Landslides are one of the most devastating and recurring natural disasters and have affected several mountainous regions across the globe. The Indian Himalayan region is no exception to landslide incidences affecting key economic sectors such as transportation and agriculture and often leading to loss of lives. As reflected in the global landslide dataset, most of the landslides in this region are rainfall triggered. The region is prone to 15% of the global rainfall-induced landslides, and thereby a review of the studies in the region is inevitable. The high exposure to landslide risk has made the Indian Himalayas receive growing attention by the landslides community. A review of landslides studies conducted in this region is therefore important to provide a general picture of the state-of-the-art, a reference point for researchers and practitioners working in this region for the first time, and a summary of the improvements most urgently needed to better address landslide hazard research and management. This article focuses on various studies ranging from forecasting and monitoring to hazard and susceptibility analysis. The various factors used to analyze landslide are also studied for various landslide zones in the region. The analysis reveals that there are several avenues where significant research work is needed such as the inclusion of climate change factors or the acquisition of basic data of highest quality to be used as input data for computational models. In addition, the review reveals that, despite the entire region being highly landslide prone, most of the studies have focused on few regions and large areas have been neglected. The aim of the review is to provide a reference for stakeholders and researchers who are currently or looking to work in the Indian Himalayas, to highlight the shortcomings and the points of strength of the research being conducted, and to provide a contribution in addressing the future developments most urgently needed to obtain a consistent advance in landslide risk reduction of the area.

Landslide susceptibility mapping has significantly progressed with improvements in machine learning techniques. However, the inventory/data imbalance (DI) problem remains one of the challenges in this domain. This problem exists as a good quality landslide inventory map, including a complete record of historical data, is difficult or expensive to collect. As such, this can considerably affect one’s ability to obtain a sufficient inventory or representative samples. This research developed a new approach based on generative adversarial networks (GAN) to correct imbalanced landslide datasets. The proposed method was tested at Chukha Dzongkhag, Bhutan, one of the most frequent landslide prone areas in the Himalayan region. The proposed approach was then compared with the standard methods such as the synthetic minority oversampling technique (SMOTE), dense imbalanced sampling, and sparse sampling (i.e., producing non-landslide samples as many as landslide samples). The comparisons were based on five machine learning models, including artificial neural networks (ANN), random forests (RF), decision trees (DT), k-nearest neighbours (kNN), and the support vector machine (SVM). The model evaluation was carried out based on overall accuracy (OA), Kappa Index, F1-score, and area under receiver operating characteristic curves (AUROC). The spatial database was established with a total of 269 landslides and 10 conditioning factors, including altitude, slope, aspect, total curvature, slope length, lithology, distance from the road, distance from the stream, topographic wetness index (TWI), and sediment transport index (STI). The findings of this study have shown that both GAN and SMOTE data balancing approaches have helped to improve the accuracy of machine learning models. According to AUROC, the GAN method was able to boost the models by reaching the maximum accuracy of ANN (0.918), RF (0.933), DT (0.927), kNN (0.878), and SVM (0.907) when default parameters used. With the optimum parameters, all models performed best with GAN at their highest accuracy of ANN (0.927), RF (0.943), DT (0.923) and kNN (0.889), except SVM obtained the highest accuracy of (0.906) with SMOTE. Our finding suggests that RF balanced with GAN can provide the most reasonable criterion for landslide prediction. This research indicates that landslide data balancing may substantially affect the predictive capabilities of machine learning models. Therefore, the issue of DI in the spatial prediction of landslides should not be ignored. Future studies could explore other generative models for landslide data balancing. By using state-of-the-art GAN, the proposed model can be considered in the areas where the data are limited or imbalanced.

Consistently over the years, particularly during monsoon seasons, landslides and related geohazards in Bhutan are causing enormous damage to human lives, property, and road networks. The determination of thresholds for rainfall triggered landslides is one of the most effective methods to develop an early warning system. Such thresholds are determined using a variety of rainfall parameters and have been successfully calculated for various regions of the world at different scales. Such thresholds can be used to forecast landslide events which could help in issuing an alert to civic authorities. A comprehensive study on the determination of rainfall thresholds characterizing landslide events for Bhutan is lacking. This paper focuses on defining event rainfall–duration thresholds for Chukha Dzongkhag, situated in south-west Bhutan. The study area is chosen due to the increase in frequency of landslides during monsoon along Phuentsholing-Thimphu highway, which passes through it and this highway is a major trade route of the country with the rest of the world. The present threshold method revolves around the use of a power law equation to determine event rainfall–duration thresholds. The thresholds have been established using available rainfall and landslide data for 2004–2014. The calculated threshold relationship is fitted to the lower boundary of the rainfall conditions leading to landslides and plotted in logarithmic coordinates. The results show that a rainfall event of 24 h with a cumulated rainfall of 53 mm can cause landslides. Later on, the outcome of antecedent rainfall varying from 3–30 days was also analysed to understand its effect on landslide incidences based on cumulative event rainfall. It is also observed that a minimum 10-day antecedent rainfall of 88 mm and a 20-day antecedent rainfall of 142 mm is required for landslide occurrence in the area. The thresholds presented can be improved with the availability of hourly rainfall data and the addition of more landslide data. These can also be used as an early warning system especially along the Phuentsholing–Thimphu Highway to prevent any disruptions of trade.

Landslides are major natural hazards that have a wide impact on human life, property, and natural environment. This study is intended to provide an improved framework for the assessment of landslide vulnerability mapping (LVM) in Chukha Dzongkhags (district) of Bhutan. Both physical (22 nos.) and social (9 nos.) conditioning factors were considered to model vulnerability using deep learning neural network (DLNN), artificial neural network (ANN) and convolution neural network (CNN) approaches. Selection of the factors was conceded by the collinearity test and information gain ratio. Using Google Earth images, official data, and field inquiry a total of 350 (present and historical) landslides were recorded and training and validation sets were prepared following the 70:30 ratio. Nine LVMs were produced i.e. a landslide susceptibility (LS), one social vulnerability (SV) and a relative vulnerability (RLV) map for each model. The performance of the models was evaluated by area under curve (AUC) of receiver operating characteristics (ROC), relative landslide density index (R-index) and different statistical measures. The combined vulnerability map of social and physical factors using CNN (CNN-RLV) had the highest goodness-of-fit and excellent performance (AUC = 0.921, 0.928) followed by DLNN and ANN models. This approach of combined physical and social factors create an appropriate and more accurate LVM that may—support landslide prediction and management.