Multi-geohazards susceptibility mapping based on machine learning—a case study in Jiuzhaigou,China

Jiuzhaigou, located in the transitional area between the Qinghai–Tibet Plateau and the Sichuan Basin, is highly prone to geological hazards (e.g., rock fall, landslide, and debris flow). High-performance-based hazard prediction models, therefore, are urgently required to prevent related hazards and manage potential emergencies. Current researches mainly focus on susceptibility of single hazard but ignore that different types of geological hazards might occur simultaneously under a complex environment. Here, we firstly built a multi-geohazard inventory from 2000 to 2015 based on a geographical information system and used satellite data in Google earth and then chose twelve conditioning factors and three machine learning methods—random forest, support vector machine, and extreme gradient boosting (XGBoost)—to generate rock fall, landslide, and debris flow susceptibility maps. The results show that debris flow models presented the best prediction capabilities [area under the receiver operating characteristic curve (AUC 0.95)], followed by rock fall (AUC 0.94) and landslide (AUC 0.85). Additionally, XGBoost outperformed the other two methods with the highest AUC of 0.93. All three methods with AUC values larger than 0.84 suggest that these models have fairly good performance to assess geological hazards susceptibility. Finally, evolution index was constructed based on a joint probability of these three hazard models to predict the evolution tendency of 35 unstable slopes in Jiuzhaigou. The results show that these unstable slopes are likely to evolve into debris flows with a probability of 46%, followed by landslides (43%) and rock falls (29%). Higher susceptibility areas for geohazards were mainly located in the southeast and middle of Jiuzhaigou, implying geohazards prevention and mitigation measures should be taken there in near future.

     

Snow avalanche susceptibility evaluation in the central Shaluli Mountains of Tibetan Plateau based on machine learning method北大核心CSCD

Snow avalanches,which are widely and frequently developed at high elevations,seriously threatens the built traffic corridors in the Tibetan Plateau. Susceptibility evaluation of snow avalanche via machine learning model with a high forecast accuracy can be appled to quickly and effectively assess the regional avalanche risk. This paper took the central Shaluli Mountain region as the study area,in which the snow avalanche inventory was established through remote sensing interpretation and field investigation verification. We quantitatively extracted 17 evaluation factors via GIS-based analysis,and these factors were selected through the variance expansion factor（VIF）. Four machine learning models containing SVM,DT,MLP and KNN were used to compile the susceptibility index map of snow avalanches,and kappa coefficient and ROC curve were used to verify the accuracy. The results suggested that the susceptibility indexes obtained from SVM,DT,MLP and KNN were in the range of［0,0. 964］,［0,815］,［0,0. 995］and［0,1］,respectively. The accuracy test results show that these four models all have good prediction accuracy. Among them,the SVM model is the best. The results also indicated that the areas with the high snow avalanche susceptibility mainly distributed in Genie Mountain and Rigong Mountain,most of which were above the planation surface of the Tibetan Plateau. The average altitude of the extremely high snow-avalanche-prone areas is 4 939 m,while the average altitude of the high snow avalanche-prone areas is 4 859 m. The snow avalanche has low perniciousness on the Sichuan-Tibet Highway and the Sichuan-Tibet Railway in the study area. This study can provide theoretical basis and method reference for disaster prevention and mitigation of snow avalanche along Sichuan-Tibet Railway and other major projects across Shaluli Mountains region. © 2022 Science Press (China).     

How do uncertainties of structure refinements influence the accuracy of the prediction of enthalpy of formation? Examples on muscovite and natrolite

A method to predict the enthalpy of formation is based on the parameter _HO^2-cation (comp) (Vieillard 1982; Vieillard and Tardy 1988a). The calculation of this parameter requires the knowledge of cristallographical and optical properties of minerals (refractive index, molar volume, average distances and shortest bond lengths of cation-oxygens in all sites, existence of hydrogen bonding, presence of a order-disorder in a site of the involved compound). This method has been tested on a great number of minerals and artificial compounds belonging to the system: Li₂O-Na₂O-K₂O-BeO-MgO-CaO-MnO-FeO-Fe₂O₃-Al₂O₃-SiO₂-H₂O (180 minerals) (Vieillard 1994a, b). When several structure refinements of the same mineral are present, several factors such as precision of crystal structure determination, chemical composition, existence of any disorder in a site, presence of hydrogen bonding and estimated standard deviation on bond lengths seem to play an important role in the predication of enthalpy of formation from oxides. Muscovite and natrolite, for which numerous structure refinements are available, are used and discussed in this study.     

