Assessing future changes in flood susceptibility under projections from the sixth coupled model intercomparison project: case study of Algiers City (Algeria)

This research investigates the changes in flash flood susceptibility in Algiers, Northern Algeria between current and future climatic conditions based on two Shared Socio-economic Pathways (SSP2-4.5 and SSP3-7.0) from the CMIP6 dataset. Three machine-learning models, namely the Generalized Linear Model (GLM), Random Forest (RF) and Gradient Boosting Machine (GBM), were employed to assess flash flood susceptibility by capturing the relationships between a set of predictive variables and historical flash flood events in the study area. The validity of the used models was assessed using the receiver operating characteristic (ROC) model and its area under the curve (AUC). This yielded excellent performance for all models with a slight superiority to GBM (AUC = 96.4%) compared to RF (AUC = 96.1%) and GLM (AUC = 93.9%). With respct to the year 2018, SSP 2–4.5 revealed a future evolution of high to very high flash flood susceptibility of + 2.9% by the year 2040, + 1.6% by 2060 and + 5.1% by 2080. Under SSP3-7.0, the spatial coverage of high and very high susceptibility classes showed more significant increase of 3.6% by 2040, + 4.9% by 2060, and + 4.7% by 2080. Overall, this research provided insights into the changes in flash flood susceptibility between current and two future climate change scenarios. This can help decision makers and urban planners in Algiers in developing adequate strategies to improve resilience against future flash floods.