Landslide susceptibility zonation using statistical and machine learning approaches in Northern Lecco,Italy

This study deals with landslide susceptibility mapping in the northern part of Lecco Province, Lombardy Region, Italy. In so doing, a valid landslide inventory map and thirteen predisposing factors (including elevation, slope aspect, slope degree, plan curvature, profile curvature, distance to waterway, distance to road, distance to fault, soil type, land use, lithology, stream power index, and topographic wetness index) form the spatial database within geographic information system. The used predictive models comprise a bivariate statistical approach called frequency ratio (FR) and two machine learning tools, namely multilayer perceptron neural network (MLPNN) and adaptive neuro-fuzzy inference system (ANFIS). These models first use landslide and non-landslide records for comprehending the relationship between the landslide occurrence and predisposing factors. Then, landslide susceptibility values are predicted for the whole area. The accuracy of the produced susceptibility maps is measured using area under the curve (AUC) index, according to which, the MLPNN (AUC?=?0.916) presented the most accurate map, followed by the ANFIS (AUC?=?0.889) and FR (AUC?=?0.888). Visual interpretation of the susceptibility maps, FR-based correlation analysis, as well as the importance assessment of predisposing factors, all indicated the significant contribution of the road networks to the crucial susceptibility of landslide. Lastly, an explicit predictive formula is extracted from the implemented MLPNN model for a convenient approximation of landslide susceptibility value.

     

Landslide identification using machine learning

Landslide identification is critical for risk assessment and mitigation.This paper proposes a novel machinelearning and deep-learning method to identify natural-terrain landslides using integrated geodatabases.First,landslide-related data are compiled,including topographic data,geological data and rainfall-related data.Then,three integrated geodatabases are established;namely,Recent Landslide Database(Rec LD),Relict Landslide Database(Rel LD)and Joint Landslide Database(JLD).After that,five machine learning and deep learning algorithms,including logistic regression(LR),support vector machine(SVM),random forest(RF),boosting methods and convolutional neural network(CNN),are utilized and evaluated on each database.A case study in Lantau,Hong Kong,is conducted to demonstrate the application of the proposed method.From the results of the case study,CNN achieves an identification accuracy of 92.5%on Rec LD,and outperforms other algorithms due to its strengths in feature extraction and multi dimensional data processing.Boosting methods come second in terms of accuracy,followed by RF,LR and SVM.By using machine learning and deep learning techniques,the proposed landslide identification method shows outstanding robustness and great potential in tackling the landslide identification problem.     

Mapping horizontal displacement of soil nail walls using machine learning approaches

Machine learning (ML) approaches have stormed nearly all engineering fields since recent years. However, the situation is somehow subtle in civil engineering practice, especially in the sub-field of geotechnical engineering where data from real-life projects are usually scarce, which in turn prevents development of meaningful mapping tools based on ML techniques. This study first shares a database containing a total of 376 measured horizontal displacements from instrumented soil nail walls reported in the literature. Then, these data are utilized to develop three types of ML models for mapping the wall horizontal displacement along depth, including artificial neural network (ANN), random forest (RF), and support vector machine (SVM). The uncertainties of the ANN, RF, and SVM models are then quantitatively evaluated using bias statistics where bias is defined as the ratio of measured to predicted horizontal displacement. The three ML models are proved to be accurate on average with medium dispersions in prediction, which outperform the existing simple empirical regression models. Probability distribution functions for those biases are also characterized. This study demonstrates that introduction of machine learning approaches into the reliability-based design framework for soil nail walls and other geotechnical structures is promising.

