Fig 11 - uploaded by P. Frattini
Content may be subject to copyright.
Frequency of hazard values of hazard maps with 10% exceedance probabilities in 50 years, for earthquakes with 500 and 2500 years return time.
Source publication
We present a general framework for probabilistic landslide hazard analysis. With respect to other quantitative hazard assessment approaches, this probabilistic landslide hazard analysis has the advantage to provide hazard curves and maps, and to be applicable to all typologies of landslides, if necessary accounting for both their onset and transit...
Context in source publication
Context 1
... probability of 10% in 50 years are about 60 kJ and 8 kJ for 500 and 2500 years return time, respectively ( Figure 11). With respect to other quantitative hazard assessment approaches (e.g. Hovius et al, 1997; Hungr et al. 1999; Stark and Hovius, 2001; Dussauge et al, 2003; Guthrie and Evans, 2004; Malamud et al., 2004; Marchi and D ’ Agostino, 2004; Jakob and Friele, 2010; Santana et al, 2012), the probabilistic landslide hazard analysis here proposed explicitly expresses hazard as a function of landslide destructive power, considering landslide intensity rather than magnitude. This supports the use of vulnerability and fragility functions, and the assessment of risk. With a few exceptions (Straub and Schubert, 2008; Spadari et al., 2013), the above mentioned approaches do not account explicitly for uncertainty. For intensity, simple statistics of the expected value are used (e.g. arithmetic average, maximum value, or a speci fi c percentile), neglecting the dispersion and introducing strong assumptions about its distribution. In the proposed methodology, we explicitly account for the uncertainty by using the whole intensity distribution in the hazard analysis. This also allows to obtain hazard curves, which explicitly represent the probability of exceeding a certain level of landslide intensity within a de fi ned time period, accounting for uncertainty and integrating different magnitude scenarios. The proposed probabilistic landslide hazard analysis is a fl exible approach, applicable to all typologies of landslides, if necessary accounting for both their onset and transit probability. However, its mathematical formulation has to be declined according to the distribution of landslide intensity values, which has to be veri fi ed in each speci fi c ...
Similar publications
Unpublished PhD Thesis. University of Canterbury, Department of Geological Sciences. Accepted 2014.
The 2016 Mw 7.8 Kaikoura earthquake and consequent tsunami have been controversial because of uncertainty over whether and where the plate interface ruptured and the incapability of the proposed source models to reproduce the near-field runup of 7 m. Existing models identify a wide range of locations for the interface rupture , from on land to offs...
Landsliding resulting from large earthquakes in mountainous terrain presents a substantial hazard and plays an important role in the evolution of mountain ranges. However estimating the scale and effect of landsliding from an individual earthquake prior to its occurrence is difficult. This study presents first order estimates of the scale and effec...
Citations
... Following the hazard definition proposed by Lari et al. (2014), the ELHA model (Eq. (9)) holistically incorporates both landslide probability and seismic landslide intensity. ...
Earthquake-induced landslides cause extensive harm, necessitating accurate predictions for effective risk management. To tackle the dual challenges posed by inadequate model accuracy and the absence of frequency-derived landslide intensity as critical information, this paper presents a robust modelling method that generates high-resolution landslide hazard map for risk assessment. Firstly, the Frequency Ratio (FR) model is employed to quantitatively explore the correlation between geo-environmental factors and landslide occurrences, enabling the exclusion of less significant factors. Subsequently, these FR values are integrated into the Light Gradient Boosting Machine (LightGBM) model, resulting in the composite model, FR-LightGBM. To test the proposed model, data from the areas affected by the Luding earthquake based on the seismic landslide intensity model following the 2022 Ms 6.8 earthquake in Sichuan, China, is examined. Model validation, using the area under the curve (AUC) of the receiver operating characteristic curve, highlighted the superior predictive accuracy of the FR-LightGBM model, marking a 3.5 % improvement in AUC value compared to both the Convolutional Neural Network (CNN) and Logistic Regression (LR) models. The results have identified high-risk areas along the Dadu riverside and the Xianshuihe fault zone correspond well with the actual landslide distribution. Overall, the practical applicability of the FR-LightGBM model is significantly enhanced by its capacity to prioritize lithology, faults, and aspect as crucial factors. Additionally, based on an extended landslide inventory established using high spatial resolution (2 m) satellite imagery from the Chinese Gaofen satellite series, the proposed approach delivered a high-resolution (12.5 m) hazard map across the Luding earthquake area, contributing to relevant studies and risk management.
