Flood scaling theory is important for flood predictions in data-scarce regions but is often applied to quantile-based floods that have no physical mechanisms. In this study, we propose a framework for flood prediction in data-scarce regions by event-based flood scaling. After analyzing the factors controlling the flood scaling, flood events are first simulated by a hydrological model with different areally averaged rainfall events and curve number (CN) values as inputs, and the peak discharge of each subcatchment is obtained. Then, the flood scaling is analyzed according to the simulated peak discharge and subcatchment area. Accordingly, the relationship curves between the scaling exponent and the two explanatory factors (rainfall intensity and CN) can be drawn. Assuming that the flood and the corresponding rainfall event have the same frequency, the scaling exponent with a specific flood frequency can be interpolated from these curves.
Hydrological modeling was used to conduct flood scaling analysis.
Rainfall intensity and curve number were identified to be the two main factors affecting flood scaling.
A framework of flood prediction by event-based flood scaling in data-scarce regions was proposed.