     

A hybrid model using machine learning methods and GIS for potential rockfall source identification from airborne laser scanning data

The main objectives of this paper are to design and evaluate a hybrid approach based on Gaussian mixture model (GMM) and random forest (RF) for detecting rockfall source areas using airborne laser scanning data. The former model was used to calculate automatically slope angle thresholds for different type of landslides such as shallow, translational, rotational, rotational-translational, complex, debris flow, and rockfalls. After calculating the slope angle thresholds, a homogenous morphometric land use area (HMLA) was constructed to improve the performance of the model computations and reduce the sensitivity of the model to the variations in different conditioning factors. After that, the support vector machine (SVM) was applied in addition to backward elimination (BE) to select and rank the conditioning factors considering the type of landslides. Then, different machine learning methods [artificial neural network (ANN), logistic regression (LR), and random forest (RF) were trained with the selected best factors and previously prepared inventory datasets. The best fit method (RF) was then used to generate the probability maps and then the source areas were detected by combining the slope raster (reclassified according to the thresholds found by the GMM model) and the probability maps. The accuracy assessment shows that the proposed hybrid model could detect the potential rockfalls with an accuracy of 0.92 based on training data and 0.96 on validation data. Overall, the proposed model is an efficient model for identifying rockfall source areas in the presence of other types of landslides with an accepted generalization performance.     

Structural and textural development in Singhbhum shear zone,eastern India

The rocks within the Singhbhum shear zone in the North Singhbhum fold belt, eastern India, form a tectonic melange comprising granitic mylonite, quartz-mica phyllonite, quartz-tourmaline rock and deformed volcanic and volcaniclastic rocks. The granitic rocks show a textural gradation from the least-deformed variety having coarse-to medium-grained granitoid texture through augen-bearing protomylonite and mylonite to ultramylonite. Both type I and type II S-C mylonites are present. The most intensely deformed varieties include ultramylonite. The phyllosilicate-bearing supracrustal rocks are converted to phyllonites. The different minerals exhibit a variety of crystal plastic deformation features. Generation of successive sets of mylonitic foliation, folding of the earlier sets and their truncation by the later ones results from the progressive shearing movement. The shear sense indicators suggest a thrust-type deformation. The microstructural and textural evolution of the rocks took place in an environment of relatively low temperature, dislocation creep accompanied by dynamic recovery and dynamic recrystallization being the principal deformation mechanisms. Palaeostress estimation suggests a flow stress within the range of 50–190 MPa during mylonitization.     

Back-analysis for initial ground stress field at a diamond mine using machine learning approaches

Exact knowledge for ground stress field guarantees the construction of various underground engineering projects as well as prediction of some geological hazards such as the rock burst. Limited by costs, field measurement for initial ground stresses can be only conducted on several measure points, which necessitates back-analysis for initial stresses from limited field measurement data. This paper employed a multioutput decision tree regressor (DTR) to model the relationship between initial ground stress field and its impact factor. A full-scale finite element model was built and computed to gain 400 training samples for DTR using a submodeling strategy. The results showed that correlation coefficient r between field measurement values and back-analysis values reached 0.92, which proved the success of DTR. A neural network was employed to store the global initial ground stress field. More than 600,000 node data extracted from the full-scale finite element model were used to train this neural network. After training, the stresses on any location can be investigated by inputting corresponding coordinates into this neural network.

     

Multivariate forecast of winter monsoon rainfall in India using SST anomaly as a predictor: Neurocomputing and statistical approaches

In this article, the complexities in the relationship between rainfall and sea surface temperature (SST) anomalies during the winter monsoon over India were evaluated statistically using scatter plot matrices and autocorrelation functions. Linear, as well as polynomial trend equations were obtained, and it was observed that the coefficient of determination for the linear trend was very low and it remained low even when polynomial trend of degree six was used. An exponential regression equation and an artificial neural network with extensive variable selection were generated to forecast the average winter monsoon rainfall of a given year using the rainfall amounts and the SST anomalies in the winter monsoon months of the previous year as predictors. The regression coefficients for the multiple exponential regression equation were generated using Levenberg-Marquardt algorithm. The artificial neural network was generated in the form of a multilayer perceptron with sigmoid non-linearity and genetic-algorithm based variable selection. Both of the predictive models were judged statistically using the Willmott's index, percentage error of prediction, and prediction yields. The statistical assessment revealed the potential of artificial neural network over exponential regression.     