... Landslide size, which controls the destructive power of landslides, is an essential component of landslide hazard assessments (Lari et al., 2014;Valagussa et al., 2019). Thus, the research on the relation between co-seismic landslide size and distance to fault is meaningful. ...
Major earthquakes can cause extensive landsliding that poses a major threat to both property and human lives. In addition to co‐seismically triggered ground failure, the earthquake‐affected region remains vulnerable to landslides due to loosened and unstable materials and structures. Many researchers have studied landslide distributions and their controlling factors after earthquakes, but the function of ground motion is unclear. To investigate the connection in a strike‐slip earthquake, we analyzed the 5 September 2022 Luding earthquake (Mw 6.6) in Sichuan Province, China. We interpreted remote‐sensing images to obtain the landslide distribution before and after the earthquake, calculated surface deformation from D‐InSAR data (pre‐ and post‐earthquake), utilized a point‐source model for the focal mechanism inversion, and then constructed a finite fault model for the rupture slip. There are clear differences in the landslide distributions on the two sides of the fault before and after the earthquake. The density of co‐seismic landslides on the west side of the fault exceeded that on the east side. The patterns of surface deformation and ground motion indicated that the areas with larger deformation and motion were associated with more landslides. Furthermore, the landslide size decreased with distance from the fault. A new finding is that co‐seismic landslides induced by strike‐slip earthquakes result in high landslide concentration on both sides of the fault, while previous studies find that co‐seismic landslides triggered by thrust earthquakes present a hanging wall concentrated distribution pattern. These findings contribute to a more comprehensive understanding of the connection between ground movement patterns and landslide distributions.
... (i) Difficult access conditions; (ii) Often very cloudy satellite coverage; (iii) Low coverage of aerial photographs (Monsieurs et al., 2018;Jacobs et al., 2018); (iv) Rapid vegetation regrowth, masking the areas where the phenomena were triggered (Sidle et al., 2006;Jacobs et al., 2016;Robinson et al., 2019); (v) Lack of knowledge on geotechnical data (characteristics, thickness) because large variety of heterogeneous, juxtaposed or overlapping, and often highly weathered materials (Baecher and Christian, 2003); (vi) Insufficient information about the temporality (date or period of occurrence) of landslides (Witt et al., 2010;Lari et al., 2014). This information is essential for assessing landslide hazard, either through a direct approach (analyzing historical data and calculating associated probabilities) or an indirect approach (analyzing precipitation or earthquake data and computing triggering thresholds; Corominas and Moya, 2008;Witt et al., 2010;Havenith et al., 2016;Melillo et al., 2018;Monsieurs et al., 2018;Shao and Xu, 2022). ...
... This approach allows establishing scenarios for different time periods (Crovelli, 2000;Lari et al., 2014). Despite the potential for a temporal effect of post-event clustering that can elevate probabilities over a short period, the outcomes accurately reflect the temporal probability of events (Lari et al., 2014). ...
... This approach allows establishing scenarios for different time periods (Crovelli, 2000;Lari et al., 2014). Despite the potential for a temporal effect of post-event clustering that can elevate probabilities over a short period, the outcomes accurately reflect the temporal probability of events (Lari et al., 2014). ...
