Mapping soil vulnerability to floods under varying land use and climate

Floods may only last few hours and can cause considerable damage and a possible threat to life. Flood prediction requires quantitative knowledge about infiltration and runoff dynamics, which is generally gained at the local scale. When scaling up local measurements to the catchment scale, account needs to be taken off the catchment’s organization (connectivity and patchiness). For this purpose, we developed a new method to map soil vulnerability to floods based on two steps: (1) identification of the flow processes at the plot scale, and (2) up-scaling this knowledge to the catchment scale. Excess surface runoff was scaled up by means of terrain analysis using digital elevation models (TauDEMs) calibrated with in situ sprinkling experiments of three rainfall-simulation intensities carried out on 57 plots under grassland and forest that dominate in the investigated area. The marked differences in textural and structural porosities between forest and grassland plots appear to control runoff processes. On the one hand, forest soil has a higher storage capacity than grassland soil, probably caused by a high unsaturated hydraulic conductivity and root water uptake, and resulting in lower surface runoff. On the other hand, fine material in the topmost 10 cm of grassland soil helps to build a structure that impedes vertical downward percolation and thus enhances surface runoff. However, within each soil category, slope plays an important role in generating surface runoff. In addition, raising the rainfall-simulation intensity from 24 to 48 mm h−1 increases the risk of predisposition to surface runoff from middle to high in major parts of the catchment under grassland, whereas forest soils showed vertical percolation in all cases except on slopes steeper than 31.3 degrees. Scaling up runoff processes using TauDEM based on sprinkling experiments provided new quantitative insights into flow processes and enabled us to trace the hydrological connectivity between zones of various predispositions to excess surface runoff under different land uses. These promising results indicate that the approach is suited for mapping soil vulnerability to floods under varying land use and climate at any scale. Our study showed that the peak discharge can be significantly reduced if the succession and connectivity of land use are well planned.