Impact of rainfall pattern on the occurrence of flash floods in Hungary

Despite the country's lowland character, flash floods have frequently been reported from the hilly and low mountain catchments of Hungary (e.g.: VASS 1997; HORVATH 2005; SZLAVIK & KLING 2007). To avoid flash-flood damage, development of a flash flood guidance (FFG) system was Suggested by the concerned authorities. To generate a rapid screening based nationwide flash flood hazard mapping as part of the FFG system, in the present study we considered and included both passive environmental factors (PFB model), and randomly, occurring active environmental Factors (precipitation). Currently, we determined the (i) spatial and (ii) the temporal patterns and behavior of rainfall events in Hungary and prepared a hazard map that is based on both passive and active environmental factors (APFB model). Extreme rainfall pattern was analyzed both in time and space using data provided by the National Oceanic and Atmospheric Administration (NOAA) and the Hungarian Meteorological Services (HMS). The hazard map was then validated using HEC-HMS simulations and calculated threshold discharge values. The spatial pattern of extreme events indicates a strong orographic effect. The frequency of extreme precipitation is always higher on the watersheds of the PFB model with highest hazard, than in both the zone of the PFB model with the lowest hazard and the national average. Consequently, the vulnerability of the passive factor-based most endangered areas will be further enhanced by the the higher frequency of extreme events. Temporal changes are ambiguous based on the long term observed data of extreme rainfall. Model indicated increase of extreme rainfall events is not clearly shown based on the employed HMS and NOAA data. However, a fluctuating cyclic pattern is clearly identifiable. The majority of the observed years was characterized with a lower-than-average number of extreme events. However, these long periods are interrupted with short spells (usually lasting a year or two) when the number of extreme events significantly exceeded the mean value. The temporal frequency of flash flood events is lower than expected based on model validation processes. This discrepancy is likely to be explained by (i) the duration difference between the simulated rainfall period and the observed rainfall period, (ii) the overestimation of total runoff by the runoff model, (iii) the fact not all flash flood-generated damages were reported to the insurance companies and (iv) the inappropriate calculation of the discharge threshold values.