Thursday, 1 February 2024: 4:30 PM
340 (The Baltimore Convention Center)
Our improved understanding of the drivers responsible for the observed year-to-year variability in flooding can provide the required information to assess future changes in flood hazard. Here we use parsimonious statistical models to show that much of the interannual variability in annual maximum daily discharge can be described in terms of aggregated climate variables (i.e., basin- and season-averaged precipitation and temperature). Instead of modeling the annual maximum time series directly, which would imply that all the flood peaks are coming from the same population, we model the seasonal maxima as a way of capturing different flood generating mechanisms; we then use a Monte Carlo approach to mix the seasonal models to get to the annual maximum time series, moving us toward a process-driven flood frequency analysis. We present results for 7,886 stream gages across the globe with at least 30 complete years of seasonal maximum daily discharge. Despite their simplicity, these models can capture the interannual variability in annual maximum discharge across a wide range of climate regimes; moreover, they enable the assessment of the sensitivity of flood peaks to a broad range of changes in precipitation and temperature.

