Landslide Scars Research Articles

Artificial neural networks applied to landslide susceptibility: The effect of sampling areas on model capacity for generalization and extrapolation

Artificial neural networks (ANNs) have been used to identify areas susceptible to landslides and constitute one of the most widely used methods for this purpose. Several factors can interfere in the performance of the models and their resulting maps (especially sampling). This research evaluated the influence of sampling areas on landslide susceptibility modelling and the capacity for generalization and spatial extrapolation of data. Based on an inventory of landslide scars, distributed in five areas of southern Brazil, non-occurrence samples were defined by means of different buffers (2–40 km) in relation to the landslides in order to test the effect of the spatial distribution of the non-occurrence samples on the modeling results. A total of 16 morphometric attributes of the terrain (extracted from a digital elevation model) were used as input variables of the model. Multilayered network training was carried out using a backpropagation algorithm and accuracy was calculated by means of the Area Under the Receiver Operating Characteristic Curve (AUROC). Model accuracy was between 0.739 and 0.931. This variation was explained mainly by the buffer used. The susceptibility map resulting from the model of greater accuracy was obtained with a 40-km buffer in order to collect non-occurrence samples. The great distance between the occurrence and non-occurrence samples facilitates the modelling, since it increases the morphometric differences between the sampling groups. When we used samples from only one of the sample areas, the spatial extrapolation of the susceptibility map to the other areas showed high performance. We conclude that the ANN model for landslides susceptibility mapping can be extrapolated spatially, considering the limits of the geomorphological unit or numerical domain of the data.

Rapid vegetation recovery at landslide scars detected by multitemporal high-resolution satellite imagery at Aso volcano, Japan

Vegetation cover is an important factor in hillslope stability and landslide susceptibility. The objective of this study is to explore vegetation recovery in rainfall-induced and coseismic landslide scars at the Aso volcano in Japan. We focus on the temporal changes and effect of topography on vegetation recovery using multitemporal satellite images of RapidEye from 2010 to 2019 and PlanetScope from 2016 to 2020 and finer spatial resolution images of unmanned aerial vehicles (UAVs). The study area is characterized by semi-natural grassland and frequent landslide events triggered by heavy rainfall in 2012 and the 2016 Kumamoto earthquake. This study validates grass vegetation recovery using the normalized difference vegetation index (NDVI) and NDVI ratio, which represents the relative difference in NDVI between the landslide scars and grass areas. The results show a distinct decrease in the NDVI ratio at landslide scars after the heavy rainfall and Kumamoto earthquake. The NDVI ratio rapidly increases in landslide scars after the events. The results indicate that NDVI at both landslide scars will reach the same level as the pre-landslide grass vegetation within 12 years. The slope aspect direction determines the topographic effects of grass vegetation recovery. The north- to west-facing landslide scars seemed to experience rapid grass vegetation recovery.

Quantifying effectiveness of trees for landslide erosion control

We developed a landslide susceptibility model using binary logistic regression for silvopastoral landscapes, which for the first time includes spatial distribution models for individual trees of different vegetation types. Models were trained and tested using a landslide inventory consisting of 43,000 landslide scars mapped across an 843 km2 area. Model performance was very good, with a median AUROC of 0.95 in the final model used for predictions, which equates to an accuracy of 88.7% using a cut-off of 0.5. We investigate the effect of highly skewed continuous tree variables on the maximum likelihood estimator by testing different sampling strategies aimed at reducing positive skewness. With an adequate sample size, we found that highly skewed continuous predictor variables do not result in an inflation of effect size.Using two farms in the study area, we illustrate application of the landslide susceptibility model for quantifying the reduction in shallow landslide erosion due to trees. Landslide erosion was reduced by 16.6% at Site 1 and 42.9% at Site 2 due to all existing vegetation. The effectiveness of individual trees on reducing landslide erosion was shown to be less a function of species than that of targeting highly susceptible areas with adequate plant densities. We found 80% of landslides are triggered in 12.1% and 7.3% of the area of Sites 1 (1700-ha) and 2 (462-ha), respectively, suggesting there is great potential for smarter targeting of erosion mitigation. The high-resolution spatial information provided by the landslide susceptibility maps can be used by decision makers in land management to support the development and targeting of erosion mitigation measures.

