Understanding the role of initial soil moisture and precipitation magnitude in flood forecast using a hydrometeorological modelling system

We adapted the WRF-Hydro modelling system to Hurricane Florence (2018) and performed a series of diagnostic experiments to assess the influence of initial soil moisture and precipitation magnitude on flood simulation over the Cape Fear River basin in the United States. Model results suggest that: (1) The modulation effect of initial soil moisture on the flood peak is non-linear and weakens as precipitation magnitude increases. There is a threshold value of the soil saturation, below and above which the sensitivity of flood peak to the soil moisture differentiates substantially; (2) For model spin-up, streamflow needs longer time to reach the 'practical' equilibrium (10%) than the soil moisture and latent heat flux. The model uncertainty from spin-up can propagate through the hydrometeorological modelling chain and get amplified into the flood peak; (3) For ensemble flood modelling with a hydrometeorological system, modelling uncertainty is dominated by the precipitation forecast. Spin-up induced uncertainty can be minimized once the model reaches the 'practical' equilibrium.

Automated summary

In this study, the WRF-Hydro modelling system was used to simulate the flood simulation of Hurricane Florence in the Cape Fear River basin in the United States. A series of diagnostic experiments were conducted to assess the influence of initial soil moisture and precipitation magnitude on flood simulation. The results showed that the modulation effect of initial soil moisture on the flood peak is non-linear and weakens as precipitation magnitude increases. Model spin-up process contributes to the model uncertainty, which can be minimized once the model reaches the 'practical' equilibrium.