| 11 Feb 2026
runoutSIM v1.0: An R package for regionally simulating landslide runout and connectivity using random walks
Abstract. Regional-scale runout modelling for landslide hazard assessment and land-use planning helps us understand not only the general likelihood of being impacted by their runout, but also how runout paths and distances vary under different environmental conditions. While R is widely used in geosciences for spatial prediction and susceptibility modelling, most existing runout models are not implemented directly in R, often requiring coupling with external software. This creates barriers for model development, modification, and integration with other geospatial and statistical tools.
To address this, runoutSIM is presented, an open-source R package for simulating the spatial extent, velocity, and connectivity of landslide runout at a regional scale. The model combines random walks to represent flow paths with a process-based approach to control runout distance and includes functionality to estimate the connectivity probability of runout from source areas intersecting with downslope features. In this model, the runout path and connectivity probabilities can also be adjusted by using spatial likelihoods of source cell predictions, such as those derived from statistical or machine learning models. In addition, runoutSIM provides an interactive map viewing environment within R that allows users to explore and query simulation results and related spatial data.
By implementing these algorithms natively in R, runoutSIM lowers technical barriers, supports flexible model development, and enables integration with data-driven approaches. We demonstrate the package in the Río Olivares basin, Chile, where a regional runout model optimized using a random grid search, machine-learning prediction of source areas, and simulation of runout connectivity help identify areas most susceptible to hazardous runout and potential source locations. runoutSIM provides a transparent and reproducible framework for regional runout modelling, supporting hazard assessment and enabling further development within R, a widely used geoscientific environment.
Runoff modelling is one of the most important issues in hydrology. In this submission, an R package named runoffSIM is developed to establish a relationship between the runoff process and the probability of landslides. In general, the modelling process consists of two stages: first, runoff is calculated, and then landslide probability is evaluated.
Four comments are provided below to facilitate further improvements of the paper:
1. A wide variety of runoff simulation algorithms exist in the literature. In the first stage of the proposed modelling framework, are other existing runoff simulation algorithms applicable to runoff calculation? If yes, the authors are advised to compare the effectiveness of runoffSIM with that of these existing algorithms. If not, the authors should elaborate on the unique advantages of runoffSIM that make it superior or more suitable for the proposed application.
2. The formulation of landslide probability within runoffSIM is only briefly described. Given the central role of landslide analysis in this package, more detailed explanations are required. In particular, the intricate relationships between runoff dynamics and landslide occurrence (e.g., how runoff parameters influence landslide probability) have not yet been sufficiently elaborated.
3. Observational datasets that include both runoff and landslide data are relatively scarce in peer-reviewed studies. Is the dataset used in the current analysis publicly available? The authors are recommended to provide more detailed information about the dataset, such as its source, spatial-temporal coverage, data collection methods, and basic descriptive statistics.
4. Verification is a critical component for evaluating the effectiveness and reliability of runoffSIM. The authors are encouraged to refer to commonly used verification metrics and diagnostic plots in the field of forecast verification (e.g., bias, root mean square error, receiver operating characteristic curves). Additionally, the verification experiments designed to test runoffSIM, as well as the specific metrics employed, should be described in greater detail to enhance the credibility of the results.