‘Isolated base‐of‐slope aprons’: An oxymoron for shallow‐marine fan‐shaped, temperate‐water, carbonate bodies along the south‐east Salento escarpment (Pleistocene, Apulia, southern Italy)

AbstractThis paper regards the lower Pleistocene temperate‐water carbonate deposits disconformably overlying an escarpment made up of faulted Cretaceous to Miocene limestones of the Apulia Foreland (southern Italy). Study deposits discontinuously crop out along the present‐day eastern Salento sea cliff, and form isolated fan‐shaped bodies, up to 1 km wide and up to 40 to 50 m thick, each of them covering an area of a few square kilometres. The internal arrangement of beds is represented by up to 25° to 30° lobate, seaward dipping clinobeds thinning and onlapping onto a rocky foreslope in the proximal sector and passing to gently inclined to sub‐horizontal strata in the distal sector. Seven facies were distinguished, mainly composed of coarse‐grained skeletal carbonates made up of a heterozoan association including coralline algae, large and small benthic foraminifera, echinoids, molluscs, bryozoans and serpulids. Since clinobeds were formed thanks to hyperconcentrated density flows (grain flows) bypassing the upper part of the inherited escarpment, these skeletal grains represent ex situ deposits whose shallow‐marine factory was located upward (landward) with respect to the bypassed zone, likely in the almost flat area on top of the Salento Peninsula. Clinobeds are often affected by tens of metres wide and long channel‐like structures interpreted as landslide scars. Inside these gullies, contorted beds (slumps) or matrix‐supported intra‐bioclastic floatstone/rudstone (massive deposits) are present. The occurrence of supercritical‐flow structures (for example, backset‐bedded beds) indicates the development of hydraulic jumps along the steep slope of gullies. Since these clinostratified, fan‐shaped carbonate bodies represent carbonate slopes, and that the latter are known as aprons, normally related to linear sourced sediments, an acceptable oxymoron for studied fan‐shaped carbonate bodies is suggested: ‘isolated base‐of‐slope aprons’.

Geo-structures and deformation-failure characteristics of rockslide areas near the Baige landslide scar in the Jinsha River tectonic suture zone

Geo-structures and deformation-failure characteristics of rockslide areas near the Baige landslide scar in the Jinsha River tectonic suture zone

Beyond 2D landslide inventories and their rollover: synoptic 3D inventories and volume from repeat lidar data

