Variability in river channel conveyance reshapes global flood hazard and exposure

Global flood hazard models are central to assessing flood risk, informing adaptation planning, and interpreting the impacts of climate change on hydrological extremes. However, these models rely on structural assumptions that introduce substantial but poorly quantified uncertainty. One such assumption is that bankfull discharge corresponds to a fixed two-year return period, effectively prescribing a uniform river channel conveyance capacity globally and shaping discharge-inundation relationships across flood magnitudes.

Here, we quantify how uncertainty in river channel conveyance propagates through global flood hazard and population exposure estimates and assess its magnitude relative to climate-driver changes in discharge. Using global bankfull discharge estimates from a geomorphological-hydrological modelling framework, we derive change factors that adjust discharge-inundation relationships within a global flood hazard model. This enables adjusted estimates of flood hazard and exposure that reflect regional channel-floodplain interactions rather than a uniform global assumption.

Accounting for bankfull variability leads to systematic and spatially coherent changes in global flood exposure. Sub-Saharan Africa shows a robust net increase in exposure across return periods, including a 1.4 million increase (11%) for the 20-year flood. In contrast, several large river basins exhibit net reductions in exposure, such as a 7.5 million decrease across the Ganges-Brahmaputra basin. However. This basin-scale signal masks substantial internal variability, with Bangladesh seeing a net increase in exposure of 10% for the 20-year flood.

We compare these effects with climate-driven changes in flood hazard and assess how the uncertainty in river channel conveyance compares with the magnitude of projected climate signals. These results highlight how structural uncertainties in global flood hazard models cascade into risk assessments, with important implications for interpreting present day flood exposure and future climate impacts.