There is a growing demand for detailed and accurate landslide maps and inventories around the globe, but particularly in hazard-prone regions such as the Himalayas. Most standard mapping methods require expert knowledge, supervision and fieldwork. In this study, we use optical data from the Rapid Eye satellite and topographic factors to analyze the potential of machine learning methods, i.e., artificial neural network (ANN), support vector machines (SVM) and random forest (RF), and different deep-learning convolution neural networks (CNNs) for landslide detection. We use two training zones and one test zone to independently evaluate the performance of different methods in the highly landslide-prone Rasuwa district in Nepal. Twenty different maps are created using ANN, SVM and RF and different CNN instantiations and are compared against the results of extensive fieldwork through a mean intersection-over-union (mIOU) and other common metrics. This accuracy assessment yields the best result of 78.26% mIOU for a small window size CNN, which uses spectral information only. The additional information from a 5 m digital elevation model helps to discriminate between human settlements and landslides but does not improve the overall classification accuracy. CNNs do not automatically outperform ANN, SVM and RF, although this is sometimes claimed. Rather, the performance of CNNs strongly depends on their design, i.e., layer depth, input window sizes and training strategies. Here, we conclude that the CNN method is still in its infancy as most researchers will either use predefined parameters in solutions like Google TensorFlow or will apply different settings in a trial-and-error manner. Nevertheless, deep-learning can improve landslide mapping in the future if the effects of the different designs are better understood, enough training samples exist, and the effects of augmentation strategies to artificially increase the number of existing samples are better understood.

Event-based landslide inventories are essential sources to broaden our understanding of the causal relationship between triggering events and the occurring landslides. Moreover, detailed inventories are crucial for the succeeding phases of landslide risk studies like susceptibility and hazard assessment. The openly available inventories differ in the quality and completeness levels. Event-based landslide inventories are created based on manual interpretation, and there can be significant differences in the mapping preferences among interpreters. To address this issue, we used two different datasets to analyze the potential of U-Net and machine learning approaches for automated landslide detection in the Himalayas. Dataset-1 is composed of five optical bands from the RapidEye satellite imagery. Dataset-2 is composed of the RapidEye optical data, and ALOS-PALSAR derived topographical data. We used a small dataset consisting of 239 samples acquired from several training zones and one testing zone to evaluate our models’ performance using the fully convolutional U-Net model, Support Vector Machines (SVM), K-Nearest Neighbor, and the Random Forest (RF). We created thirty-two different maps to evaluate and understand the implications of different sample patch sizes and their effect on the accuracy of landslide detection in the study area. The results were then compared against the manually interpreted inventory compiled using fieldwork and visual interpretation of the RapidEye satellite image. We used accuracy assessment metrics such as F1-score, Precision, Recall, and Mathews Correlation Coefficient (MCC). In the context of the Nepali Himalayas, employing RapidEye images and machine learning models, a viable patch size was investigated. The U-Net model trained with 128 × 128 pixel patch size yields the best MCC results (76.59%) with the dataset-1. The added information from the digital elevation model benefited the overall detection of landslides. However, it does not improve the model’s overall accuracy but helps differentiate human settlement areas and river sand bars. In this study, the U-Net achieved slightly better results than other machine learning approaches. Although it can depend on architecture of the U-Net model and the complexity of the geographical features in the imagery, the U-Net model is still preliminary in the domain of landslide detection. There is very little literature available related to the use of U-Net for landslide detection. This study is one of the first efforts of using U-Net for landslide detection in the Himalayas. Nevertheless, U-Net has the potential to improve further automated landslide detection in the future for varied topographical and geomorphological scenes.

Accurate early warning systems for landslides are a reliable risk-reduction strategy that may significantly reduce fatalities and economic losses. Several machine learning methods have been examined for this purpose, underlying deep learning (DL) models’ remarkable prediction capabilities. The long short-term memory (LSTM) and gated recurrent unit (GRU) algorithms are the sole DL model studied in the extant comparisons. However, several other DL algorithms are suitable for time series forecasting tasks. In this paper, we assess, compare, and describe seven DL methods for forecasting future landslide displacement: multi-layer perception (MLP), LSTM, GRU, 1D convolutional neural network (1D CNN), 2xLSTM, bidirectional LSTM (bi-LSTM), and an architecture composed of 1D CNN and LSTM (Conv-LSTM). The investigation focuses on four landslides with different geographic locations, geological settings, time step dimensions, and measurement instruments. Two landslides are located in an artificial reservoir context, while the displacement of the other two is influenced just by rainfall. The results reveal that the MLP, GRU, and LSTM models can make reliable predictions in all four scenarios, while the Conv-LSTM model outperforms the others in the Baishuihe landslide, where the landslide is highly seasonal. No evident performance differences were found for landslides inside artificial reservoirs rather than outside. Furthermore, the research shows that MLP is better adapted to forecast the highest displacement peaks, while LSTM and GRU are better suited to model lower displacement peaks. We believe the findings of this research will serve as a precious aid when implementing a DL-based landslide early warning system (LEWS).

