Search

A common and largely unresolved problem of national-scale landslide early warning systems is their independent evaluation. In a recent paper, Mondini et al. (Nat Commun 14:2466, 2023) proposed a deep-learning system for short-term forecasting of rain-induced shallow landslides in Italy. Here, we independently evaluate the performance of this national-scale system by demonstrating its application between 1 January and 31 May 2021. For the purpose, we use hourly rainfall measurements from the same rain gauge network and different and independent information on the timing and location of 163 rain-induced landslides obtained from the FraneItalia catalogue that occurred in Italy in a period non considered in the construction of the system (https://zenodo.org/records/7923683). Independent demonstration confirmed the good predictive performance of the forecasting system and revealed no geographical or temporal bias in the forecasts. The analysis also showed that the system was more effective at predicting multiple landslides in the same general area than single landslides. This was a good result as multiple landslides are inherently more dangerous than single failures. Analysis of the few misclassified landslides showed that approximately one-third of the landslides were rockfalls, and for approximately another third there was uncertainty about when or where the landslides occurred. We conclude that, despite the inevitable misclassifications inherent in any probabilistically based national-scale landslide forecasting system, the deep-learning system analysed is well suited for short-term operational forecasting of rain-induced shallow landslides in Italy. [ABSTRACT FROM AUTHOR]

Rainfall triggered landslides occur in all mountain ranges posing threats to people and the environment. Given the projected climate changes, the risk posed by landslides is expected to increase, and the ability to anticipate their occurrence is key for effective risk reduction. Empirical thresholds and physically-based models are used to anticipate the short-term occurrence of rainfall-induced shallow landslides. But, evidence suggests that they may not be effective for operational forecasting over large areas. We propose a deep-learning based strategy to link rainfall to landslide occurrence. We inform and test the system with rainfall and landslide data available for the last 20 years in Italy. Our results indicate that it is possible to anticipate effectively the occurrence of rainfall-induced landslides over large areas, and that their location and timing are controlled primarily by the precipitation, opening to the possibility of operational landslide forecasting based on rainfall measurements and quantitative meteorological forecasts. How much rain does it take to trigger a landslide? This work shows that deep learning can identify the driving forces that can cause rainfall induced landslides, opening up the possibility of forecasting landslide events over large areas [ABSTRACT FROM AUTHOR]

We present a procedure to detect landslide events by analyzing in-sequence data acquired from regional broadband seismic networks and spaceborne radar imagery. The combined used of these techniques is meant to exploit their complementary elements and mitigate their limitations when used singularly. To test the method, we consider a series of six slope failures associated to the Piz Cengalo rock avalanche that recently occurred in the Swiss Alps, a region where we can benefit from high spatial density and quality of seismic data, as well as from the high spatial and temporal resolution of the European Space Agency (ESA) Copernicus Sentinel-1 radar satellites. The operational implementation of the proposed approach, in combination with the future increase in availability of seismic and satellite data, can offer a new and efficient solution to build and/or expand landslide catalogues based on quantitative measurements and, thus, help in hazard assessments and the definition of early warning systems at regional scales. [ABSTRACT FROM AUTHOR]

We propose a framework to systematically generate event landslide inventory maps from satellite images in southern Taiwan, where landslides are frequent and abundant. The spectral information is used to assess the pixel land cover class membership probability through a Maximum Likelihood classifier trained with randomly generated synthetic land cover spectral fingerprints, which are obtained from an independent training images dataset. Pixels are classified as landslides when the calculated landslide class membership probability, weighted by a susceptibility model, is higher than membership probabilities of other classes. We generated synthetic fingerprints from two FORMOSAT-2 images acquired in 2009 and tested the procedure on two other images, one in 2005 and the other in 2009. We also obtained two landslide maps through manual interpretation. The agreement between the two sets of inventories is given by the Cohen’s k coefficients of 0.62 and 0.64, respectively. This procedure can now classify a new FORMOSAT-2 image automatically facilitating the production of landslide inventory maps. [ABSTRACT FROM AUTHOR]