Identification of landslide susceptibility zones in Gish River basin,West Bengal,India

The occurrence of landslide in the hilly region of Darjeeling during monsoon season is a matter of serious concern. Every year this natural hazard damages the major roads at several places and thus disrupts the transport and communication system in this region. This paper tries to prepare a landslide susceptibility zone (LSZ) map for the Gish River basin. A total number of 16 spatial parameters have been taken for this study and these are categorised under six factor clusters or groups for example, triggering factors, protective factor, lithological factors, morphometric factors, hydrological factors and anthropogenic factors. The LSZ map is prepared by integrating all the parameters adopting the weighting base as logistic regression. The landslide susceptibility map shows that nearly 9.11% of the area falls under the very high landslide-susceptible zone while 40.28% of the area of the total basin lies under the very low landslide-susceptible zone. The landslide-susceptible model is validated through the receiver operating characteristic curve. This curve shows 86% success rate in defining landslide-susceptible zones and 83.40% prediction rate for the occurrence of landslides. The spatial relationship between the landslide susceptibility model and other factors’ groups shows that the morphometric factors’ cluster (mainly slope) is the focalone for the determination of landslide-susceptible zone.     

Flood hazard mapping in western Iran: assessment of deep learning vis-à-vis machine learning models

Natural Hazards - On March 25, 2019, widespread flood events occurred across Iran’s provinces and set a new record for socioeconomic losses and casualties. In hindsight, it opened an...     

A comparison of landslide susceptibility maps produced by logistic regression, multi-criteria decision, and likelihood ratio methods: a case study at ?zmir, Turkey

The main purpose of this study is to compare the use of logistic regression, multi-criteria decision analysis, and a likelihood ratio model to map landslide susceptibility in and around the city of İzmir in western Turkey. Parameters, such as lithology, slope gradient, slope aspect, faults, drainage lines, and roads, were considered. Landslide susceptibility maps were produced using each of the three methods and then compared and validated. Before the modeling and validation, the observed landslides were separated into two groups. The first group was for training, and the other group was for validation steps. The accuracy of models was measured by fitting them to a validation set of observed landslides. For validation process, the area under curvature (AUC) approach was applied. According to the AUC values of 0.810, 0.764, and 0.710 for logistic regression, likelihood ratio, and multi-criteria decision analysis, respectively, logistic regression was determined to be the most accurate method among the other used landslide susceptibility mapping methods. Based on these results, logistic regression and likelihood ratio models can be used to mitigate hazards related to landslides and to aid in land-use planning.     

How to manipulate interactive E-book on learning natural catastrophe—An example of structural mechanics using power machine

Economic losses caused by natural disasters have increased rapidly in recent years. Therefore, learning about natural catastrophe will be helpful for saving life and reducing cost. This study explores the effect of using the interactive electronic book available from Lego Simple Power Machine sets for learning. The applications for Power Machine are wide, such as in robots manufacturing, artificial intelligence, and intelligent devices. Based on the theory of Human–Computer Interaction, user-centered design is a modern and widely practiced design philosophy rooted in the idea that users must take center-stage in the design of any computer system. An interactive electronic book is developed to test learning achievement with the Lego Simple Power Machine set. The results reveal that the attention of users can be improved and the image can also be improved during the interaction and operation process. Finally, this study proves that the interactive electronic book makes learning easier and faster than the traditional book.     

Different ways of landslide geometry interpretation in a process of statistical landslide susceptibility and hazard assessment: Horná Súča (western Slovakia) case study

The article draws a comparison between different ways of landslide geometry interpretation in the scope of the statistical landslide hazard and risk assessment processing. The landslides are included as a major input variable, which are compared with all of the input parametric factors. Based on the above comparison the input data are classified and the final map of landslide susceptibility is constructed. Methodology of multivariate conditional analysis has been used for the construction of final maps. Unique condition units was developed by combination of geological map (lithological units) and slope angle map. Lithological units were derived from geological map and subsequently reclassified into 22 classes. Slope angle map was calculated from digital elevation model (contour map at a scale 1:10,000) and reclassified into nine classes. As a case study, a wide area of Horná Súča (western Slovakia) strongly affected by landsliding (predominantly made of Flysch) has been chosen. Spatial data in the form of parametric maps, as well as final statistical data set were processed in GIS GRASS environment. Four different approaches are used for landslides interpretation: (1) area of landslide body including accumulation zone, (2) area of depletion zone, (3) lines of elongated main scarps, (4) lines of main scarp upper edge. For each approach, a zoning map of landslide susceptibility was compiled and these were compared with each other. Depending on the interpretation approach, the final susceptibility zones are markedly different (in tens of percent).     

