Flash-Flood Susceptibility Assessment Using Multi-Criteria Decision Making and Machine Learning Supported by Remote Sensing and GIS Techniques

Abstract

Concerning the significant increase in the negative effects of flash-floods worldwide, the main goal of this research is to evaluate the power of the Analytical Hierarchy Process (AHP), fi (kNN), K-Star (KS) algorithms and their ensembles in flash-flood susceptibility mapping. To train the two stand-alone models and their ensembles, for the first stage, the areas affected in the past by torrential phenomena are identified using remote sensing techniques. Approximately 70% of these areas are used as a training data set along with 10 flash-flood predictors. It should be remarked that the remote sensing techniques play a crucial role in obtaining eight out of 10 flash-flood conditioning factors. The predictive capability of predictors is evaluated through the Information Gain Ratio (IGR) method. As expected, the slope angle results in the factor with the highest predictive capability. The application of the AHP model implies the construction of ten pair-wise comparison matrices for calculating the normalized weights of each flash-flood predictor. The computed weights are used as input data in kNN–AHP and KS–AHP ensemble models for calculating the Flash-Flood Potential Index (FFPI). The FFPI also is determined through kNN and KS stand-alone models. The performance of the models is evaluated using statistical metrics (i.e., sensitivity, specificity and accuracy) while the validation of the results is done by constructing the Receiver Operating Characteristics (ROC) Curve and Area Under Curve (AUC) values and by calculating the density of torrential pixels within FFPI classes. Overall, the best performance is obtained by the kNN–AHP ensemble model.

Assessment of landslide susceptibility at a local spatial scale applying the multi-criteria analysis and GIS: a case study from Slovakia

“…On the other hand, the general applicability of multi-criteria-based methodologies (particularly the AHP technique) has been proved in a number of landslide studies, for example, Ayalew et al (2004), Kouli et al (2010), Kayastha et al (2013), Feizizadeh et al (2014, Ahmed (2015), Chen et al (2017), Kumar et al (2018), Mallick et al (2018), Nicu (2018), Bera et al (2019), but also in studies which focused on other types of natural hazards (Skilodimou et al 2019;Santos et al 2019;Costache et al 2020). In addition, application of MCDA across different spatial scales (from local to global) is also appropriate, as evidenced by these studies.…”

Linking the Remote Sensing of Geodiversity and Traits Relevant to Biodiversity—Part II: Geomorphology, Terrain and Surfaces

“…In fact, it was the use of RADAR that revolutionized the monitoring of flash flood hazards [194]. Costache et al [195] conducted research on flash flood susceptibility assessments using multi-criteria decision making and machine learning approaches based on SRTMand GIS techniques. With the open access of the RS time series for Sentinel-1 data these techniques are now widely implemented for flood detection and mapping [192,195,196].…”

“…Costache et al [195] conducted research on flash flood susceptibility assessments using multi-criteria decision making and machine learning approaches based on SRTMand GIS techniques. With the open access of the RS time series for Sentinel-1 data these techniques are now widely implemented for flood detection and mapping [192,195,196]. Such techniques enabled the morphological characterization of the Kyagar glacier and the monitoring of glacier lake outburst floods based on a time series in 2018 Sentinel-1A data [197].…”

Spatial Prediction of Landslide Susceptibility Based on GIS and Discriminant Functions

“…These methods need landslide inventories expressed as landslide density maps to produce functional relationships with causative factors [34]. These methods have also been used in other study fields and exhibit higher spatial accuracy, such as badland susceptibility assessment [35], land subsidence susceptibility mapping [36,37], ground subsidence susceptibility mapping [38,39], flood susceptibility mapping [40][41][42], and groundwater potential mapping [43,44]. Recently, scientists have done a lot of research on hybrid machine learning methods for better assessment and prediction of landslides [45].…”