ALBERT

Description: On 10 October 2011, a rain-on-snow flood occurred in the Bernese Alps, Switzerland, and caused significant damage. As this flood peak was unpredicted by the flood forecast system, questions were raised concerning what has caused this flood and whether it was predictable at all. In this study, we focused on one valley that was heavily hit by the event, the Loetschen valley (160 km2), and aimed to reconstruct the anatomy of this rain-on-snow flood from the synoptic conditions represented by European Centre for Medium-Range Weather Forecasts ECWMF analysis data, and the local meteorology within the valley recorded by an extensive met-station network. In addition, we applied the hydrological model WaSiM-ETH to improve our hydrological process understanding about this event and to demonstrate the predictability of this rain-on-snow flood. We found an atmospheric river bringing moist and warm air to Switzerland that followed an anomalous cold front with sustained snowfall to be central for this rain-on-snow event. Intensive rainfall (average 100 mm day−1) was accompanied by a drastic temperature increase (+8 K) that shifted the zero degree line from 1500 m a.s.l. to 3200 m a.s.l. in 12 h. The northern flank of the valley received significantly more precipitation than the southern flank, leading to an enormous flood in tributaries along the northern flank, while the tributaries along the southern flank remained nearly unchanged. We hypothesized that the reason for this was a cavity circulation combined with a seeder-feeder-cloud system enhancing both local rainfall and snow melt by condensation of the warm, moist air on the snow. Applying and adjusting the hydrological model, we show that both the latent and the sensible heat fluxes were responsible for the flood and that locally large amounts of precipitation (up to 160 mm rainfall in 12 h) was necessary to produce the estimated flood peak. With considerable adjustments to the model and meteorological input data, we were able to reproduce the flood peak, demonstrating the ability of the model to reproduce the flood. However, driving the optimized model with COSMO-2 forecast data, we still failed to simulate the flood precisely because COSMO-2 forecast data underestimated both the local precipitation peak and the temperature increase. Thus, this rain-on-snow flood was predictable, but requires a special model set up and extensive and locally precise meteorological input data, especially in terms of both precipitation and temperature.