A probabilistic analysis of compound flooding in river confluences under different hydroclimatic and geospatial conditions.

River flooding impacts more people globally than any other natural disaster. River confluences are key nodes of a river network, characterized by complex hydrodynamic conditions. Hence, a general framework to investigate the sensitivity of confluences to extreme peak flows upstream, climate, and geomorphological characteristics is of great importance. Currently, a systematic large-scale investigation of peak flows of main river upstream and tributary and their interactions is missing in the literature. Here we analyse upstream and downstream peak flows and their relative occurrence in 153 catchments (51 confluences) across the world.  The results indicate that the time lag between the upstream and downstream discharge can be explained by their seasonality of the climate zone. This leads to different patterns of flood events in confluences, which can occur even when upstream discharges are not hazardous per se. The probabilistic characterization of the co-occurrence of peak discharge downstream of the confluence for different conditions upstream shows that the co-occurrence of extreme discharges upstream and downstream is not the most likely scenario, whilst the probability of peak discharge downstream with corresponding moderate discharges upstream cannot be neglected. These outcomes provide a general framework for improving the flood resilience in the proximity of the confluence.