Improving the coastal aquifers’ vulnerability assessment using SCMAI ensemble of three machine learning approaches

The main objective of this study is to integrate adaptive neuro-fuzzy inference system (ANFIS), support vector machine (SVM) and artificial neural network (ANN) to design an integrated supervised committee machine artificial intelligence (SCMAI) model to spatially predict the groundwater vulnerability to seawater intrusion in Gharesoo-Gorgan Rood coastal aquifer placed in the northern part of Iran. Six hydrological GALDIT parameters (i.e., G groundwater occurrence, A aquifer hydraulic conductivity, L level of groundwater above sea level, D distance from the shore, I impact of the existing status of seawater intrusion in the region, and T thickness of the aquifer) were considered as inputs for each model. In the training step, the values of GALDIT’s vulnerability index were conditioned by using the values of TDS concentration in order to obtain the conditioned vulnerability index (CVI). The CVI was considered as the target for each model. After training the models, each model was tested using a separate TDS dataset. The results indicated that the ANN and ANFIS algorithms performed better than the SVM algorithm. The values of correlation were obtained as 88, 87, and 80% for ANN, ANFIS, and SVM models, respectively. In the testing step of the SCMAI model, the values of RMSE, R², and r were obtained as 6.4, 0.95, and 97%, respectively. Overall, SCMAI model outperformed other models to spatially predicting vulnerable zones. The result of the SCMAI model confirmed that the western zones along the shoreline had the highest vulnerability to seawater intrusion; therefore, it seems critical to consider emergency protection plans for study area.

Graphic abstract

     

基于多元统计和傅立叶和谱分析的土壤重金属的来源解析及其风险评价

分析珠江三角洲腹地佛山顺德区208个蔬菜地表层土样Cu、Ni、Cr、As、Pb、Zn、Cd和Hg等8种重金属的全量,结果表明,8种重金属的平均浓度高于广东省土壤背景值。Cd和Hg的最高浓度和变异系数分别为6.54mg/kg、115%和4.82mg/kg、151%,暗示Cd和Hg的人为来源。多元统计与傅立叶和谱分析的结合,解释了Cr、Ni和Cu的自然来源,Pb、Zn、As、Cd和Hg的人为来源;傅立叶和谱分析进一步阐释了Zn与Cu的双重来源,并推断土壤Hg来源于大气沉降。研究区内大约21.7%的土壤受重金属污染,表明需要调整该区的农业生产活动。     

Ravine erosion and reclamation in India

Ravine and gully erosion affects 1% of India's land area. Zones of severe ravine trenching are found along the margins of the Gangetic Basin and in the semi-arid northwest. Ravine reclamation is currently rated as a high national priority, and India has the Third World's leading soil conservation movement. This paper reviews the technical contributions made to the study of ravine origins and genesis by government soil conservation research workers in the light of the dissident views expressed by academic geoscientists. It also reviews the soil conservation establishment's major contributions to cost-effective ravine reclamation planning. However, it is emphasised that while great advances have been made towards technological remedies for ravine erosion, relatively little has been accomplished in the realm of social science. Successful ravine reclamation requires the support and involvement of the local cultivator and local community and, probably, reform of local land tenure and social arrangements. Towards this end, much experience has been accumulated in non-government circles through the activities of Gandhian sarvodaya groups. It is recommended that their ideas and methods are integrated into ravine reclamation activities.     

