Patterns of extreme high flows in relation to their dominant generating processes across Sweden

Understanding the nature of flooding of a region is key for flood management. The impact of flooding depends on how spatially extensive it is and this, in turn, is influenced by the processes generating the flood. In this study, we investigated the relative importance of rainfall and snowmelt in the generation of floods of different magnitudes and characterized their spatial patterns in different climate regions of Sweden. We generated a large number of spatially diverse extreme river flow scenarios across Sweden that are statistically consistent with the observations by employing a multi-site weather generator and a highly resolved semi-distributed hydrological model. The extreme flows within each of the main rivers were classified based on their generating meteorological forcing and the spatial distribution of the flow magnitudes was assessed. The results reveal that rainfall is the main contributor of extreme flows of all magnitudes in the southern part followed by rain-on-snow, while in the northern part, rain-on-snow is the main process resulting in extreme flows followed by rainfall. Pure snowmelt is the least contributor of extremes in all regions and its contribution decreases with increasing magnitude of the flow. The proportion of events generated by rainfall increases with the magnitude of the flow in all regions. Extremes of lower magnitudes are generally more spatially widespread than the higher extremes and events generated by snowmelt and rain-on-snow are spatially more widespread than events generated by rainfall.

The possible impact of climate change was also assessed by generating extreme flows for end-of-century climate change scenarios by perturbing the weather inputs generated by the weather generator using data from a set of regional climate models and using them to force the hydrologic model. The results show that the main generating processes in each region remain the same. However, the proportion of rainfall generated events will be markedly higher than under the present climate.