Malawi, a landlocked country characterized by its mountainous terrain crisscrossed by the Great Rift Valley and Lake Malawi, is facing an increasing threat of landslides primarily triggered by heavy rainfall from tropical cyclones and depressions. While there are general landslide susceptibility maps available on a global and continental scale, Malawi lacks its own specific landslide hazard maps that take into account regional nuances, different types of landslides, and their triggering factors. However, factoring in these parameters is essential for accurately quantifying hazards. This contribution aims to fill this gap by proposing hazard maps that consider both spatial and temporal probabilities of landslide events. The methodology employed here is based on quantifying the probability of failure at both spatial and temporal levels, following the guidelines set forth by the Joint Technical Committee – 1 for slopes and landslides (JTC-1). To enhance the accuracy of the landslide inventory, a combination of literature review, visual remote sensing, and field surveys was used. This comprehensive data collection approach included information on the types of landslides, their activity levels, and the periods during which they are most likely to be triggered. Subsequently, susceptibility analyses were conducted for various types of landslides using a data-driven approach. Temporal analyses were carried out, taking into consideration two key factors: (i) the recurrence time of different phenomena, such as debris-flows, debris-slides, and slides from 1946 to 2019, and (ii) the rainfall patterns induced by various tropical meteorological events, as defined by the World Meteorological Organization and Meteo-France. the computation of exceedance probabilities based on the Poisson distribution, predicting the likelihood of landslide reactivation for six different return periods, ranging from 1 to 100 years, following various typical meteorological events. Subsequently, susceptibility analyses were conducted for various types of landslides using a data-driven approach. Temporal analyses were carried out, taking into consideration two key factors: (i) the recurrence time of different phenomena, such as debris-flows, debris-slides, and slides from 1946 to 2019, and (ii) the rainfall patterns induced by various tropical meteorological events, as defined by the World Meteorological Organization and Meteo-France. the computation of exceedance probabilities based on the Poisson distribution, predicting the likelihood of landslide reactivation for six different return periods, ranging from 1 to 100 years, following various typical meteorological events. Ultimately, this methodology facilitates the development of various spatio-temporal landslide risk scenarios on a national scale.
... For this reason, few recent alternatives have been proposed to express landslide intensity over entire landscapes. Specifically, Lari et al. (2014) have proposed an intensity measure for rock falls under the assumption that failure sources are distributed according to a Poisson exponential family. Similarly, Lombardo et al. (2018) proposed a doubly stochastic structure (via a Log-Gaussian Cox Process -LGCP) to model the expected number of landslides per mapping unit. ...
... Therefore, the point we are trying to raise here is questioning whether the susceptibility framework shouldn't be considered largely solved (Ozturk et al. 2021), whenever heavily non-linear models are tasked with distinguishing the distribution of landslides purely in space. Conversely, the data-driven estimation of landslide intensity (Lari et al., 2014), whether it is spatially (Moreno et al., 2023), temporally , or spatiotemporally (Fang et al., 2024) addressed it is still at an infancy stage where few contributions are available and much may still be gained from a common geoscientific effort. ...
At the time of its development, GeoSure was created using expert knowledge based on a thorough understanding of the engineering geology of the rocks and soils of Great Britian. The ability to use a data-driven methodology to develop a national scale landslide susceptibility was not possible due to the relatively small size of the landslide inventory at the time. In the intervening 20 years the National Landslide Database has grown from around 6000 points to over 18,000 records today and continues to be added to. With the availability of this additional inventory, new data-driven solutions could be utilised. Here, we tested a Bernoulli likelihood model to estimate the probability of debris flow occurrence and a log-Gaussian Cox process model to estimate the rate of debris flow occurrence per slope unit. Scotland was selected as the test site for a preliminary experiment, which could potentially be extended to the whole British landscape in the future. Inference techniques for both of these models are applied within a Bayesian framework. The Bayesian framework can work with the two models as additive structures, which allows for the incorporation of the spatial and covariate information in a flexible way. The framework also provides uncertainty estimates with model outcomes. We also explored consideration on how to communicate uncertainty estimates together with model predictions in a way that would ensure an integrated framework for master planners to use with ease, even if administrators do not have a specific statistical background. Interestingly, the spatial predictive patterns obtained do not stray away from those of the previous GeoSure methodology, but rigorous numerical modelling now offers objectivity and a much richer predictive description.