Sensitivity analysis of parameters influencing the ice–seabed interaction in sand by using extreme learning machine

Ice gouging problem is a significant challenge threatening the integrity of subsea pipelines in the Arctic (e.g., Beaufort Sea) and even non-Arctic (e.g., Caspian Sea) offshore regions. Determining the seabed response to ice scour through the subgouge soil deformations and the keel reaction forces are important aspects for a safe and cost-effective design. In this study, the subgouge soil deformations and the keel reaction forces were simulated by the extreme learning machine (ELM) for the first time. Nine ELM models (ELM 1–ELM 9) were developed using the key parameters governing the ice–seabed interaction. The number of neurons in the hidden layer was optimized and the best activation function for the ELM network was identified. The premium ELM model, resulting in the lowest level of inaccuracy and complexity and the highest level of correlation with experimental values was identified by performing a sensitivity analysis. The gouge depth ratio and the shear strength of the seabed soil were found to be the most influential input parameters affecting the subgouge soil deformations and the keel reaction forces. A set of the ELM-based equations were proposed to approximate the ice gouging parameters. The uncertainty analysis showed that the premium ELM model slightly underestimated the subgouge soil deformation.

     

Development and application of Shannon’s entropy integrated information value model for landslide susceptibility assessment and zonation in Sikkim Himalayas in India

Natural Hazards - The state of Sikkim in India has many steep slopes and has been susceptible to landslides. Since 1968 there have been innumerable losses of lives and properties due to landslides....     

A comparative evaluation of heuristic and bivariate statistical modelling for landslide susceptibility mappings in Ghurmi–Dhad Khola, east Nepal

Landslides are one of the major natural disasters that occur in the Himalayan range with recurring frequency, causing enormous loss of life and property every year. Preparation of landslide inventory maps and landslide susceptibility zonation maps are the important tasks to be taken into account initially for safe mitigation measures. The present paper focuses on landslide susceptibility maps of the Ghurmi–Dhad Khola area, east Nepal, using Geographic Information System. For this purpose, the landslide susceptibility maps are prepared by using the heuristic and bivariate statistical methods. The parameters considered for the study are slope angle, slope aspect, elevation, distance from drainage, geology, land cover, rock and soil type, and distance from faults and folds. The landslide susceptibility zonation map produced from the heuristic method shows that 42.59 % of the observed landslide falls under the very high susceptible zone and 33.00 % under the high susceptible zone. Likewise, the landslide susceptibility zonation map produced from the bivariate method depicts that 44.19 % of the observed landslide falls under the very high susceptible zone and 31.59 % under the high susceptible zone. Both the landslide susceptibility zonation maps are identical, and success rates of both the maps are above 80 %. While comparing the landslide susceptibility maps obtained from two different methods, about 78 % of the study area falls in the identical susceptible zones. Special attention should be taken into consideration for the construction works in the areas which have been spatially agreed as very high and high susceptible zones from both techniques. Moreover, these maps can be used for slope management, land use planning, disaster management planning, etc., by the concerned authorities.     

Is spatial autocorrelation introducing biases in the apparent accuracy of paleoclimatic reconstructions?

We address the issue of spatial autocorrelation, an occurrence that may introduce biases in the evaluation of the performance of transfer functions, by using two fundamentally different approaches, one based on calibration (weighted averaging partial least squares regressions; WA-PLS) and the other based on similarities (modern analogue technique; MAT). The tests were made after spatial standardization of 700 North Atlantic surface data points, which include 29 dinocyst taxa and 4 climate parameters. The evaluation of transfer function performance was made by defining a verification dataset that was gradually isolated from the calibration or comparison datasets. Although strong spatial autocorrelation characterizes the original climate parameter distribution, the results show that the spatial structure of data has relatively low effect on the calculation of the error of prediction. They also show that the performances of MAT are generally better than those of WA-PLS, with lower error of prediction. The better performance of MAT in the present study can be explained by the non-modal distribution of salinity and temperature in the studied marine environments, which is not appropriate for the application of WA-PLS. The two methods yield equivalent results about the spatial structure of the residuals based on empirical semi-variograms. The analyses we performed include the comparison of reconstructions based on original raw data and gridded data. Results suggest that the gridding of the reference database may reduce the noise and thus improve the performance of the techniques.     

How to avoid false interpretations of Sentinel-1A TOPSAR interferometric data in landslide mapping? A case study: recent landslides in Transdanubia,Hungary

Natural Hazards - It is a crucial issue to better understand the usability of Sentinel-1 satellites in geomorphologic applications, since Sentinel-1 and the Copernicus Program are considered to be...     

Remote sensing GIS-based landslide susceptibility & risk modeling in Darjeeling–Sikkim Himalaya together with FEM-based slope stability analysis of the terrain

Natural Hazards - Landslide susceptibility (LSI) modeling of Darjeeling–Sikkim Himalaya is performed by integrating 28 causative factors on 28C28 combinations on Geographical Information...     