Abstract. Efficient and robust landslide mapping and volume estimation is essential to rapidly infer landslide spatial distribution, to quantify the role of triggering events on landscape changes, and to assess direct and secondary landslide-related geomorphic hazards. Many efforts have been made to develop landslide mapping methods, based on 2D satellite or aerial images, and to constrain the empirical volume–area (V–A) relationship which, in turn, would allow for the provision of indirect estimates of landslide volume. Despite these efforts, major issues remain, including the uncertainty in the V–A scaling, landslide amalgamation and the underdetection of landslides. To address these issues, we propose a new semiautomatic 3D point cloud differencing method to detect geomorphic changes, filter out false landslide detections due to lidar elevation errors, obtain robust landslide inventories with an uncertainty metric, and directly measure the volume and geometric properties of landslides. This method is based on the multiscale model-to-model cloud comparison (M3C2) algorithm and was applied to a multitemporal airborne lidar dataset of the Kaikōura region, New Zealand, following the Mw 7.8 earthquake of 14 November 2016. In a 5 km2 area, the 3D point cloud differencing method detects 1118 potential sources. Manual labeling of 739 potential sources shows the prevalence of false detections in forest-free areas (24.4 %), due to spatially correlated elevation errors, and in forested areas (80 %), related to ground classification errors in the pre-earthquake (pre-EQ) dataset. Combining the distance to the closest deposit and signal-to-noise ratio metrics, the filtering step of our workflow reduces the prevalence of false source detections to below 1 % in terms of total area and volume of the labeled inventory. The final predicted inventory contains 433 landslide sources and 399 deposits with a lower limit of detection size of 20 m2 and a total volume of 724 297 ± 141 087 m3 for sources and 954 029 ± 159 188 m3 for deposits. Geometric properties of the 3D source inventory, including the V–A relationship, are consistent with previous results, except for the lack of the classically observed rollover of the distribution of source area. A manually mapped 2D inventory from aerial image comparison has a better lower limit of detection (6 m2) but only identifies 258 landslide scars, exhibits a rollover in the distribution of source area of around 20 m2, and underestimates the total area and volume of 3D-detected sources by 72 % and 58 %, respectively. Detection and delimitation errors in the 2D inventory occur in areas with limited texture change (bare-rock surfaces, forests) and at the transition between sources and deposits that the 3D method accurately captures. Large rotational/translational landslides and retrogressive scars can be detected using the 3D method irrespective of area's vegetation cover, but they are missed in the 2D inventory owing to the dominant vertical topographic change. The 3D inventory misses shallow (< 0.4 m depth) landslides detected using the 2D method, corresponding to 10 % of the total area and 2 % of the total volume of the 3D inventory. Our data show a systematic size-dependent underdetection in the 2D inventory below 200 m2 that may explain all or part of the rollover observed in the 2D landslide source area distribution. While the 3D segmentation of complex clustered landslide sources remains challenging, we demonstrate that 3D point cloud differencing offers a greater detection sensitivity to small changes than a classical difference of digital elevation models (DEMs). Our results underline the vast potential of 3D-derived inventories to exhaustively and objectively quantify the impact of extreme events on topographic change in regions prone to landsliding, to detect a variety of hillslope mass movements that cannot be captured by 2D landslide mapping, and to explore the scaling properties of landslides in new ways.

Megatsunamis Induced by Volcanic Landslides in the Canary Islands: Age of the Tsunami Deposits and Source Landslides

Evidence for frequent, large landslides on the flanks of the volcanic edifices forming the Canary Islands include outstanding landslide scars and their correlative submarine and subaerial rock and debris avalanche deposits. These landslides involved volumes ranging from tens to hundreds of km3. The sudden entry of large volumes of rock masses in the sea may have triggered tsunamis capable of affecting the source and neighboring islands, with the resulting huge waves dragging coastal and seabed materials and fauna and redepositing them inland. Here, we present new geological evidence and geochronological data of at least five megatsunamis in Tenerife, Lanzarote, and Gran Canaria, triggered by island flank megalandslides, and occasionally explosive eruptions, during the last 1 million years. The exceptional preservation of the megatsunami deposits and the large area they cover, particularly in Tenerife, provide fundamental data on the number of tsunami events and run-ups, and allow proposals on the sources and age of the tsunamis. Tsunami run-up heights up to 290 m above coeval sea level, some of the highest known on Earth in recent geological times, were estimated based on sedimentological, geomorphological, paleontological, and geochronological data. The research results made it possible to estimate the recurrence of tsunamis in the archipelago during the last hundreds of thousands of years, and to establish relationships between tsunami deposits and the probable triggering island flank landslides.

Intra-Annual Identification of Local Deforestation Hotspots in the Philippines Using Earth Observation Products

