Geomorphic Risk Reduction Using Geospatial Methods and Tools

In this chapter, we will introduce the landslide susceptibility assessment (LSA) methods based on machine learning techniques. The economic loss or even casualties caused by landslides indicate the significance of LSA. LSA can be regarded as either regression or classification problems, which can be processed by machine learning techniques. LSA provides administrators or researchers with information on potential disaster areas, which can be an efficient way to relieve the pressure of disaster reduction and mitigation. Several landslide inventories and disaster-related geo-environmental variable datasets were recommended. A total of 9 machine learning methods applied in LSA were simply introduced. The advantages and future work of LSA based on machine learning techniques were summarized from the aspects of scale, performance, modeling, and interpretability.

The concept of gully erosion susceptibility has received more focus in recent years, and the attention has been drown by researchers for the implementation of policy and practices. Soil erosion through gully development is a natural geomorphic process that controlled by human activities and highly effected environmental quality, ecosystem, natural resources, and agricultural activities; and it promotes hazards. So, gully erosion management is a key attempt for sustainable land use practices to assess and monitor the soil quality. In this work, the researchers try to employ a gully erosion susceptibility model along with management strategies and its controlling factors in Pathro River Basin, India. Authors follow a well-accepted machine learning-based boosted regression tree (BRT) model for the assessment of urgent management within the study area. Twelve predisposing factors were used here for the development of susceptibility map to find the areas that urgently required to take a robust management policy. The model depicts high prediction capacity with  a strong area under the curve (AUC is 87.40%). Finally, the dynamic nature of ecosystem service value (ESVs) and its sensitivity to land use have been examined by implementing an elasticity indicator. Badland areas were converted to forestland during 2010–2020 to manage the land degradation, but in some areas the gully erosion processes and their increasing trend were found due to unplanned land use practices. Also, the causes of erosive agents were evaluated by fitted function. For this study basin, forest is pivotal land use for both management of land degradation and ESVs so, it should be managed and conserved by afforestation programmes. The outcome of this work provides a new window for policy makers to initiate appropriate dimension about the land degradation and ecosystem management in prioritized areas of humid tropics.

Landslide is one of the important problems in the Mirik region of West Bengal. For managing this problem it is important to delineate the areas which are highly susceptible to landslide. In the present study ensemble of ANN, general linear model (GLM), and ensemble ANN-GLM machine learning methods were applied for producing the landslide susceptibility maps (LSMs) of the Mirik region. A total of 373 landslide locations and twelve landslide conditioning factors (LCFs) are retrieved from the spatial database and used for modeling the landslide susceptibility. Multicollinearity between the LCFs was carried out in order to select suitable LCFs. The built-in models were validated using ROC-AUC, mean absolute error (MAE), root mean square error (RMSE), and kappa coefficient. Using the 70:30 ratio landslide locations were classified into training and testing datasets. The ANN-GLM model got the lowest RMSE and the highest ROC-AUC (0.864) and kappa index (0.889) during the validation phase (0.086). As per the result of ensemble model 20.99% area of the Mirik region is very highly susceptible for landslide. The anticipated model is reliable in lowering the danger of landslide risks for prospective land use planning in the Mirik region of West Bengal 112.

The ambition of the current study was to generate landslide susceptibility maps (LSMs) for the Meenachil river basin’s upper catchment using the ensemble NBT-RTF, Naive Bayes tree (NBT), and rotation forest (RTF). For landslide susceptibility modelling, 189 landslide sites and 12 landslide conditioning factors (LCFs) were gathered. Multi-collinearity analysis was done among the LCFs to determine the best LCFs to use. The metrics utilized to assess the predictive power of the employed models are ROC-AUC, mean-absolute-error (MAE), root-mean-square-error (RMSE), and kappa coefficient. Almost 14% of the studied region has very high landslide susceptibility, according to the results of the best-performed model. The NBT-RTF model got the lowest RMSE and the greatest ROC-AUC (0.867) and kappa index (0.884) during the validation phase (0.234). The anticipated model is reliable for minimizing the impact of landslides in the research region and planning land development.