... As for the "when" or "how frequently" components, the scientific efforts have consolidated into near-real-time or early-warning systems such as the Ground Failure module of the USGS (Nowicki Jessee et al., 2018) for co-seismic landslides or the LHASA product of NASA (Stanley et al., 2021) for rainfall-induced landslides. Finally, regarding the intensity component (Corominas et al., 2014), this is the element that has historically hardly received any attention from the statistical perspective, with limited models tested for rockfalls (Lari et al., 2014) and debris flows (Lombardo, Saia, et al., 2018;Lombardo, Opitz, & Huser, 2018. ...
The most adopted definition of landslide hazard combines spatial information about landslide location (susceptibility), threat (intensity), and frequency (return period). Only the first two elements are usually considered and estimated when working over vast areas. Even then, separate models constitute the standard, with frequency being rarely investigated. Frequency and intensity are intertwined and depend on each other because larger events occur less frequently and vice versa. However, due to the lack of multi-temporal inventories and joint statistical models, modelling such properties via a unified hazard model has always been challenging and has yet to be attempted. Here, we develop a unified model to estimate landslide hazard at the slope unit level to address such gaps. We employed deep learning, combined with a model motivated by extreme-value theory to analyse an inventory of 30 years of observed rainfall-triggered landslides in Nepal and assess landslide hazard for multiple return periods. We also use our model to further explore landslide hazard for the same return periods under different climate change scenarios up to the end of the century. Our results show that the proposed model performs excellently and can be used to model landslide hazard in a unified manner. Geomorphologically, we find that under both climate change scenarios (SSP245 and SSP885), landslide hazard is likely to increase up to two times on average in the lower Himalayan regions while remaining the same in the middle Himalayan region whilst decreasing slightly in the upper Himalayan region areas.
... Landslide hazard assessment aims to identify which areas are most prone to trigger landslides with a certain intensity within a given period of time (Guzzetti et al., 2005;van Westen et al., 2006;Corominas et al., 2014;Lari et al., 2014). Therefore, landslide hazard evaluation is carried out by means of the analysis of three probabilities: probability of landslide size, temporal probability of landslides and spatial probability of landslides also known as landslide susceptibility. ...
Landslides are widespread phenomena that occur in any terrestrial area with slopes, causing massive property damage and, in the worst-case scenario, loss of human life. This propensity to suffer losses is particularly high for developing countries due to their urban development, population growth and drastic land use changes. Social and economic consequences of landslides can be reduced through detailed planning and management strategies, which can be aided by risk analysis. In this study, we performed a detailed quantitative risk analysis for landslides in the whole of Central Asia (4 000 000 km2). Landslide-induced risk was computed, using a 200 m spatial resolution, in terms of exposed population and expected economic losses to buildings and linear infrastructures (roads and railways). The purpose of our study is to produce the first regional-scale landslide risk assessment for Central Asia in order to inform regional-scale risk mitigation strategies, and it represents an advanced step in the landslide risk analysis for extremely broad areas.
... As a result, existing and new thresholds are difficult to compare and evaluate quantitatively (Guzzetti et al. 2008;Brunetti et al. 2010). These methods focus on the temporal scale, considering individual landslides as discrete point events, and thus do not provide spatial information, while landslide hazard management would require the combined anticipation of "when", but also "where" and "how large" a landslide is expected to be (Lari et al. 2014). Recent studies have worked in combining the spatial and temporal occurrence prediction of landslides, by applying statistical or machine learning algorithms (Lombardo et al. 2020). ...