Landslides cause damages and affect victims worldwide, but landslide information is lacking. Even large events may not leave records when they happen in remote areas or simply do not impact with vulnerable elements. This paper proposes a procedure to measure spatial autocorrelation changes induced by event landslides in a multi-temporal series of synthetic aperture radar (SAR) intensity Sentinel-1 images. The procedure first measures pixel-based changes between consecutive couples of SAR intensity images using the Log-Ratio index, then it follows the temporal evolution of the spatial autocorrelation inside the Log-Ratio layers using the Moran's I index and the semivariance. When an event occurs, the Moran's I index and the semivariance increase compared to the values measured before and after the event. The spatial autocorrelation growth is due to the local homogenization of the soil response caused by the event landslide. The emerging clusters of autocorrelated pixels generated by the event are localized by a process of optimal segmentation of the log-ratio layers. The procedure was used to intercept an event that occurred in August 2015 in Myanmar, Tozang area, when strong rainfall precipitations triggered a number of landslides. A prognostic use of the method promises to increase the availability of information about the number of events at the regional scale, and to facilitate the production of inventory maps, yielding useful results to study the phenomenon for model tuning, landslide forecast model validation, and the relationship between triggering factors and number of occurred events. [ABSTRACT FROM AUTHOR]

Abstract: Landslide mapping is important for emergency management and landslide risk assessment. In this study, we propose a new mapping method, which combines the spectral signal contained in a satellite image and the geoenvironmental information included in a landslide susceptibility map. The image analysis captures areas with spectral signatures of event landslides in the image, while the landslide susceptibility map filters sub-areas, which do not have landslide prone conditions. The method assigns to every pixel of the satellite image a combined probability of landslide presence. We mapped typhoon-triggered landslides in southern Taiwan using the method. To compare with a landslide inventory prepared from orthophotos, we converted the probability map to a binary map of landslides and landslide free areas. Map comparison resulted in an overall accuracy of 0.93, an area percentage of overlapped landslides of 0.84, a modified success rate of 0.89, and a kappa statistic of 0.73. The method is fast, flexible, and relatively easy to use, and the probability map the method produces is useful by itself. We expect that the method can facilitate the rapid production of event landslide inventory maps, which in turn can assist emergency management. [Copyright &y& Elsevier]

Abstract: We propose a semi-automatic approach to detect, map and classify rainfall-induced shallow landslides. The approach combines the classification of a post-event multispectral satellite image with information on the morphometric signature of landslides in a Bayesian framework. We apply the approach in two steps. First, we detect and map the rainfall-induced landslides separating the stable ground from the failed areas. Next, we classify internally the landslides separating the source from the run out areas. We obtain the prior probability from the Mahalanobis discriminant function used to classify the satellite image, and the likelihood from the frequency distribution of terrain slope and cross section convexity in the pre-existing shallow landslides. We tested the approach in southern Taiwan, in a catchment where Typhoon Morakot caused abundant landslides in August 2009. Using the semi-automatic approach, we obtained a detailed event landslide inventory map that we compared to an inventory obtained through the visual interpretation of post-event ortho-photographs taken a few days after the landslide triggering rainfall event. Quantitative comparison in a Geographical Information System revealed a degree of matching between the two event inventories exceeding 90%. The approach is general and flexible, and can be used with different satellite imagery and topographic data. Best suited in landscapes where shallow landslides leave distinct radiometric and topographic signatures, the approach is expected to facilitate the production of event landslide inventory maps with positive consequences for geomorphological investigations, landslide hazard and risk modeling, and for post event recovery efforts. [Copyright &y& Elsevier]

