Landslide Susceptibility Mapping
Also known as: Landslide Susceptibility Modeling, Slope-Failure Susceptibility Mapping, Statistical Landslide Hazard Mapping, Landslide Probability Mapping
Landslide susceptibility mapping estimates where slope failures are likely to occur by statistically relating a mapped inventory of past landslides to the terrain conditions that predispose a slope to fail. The premise, articulated across the statistical landslide literature that Guzzetti, Reichenbach, and colleagues helped systematize, is that landslides recur under geological and morphological conditions similar to those that produced them before, so the spatial pattern of past failures reveals the susceptibility of as-yet unfailed terrain. The analyst partitions the landscape into mapping units, characterizes each by conditioning factors such as slope, aspect, lithology, and land cover, and fits a classifier — logistic regression, discriminant analysis, or machine learning — to predict the probability of failure. Reichenbach and co-authors' 2018 review of 565 studies catalogued the methods, factors, and validation practices, while Guzzetti and co-workers' 2006 paper established how to rigorously assess model quality. The output is a zonation ranking terrain from low to high susceptibility. Susceptibility maps describe spatial likelihood, not when or how large a failure will be.
Key highlights
- Exploits the empirical regularity that landslides recur under terrain conditions similar to past failures, requiring no detailed mechanical modeling.
- Produces a continuous, region-wide spatial ranking from widely available DEM, geological, and land-cover data.
- Accommodates many algorithms, from interpretable logistic regression to high-performing machine learning, within one framework.
- Has a mature, standardized validation culture (independent-data AUC) that makes predictive skill explicit and comparable across studies.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use landslide susceptibility mapping when you need to know where across a region slope failures are most likely, to support land-use planning, infrastructure routing, hazard zonation, and the targeting of detailed site investigations. It is appropriate where a reasonably complete landslide inventory exists and where terrain conditioning factors — at minimum a digital elevation model and a geological map — can be assembled at a resolution matching the failures of interest. It works best for the failure type represented in the inventory, since susceptibility for shallow debris flows and for deep-seated rotational slides reflects different conditioning factors and should be modeled separately. It is not the right tool when you need the temporal probability of failure (how often) or the expected size and runout (how large), which require landslide frequency and magnitude analysis; nor when the inventory is too sparse, biased, or poorly located to support statistical learning, in which case heuristic or physically based slope-stability approaches may be preferable.
Strengths & limitations
- Exploits the empirical regularity that landslides recur under terrain conditions similar to past failures, requiring no detailed mechanical modeling.
- Produces a continuous, region-wide spatial ranking from widely available DEM, geological, and land-cover data.
- Accommodates many algorithms, from interpretable logistic regression to high-performing machine learning, within one framework.
- Has a mature, standardized validation culture (independent-data AUC) that makes predictive skill explicit and comparable across studies.
- Estimates only spatial likelihood, not the temporal frequency or the magnitude and runout of future failures.
- Is entirely dependent on the completeness, accuracy, and bias of the landslide inventory used to train and test it.
- Results vary substantially with the choice of mapping unit, conditioning factors, and algorithm, complicating comparison and transfer.
- Assumes future failures occur under conditions like past ones, an assumption strained by land-use change, climate change, and rare triggers.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
What is the difference between landslide susceptibility and landslide hazard?
Susceptibility is the spatial likelihood that a landslide will occur in a given place, given the terrain conditions; it answers 'where'. Hazard adds the temporal dimension and magnitude — the probability of a landslide of a given size occurring within a given period — answering 'where, when, and how big'. As Guzzetti and colleagues emphasize, most statistical models map susceptibility because the static conditioning factors capture spatial predisposition but not occurrence frequency. Converting susceptibility to hazard requires additional information on landslide recurrence and size distributions, which is far harder to obtain than the conditioning factors.
Why is independent validation with AUC so emphasized?
A susceptibility model can fit its training landslides almost perfectly yet generalize poorly, so fit alone is a misleading measure of quality. Guzzetti and co-authors argued that models must be tested on landslides not used in fitting — held out spatially or temporally — and their prediction skill quantified, most commonly by the area under the receiver-operating-characteristic curve, which measures how well the model ranks failed above unfailed locations across all thresholds. An AUC near 0.5 means no skill and near 1.0 means excellent ranking. Reporting independent-data AUC together with an account of uncertainty is now the expected standard for credible susceptibility maps.
Does the choice of mapping unit really matter?
Yes, substantially. The mapping unit sets the spatial element that receives a susceptibility value and the scale at which conditioning factors are summarized. Reichenbach and co-authors document that grid cells are simple and ubiquitous but somewhat arbitrary and resolution-dependent, while slope units, delineated by drainage and divide lines, are more geomorphologically meaningful and align better with the physical bodies that actually fail. Changing the unit can change both the apparent performance and the spatial pattern of the result, so the unit should be chosen to match the failure type and the intended use of the map, and reported transparently.
Sources
- 1.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.
- 2.Guzzetti, F., Reichenbach, P., Ardizzone, F., Cardinali, M., & Galli, M. (2006). Estimating the quality of landslide susceptibility models. Geomorphology, 81(1-2), 166-184.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). Landslide Susceptibility Mapping. ScholarGate. https://scholargate.app/disaster-studies/landslide-susceptibility-mapping