Monte Carlo chains based on 2000 iterations including 500 warm-up samples and checked each chain for convergence. We assessed the performance of this classifier based on its posterior predictive distribution and recorded the fraction of correct classifications compared to the observed frequency of landslides in all study areas and for all landform types.
We find that higher crown openness and wind speeds credibly predict higher probabilities of detecting landslides regardless of topographic location, though much better in low-order channels and on midslope locations than on open slopes. Wind speed has less predictive power in areas that were impacted by tephra fall from recent volcanic eruptions, while the influence of forest cover in terms of crown openness remains.