Application of InSAR satellite method for mapping of active landslides in Bulgaria – opportunities and perspectives

Landslides are geological phenomena that are spread on Bulgarian territory mainly along the northern Black Sea coast and on the right banks of the Danube in the western part of the country. Mitigation of the negative effects of these destructive geological phenomena is the compilation of inventory maps of their distribution and registers with the main characteristics of the individual landslides. Conventional methods for making such maps are time-consuming and resource-intensive. Modern satellite, air and ground-based remote sensing technologies facilitate the production of landslide maps, reducing the time and resources required to compile and systematically update them. In this paper, we demonstrate the applicability of Differential Sentinel-1A satellite SAR interferometry (DInSAR) to assess the movement activity and use the information for further updating the national landslide inventories in Bulgaria. We perform several analyses based on multi-temporal InSAR techniques of Sentinel-1A data over selected areas prone to landslides. The use of new opportunities for free access to satellite images, which can be applied in conjunction with other methods, greatly facilitates the processes of inventory, mapping and study of landslides.

Effect of the Slope Angle and Its Classification on Landslide

The phase after the determination of the landslide area in landslide susceptibility studies is the selection of methods and parameters to be used. Approximately 1500 randomly selected publications show that it is necessary to select a parameter based on the area. Research has shown that the parameter of slope is greatly preferred. There is nearly consensus of opinion among researchers regarding the use of the parameter. The research included the definition of slope made by different researchers, the advantages and disadvantages of the use of the parameter, different classifications that are used, the formation intervals of landslides, their use together with other parameters, and its effect on the formation of landslides. Classifications were studied based on the slope values at which landslides. Generally, automatic slope classifications are used in the preparation of landslide maps. There isn’t standard in parameter maps. Therefore, there isn’t class range that is referenced when preparing slope maps. In this study, preferred class ranges and slope values where landslides occur were determined in the literature. 40 landslides area has been selected in Turkey. These were evaluated in the slope classes determined according to the literature. The results compared with the literature were found to be compatible.

Landslide Detection Using Multi-Scale Image Segmentation and Different Machine Learning Models in the Higher Himalayas

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.

ASSESSING THE AGREEMENT BETWEEN EO-BASED SEMI-AUTOMATED LANDSLIDE MAPS WITH FUZZY MANUAL LANDSLIDE DELINEATION

Landslide mapping benefits from the ever increasing availability of Earth Observation (EO) data resulting from programmes like the Copernicus Sentinel missions and improved infrastructure for data access. However, there arises the need for improved automated landslide information extraction processes from EO data while the dominant method is still manual delineation. Object-based image analysis (OBIA) provides the means for the fast and efficient extraction of landslide information. To prove its quality, automated results are often compared to manually delineated landslide maps. Although there is awareness of the uncertainties inherent in manual delineations, there is a lack of understanding how they affect the levels of agreement in a direct comparison of OBIA-derived landslide maps and manually derived landslide maps. In order to provide an improved reference, we present a fuzzy approach for the manual delineation of landslides on optical satellite images, thereby making the inherent uncertainties of the delineation explicit. The fuzzy manual delineation and the OBIA classification are compared by accuracy metrics accepted in the remote sensing community. We have tested this approach for high resolution (HR) satellite images of three large landslides in Austria and Italy. We were able to show that the deviation of the OBIA result from the manual delineation can mainly be attributed to the uncertainty inherent in the manual delineation process, a relevant issue for the design of validation processes for OBIA-derived landslide maps.