The landslide risk assesses how many significant landslides are expected each year per km² around the site, triggered by rainfall or by earthquakes. It uses the World Bank global landslide hazard map (Arup 2021), built on how prone slopes are to sliding and on the rainfall observed from 1980 to 2018. The class is the same in every scenario, because the IPCC does not consider landslide projections reliable.
Indicator. Annual frequency of significant landslides per km² in the site's 1 km cell: the sum of the rainfall-triggered and the earthquake-triggered part. The published class (ThinkHazard!) is reported as well.
Data. Global Landslide Hazard Map of the World Bank and GFDRR, produced by Arup (final report 2020, data 2021), 1 km resolution, between 60° S and 72° N: (a) rainfall-triggered landslides, median of the yearly totals 1980–2018 of a rare-events logistic regression (King and Zeng 2001) on observed daily rainfall and the landslides of the NASA catalogue, one model for each NASA susceptibility class; (b) earthquake-triggered landslides, NASA susceptibility combined with the expected seismic shaking after Nadim et al. (2006; 2013); (c) the qualitative class of their sum. Scenarios: SSP1-2.6 / SSP2-4.5 / SSP3-7.0 / SSP5-8.5, with the same class.
Calculation.
(1) The three published values are read in the 1 km cell that contains the site; if the cell is sea (coastal sites) the closest of the 8 adjacent land cells is used.
(2) The two frequencies are added, as in the Arup report (section 3.5.1).
(3) The class is the same in every scenario and year: IPCC AR6 (WGI, Summary for Policymakers, C.2.4) has low confidence in future changes of landslides in most regions.
Note: the map describes 1980–2018; for some mountain areas (the Alps, the Himalaya, northern Taiwan, permafrost areas) the IPCC (chapter 12) expects more landslides, which the class does not include. The 1 km cell can contain slopes steeper than the site itself.
Thresholds → level A–F. Significant landslides per km² per year.
• from 0.03 → F
• 0.01–0.03 → E
• 0.003–0.01 → D
• 0.001–0.003 → C
• 0.0001–0.001 → B
• below 0.0001 → A.
Limits 0.0001, 0.001 and 0.01 from the ThinkHazard! classes of the report (Table 7), steps 0.003 and 0.03 from the Nadim et al. 2013 scale (Table 6); the six-level ladder is a xNatura choice. The rainfall raster publishes values below 0.001 as 0, so A and B come from the published class.
From physical to impact. The physical level (A–F) is combined with the production process's sensitivity through the risk matrix (6 physical levels by 5 sensitivity bands) → impact level. The euro amount of this risk is the one described in the "Economic damage" paragraph (Version 2026): the damage factor per class of Version 2025 is no longer used for this risk.
Economic damage. Besides the class, the expected annual loss in euro is estimated as a share of the asset value. Basis: "xNatura estimate".
• Frequency: the significant landslides per km² per year of the site's cell, the same map used for the class.
• Area: 300 m² per landslide, inside the 100-1,000 m² range of the Arup report's definition of a significant landslide (xNatura choice: the map does not publish the area).
• Damage: 30% of the value when a landslide reaches the building (xNatura choice: there is no published loss ratio for reinforced concrete or steel industrial buildings; vulnerability studies such as Uzielli et al. 2008 rank reinforced structures as the least vulnerable).
• Expected annual loss = frequency × landslide area × 30% × value. No loss at a return period: at a single site the frequency is far below once in 100 years. Same value in every scenario, as the class.
The rainfall layer publishes frequencies below 0.001 as 0, so a very low hazard gives no euro.
Sources.
• [The Global Landslide Hazard Map](The Global Landslide Hazard Map)
• [Arup 2020, The Global Landslide Hazard Map: Final Project Report](Arup 2020, The Global Landslide Hazard Map: Final Project Report)
• [Global landslide and avalanche hotspots](Global landslide and avalanche hotspots)
• [Nadim et al. 2013, Assessment of global landslide hazard hotspots](Nadim et al. 2013, Assessment of global landslide hazard hotspots)
• [A heuristic approach to global landslide susceptibility mapping](A heuristic approach to global landslide susceptibility mapping)
• [A global landslide catalog for hazard applications: method, results, and limitations](A global landslide catalog for hazard applications: method, results, and limitations)
• [King and Zeng 2001, Logistic regression in rare events data](King and Zeng 2001, Logistic regression in rare events data)
• [ThinkHazard! methodology report (GFDRR)](ThinkHazard! methodology report (GFDRR))
• [IPCC AR6 WGI, Summary for Policymakers](IPCC AR6 WGI, Summary for Policymakers)
• [IPCC AR6 WGI, Chapter 12: Climate Change Information for Regional Impact and for Risk Assessment](IPCC AR6 WGI, Chapter 12: Climate Change Information for Regional Impact and for Risk Assessment)
• [Uzielli et al. 2008, A conceptual framework for quantitative estimation of physical vulnerability to landslides](Uzielli et al. 2008, A conceptual framework for quantitative estimation of physical vulnerability to landslides).
