Extreme flooding controlled by stream network organization and flow regime

Abstract River floods are among the most common natural disasters worldwide, with substantial economic and humanitarian costs. Despite enormous efforts, gauging risk of extreme unprecedented magnitude is an outstanding challenge. Limited observational data from very high-magnitude flood events hinders prediction efforts identification discharge thresholds marking rise progressively larger floods, termed divides. Combining long hydroclimatic records a process-based model for hazard assessment, here we demonstrate that spatial organization stream networks river flow regime control appearance divides floods. In contrast their ubiquitous attribution to rainfall anomalous antecedent conditions, show propensity generate well predicted by intrinsic properties basins. Most importantly, it can be assessed prior occurrence catastrophes through measurable metrics these derived commonly available data, namely hydrograph recession exponent coefficient variation daily flows. These results highlight certain rivers generating importance using mapping tools that, rather than solely relying on past records, identify regions susceptible ordinary dynamics.