Landslides represent a severe hazard in many areas of the world. Accurate landslide maps are needed to document the occurrence and extent of landslides and to investigate their distribution, types, and the pattern of slope failures. Landslide maps are also crucial for determining landslide susceptibility and risk. Satellite data have been widely used for such investigations—next to data from airborne or unmanned aerial vehicle (UAV)-borne campaigns and Digital Elevation Models (DEMs). We have developed a methodology that incorporates object-based image analysis (OBIA) with three machine learning (ML) methods, namely, the multilayer perceptron neural network (MLP-NN) and random forest (RF), for landslide detection. We identified the optimal scale parameters (SP) and used them for multi-scale segmentation and further analysis. We evaluated the resulting objects using the object pureness index (OPI), object matching index (OMI), and object fitness index (OFI) measures. We then applied two different methods to optimize the landslide detection task: (a) an ensemble method of stacking that combines the different ML methods for improving the performance, and (b) Dempster–Shafer theory (DST), to combine the multi-scale segmentation and classification results. Through the combination of three ML methods and the multi-scale approach, the framework enhanced landslide detection when it was tested for detecting earthquake-triggered landslides in Rasuwa district, Nepal. PlanetScope optical satellite images and a DEM were used, along with the derived landslide conditioning factors. Different accuracy assessment measures were used to compare the results against a field-based landslide inventory. All ML methods yielded the highest overall accuracies ranging from 83.3% to 87.2% when using objects with the optimal SP compared to other SPs. However, applying DST to combine the multi-scale results of each ML method significantly increased the overall accuracies to almost 90%. Overall, the integration of OBIA with ML methods resulted in appropriate landslide detections, but using the optimal SP and ML method is crucial for success.

Rainfall-induced landslide inventories can be compiled using remote sensing and topographical data, gathered using either traditional or semi-automatic supervised methods. In this study, we used the PlanetScope imagery and deep learning convolution neural networks (CNNs) to map the 2018 rainfall-induced landslides in the Kodagu district of Karnataka state in the Western Ghats of India. We used a fourfold cross-validation (CV) to select the training and testing data to remove any random results of the model. Topographic slope data was used as auxiliary information to increase the performance of the model. The resulting landslide inventory map, created using the slope data with the spectral information, reduces the false positives, which helps to distinguish the landslide areas from other similar features such as barren lands and riverbeds. However, while including the slope data did not increase the true positives, the overall accuracy was higher compared to using only spectral information to train the model. The mean accuracies of correctly classified landslide values were 65.5% when using only optical data, which increased to 78% with the use of slope data. The methodology presented in this research can be applied in other landslide-prone regions, and the results can be used to support hazard mitigation in landslide-prone regions.

Slope failures occur when parts of a slope collapse abruptly under the influence of gravity, often triggered by a rainfall event or earthquake. The resulting slope failures often cause problems in mountainous or hilly regions, and the detection of slope failure is therefore an important topic for research. Most of the methods currently used for mapping and modelling slope failures rely on classification algorithms or feature extraction, but the spatial complexity of slope failures, the uncertainties inherent in expert knowledge, and problems in transferability, all combine to inhibit slope failure detection. In an attempt to overcome some of these problems we have analyzed the potential of deep learning convolutional neural networks (CNNs) for slope failure detection, in an area along a road section in the northern Himalayas, India. We used optical data from unmanned aerial vehicles (UAVs) over two separate study areas. Different CNN designs were used to produce eight different slope failure distribution maps, which were then compared with manually extracted slope failure polygons using different accuracy assessment metrics such as the precision, F-score, and mean intersection-over-union (mIOU). A slope failure inventory data set was produced for each of the study areas using a frequency-area distribution (FAD). The CNN approach that was found to perform best (precision accuracy assessment of almost 90% precision, F-score 85%, mIOU 74%) was one that used a window size of 64 × 64 pixels for the sample patches, and included slope data as an additional input layer. The additional information from the slope data helped to discriminate between slope failure areas and roads, which had similar spectral characteristics in the optical imagery. We concluded that the effectiveness of CNNs for slope failure detection was strongly dependent on their design (i.e., the window size selected for the sample patch, the data used, and the training strategies), but that CNNs are currently only designed by trial and error. While CNNs can be powerful tools, such trial and error strategies make it difficult to explain why a particular pooling or layer numbering works better than any other.