Spatial prediction of landslide susceptibility using GIS-based statistical and machine learning models in Wanzhou County,Three Gorges Reservoir,China

Landslide susceptibility mapping is vital for landslide risk management and urban planning. In this study, we used three statistical models [frequency ratio, certainty factor and index of entropy (IOE)] and a machine learning model [random forest (RF)] for landslide susceptibility mapping in Wanzhou County, China. First, a landslide inventory map was prepared using earlier geotechnical investigation reports, aerial images, and field surveys. Then, the redundant factors were excluded from the initial fourteen landslide causal factors via factor correlation analysis. To determine the most effective causal factors, landslide susceptibility evaluations were performed based on four cases with different combinations of factors (“cases”). In the analysis, 465 (70%) landslide locations were randomly selected for model training, and 200 (30%) landslide locations were selected for verification. The results showed that case 3 produced the best performance for the statistical models and that case 2 produced the best performance for the RF model. Finally, the receiver operating characteristic (ROC) curve was used to verify the accuracy of each model’s results for its respective optimal case. The ROC curve analysis showed that the machine learning model performed better than the other three models, and among the three statistical models, the IOE model with weight coefficients was superior.

     

Landslide detection in the Himalayas using machine learning algorithms and U-Net

Landslides - Event-based landslide inventories are essential sources to broaden our understanding of the causal relationship between triggering events and the occurring landslides. Moreover,...     

Soil erosion modeling of a Himalayan watershed using RS and GIS

Employing the remote sensing (RS) and geographical information system (GIS), an assessment of sediment yield from Dikrong river basin of Arunachal Pradesh (India) has been presented in this paper. For prediction of soil erosion, the Morgan-Morgan and Finney (MMF) model and the universal soil loss equation (USLE) have been utilized at a spatial grid scale of 100 m × 100 m, an operational unit. The average annual soil loss from the Dikrong river basin is estimated as 75.66 and 57.06 t ha⁻¹ year⁻¹ using MMF and USLE models, respectively. The watershed area falling under the identified very high, severe, and very severe zones of soil erosion need immediate attention for soil conservation.     

Predicting permeability of tight carbonates using a hybrid machine learning approach of modified equilibrium optimizer and extreme learning machine

Acta Geotechnica - It is a problematic task to perform petro-physical property prediction of carbonate reservoir rocks in most cases, specifically for permeability prediction since a carbonate rock...     

A machine learning approach to tracking crustal thickness variations in the eastern North China Craton

The variation of crustal thickness is a critical index to reveal how the continental crust evolved over its four billion years. Generally, ratios of whole-rock trace elements, such as Sr/Y,(La/Yb)nand Ce/Y, are used to characterize crustal thicknesses. However, sometimes confusing results are obtained since there is no enough filtered data. Here, a state-of-the-art approach, based on a machine-learning algorithm, is proposed to predict crustal thickness using global major-and trace-element geochemical data of intermediate arc rocks and intraplate basalts, and their corresponding crustal thicknesses. After the validation processes, the root-mean-square error(RMSE) and the coefficient of determination(R~2) score were used to evaluate the performance of the machine learning algorithm based on the learning dataset which has never been used during the training phase. The results demonstrate that the machine learning algorithm is more reliable in predicting crustal thickness than the conventional methods. The trained model predicts that the crustal thickness of the eastern North China Craton(ENCC) was ～45 km from the Late Triassic to the Early Cretaceous, but ～35 km from the Early Cretaceous, which corresponds to the paleo-elevation of 3.0 ± 1.5 km at Early Mesozoic, and decease to the present-day elevation in the ENCC. The estimates are generally consistent with the previous studies on xenoliths from the lower crust and on the paleoenvironment of the coastal mountain of the ENCC, which indicates that the lower crust of the ENCC was delaminated abruptly at the Early Cretaceous.     

