No abstract

No abstract

In this work, we present a whole system model of megafloods from catastrophic ice-dam failure in the late Pleistocene that comprises the study of the dynamics of the glacial lake, the propagation of the flood wave downstream of the dam, and an approximation to the ice breach process. The ice-dam incision rate was simply considered an unknown constant, which was varied systematically to best fit the maximum altitude of the simulated water surface and the paleostage indicators in the downstream valley during the transient megaflood. Hence, the hydrograph resulting from the breach of the ice dam was not prescribed but was an output of the paleohydraulic reconstruction. By considering two possible configurations of the breach in the ice dam, i.e. full or partial removal of the ice, we constrained the incision rate in the narrow range of 28 − 42 m•h-1. Two connected glacial lakes, Kuray and Chuja, released 95% of the stored water volume (i.e., 564 km 3) in 33.8 hours. A peak discharge of 10.5 M m 3 •s-1 was required to form numerous giant bars and run-up deposits in the Chuja and Katun valleys. The peak streamflow occurred after 11 hours when 45% of the available lake volume had been evacuated from the Kuray and Chuja basins. Further verification of the reconstructed megaflood was achieved by studying the computed hydraulic conditions during the lake draining that justify the existence and orientation of several fields of subaqueous gravel-dunes in the glacial lake. Complex spatiotemporal patterns during the recession stage of the flood built most of the fields of bedforms. In terms of nondimensional parameters, the Froude and Shields numbers that formed the dune fields were similar to those observed in large sandy rivers, but the flow was undoubtedly unsteady and two-dimensional. We conclude by noting that the extensions of the simulated area cannot be cropped or analysed by independent parts in order to predict the formation of the most relevant geological records due to the unsteady, two-dimensional nature of the flow motion and the development of backwater effects in the drainage network. Lastly, the paleohydrological reconstruction of a megaflood has helped not only to infer the dynamics of the event but also to retrodict the mean parameters of the ice-dam failure mechanism.

We present a basin-scale method to assimilate hydrological data from remote-sensed flood evidence and map civil infrastructures with risk of flooding. As in many rural areas with a semi-arid climate, the studied catchments do not contain stream gauge, and precipitation data does not capture the spatial variability of extreme hydrological events. Remote-sensed flood evidence as slackwater sediments were available at the whole basin, allowing the paleohydrological reconstruction at many sites across the catchment. The agreement between the predicted and observed inundation area was excellent, with an error lower than 15% on average. In addition, the simulated elevations overlapped the observed values in the flooded areas, showing the accuracy of the method. The peak discharges that provoked floods recorded the spatial variability of the precipitation. The variation coefficients of the rainfall intensity were 30% and 40% in the two studied basins with a mean precipitation rate of 3.1 and 4.6 mm/h, respectively. The assumption of spatially uniform precipitation leads to a mean error of 20% in evaluating the local water discharges. Satellite-based rainfall underpredicted the accumulated precipitation by 30–85.5%. Elaborating an inventory of the civil infrastructures at risk was straightforward by comparing the water surface elevation and transport network. The reconstructed maps of rainfall rate were used in the distributed hydrological model IBERPLUS to this end. Recent flood events that overtopped the infrastructures at risk verified our predictions. The proposed research methods can be easily applied and tested in basins with similar physical characteristics around the Mediterranean region.