Incorporating probabilistic approach into local multi-criteria decision analysis for flood susceptibility assessment

Flood risk management can be enhanced by integrating geographic information system (GIS) with multi-criteria decision analysis (MCDA). However, the conventional, deterministic MCDA methods ignore uncertainty in the decision-making process and fail to account for local variability in criteria values and preferences. Therefore, a spatially explicit MCDA model which effectively incorporates spatial heterogeneity is required. In this paper, a probabilistic or stochastic MCDA method which incorporates the uncertainty into a local weighted linear combination (WLC) was utilized to evaluate flood susceptibility; and an application case in Gucheng County, Central China, was developed. A GIS database of geomorphological and hydro-meteorological criteria contributing to flood susceptibility analysis was constructed using six conditioning factors: digital elevation model (DEM), slope (SL), maximum three-day precipitation (M3DP), topographic wetness index (TWI), distance from the river (DR), and Soil Conservation Service Curve Number (SCS-CN). The results of local WLC were compared with those of the global WLC. It shows that the local WLC model can provide much more valuable information about the spatial patterns of criterion values, ranges, weights, trade-offs and overall scores, whereas the global WLC can only depict the spatial distribution of criterion values and overall scores. The local WLC can also help to prioritize the most susceptible locations within a neighborhood when navigating the disaster assistance process. Moreover, the uncertainty analysis of criteria weights increases the degree of confidence in the model output. It is concluded that the presented approach can provide more insights and understanding of the nature of the flood susceptibility than global WLC.