Delineation of groundwater potential zone in Chhatna Block, Bankura District, West Bengal, India using remote sensing and GIS techniques

Groundwater is one of the most valuable natural resources, which is an immensely important and dependable source of water supply in all climatic regions over the world. Groundwater is in demand in areas where surface water supply is inadequate and nonsexist in the Chhatna Block, Bankura district and is located on the eastern slope of Chotonagpur Plateau, which is mapped on 73 I/15, 73 I/16 and 73 M/3, and falls between latitude 23°10′–23°30′N and longitude 86°47′–87°02′E. It represents plain land and gentle slope, which is responsible for infiltration and groundwater recharge. The groundwater in this region is confined within the fracture zones and weathered residuum. The present investigation is, therefore, undertaken to delineate potential zones for groundwater development with the help of a remote-sensing study. IRS–LISS-III data along with other data sets, e.g., existing toposheets and field observation data, have been utilized to extract information on the hydrogeomorphic features which include valley fills, buried pediment moderate, buried pediment shallow and structural hills, lineament density contour and slope map of this hard rock terrain. The target of this study is to delineate the groundwater potential zones in Chhatna block, Bankura District, West Bengal. Satellite imagery, along with other data sets, has been utilized to extract information on the groundwater controlling features of this study area. Three features (hydrogeomorphology, slope, and lineaments) that influence groundwater occurrences were analyzed and integrated. All the information layers have been integrated through GIS analysis and the groundwater potential zones have been delineated. The weighted index overlay method has been followed to delineate groundwater potential zones. The results indicate that good to excellent groundwater potential zones are available in almost the entire block. The results show that there is good agreement between the predicted groundwater potential map and the existing groundwater borehole databases. The area is characterized by hard rock terrain—still due to the presence of planation surface along valley fills; it became the prospective zone. The area has been categorized into four distinct zones: excellent, good, fair and poor. Excellent groundwater potential zones constitute 30–35 % of the total block area; good groundwater potential zones occupy a majority of the block, covering approximately 55–60 % and the fair potential zones occupy about 10–15 % of the total block. Poor potential zones occupy a very insignificant portion (less than 1 %).     

Earthquake risk assessment in NE India using deep learning and geospatial analysis

Earthquake prediction is currently the most crucial task required for the probability, hazard, risk mapping, and mitigation purposes. Earthquake prediction attracts the researchers' attention from both academia and industries. Traditionally, the risk assessment approaches have used various traditional and machine learning models. However, deep learning techniques have been rarely tested for earthquake probability mapping. Therefore, this study develops a convolutional neural network (CNN) model for earthquake probability assessment in NE India. Then conducts vulnerability using analytical hierarchy process (AHP), Venn's intersection theory for hazard, and integrated model for risk mapping. A prediction of classification task was performed in which the model predicts magnitudes more than 4 Mw that considers nine indicators. Prediction classification results and intensity variation were then used for probability and hazard mapping, respectively. Finally, earthquake risk map was produced by multiplying hazard, vulnerability, and coping capacity. The vulnerability was prepared by using six vulnerable factors, and the coping capacity was estimated by using the number of hospitals and associated variables, including budget available for disaster management. The CNN model for a probability distribution is a robust technique that provides good accuracy. Results show that CNN is superior to the other algorithms, which completed the classification prediction task with an accuracy of 0.94, precision of 0.98, recall of 0.85, and F1 score of 0.91. These indicators were used for probability mapping, and the total area of hazard (21,412.94 km²), vulnerability (480.98 km²), and risk (34,586.10 km²) was estimated.     

矿区地下水系统水质分类判别的多元统计分析

以某矿区地下水系统为例,对该矿区地下水水化学资料进行了多元统计分析方法耦合应用研究,主要包括利用因子分析对存在相关关系的离子变量进行空间降维处理,找出能够反映众多离子信息的基础变量(正交因子),以其作为系统聚类变量;运用系统聚类法获取能代表各地下水子系统水化学特征的典型水样;使用贝叶斯逐步线性判别分析建立地下水各子系统水化学判别模型(判别函数),并对随机检验样品进行判别归属和判别模型统计检验。结果表明:这是一种稳定性较好且切实有效的、适用于矿区地下水系统水化学分类及水源水化学判别的方法。