Abstract: An important part of landslide research is the interpretation and delineation of landslides, which has increasingly been based on high-resolution satellite images in recent years. Using pre- and post-event FORMOSAT-2 satellite images as the data sources, this study presents a new method that combines four change detection techniques for mapping shallow landslides triggered by typhoons in Taiwan. The four techniques are normalized differential vegetation index (NDVI), spectral angle, principal component analysis, and independent component analysis. We apply the multiple change detection (MCD) technique to map landslides triggered by two typhoons of vastly different magnitudes. Comparisons are then made between MCD results with landslide inventory maps compiled by using a single index (change in NDVI) in one case study and visual analysis in another. Comparison results show that MCD can perform better than change in NDVI in dealing with old landslides and landslides with non-homogeneous spectral responses. MCD is also able to detect small landslides, which are often missed by visual analysis. Additionally, landslide maps prepared by MCD include runout features of sediment deposits from debris flows. A relatively fast processing chain, MCD is expected to become a useful new tool for emergency management after a typhoon event, which occurs on average four to five times a year in Taiwan. [Copyright &y& Elsevier]

Based on a minimum amount of rainfall that when reached or exceeded can trigger landslides, rainfall thresholds are used to predict potential landslide occurrence and are essential parts of many landslide early warning systems. Despite the extensive literature on the definition and use of rainfall thresholds, little attention has been given to examining and comparing the mathematical methods that can be used to define thresholds as lower bounds of clouds of empirical rainfall conditions known to have triggered landslides. When multiple thresholds are available, it is unclear how to combine them. Here, we address both issues. We test and compare four mathematical methods to define event cumulated rainfall—rainfall duration, ED thresholds using 2259 measurements of rainfall duration (D, in hours) and cumulated rainfall (E, in mm) that resulted in mostly shallow landslides in Italy between January 2002 and December 2012. The methods cover a broad spectrum of data driven approaches, including a frequentist least square method, a frequentist quantile regression method, a Bayesian quantile regression method, and a machine-learning symbolic regression method. We apply and compare the methods for three non-exceedance probability levels, p = 0.01, 0.05, 0.10, and we propose a voting strategy to combine the predictions into a single, dichotomous—i.e. ‘sharp’—non-probabilistic landslide prediction that we apply to the available dataset of rainfall measurements. [ABSTRACT FROM AUTHOR]

Despite landslides impact the society worldwide every day, landslide information is inhomogeneous and lacking. When landslides occur in remote areas or where the availability of optical images is rare due to cloud persistence, they might remain unknown, or unnoticed for long time, preventing studies and hampering civil protection operations. The unprecedented availability of SAR C-band images provided by the Sentinel-1 constellation offers the opportunity to propose new solutions to detect landslides events. In this work, we perform a systematic assessment of Sentinel-1 SAR C-band images acquired before and after known events. We present the results of a pilot study on 32 worldwide cases of rapid landslides entailing different types, sizes, slope expositions, as well as pre-existing land cover, triggering factors and climatic regimes. Results show that in about eighty-four percent of the cases, changes caused by landslides on SAR amplitudes are unambiguous, whereas only in about thirteen percent of the cases there is no evidence. On the other hand, the signal does not allow for a systematic use to produce inventories because only in 8 cases, a delineation of the landslide borders (i.e., mapping) can be manually attempted. In a few cases, cascade multi-hazard (e.g., floods caused by landslides) and evidences of extreme triggering factors (e.g., strong earthquakes or very rapid snow melting) were detected. The method promises to increase the availability of information on landslides at different spatial and temporal scales with benefits for event magnitude assessment during weather-related emergencies, model tuning, and landslide forecast model validation, in particular when accurate mapping is not required. [ABSTRACT FROM AUTHOR]

