Flood source areas: can we map areas in a catchment contributing to flood peaks?

The high temporal and spatial variability of runoff generation processes makes it difficult to identify runoff source areas (partial or variable source areas), which may contribute to flooding, especially to the flood peak. Several methods have been introduced to model runoff process or map dominant runoff processes. However, no method can map areas contributing to the flood peak. This study introduces and defines flood source areas (FSA), presents the Flood Peak Source Area Index (FSAI) for quantification and comparison, and evaluates the effectiveness of classifying these areas by a new law in Germany, which is supposed to improved flood protection and risk reduction.

The distributed process-based hydrological model RoGeR was used to simulate runoff generation processes like Hortonian overland flow, saturation overland flow, subsurface stormflow, and deep percolation triggering groundwater flow to calculate the FSA. We simulated observed flood-generating rainfall-runoff events and design rainfall events with a 50-year return period, three durations (1h, 6h, 24h), and two initial soil moisture conditions (dry and wet) in six meso-scale catchments in south-west Germany representing the main soil types and geological settings in Germany. The analysis has three steps. For each scenario, the peak discharge period was determined based on the time between the "peak value -10%" before and after the peak. The second step finds source areas for each runoff generation process within the defined peak period based on travel times to the catchment outlet. These defined areas were intersected with RoGeR's spatial runoff generation maps for each runoff component and time step in the third step.  To define the FSAI, we divided the maps of contributing runoff (mm) for each runoff component by the total catchment runoff (mm) during the flood peak period. This is repeated for all runoff components and added to get the quotient of total runoff to the runoff peak volume. Areas with values >1 significantly contribute to flood peak, while those with values < 1 contribute less. With overall a value of one, the entire catchment would contribute equally to the flood peak.

Results show that FSAI > 1 are occurring on 10-60% of the catchment area, depending on event and catchment. On the other hand, 15% to 90% of the catchment area have an FSAI of zero, indicating no flood peak contribution, but this is highly variable by catchment and event characteristics. FSA vary in size and location depending on the event, making them non persistent in space. The FSA patterns vary depending on initial soil moisture, precipitation intensity and duration, spatial distribution, and flood peak shape. Scale dependence matters too. FSAs vary in extent and location depending on the flood hydrograph reference point (catchment outlet). This study found no clear FSA in a watershed to map. FSA can occur anywhere in a catchment, making retention measures to reduce flood risk difficult to establish. But the study also found that roads, urban areas, and wetlands have a disproportionally higher FSAI, indicating their high sensitivity for flood genesis and making runoff reduction in these areas most effective.