Towards flood risk mapping based on multi-tiered decision making in a densely urbanized metropolitan city of Istanbul

Abstract

• The AHP and ANP were performed to assess hazard and vulnerability, respectively. • DEMATEL technique was integrated to define the ANP structure for criteria assessment. • Focus group discussions were conducted at each level of the framework refinement. • Income level and education level have highest weights in vulnerability cluster. • Return period of a storm event is the most significant criteria in hazard cluster. Multi-dimensional risk assessment strategies aided by the active participation of stakeholders are essential to combat flood events for metropolitan cities. This study aims to prioritize the districts of Istanbul regarding flood hazard, flood vulnerability, and flood risk with a multi-tiered comprehensive decision-making procedure. The proposed framework was enriched by focus group discussion, which was conducted through the professionals dealing with flood risk management in Istanbul. Analytical hierarchy process (AHP) was used to evaluate hazard criteria while the vulnerability criteria were subjected to the analytical network process (ANP). The structure of the ANP was adjusted by adopting the decision-making trial and evaluation laboratory (DEMATEL) technique. VIKOR (Visekriterijumska Optimizacija I Kompromisno Resenje) method was employed to obtain district scores. Findings highlighted the income level, education level, and population density criteria regarding vulnerability cluster, while the return period of a storm event, storm water pipe network and imperviousness were the most significant hazard criteria. Sensitivity analysis of the VIKOR method also showed the robustness of the adopted methodology and the stability of the computed results under diverging conditions. The proposed generic framework can be replicated in other densely urbanized metropolitan cities and corresponding recommendations could assist decision-makers in achieving sustainable development goals.