How do “numbers” construct social reality in disaster-stricken areas?: a case of the 2008 Wenchuan earthquake in Sichuan,China

This article focuses on the dynamics of using numbers to construct an image of social reality in disaster areas. Numbers are neither objective nor value-neutral but are rather generated, transmitted and shared with social signification. In other words, numbers can be thought of as simply socially constructed information. Statistics and other numbers usually work in positive ways. However, it is also possible that using numbers in the media can lead to unintended messages that could produce negative consequences. We conducted field studies in disaster-stricken areas of the 2008 Wenchuan earthquake in China and compared findings to the case of the 1995 Kobe earthquake in Japan in order to examine how numbers—in terms of the amount of donations, the timeline of reconstruction projects and casualty figures—construct social reality and cause a variety of social dysfunctions.     

Impacts of future climate and land cover changes on landslide susceptibility: regional scale modelling in the Val d’Aran region (Pyrenees,Spain)

It is widely accepted that future environmental changes will affect rainfall-induced shallow slides in high-mountain areas. In this study, the Val d’Aran region located in the Central Pyrenees was selected to analyze and quantify the impacts of land use and land cover (LULC) and climate changes on regional landslides susceptibility. We analyzed 26 climate models of the EURO-CORDEX database focussing on the future rainfall conditions. The IDRISI TerrSet software suite was used to create the future LULC maps. These two inputs were analyzed individually and in a combined way defining 20 different scenarios. All these scenarios were incorporated in a physically based stability model to compute landslides susceptibility maps. The results showed that both environmental conditions will considerably change in the future. The daily rainfall will increase between 14 and 26% assuming a return period of 100 years. This intensification of precipitation will produce an overall decrease of the stability condition in the study area. Regarding the LULC prediction, the forest area will significantly increase, while in particular grassland, but also shrubs decrease. As a consequence, the overall stability condition improves, because the root strength is higher in forest than in grassland and shrubs. When we analyzed the combined impacts, the results showed that the positive effect of LULC changes is larger than the negative influence of rainfall changes. Hence, when combining the two aspects in the future scenarios, the stability condition in the study area will improve.

     

A comparative study of different machine learning algorithms in predicting EPB shield behaviour: a case study at the Xi’an metro,China

Acta Geotechnica - Complex geological conditions and/or inappropriate shield tunnel boring machine (TBM) operation can significantly degrade both the excavation and safety of tunnel construction....     

Why do smallholders plant biofuel crops? The ‘politics of consent’ in Mexico

Recent studies have addressed the social and environmental impacts of biofuel crops but seldom the question as to why rural producers engage in their production. It is particularly unclear how governments worldwide, especially in middle-income countries such as Brazil, Thailand, and Mexico, could enroll so many smallholders in biofuel cropping projects. Conventional views see yields and economic returns as main drivers for smallholder participation in biofuel production but ignore the role played by power and politics. This paper analyses the rapid biofuel expansions (oil palm, jatropha) in the southern Lacandon rainforest in Chiapas (Mexico) and their partial failure (jatropha) from a political ecology perspective. Our findings indicate that biofuel expansions in this region not only occurred for productive reasons, but also because biofuel programmes provided prospects for political gains through strengthened rural organisations. In contrast with emphasis on state coercion and local resistance—common in political ecology—the biofuel expansion relied, in this case, upon a ‘politics of consent’ in which both the state and rural organisations, albeit in a power-laden relationship, sought to achieve their own goals by supporting the planting of biofuel crops. These findings suggest the need to rethink how particular approaches within political ecology apply Gramsci’s notions of power and hegemony and, more broadly, to consider the importance of politics in explaining why certain forms of agricultural production become dominant.     

How do lineations reflect the strain history of transpressive shear zones? The example of the active Alpine Fault zone,New Zealand

Lineations within mylonites exhumed in the hanging wall of New Zealand's active Alpine Fault zone have a complicated relationship to contemporary plate kinematics. The shear zone is triclinic and macroscopic object lineations are not usually parallel to the simple shear direction, despite high total simple shear strains (γ ≥ 150). This is mostly because the lineations are inherited from pre-mylonitic fabrics, and have not been rotated into parallelism with the mylonitic stretching direction (which pitches c. 44° in the fault plane). Furthermore, some lineations have been variably rotated depending on whether they are present in shear bands or microlithons, which accommodated bulk strains with different vorticities. Total strains required to obtain parallelism between the finite maximum principal stretching direction calculated from transpression models and these mylonitic lineations, are pure shear stretch, S₁ ∼ 3.5; simple shear 11.7 < γ < 150. The observations and numerical models also show that linear features are not rotated much during simple shear because they initially lie within the shear plane, and that inherited fabric components may not be destroyed until very high simple shear strains have been attained.