Like many other tropical countries, the Philippines has suffered from decades of deforestation and forest degradation during and even after the logging era. Several open access Earth Observation (EO) products are increasingly being used for deforestation analysis in support of national and international initiatives and policymaking on forest conservation and management. Using a combination of annual forest loss and near-real time forest disturbance products, we provide a comprehensive analysis of the deforestation events in three forest frontiers of the Philippines. A space-time pattern mining approach was used to map quarterly deforestation hotspots at 1 km pixel size (100 hectares), where hotspots are classified according to the spatial and temporal variability of the 2000–2020 deforestation in the study area. Our results revealed that 79–81% of the hotspots overlap with primary forests and 27–29% are inside the state-declared protected areas. The intra-annual analysis of deforestation in 2020 revealed an alarming trend, where most deforestation occurred between the 1st and 2nd quarter (92–94% in hotspot forests; 87–89% in non-hotspot forests), highly overlapping within the slash-and-burn farming season. We also found “new” hotspots (2020) formed mostly from landslide scars and partly from selective logging, the latter is believed to be underestimated. Our study paves the way for rapid and regular assessment of the country’s deforestation, useful for the respective environmental institutions who convene several times a year. Moreover, our findings assert the imperative of alternative livelihoods to upland farmers, efficient forest protection activities, and even the mitigation of landslide risks.

Modeling of the Dec. 22nd 2018 Anak Krakatau volcano lateral collapse and tsunami based on recent field surveys: Comparison with observed tsunami impact

The Dec. 22, 2018 lateral collapse of the Anak Krakatau (AK) volcano in the Sunda Straits of Indonesia discharged volcaniclastic material into the 250 m deep caldera southwest of the volcano and generated a large tsunami, causing runups of up to 85 m in the near-field, and 13.5 m in the far-field, on the nearby coasts of Sumatra and Java. The tsunami caused 437 fatalities, the greatest number from a volcanically-induced tsunami since the catastrophic explosive caldera-forming eruption of Krakatau in 1883 and the sector collapse of Ritter Island in 1888. For the first time in over 100 years, the 2018 AK event provides an opportunity to study a major volcanically-generated tsunami that caused widespread loss of life and significant damage. Here, we present numerical simulations of the collapse and tsunami generation, propagation, and coastal impact, with state-of the-art numerical models, using both a new parametrization of the collapse and a near-field bathymetric dataset based on our 2019 field surveys and satellite images. These subaerial and submarine data sets are used to constrain the geometry and magnitude of the landslide mechanism, which show that the primary landslide scar bisected the AK edifice, cutting behind the central vent and removing 50% of its subaerial volume. The primary landslide volume is estimated to range from 0.175–0.313 km3, based on uncertainties in the shape of the submerged part of the failure plane. This is supported by an independent estimate of the primary landslide deposit volume of 0.214 ± 0.036 km3. Given uncertainties in the failure volume, we define a range of potential failure surfaces that span these values in 4 collapse scenarios of volume ranging from 0.175 to 0.313 km3. These AK collapses are modeled, assuming either a granular or viscous fluid rheology, together with their corresponding tsunami generation and propagation. Observations of a single tsunami, with no subsequent waves, are consistent with our interpretation of landslide failure in a rapid, single phase of movement rather than a more piecemeal process, generating a tsunami which reached nearby coastlines within ~30 min. For both modeled rheologies, the 0.224 km3 collapse (second and preferred scenario) most successfully reproduces the near- and far-field tsunami flow depth and runup observed in all post-event field survey results, tide gauge records, and eyewitness reports to date, suggesting our estimated landslide volume range is appropriate. This event highlights the significant hazard posed by relatively small-scale lateral volcanic collapses, which can occur en-masse, without any precursory signals, and are an efficient and unpredictable tsunami source. Our successful simulations demonstrate that current numerical models can accurately forecast tsunami hazards from these events. In cases such as Anak Krakatau's, the absence of precursory warning signals, together with the short travel time following tsunami initiation present a major challenge for mitigating tsunami coastal impact, stressing the need to develop and install early warning systems for such events.

