Uncertainty of internal climate variability in probabilistic flood simulations using d4PDF

Abstract

Many studies have proved that hydrological extreme val-ues estimated from decadal observation data and river inundation simulations are associated with various uncertain-ties; however, few studies have evaluated the uncertainties associated with internal climate variability. We used large-ensemble river inundation simulations to quantitatively evaluate uncertainties in river depth at the Takahama moni-toring station and flood extent in the Yodo River basin. Using a single 60-year ensemble, the river depth for a 1,000-year return period (RP) flood scale have uncertainty between –11.7% and +9.2% in a 3,000-year flood simula-tion. Thus, the RP of the simulated river depth ranges from 207–3,441 years. To maintain the RP uncertainty within ±300 years would require a simulation of ≥1,200 years. The flood extent uncertainty with an RP of 1,000 years was found to be –8.4% and +7.6% based on a 3,000-year simulation for the lower Yodo River basin. According to this result, the RP of the simulated flood extent ranges from 340–3,060 years. These results suggest that the decadal data used in conventional flood risk analyses potentially contain large uncertainty related to internal climate variability in the RP for water depth and flood extent by approximately 0.3–3-fold.