Abstract: A coupled model has been developed to simulate, at watershed level, landslides and debris flows induced by a severe typhoon (tropical cyclone) in Taiwan. The model comprises a landslide susceptibility model to predict landslide occurrence, an empirical model to select debris-flow initiation points, and a debris flow model to simulate the transport and deposit of failed materials from the identified source areas. In raster format with a 10m spatial resolution, the model output includes unstable cells, debris-flow initiation cells, debris-flow velocities, runout paths, and deposition zones. The model was first tested and calibrated in a small area, where the damage by landslides had been investigated and recorded. It was then applied to a watershed, and the simulation results were validated by comparing them with a landslide/debris-flow inventory map prepared from satellite images using a multiple change detection technique. Model test and validation results confirm the usefulness of the model in predicting the number and size of affected areas (landslides and runouts combined), runout path, and volume of runout deposits. It is a common practice in Taiwan to separate landslide and debris-flow inventories and to study debris flows only in select drainage basins. This study suggests that landslide and debris flow should be modeled as a sequential process for efficient watershed management. [Copyright &y& Elsevier]

Abstract: Many shallow landslides in Taiwan are triggered by typhoons (tropical cyclones) during the summer months. Each typhoon produces a different rainfall distribution, depending on its track and position and the atmospheric conditions. This study investigated whether the additional rainfall data in a landslide susceptibility model can improve its performance in predicting typhoon-triggered landslides, and whether information on past typhoon events, combined with an event-based landslide inventory, can help predict landslides triggered by a typhoon. To answer these questions, the study adopted a method that integrates rainfall data with the critical rainfall model (a landslide susceptibility model based on geoenvironmental factors) to derive a logistic regression model for predicting landslide occurrence. Results of a back analysis of landslides triggered by nine typhoons from 2001 to 2009 reveal that, by including rainfall data, the integrated method performs better than the critical rainfall model in the average overall accuracy rate (0.78 vs. 0.45) and the average modified success rate (0.75 vs. 0.68). Our preliminary results also suggest that it is possible to predict landslides triggered by a typhoon by using a catch-all model developed from all other typhoon events in an inventory, or a group model developed from other typhoon events of similar rainfall characteristics in an inventory. This study opens up a new research direction in analyzing rainfall-triggered landslides in Taiwan and elsewhere. [Copyright &y& Elsevier]

Accurate landslide inventory mapping is essential for quantitative hazard and risk assessment. Although multi-temporal change detection techniques have contributed greatly to landslide inventory preparation, it is still challenging to generate quality change detection images (CDIs) for accurate landslide mapping. The recently proposed change detection-based Markov random field (CDMRF) provides an effective approach for rapid mapping of landslides with minimum user interventions. However, when CDI is generated by change vector analysis (CVA) alone, the CDMRF method may suffer from noise especially when the pre- and post-event remote sensing images are acquired under different atmospheric, illumination, and phenological conditions. This paper improved such CDMRF approach by integrating normalized difference vegetation index (NDVI), principal component analysis (PCA), and independent component analysis (ICA) generated CDIs with MRF for landslide inventory mapping from multi-sensor data. To justify the effectiveness and applicability, the improved methods were applied to map rainfall-, typhoon-, and earthquake-triggered landslides from the pre- and post-event satellite images acquired by very high resolution QuickBird, high resolution FORMOSAT-2, and moderate resolution Sentinel-2. Moreover, they were tested on pre-event Landsat-8 and post-event Sentinel-2 datasets, indicating that they are operational for landslide inventory mapping from combined multi-temporal and multi-sensor data. The results demonstrate that the improved δ NDVI-, PCA-, and ICA-based approaches perform much better than CVA-based CDMRF in terms of completeness, correctness, Kappa coefficient, and F -measures. To the best of our knowledge, it is the first time that NDVI, PCA, and ICA are integrated with MRF for landslide inventory mapping from multi-sensor data. It is anticipated that this research can be a starting point for developing new change detection techniques that can readily generate quality CDI and for applying advanced machine learning algorithms (e.g., deep learning) to automatic detection of natural hazards from multi-sensor time series data. • NDVI, PCA, and ICA are integrated into MRF for landslide mapping. • Improved CDMRF for landslide mapping from multi-sensor data • NDVI-, PCA-, and ICA-based MRF outperform CVA-based MRF significantly • Applicable to map rainfall-, typhoon-, and earthquake-triggered landslides [ABSTRACT FROM AUTHOR]