Assessment and Analysis of a Rainfall–Time-Lagging Water-Related Disaster in Mountainous Areas

Both global climate change and human activities are continuously impacting the abruptness and frequency of water-related natural disasters such as flash floods, debris flows, and landslides in mountainous areas, greatly threatening the safety of lives and properties. A recent rainfall-induced debris flow event happened on July 6, 2020 in the Chenghuangmiao Gully, in Sichuan Province, China, resulting in severe damage to buildings at the outlet. An integrated analysis of the consequence and triggering mechanism of this debris flow event was conducted with hydrologic information, topographic details, vegetation regimes, and drone aerial imagery. The result shows that the entire runout of the debris flow differs from that of common ones (debris flow and rainfall were highly related and synchronized), which happened 4 h after the stop of the rainfall. The hysteretic feature increases the difficulty of the prediction and warning of the debris flow due to lack of a responsible triggering mechanism. The hillslope surface is well covered by vegetation, hindering regular observation and cleaning up of long-term deposited wood and sediment debris. This effect increases the crypticity and abruptness of potential debris flows. With field evidence and analysis, it is speculated that long-term accumulative processes of dead wood sand sediment deposition formed a small-scale debris dam, and the continuous water release from the watershed led to dam breaching, subsequently triggering the initiation of the debris flow. Multiple steps distributing along the gully of an average slope of 15.65° contributed to the amplification of the debris flow once the breach of the upstream wood and sediment dam occurred. Along the gully, small-scale landslide scars can be observed, possibly amplifying the scale of the debris flow and disaster impact. This debris event gives a lesson of necessary demands of predicting and managing the risks of a low-frequency debris flow non-synchronized with rainfall events.

A massive ancient river-damming landslide triggered by buckling failure in the upper Jinsha River, SE Tibetan Plateau

Large-scale landsliding is an extremely important geological process in shaping landscapes in the Tibetan Plateau. In this research, an ancient river-damming landslide with an estimated debris volume of 4.9 × 107 m3, located in the upper Jinsha River, SE Tibetan Plateau, was studied. The landslide once formed a dam over 60 m high and blocked the river. Lacustrine sediments, composed of silty clay with particle sizes of 0.002–0.25 mm, are intermittently distributed along both banks, extending about 6.5 km upstream. The OSL dating indicates that the lacustrine sediments have an age of 2.6 ± 0.2 ka. Detailed field investigation and theoretical analysis was performed to investigate the characteristics, potential cause and mechanism of the landslide. The results suggest that the landslide was most likely triggered by buckling of planar rock slabs under gravity. It may start as a translational sliding along the weak interlayer composed of mica schist at the upper part of the slope and then formed buckles by curving amphibolite rock beds near the slope toe. The hillslope has still been affected by gravitational deformations, with geomorphology characterized by tension cracks, buckle folds, and small landslide scars distributed on the slope surface, suggesting that the evolution of the river valley caused by buckling deformation has not achieved equilibrium.

New chronological constraints on intense Holocene eruptions and landslide activity at Tacaná volcanic complex (Mexico)

Usual methods are unfortunately unsuitable to accurately date many Holocene deposits and landforms from active Mexican volcanoes. This is notably the case for the Tacaná Volcanic Complex (TVC), located in the State of Chiapas in southern Mexico, and the San Marcos Department in Guatemala. The complex consists of the Chichuj, Tacaná, and San Antonio volcanoes and the Ardillas dome. Tacaná volcano, the main summit of the TVC, collapsed 15 ± 5 ka ago ( 40 Ar/ 39 Ar), producing the Agua Caliente Debris Avalanche. The result was the formation of a horseshoe-shaped crater 600 m wide open to the northwest. Afterward, several undated effusive and explosive eruptions occurred. All these eruptions are likely Late Pleistocene to Holocene in age because the landforms derived from that activity overlap the horseshoe-shaped crater. To corroborate this hypothesis, we used surface exposure dating with in situ- produced cosmogenic 36 Cl combined with lichenometry and dendrochronology in the summit domes, the Ardillas dome, the horseshoe-shaped crater's cliff, and an andesitic lava. The 36 Cl exposure ages (9.3 ± 1.9 ka; 8.9 ± 0.9 ka; 8.6 ± 1.7 ka) of the summit domes are statistically indistinguishable within the associated uncertainties and suggest that the domes formed during Early Holocene. Instead, the southern part of the Late Pleistocene collapsed crater yielded significantly younger ages ranging from 5.3 ± 0.6 to 7.0 ± 0.8 ka. These ages are assumed to be minimum and might indicate later rockfall activity and instability, affecting the horseshoe-shaped crater. The 36 Cl exposure age of an andesitic lava (0.4 ± 0.1 ka), emplaced to the SW of Tacaná, is in good agreement with lichenometry (>347 yr). These ages could represent a gravitational collapse event associated with phreatic explosions vented close to a scar collapse. The 36 Cl exposure ages derived from the Ardillas lava dome (0.3 ± 0.1 ka) do not correspond to its emplacement age. The dome must be older than pyroclastic deposits dated at 760 ± 30 yr BP ( 14 C) that cover its surface. Strikingly, the 36 Cl, lichenometrical and dendrochronological data of the “Andesitic” lava and the Ardillas dome are associated with NE-SW landslide scars, explosion craters, phreatic vents, fumarolic activity, and tensional fractures. All these features are perpendicular to the NW-SE minimum stress regime (σ3), affecting the TVC and generating gravitational activity. Nowadays, gravitational collapses represent a real threat to the surrounding populations that can be triggered not only by volcanic activity but also by seismicity and extraordinary rains.

