GIS-based comparative assessment of flood susceptibility mapping using hybrid multi-criteria decision-making approach, naïve Bayes tree, bivariate statistics and logistic regression: A case of Topľa basin, Slovakia

• Published in: Ecological indicators Vol. 117; p. 106620
• Main Authors: Ali, Sk Ajim, Parvin, Farhana, Pham, Quoc Bao, Vojtek, Matej, Vojteková, Jana, Costache, Romulus, Linh, Nguyen Thi Thuy, Nguyen, Hong Quan, Ahmad, Ateeque, Ghorbani, Mohammad Ali
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-10-2020
• Subjects: Bivariate statistics; Flood susceptibility; Hybrid multi-criteria approach; Machine learning; Topľa river basin, Slovakia, GIS; Hybrid multi-criteria approach; Bivariate statistics; Topľa river basin, Slovakia, GIS; Machine learning; Flood susceptibility
• ISSN: 1470-160X; 1872-7034
• DOI: 10.1016/j.ecolind.2020.106620

[Display omitted]
•Flood susceptibility is assessed using bivariate statistics, MCDSA, and machine learning.•Different physical-geographical factors were integrated and mapped.•The study revealed that very high and high flood susceptibility span on large areas.•The validation of results showed that DEMATELANP was the best model.
Flood is a devastating natural hazard that may cause damage to the environment infrastructure, and society. Hence, identifying the susceptible areas to flood is an important task for every country to prevent such dangerous consequences. The present study developed a framework for identifying flood-prone areas of the Topľa river basin, Slovakia using geographic information system (GIS), multi-criteria decision making approach (MCDMA), bivariate statistics (Frequency Ratio (FR), Statistical Index (SI)) and machine learning (Naïve Bayes Tree (NBT), Logistic Regression (LR)). To reach such a goal, different physical-geographical factors (criteria) were integrated and mapped. To access the relationship and interdependences among the criteria, decision-making trial and evaluation laboratory (DEMATEL) and analytic network process (ANP) were used. Based on the experts’ decisions, the DEMATEL-ANP model was used to compute the relative weights of different criteria and a GIS-based linear combination was performed to derive the susceptibility index. Separately, the flood susceptibility index computation through NBT-FR and NBT-SI hybrid models assumed, in the first stage, the estimation of the weight of each class/category of conditioning factor through SI and FR and the integration of these values in NBT algorithm. The application of LR stand-alone required the calculation of the weights of conditioning factors by analysing their spatial relation with the location of the historical flood events. The study revealed that very high and high flood susceptibility classes covered between 20% and 47% of the study area, respectively. The validation of results, using the past flood points, highlighted that the hybrid DEMATEL-ANP model was the most performant with an Area Under ROC curve higher than 0.97, an accuracy of 0.922 and a value of HSS of 0.844. The presented methodological approach used for the identification of flood susceptible areas can serve as an alternative for the updating of preliminary flood risk assessment based on the EU Floods Directive.