In Italy, rainfall represents the most common triggering factor for landslides; thus, many Italian Regional Departments of Civil Protection are setting up warning systems based on rainfall recordings. Common methods are mainly based on empirical relationships that provide the rainfall thresholds above which the occurrence of landslide phenomena is likely to be expected. In recent years, the use of machine learning approaches has gained popularity in landslide susceptibility analysis and prediction. To support the operational early warning system of Liguria Civil Protection Department for landslides hazard, we propose the implementation of a polynomial Kernel regularized least squares regression (KRLS) algorithm, for predicting the daily occurrence of shallow landslides in the five Alert Zones in Liguria (North Western Italy). The model provides an estimate of the number of landslides associated with the set of three different hydrological features, also used for the Hydrological Assessment procedure: the soil moisture, the accumulated precipitation over 12 h and the precipitation peak over 3 h. Results of the model are converted to an Alert Scenario of landslide occurrence, based on the magnitude of the expected event and identified according to the National and Regional legislation (Regional Civil Protection guidelines D.G.R. n. 1116, 23/12/2020). The performance of the predictive model (e.g. accuracy of 93%) is deemed satisfactory and the methodology is considered a valuable support to the operational early warning system of Liguria Civil Protection Department. The choice of predictive variables allows, in future development, the values obtained from historical data to be replaced by those obtained from meteorological forecast models, introducing the use of the developed model in the operational forecasting chain.
... Since the early 1980s, many researchers have attempted to identify landslide-susceptible zones by using a heuristic approach (Pachauri et al. 1998; Barredo et al. 2000;Korup 2005;Huang et al. 2020;Yu et al. 2022), semi-quantitative approach (Kumar and Anbalagan 2015;Singh et al. 2019;Psomiadis et al. 2020;Sonker et al. 2021;Shano et al. 2022;Gatto et al. 2023) and quantitative approach also known as statistical, artificial intelligence, probabilistic, and deterministic methods (Remondo et al. 2005;Guzzetti et al. 2005;Abella and Van Westen 2007;Pradhan et al. 2009;Lari et al. 2014;Vasu et al. 2016;Kumar and Anbalagan 2019;Zeng et al. 2021;Ali et al. 2022;Sun et al. 2023;Khusulio and Kumar 2023;Chang et al. 2023). To determine which model is the most accurate, several researchers have utilized a variety of models and compared them (Erener and Düzgün 2010;Conforti et al. 2011;Das and Lepcha 2019;Peethambaran et al. 2020;Pradhan et al. 2021;Jiang et al. 2023). ...
This research was to assess the terrain factors responsible for landslide and generation of landslide susceptibility map (LSM) in the Mao-Maram of Manipur, India. Using multi-source spatial data, twelve causal factors were identified slope, relative
relief, road buffer, fault/fold and thrust (FFT) buffer, aspect, stream power index (SPI), drainage buffer, profile curvature, land used/land cover (LULC), soil, and geology. In addition, using fieldwork and remote sensing technique, 234 landslide incident was recorded for training and testing the model of the study region. Three data mining models used are support vector machine, random forest, and boosted tree which was used to generate a landslide susceptibility map. The ensemble technique of these three data mining models has been used to generate a separated landslide susceptibility map by normalizing landslide susceptibility index maps as well as analyzing image data using the GIS platform. Five relative susceptibility classes (very high, high, medium, low, and very low) were created using natural break (Jenks) for the study region. Prediction accuracy for four landslide susceptibility maps was done with a receiver operating characteristic (ROC) curve area under the curve. The prediction accuracy was obtained for boosted tree (90%), support vector machine (90.6%), random forest (91.5%), and ensemble technique (92.5%). Interestingly, from the accuracy result, it is found that the ensemble technique has obtained the highest accuracy. LSM created from each of the four models can be useful for development, management, disaster prevention, and planning.