Landslides susceptibility modelling using Multivariate Logistic Regression Model in the Sahla Watershed in Northern Morocco

This study aimed to assess landslide susceptibility in the Sahla watershed in northern Morocco. Landslides hazard is the most frequent phenomenon in this part of the state due to its mountainous precarious environment. The abundance of rainfall makes this area suffer mass movements led to a notable adverse impact on the nearby settlements and infrastructures. There were 93 identified landslide scars. Landslide inventories were collected from Google Earth image interpretations. They were prepared out of landslide events in the past, and future landslide occurrence was predicted by correlating landslide predisposing factors. In this paper, landslide inventories are divided into two groups, one for landslide training and the other for validation. The Landslide Susceptibility Map (LSM) is prepared by Logistic Regression (LR) Statistical Method. Lithology, stream density, land use, slope curvature, elevation, topographic wetness index, slope aspect, and slope angle were used as conditioning factors. The Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC) was employed to examine the performance of the model. In the analysis, the LR model results in 96% accuracy in the AUC. The LSM consists of the predicted landslide area. Hence it can be used to reduce the potential hazard linked with the landslides in the Sahla watershed area in Rif Mountains in northern Morocco.

CORRELATIONS BETWEEN LANDSLIDE SCARS PARAMETERS USING REMOTE SENSING METHODS IN THE ESTRADA DE FERRO VITÓRIA-MINAS, SOUTHEASTERN BRAZIL

The remote sensing has been widely used in studies involving records and monitoring landslides, also, this tool is widely used in mapping areas of geological risk. In this paper, two remote sensing methodologies were used to identify, characterize and catalog scars generated along the Estrada de Ferro Vitória-Minas (EFVM). The first methodology allowed to estimate the area and volume of landslides scars of EFVM through the images available in Google Earth Pro. The second methodology used a drone to capture the parameters in the field of landslides scars. From the obtained parameters it was possible to compare the estimates of areas and volumes of landslides by drone image with the data obtained by images of Google Earth Pro and to test the applicability of both methods. The results obtained showed estimates scars areas with correlation of 0.92 for the two methodologies. The volume estimates presented a correlation of 0.8; with underestimation of the results obtained by Google Earth Pro.