The landslides frequently affect Kurseong and the villages around it causing loss of life and property. The current study used the M5P technique, a deep learning neural network (DLNN), and an M5P-DLNN ensemble strategy to estimate the landslide susceptibility in the Kurseong area of West Bengal. The model’s results were cross-checked using the four folds of data. General field surveys and pertinent documents were used to find the locations of current landslides. Then, historical landslide locations were gathered, shown as an inventory map, and separated into four folds in order to calibrate and validate the models. A total of twelve LCFs (landslide conditioning factors) were employed to model the susceptibility to landslides. The developed landslide models were verified using two statistical techniques, namely the mean-absolute-error (MAE) and the root-mean-square-error (RMSE), as well as the receiver operating characteristic (ROC), accuracy, and precision. The accuracy measurements’ findings showed that all models had a good chance of identifying the Kurseong region’s landslide susceptibility. The ensemble model outscored the individual models in terms of precision among these models. The results of this study could be used to reduce the probability of landslides in the Kurseong region and other nearby locations with a similar topography and geology.

In the Bageshwar district, much of its rainfall comes from convectional thunderstorms, which trigger landslides and affect the physical as well as socio-economic environment. It causes loss of human lives, property damage, hampers the transport and communication system, etc. The present study aims to demonstrate the landslide vulnerability zones in Bageshwar district through the aid of ANN and LR models. For fulfilling the objectives of this study factors like slope, elevation, aspect, curvature, TRI, SPI, TWI, land use/land cover, rainfall, distance from the river, distance from lineament and soil were considered to prepare a landslide susceptibility zone map. For identifying the Socio-Economic vulnerable zones the considerable factors used were population density, female density, marginal cultivators, workers density, and literacy, medical facilities, building type, the distance from road, settlement density and household frequency. For better accuracy of the classification here, we opted machine learning methods like ANN and statistical models such as LR. The Landslide vulnerability which is an ensemble of landslide susceptibility and socio-economic vulnerability reveals that 9.15% (ANN), 7.39% (LR) of the area is very highly vulnerable to landslides and 17.77% (ANN), 18.89% (FR) of the area is out of danger for landslide. Therefore, it is expected that the study will give a new direction to the planners which will assist them to take steps regarding this study matter. Finally, for the evaluation of the opted two models, receiver operating characteristics (ROC) were considered for verifying the outcome of this study. ANN model signifies a high accuracy level indicating 84.06% area under the curve (AUC) and LR model indicates 75.79% AUC.

Flood is a widespread geomorphic hazard that causes an immense destruction not only in physical terms but also mentally, socially, and economically. To limit its destructive effects, proper planning, cope up ideas, and mitigation strategies are required. So the present study deals with the preparation of flood susceptibility mapping in the Sundarban region of West Bengal, India. The study prepares a flood inventory map and also identifies the collinearity among the factors and their IGR values. The factors selected in the present study include drainage density, drainage proximity, SPI, TWI, annual rainfall, vulnerable embankments, flood inundation, LULC, elevation, slope, curvature, and clay content. Three models and their ensemble approach were used to prepare the flood susceptibility maps. The models used were J48DT, J48DT-AdaBoost, and J48DT-Bagging. Finally, the outputs obtained were validated using six techniques namely, Sensitivity, Specificity, AUC (%), Kappa, MAE, and RMSE.

Flash flood is a significant issue in the Beas river basin. The main objectives of this study are to map flash flood susceptibility (FFS) in the Beas River Basin of Himachal Pradesh using a random forest (RF) data-driven model, prioritie flash flood conditioning factors using this methodology, and compare it with a multivariate adaptive regression spline (MARS).To provide the best prediction performance, their ensemble (MARS-RF) is used. The findings demonstrated that while predicting the FFS, the MARS, RF, and ensemble models obtained corresponding areas under curves (AUC) of 0.828, 0.856, and 0.88. The ensemble approach, which bases priority determination on the best model, was discovered to have a significant sensitivity to flash floods. It was found that 11.81% of the total area was extremely susceptible to flash flood events. These zonations employing RS-GIS and machine learning models can help decision-makers take quick and effective action to reduce flash floods and lessen the possibility of severe loss of life and property.

