Hydrological analysis of a flash flood across a climatic and geologic gradient: The September 18, 2007 event in Western Slovenia

A Mesoscale Convective System in North-Western Slovenia produced up to 350-400 mm in 12 h on 18 September, 2007. The region impacted by the storm shows significant differences in climatic and geologic properties at short distances. Owing to such variability, extreme flooding concentrated over the Selska Sora watershed at Železniki (103.3 km2), outside the area which received the highest precipitation. Hydrometeorological analyses of the storm are based on accurate analysis of C-band weather-radar observations and data from a rain gauge network. Detailed surveys of high-water marks and channel/floodplain geometry, carried out two months after the flood, are used for hydrologic analyses of the Selska Sora flood. These include estimation of peak discharge at 21 sites. Unit peak discharges range from 5 to 7 m3 s−1 km−2 in basins characterised by size up to approximately 25 km2. Higher unit peak discharges (>10 m3 s−1 km−2), estimated in a few smaller basins, are influenced by intense sediment transport. Observed rainfall, estimated peak discharges, and observer notes on timing of peak discharge are used along with a distributed hydrologic model to reconstruct hydrographs at multiple locations. Examination of the rainfall distribution and flood response shows that the extent and the position of the karst terrain provided a major control on flood response in the region impacted by the storm. Use of the distributed hydrological model together with the post-flood survey observations is shown to provide an accurate description of the flood. Water balance and response time characteristics are examined for selected catchments, showing that event runoff coefficient ranged between 17% and 24% for different catchments. The quality of the peak discharge simulation at the 21 surveyed sites is substantially degraded when using spatially-uniform rainfall over the area covering all the surveyed sub-catchments, mainly due to rainfall volume errors introduced by using the spatially uniform value. On the other hand, the influence of rainfall spatial averaging at the scale of the sub-catchments generally has a very limited effect on runoff modelling, showing that rainfall spatial organisation was not able to overcome the catchment dampening effect for this flood.