Comparison of length and dynamics of wood pieces in streams covered with coniferous and broadleaf forests mapped using orthophotos acquired by an unmanned aerial vehicle

As wood pieces supplied by landslides and debris flows are one of the main components of ecological and geomorphic systems, the importance of quantifying the dimensions of the wood pieces is evident. However, the low accessibility of disturbed channels after debris flows generally impedes accurate and quick wood-piece investigations. Thus, remote-sensing measurements for wood pieces are necessitated. Focusing on sub-watersheds in coniferous and broadleaf forests in Japan (the CF and BF sites, respectively), we measured the lengths of wood pieces supplied by landslides (> 0.2 m length and > 0.03 m diameter) from orthophotos acquired using a small unmanned aerial vehicle (UAV). The measurement accuracy was analyzed by comparing the lengths derived from the UAV method with direct measurements. The landslides at the CF and BF sites were triggered by extremely heavy rainfalls in 2017 and 2018, respectively. UAV flights were operated during February and September 2019 at the CF site and during November 2018 and December 2019 at the BF site. Direct measurements of wood pieces were carried out on the date of the respective second flight date in each site. When both ends of a wood piece are satisfactorily extracted from an orthophoto acquired by the UAV, the wood-piece lengths at the CF site can be measured with an accuracy of approximately ±0.5 m. At the BF site, most of the extracted lengths were shorter than the directly measured lengths, probably because the complex structures of the root wad and tree crown reduced the visibility. Most wood pieces were discharged from landslide scars at the BF site, but at the CF site, approximately 750 wood pieces remained in the landslide scars approximately 19 months after the landslide occurrence. The number of wood pieces in the landslide scars of the CF site increased with increasing landslide area, suggesting that some wood pieces can be left even if large landslides occur. The lengths and locations of the entrapped wood pieces at both sites were not significantly changed between the two UAV flight dates. However, during this period, the rainfall intensities around the CF site measured by the closest rain-gauge of the Japan Meteorological Agency reached their second highest values from 1976 to 2019, which exceeded the 30-year return period. This suggests that most of the entrapped wood pieces rarely migrated even under intense rainfall.

The damage caused by landslides in socio-economic spheres within the Kigezi highlands of South Western Uganda

Abstract An assessment of the socio-economic implications of landslide occurrence in the Kigezi highlands of South Western Uganda was conducted. Landslide occurrence is on the increase and threatens community livelihoods in these highlands. Detailed field investigations were undertaken with the help of local communities between June 2018 and May 2020 to identify and map recent and visible landslide scars in Rukiga uplands of Kigezi highlands. In the course of field inventories, 85 visible landslide scars were identified and mapped using handheld GPS receivers to produce a landslide distribution map for the study area. A socio-economic analysis was conducted to establish the effects of landslide damage on people’s livelihoods as well as their existing coping and adaptation mechanisms. The assessment was administered through field observations and surveying, focus group discussions, key informants and household interviews as well as the use of Local Government Environmental Reports. The study established an increase in the spatial-temporal distribution of landslides over the Kigezi highlands in the past 40 years. The landslides have resulted in a reduction in the quality of land, loss of lives, destruction of transport infrastructures, settlements, farmlands, crops and other socio-economic infrastructures. Therefore, it is important to look for reliable and sustainable measures to prevent landslide hazards. Total landscape reforestation with deep-rooted trees can possibly reduce the landslide risk. It is also important to undertake policy implementation for preparedness and mitigation plans against landslides in this region and in the country at large. Proper soil and water conservation measures could help in enhancing soil strength against landslide hazards.

Convolutional neural networks applied to semantic segmentation of landslide scars

Landslides are considered to be among the most alarming natural hazards. Therefore, there is a growing demand for databases and inventories of these events worldwide, since they are a vital resource for landslide risk assessment applications. Given the recent advances in the field of image processing, the objective of this study is to evaluate the performance of a deep convolutional neural network architecture called U-Net for the mapping of landslide scars from satellite imagery. The question that drives the study is: can fully convolutional neural networks be successfully applied as the backbone of automatic frameworks for building landslide inventories, keeping or improving the identification accuracy and agility when compared to other methods? To seek for an answer to it, scenes from the Landsat-8 satellite of a region of Nepal were obtained and processed in order to compose a landslide image database that served as the basis for the training, validation and test of deep convolutional neural networks. The U-Net architecture was applied and the results indicate that it has the potential to identify landslide scars, improving over previously published research on the topic for the same study region. The validation process resulted in recall, precision and F1-score values of 0.74, 0.61 and 0.67, respectively, thus higher than those from previous studies using different methodologies. The results indicate the potential of the method to be applied in dynamic mapping systems for landslide scar identification, which paves the way to the composition and updating of landslide scar databases. These, in turn, can support a great deal of quantitative landslide susceptibility mapping methods that heavily rely on data to provide accurate results.