This work focuses on understanding to what degree the remote sensing tool of Interferometric Synthetic Aperture Radar (InSAR) can be used to track sub-surface ground motion in deep-seated landslides. We also consider the uncertainties that may arise out of using a remote sensing tool to track ground motion, as opposed to traditional boreholes, and how InSAR can be used to understand this uncertainty. The landslide case study of interest in this work is the El-Forn landslide in Canilllo, Andorra. These objectives of this work will be completed by focusing on the utilization of available Sentinel-1 data. Sentinel-1 data was processed and were used to generate a stack of Sentinel-1 images using small temporal and spatial baseline subsets (SBAS), which relies on many SAR acquisitions and implements a combination of the multi-look interferograms computed from the original SAR acquisitions, generating mean deformation velocity maps and time series (Berardino et al., IEEE Trans Geosci Remote Sens 40:2375–2383, 2002; Handwerger et al., Sci Rep 9: 1–12, 2019; Yunjun et al., Comput Geosci 133:104–331, 2019). From there, the interferogram pairs were inverted with the Miami Insar Time series software in Python (MintPy), InSAR Scientific Computing Environment (ISCE), and Hyp3 toolboxes/open-repositories to create displacement time series (Yunjun et al., Comput Geosci 133:104–331, 2019) over the main scarp. The displacement time series from InSAR was compared to in-situ ground motion measurements, suggesting that InSAR-based displacement data can be used to track sub-surface ground motion trends. Similarly, InSAR was found to indicate extreme sub-surface events in ground motion time seriry kriging was used as a method of geospatial interpolation over the landslide in order to visualize the quantity of remote observations necessary to minimize error in scarp reconstruction, suggesting that around 20 total observations lowers the normalized root mean squared error for the geospatially interpolated reconstruciton of the El Forn landslide surface.

The present research work is carried out within the selected stretch of river Damodar starting from Durgapur barrage to Pallaroad of West Bengal, India. Measurement of braiding index (>1.5) and sinuosity (<1.5) with the aim of analyzing river morphometric parameters along with river shifting related with erosion–deposition for sinuosity throughout the study time duration indicate that the river has a braiding and straight or sinuous nature. Island area has increased from 54%–61% of the total river area during 2001–2018. Statistical analysis reveals that braiding index is positively correlated to alpha index (r = 0.902) and negatively correlated with island area (r = −0.903). The river has shifted in the east–west direction but the overall stretch stabilizes its banks and shows a tendency toward equilibrium. The total area of erosion and deposition within the selected stretch is 14.01 km² and 36.339 km², respectively. Statistically, the depositional phenomenon shows a significant change within the studied time period whereas no such significant change is observed in the erosion phenomenon. The results obtained can be helpful in developing flood management strategies proving useful for future preparedness and mitigation measures.

River shifting is a natural process. The present study focuses on the shifting character of the Ganga River along the Malda district. The shifting of the river was observed from 1990 to 2020. This study used Satellite images and GIS tools to understand river shifting. It can be stated that the River Ganga is shifting towards its left bank. This means left bank erosion is much more than right bank erosion. The river shifting has increased with time. A more meandering shape of the river was observed in 2020 than in 2015. Farakka barrage plays an important role in flood and erosion as well as the deposition of sediments. People in Manikchak, Kaliachak-II, and Kaliachak-III blocks of Malda district West Bengal were highly affected due to this river shifting in the lower course of the Ganga River. A few portions of the Rajmahal block of Jharkhand, located on the right side of the river are also affected.

Bhutan is poised to a rugged terrain country that exhibits variant of hydrology behaviors that demands to be effective utilization of different types of hydrological databases. Due to the complicated network of drainage systems, proper forecasting needs to be studied and analyzed, so that effective management of floods/drought events on a daily basis can be implemented in the coming future. However, hydrological data in Bhutan are scarce information exclusively for a developing country, wherein, minimal gaging stations are set up currently to analyze the hydrological behaviors. This paper attempts to analyze the majority of the ungauged basins. While doing analysis, the hydraulics structure aligned to complicated hydrological behaviors precisely incorporated for the stormwater management works and allied river mitigations. This study is also to evaluate the surface runoff from the Barsa (River) Basin in Pasakha using precipitation data and curve number (CN) by performing infiltration methods. Pasakha is one of the most vibrant industrial hub areas as all kinds of industries are located, which is seemingly felt to be protected from flood from Barsa River. HEC-HMS 3.4 has been employed to prepare the Hydrological Soil Group (HSG) map and Land use map of the area. Curve Number zoning of the area has been assigned to generate the runoff amount from the HEC-HMS. HSG map has been mapped by performing numerous standard infiltration tests throughout the basin to get the infiltration rate at each basin and interpolating the obtained results by employing the Geostatistical Kriging interpolation method in ArcGIS. A land use map has been prepared by using Toposheet 78F05 (1:50,000), Phuentsholing base map (CAD data), and Recent Google images. Rainfall data for the year (1996–2019) has been acquired from the Metrological Department, Ministry of Economic Affairs. Curve Number has been manually calculated using SCS-TR55 table, those HSG map and Land use map, which later has been used as feed data in HEC-HMS for simulation of surface runoff as discharge. The study area, pasakha is grouped as A and B based on the HSG classification table, and assigning the value is validated by visiting each basin. It has been noticed that W350 has the highest runoff potential after simulation as the soil is grouped as B, a higher impervious area with an average flow of 11.6 m³/s. The lowest runoff potential is found to be W500 at 2.7 m³/s and the average surface runoff overall was 7.4 m³/s.

