Anticipating landslides has long been a defining ambition of geomorphology. Over the past four decades, the field has progressed from largely qualitative assessments of unstable slopes to probabilistic, spatially explicit, and increasingly dynamic forecasts. Yet prediction remains difficult because landslides encompass diverse processes, scales, triggers, and consequences. A useful anticipation (e.g., forecast, prediction, projection, outlook) must specify where and when failures may occur, how many are expected, how large and mobile they may become, and whom or what they may affect.
Several advances have transformed this capacity. Inventories and multi-temporal mapping show that landslides are pervasive landscape processes, enabling susceptibility models from local to global scales. Analyses of event and multi-temporal inventories have clarified recurrence, path dependence, and size distributions; empirical length, width, area, volume, and area-volume relationships support estimates of magnitude. For rainfall-induced failures, thresholds based on rainfall intensity, duration, and accumulation evolved from heuristic lines into probabilistic models describing regional variability and uncertainty. More recently, machine learning has captured nonlinear relations between antecedent and triggering rainfall, produced hourly forecasts, combined ensembles of predictors, and opened a path from short-term warnings to decadal space-time projections. Ground-based and satellite remote sensing, sensor networks, improved telecommunications, computational power, and physically based modelling have expanded monitoring and sometimes allowed the onset of individual failures to be anticipated. Risk models have also begun to quantify societal consequences and reveal how vulnerability varies with age, gender, location, and behaviour.
Progress is uneven. Landslide inventories and monitoring are extensive in parts of Europe, the United States, Japan, and New Zealand, but accessible information is scarcer across much of Africa and South America. This imbalance reflects differences in observation, reporting, and data access, not necessarily differences in hazard. Even well-mapped regions retain blind spots, as the 2023 Emilia-Romagna landslides demonstrated. Submarine landslides remain poorly observed despite their potential consequences. Sparse seabed mapping and rare direct observations limit knowledge of their frequency, triggers, and dynamics. Yet they can generate tsunamis that threaten coastal communities and damage submarine cables and other offshore infrastructure.
These achievements do not amount to deterministic predictions. Susceptibility is often confused with hazard; model scores are misinterpreted as intermediate probabilities; performance metrics can conceal where errors occur; and forecasts are commonly tested against incomplete or non-independent observations. Inventories remain temporally uneven, spatially incomplete, and biased toward visible or damaging events. Precise initiation times, essential for training and validating warning systems, are rarely available. Monitoring networks remain sparse and fragmented. Models still struggle to predict the number, exact location, size, runout, velocity, and impacts of failures. Seasonal forecasts remain uncertain, while multi-year forecasts needed for land-use planning are immature.
Climate and land-use change further challenge assumptions of stationarity. More intense rainfall may increase rapid shallow failures, whereas warmer and drier conditions may reduce some slow, deep-seated movements; snowmelt, permafrost thaw, vegetation change, fire, and human modification introduce additional, interacting controls. Robust projections therefore require representative event and multi-temporal inventories, systematic detection, improved subsurface hydrological and geotechnical data, models that assimilate monitoring observations in real time, and explicit treatment of uncertainty.
The remaining challenge is not scientific alone. Anticipations must be connected to vulnerability, exposure, and cascading hazards; data and standards must be open and interoperable; responsibilities must be coordinated across institutions; and warning information must be communicated in forms that support decisions. Because natural hazards cannot be eliminated and decisions are ultimately dichotomous even when forecasts are probabilistic, uncertainty is not a justification for inaction. The next stage of progress will depend on integrating prediction science, sustained monitoring, governance, and public engagement so that anticipation becomes more accurate, credible, actionable, and socially useful.
Recommended readings
Casagli, N., Intrieri, E., Tofani, V., Gigli, G., Raspini, F. (2023). Landslide detection, monitoring and prediction with remote-sensing techniques. Nat Rev Earth Environ 4, 51–64. https://doi.org/10.1038/s43017-022-00373-x
Gariano, S. L., & Guzzetti, F. (2016). Landslides in a changing climate. Earth-Science Reviews, 162, 227-252. https://doi.org/10.1016/j.earscirev.2016.08.011
Guzzetti, F., Berti, M., Reichenbach, P., & Tofani, V. (2025). Landslide risk management in Italy: practices, advances, and future directions. Rendiconti Lincei. Scienze Fisiche e Naturali, 36, 1165-1173. https://doi.org/10.1007/s12210-025-01382-w
Guzzetti, F., Gariano, S.L., Peruccacci, S., Brunetti, M.T., Marchesini, I., Rossi, M., Melillo, M., (2020). Geographical landslide early warning systems. Earth-Science Reviews 200, 102973. https://doi.org/10.1016/j.earscirev.2019.102973
Guzzetti, F., Mondini, A.C., Cardinali, M., Fiorucci, F., Santangelo, M., Chang, K.-T. (2012). Landslide inventory maps: New tools for an old problem. Earth-Science Reviews 112, 42–66. https://doi.org/10.1016/j.earscirev.2012.02.001
Mondini, A. C., Guzzetti, F., Melillo, M. (2023). Deep learning forecast of rainfall-induced shallow landslides. Nature Communications, 14, 2466. https://doi.org/10.1038/s41467-023-38135-y
Mondini, A.C., Guzzetti, F., Chang, K.-T., Monserrat, O., Martha, T.R., Manconi, A., 2021. Landslide failures detection and mapping using Synthetic Aperture Radar: Past, present and future. Earth-Science Reviews 216, 103574. https://doi.org/10.1016/j.earscirev.2021.103574
Reichenbach, P., Rossi, M., Malamud, B. D., Mihir, M., Guzzetti, F. (2018). A review of statistically-based landslide susceptibility models. Earth-Science Reviews, 180, 60-91. https://doi.org/10.1016/j.earscirev.2018.03.001
/filters:quality(70)/prod01/prodbucket01/media/durham-university/research-/research-institutes/institute-of-hazard-risk-and-resilience/43875.jpg)