Mamdani fuzzy inference systems and artificial neural networks for landslide susceptibility mapping

Two Artificial Intelligence (AI) methods, Fuzzy Inference System (FIS) and Artificial Neural Network (ANN), are applied to Landslide Susceptibility Mapping (LSM), to compare complementary aspects of the potentials of the two methods and to extract physical relationships from data. An index is proposed in order to rank and filter the FIS rules, selecting a certain number of readable rules for further interpretation of the physical relationships among variables. The area of study is Rolante river basin, southern Brazil. Eleven attributes are generated from a Digital Elevation Model (DEM), and landslide scars from an extreme rainfall event are used. Average accuracy and area under Receiver Operating Characteristic curve (AUC) resulted, respectively, in 81.27% and 0.8886 for FIS, and 89.45% and 0.9409 for ANN. ANN provides a map with more amplitude of outputs and less area classified as high susceptibility. Among the 40 (10%) best-ranked FIS rules, 13 have high susceptibility output, while 27 have low; a cause is that low susceptibility areas are larger on the map. Slope is highly connected to susceptibility. Elevation, when high (plateau) or low (floodplain), inhibits high susceptibility. Six attributes show the same fuzzy set for the 18 best-ranked rules, meaning this fuzzy set is common on the map. Overall findings point out that ANN is best suited for LSM map generation, but, based on them, using FIS is important to help researchers understand more about AI models for LSM and about landslide phenomenon.

Preliminary analysis of a catastrophic landslide event on 6 August 2020 at Pettimudi, Kerala State, India

The Western Ghats, the most prominent orographic feature of peninsular India, occupying about 47% of Kerala state, is vulnerable to various types of landslides. The occurrence of cataclysmic landslides along the windward slope of Western Ghats during torrential rainfall is attributed to the unique physiography and climate. A disastrous landslide occurred in the Pettimudi village of Idukki district, Kerala State at 2245 h (UTC+5.30) on 6 August 2020 due to heavy downpour. A preliminary analysis was carried out to understand the characteristics of the event and to investigate the slope failure mechanisms. The landslide was initiated on the steep slopes of Shola forest and was slid through the tea plantations. The dimensions of the landslide scar were measured as 1200m in length, 30-120m width, 3-7m depth and 125m height. An approximate area of 70125 m2 was estimated to be affected by the landslide. The heavy downpour which is thought to trigger the Pettimudi landslide was measured as 61.6 cm/day, the highest ever recorded in the area.The disaster caused 66 fatalities of residents, among which four are still missing, making it the most catastrophic landslide event reported in the state.

Woody Vegetation Succession and Prediction of the Occurrence Cycle of Shallow Landslides on Steep Granite Slopes

We investigated woody vegetation succession and the rate of topsoil development on shallow landslide scars on steep granite slopes in Kagoshima Prefecture, Japan. We used our results to predict the occurrence cycle of shallow landslides. For this study, nine shallow landslide scars of ≥40º were selected. The age of the scars ranged from 2 to 49 years. We installed 5 m × 5 m quadrats in the scars and conducted a field study. The results showed that woody vegetation in the quadrats had a smaller Fisher-Williams index of diversity and the rate of evergreen broad-leaved trees than a climax forest. Accordingly, the succession of woody vegetation in the quadrats was not at the level of a climax forest. The rate of topsoil development, based on the relationship between topsoil depth and the time since the occurrence of the shallow landslide, was 0.25 cm/year. Thus, the occurrence cycle of shallow landslides, based on the topsoil depth needed to generate a shallow landslide and the rate of topsoil development, ranged from 200 to 300 years (average 248 years). Our method was effective for predicting the occurrence of shallow landslides on steep granite slopes.