... Now the question arises of how the idea of paraglacial exhaustion can be reconciled with the power-law distribution of rockslide sizes, perhaps in combination with SOC. While size distributions of rockfalls and rockslides were addressed in several studies during the previous decade (e.g., Bennett et al., 2012;Lari et al., 2014;Valagussa et al., 2014;Strunden et al., 2015;Tanyas et al., 2018;Hartmeyer et al., 2020;Mohadjer et al., 2020), only the two latest studies refer directly to paraglacial exhaustion. Mohadjer et al. (2020) attempted to estimate the total volume of rockfalls in a deglaciated valley at different timescales up to 11 000 years. ...
Rockslides are a major hazard in mountainous regions. In formerly glaciated regions, the disposition mainly arises from oversteepened topography and decreases through time. However, little is known about this decrease and thus about the present-day hazard of huge, potentially catastrophic rockslides. This paper presents a new theoretical concept that combines the decrease in disposition with the power-law distribution of rockslide volumes found in several studies. The concept starts from a given initial set of potential events, which are randomly triggered through time at a probability that depends on event size. The developed theoretical framework is applied to paraglacial rockslides in the European Alps, where available data allow for constraining the parameters reasonably well. The results suggest that the probability of triggering increases roughly with the cube root of the volume. For small rockslides up to 1000 m3, an exponential decrease in the frequency with an e-folding time longer than 65 000 years is predicted. In turn, the predicted e-folding time is shorter than 2000 years for volumes of 10 km3, so the occurrence of such huge rockslides is unlikely at the present time. For the largest rockslide possible at the present time, a median volume of 0.5 to 1 km3 is predicted. With a volume of 0.27 km3, the artificially triggered rockslide that hit the Vaiont reservoir in 1963 is thus not extraordinarily large. Concerning its frequency of occurrence, however, it can be considered a 700- to 1200-year event.
... Furthermore, due to its unique subtropical monsoon climate with rich and concentrated rainfall, the region is considered an intense coseismic-landslide-prone zone (Cui et al., 2009). Therefore, deep scientific understandings of the spatial distribution of earthquake-induced landslides in this area, followed by near-real-time emergency assessment (Cao et al., 2019;Tanyas et al., 2019) and medium-and long-term risk assessment (Guzzetti et al., 2005;Lari et al., 2014), can effectively reduce the landslide risk after the earthquake and also serve a role in emergency rescue and town planning (Lan et al., 2022). ...
To enhance the timeliness and accuracy of spatial prediction of coseismic landslides, we propose an improved three-stage spatial prediction strategy and develop corresponding hazard assessment software named Mat.LShazard V1.0. Based on this software, we evaluate the applicability of this improved spatial prediction strategy in six earthquake events that have occurred near the Sichuan–Yunnan region, including the Wenchuan, Ludian, Lushan, Jiuzhaigou, Minxian, and Yushu earthquakes. The results indicate that in the first stage (immediately after the quake event), except for the 2013 Minxian earthquake, the area under the curve (AUC) values of the modeling performance are above 0.8. Among them, the AUC value of the Wenchuan earthquake is the highest, reaching 0.947. The prediction results in the first stage can meet the requirements of emergency rescue by immediately obtaining the overall predicted information of the possible coseismic landslide locations in the quake-affected area. In the second and third stages, with the improvement of landslide data quality, the prediction ability of the model based on the entire landslide database is gradually improved. Based on the entire landslide database, the AUC value of the six events exceeds 0.9, indicating a very high prediction accuracy. For the second and third stages, the predicted landslide area (Ap) is relatively consistent with the observed landslide area (Ao). However, based on the incomplete landslide data in the meizoseismal area, Ap is much smaller than Ao. When the prediction model based on complete landslide data is built, Ap is nearly identical to Ao. This study provides a new application tool for coseismic landslide disaster prevention and mitigation in different stages of emergency rescue, temporary resettlement, and late reconstruction after a major earthquake.