Indicator. Annual frequency of significant landslides per km² in the site's 1 km cell: the sum of the rainfall-triggered and the earthquake-triggered part. The published class (ThinkHazard!) is reported as well.
Data. Global Landslide Hazard Map of the World Bank and GFDRR, produced by Arup (final report 2020, data 2021), 1 km resolution, between 60° S and 72° N: (a) rainfall-triggered landslides, median of the yearly totals 1980–2018 of a rare-events logistic regression (King and Zeng 2001) on observed daily rainfall and the landslides of the NASA catalogue, one model for each NASA susceptibility class; (b) earthquake-triggered landslides, NASA susceptibility combined with the expected seismic shaking after Nadim et al. (2006; 2013); (c) the qualitative class of their sum. Scenarios: SSP1-2.6 / SSP2-4.5 / SSP3-7.0 / SSP5-8.5, with the same class.
Calculation.
(1) The three published values are read in the 1 km cell that contains the site; if the cell is sea (coastal sites) the closest of the 8 adjacent land cells is used.
(2) The two frequencies are added, as in the Arup report (section 3.5.1).
(3) The class is the same in every scenario and year: IPCC AR6 (WGI, Summary for Policymakers, C.2.4) has low confidence in future changes of landslides in most regions.
Note: the map describes 1980–2018; for some mountain areas (the Alps, the Himalaya, northern Taiwan, permafrost areas) the IPCC (chapter 12) expects more landslides, which the class does not include. The 1 km cell can contain slopes steeper than the site itself.
Thresholds → level A–F. Significant landslides per km² per year.
• from 0.03 → F
• 0.01–0.03 → E
• 0.003–0.01 → D
• 0.001–0.003 → C
• 0.0001–0.001 → B
• below 0.0001 → A.
Limits 0.0001, 0.001 and 0.01 from the ThinkHazard! classes of the report (Table 7), steps 0.003 and 0.03 from the Nadim et al. 2013 scale (Table 6); the six-level ladder is a xNatura choice. The rainfall raster publishes values below 0.001 as 0, so A and B come from the published class.
From physical to impact. The physical level (A–F) is combined with the production process's sensitivity through the risk matrix (6 physical levels by 5 sensitivity bands) → impact level. The euro amount of this risk is the one described in the "Economic damage" paragraph (Version 2026): the damage factor per class of Version 2025 is no longer used for this risk.
Economic damage. Besides the class, the expected annual loss in euro is estimated as a share of the asset value. Basis: "xNatura estimate".
• Frequency: the significant landslides per km² per year of the site's cell, the same map used for the class.
• Area: 300 m² per landslide, inside the 100-1,000 m² range of the Arup report's definition of a significant landslide (xNatura choice: the map does not publish the area).
• Damage: 30% of the value when a landslide reaches the building (xNatura choice: there is no published loss ratio for reinforced concrete or steel industrial buildings; vulnerability studies such as Uzielli et al. 2008 rank reinforced structures as the least vulnerable).
• Expected annual loss = frequency × landslide area × 30% × value. No loss at a return period: at a single site the frequency is far below once in 100 years. Same value in every scenario, as the class.
The rainfall layer publishes frequencies below 0.001 as 0, so a very low hazard gives no euro.
Sources.
• [The Global Landslide Hazard Map](The Global Landslide Hazard Map)
• [Arup 2020, The Global Landslide Hazard Map: Final Project Report](Arup 2020, The Global Landslide Hazard Map: Final Project Report)
• [Global landslide and avalanche hotspots](Global landslide and avalanche hotspots)
• [Nadim et al. 2013, Assessment of global landslide hazard hotspots](Nadim et al. 2013, Assessment of global landslide hazard hotspots)
• [A heuristic approach to global landslide susceptibility mapping](A heuristic approach to global landslide susceptibility mapping)
• [A global landslide catalog for hazard applications: method, results, and limitations](A global landslide catalog for hazard applications: method, results, and limitations)
• [King and Zeng 2001, Logistic regression in rare events data](King and Zeng 2001, Logistic regression in rare events data)
• [ThinkHazard! methodology report (GFDRR)](ThinkHazard! methodology report (GFDRR))
• [IPCC AR6 WGI, Summary for Policymakers](IPCC AR6 WGI, Summary for Policymakers)
• [IPCC AR6 WGI, Chapter 12: Climate Change Information for Regional Impact and for Risk Assessment](IPCC AR6 WGI, Chapter 12: Climate Change Information for Regional Impact and for Risk Assessment)
• [Uzielli et al. 2008, A conceptual framework for quantitative estimation of physical vulnerability to landslides](Uzielli et al. 2008, A conceptual framework for quantitative estimation of physical vulnerability to landslides).