Landslide susceptibility in the Jakholi region of Garhwal Himalaya was assessed applying novel ensembles of statistical and machine learning algorithms. To begin, landslides were established and a landslide inventory map was prepared. The total locations considered for this study were divided into 70% for the training datasets and 30% for the validation datasets. Following that, a total of 15 landslide conditioning factors were chosen. Furthermore, the certainty factor approach was used for conducting a study of the correlation between conditioning factors as well as landslides. Following that, the CF, CF-SVM, CF-ANN, and CF-RF ensemble models were used for landslide susceptibility modeling and zoning. Finally, the average performance of the four models was evaluated and compared using the receiver operating characteristic (ROC) curve and statistical parameters. For the four models, the area under the curve (AUC) comes out to be greater than 0.81. According to the findings. ROC results indicate that the CF-RF ensemble model gave better performance and the CF model gave comparatively low accuracy. Additionally, this research also showed that an integrated model isn’t always better as compared to a single model. This ensemble analysis can be used as a useful method for land planning and monitoring in the future. It can be successfully utilized for the simulation of other geohazards.

Among the various natural geological phenomenon affecting the Indian peninsula, landslides have had huge damaging effects on human life and infrastructure, thereby requiring researchers to delve into more in-depth landslide-related studies with a focus on finding effective methods to better delineate landslide-prone areas as well as better understand the various underlying causative factors. The district of Kullu, situated in the Indian state of Himachal Pradesh has been the witness to some of these disastrous events, possibly owing to a recent surge in tourism, erratic climatic variation, and un-scientific construction. This study is an attempt to evaluate the applicability and effectiveness of four probabilistic approaches; Frequency Ratio (FR), Shannon Entropy (SE), Information Value (IV), and Weight-of-Evidence (WoE) in a Geographic Information System (GIS) environment for producing landslide susceptibility maps, to aid local authorities, in finding better-suited land use planning and risk-reduction strategies. The nine selected causative factors: slope, aspect, land use/land cover, elevation, curvature, distance to faults/lineaments, distance to roads, distance to drainage, and lithology along with an updated landslide inventory, compiled from past incidence maps from the Geological Survey of India (GSI) and through visual interpretation of google earth imageries (2001–2019) formed the input for this research work. Three different metrics namely, landslide density index (LDI), relative landslide density index (R_index), and area under curve (AUC) were then used to validate and compare the resulting landslide susceptibility maps. The highest model fitness and predictive ability were demonstrated by the frequency ratio and shannon entropy approaches for this geographical extent.

One of the most destructive natural catastrophes in the globe is flooding. Every year flood causes huge losses of lives and properties. To reduce the losses caused by the flood it is essential to know the resilience of the local inhabitants. In the present study flood resilience has been analysed in the lower catchment area of the Mayurakshi river basin using economic, social, physical, infrastructural and natural indicators. For this study 434 flood affected villages have been selected of the lower catchment area that are frequently affected by the flood. For analysing the flood resilience fuzzy-analytical hierarchical process (F-AHP) has been used. After normalizing and calculating the weight of the variables linear sum up method has been applied for producing the social, economic, physical, infrastructural, natural and overall flood resilience maps. According to the produced maps the overall flood resilience for the villages located in the confluence area are very low. Nearly 19.94% of the selected mouzas have very low flood resilience. So, in the mouzas immediate governmental help is needed to cope up with the flood. The present work will be help full for the planner for formulating the strategies to reduce the effect of flood in the lower catchment area of Mayurakshi river basin.

In this study, post-liquefaction Standard penetration test (SPT) data from the Chi-Chi earthquake was collected and included into a Random Forest (RF) and Support Vector Machine (SVM) model by using a metaheuristic-based optimizer called the Cat Swarm Optimization (CSO). This was done in an effort to improve the accuracy of the projections across a number of different iterations. After that, the data were normalized, and a person correlation matrix and a chi-square test were used to establish the degree of link between the variables. After that, we chose the parameters for the RF and SVM models that would be used for both training and testing by using a random sampling technique. This allowed us to have complete control over the models. It was discovered that CSO not only improved the fitting of the model and the quality of the results, but it also speed up the procedure.