Late Pleistocene outburst floods from Issyk Kul, Kyrgyzstan?

Elevated shorelines and lake sediments surrounding Issyk Kul, the world's second largest mountain lake, record fluctuating lake levels during Quaternary times. Together with bathymetric and geochemical data, these markers document alternating phases of lake closure and external drainage. The uppermost level of lake sediments requires a former damming of the lake's western outlet through the Boam gorge. We test previous hypothesised ice or landslide dam failures by exploring possible links between late Quaternary lake levels and outbursts. We review and recompile the chronology of reported changes in lake site, and offer new ages of abandoned shorelines using C-14 in bivalve and gastropod shells, and plant detritus, as well as sand lenses in delta and river sediments using Infrared Stimulated Luminescence. Our dates are consistent with elevated lake levels between similar to 45ka and 22 ka. Cosmogenic Be-10 and Al-26 exposure ages of fan terraces containing erratic boulders (> 3 m) downstream of the gorge constrain the timing of floods to 20.5-18.5 ka, postdating a highstand of Issyk Kul. A flow-competence analysis gives a peak discharge of > 104 m(3) s(-1) for entraining and transporting these boulders. Palaeoflood modelling, however, shows that naturally dammed lakes unconnected to Issyk Kul could have produced such high discharges upon sudden emptying. Hence, although our data are consistent with hypotheses of catastrophic outburst floods, average lake-level changes of up to 90 mm yr(-1) in the past 150 years were highly variable without any outbursts, so that linking lake-level drops to catastrophic dam breaks remains ambiguous using sedimentary archives alone. This constraint may readily apply to other Quaternary lakes of that size elsewhere. Nonetheless, our reconstructed Pleistocene floods are among the largest reported worldwide, and motivate further research into the palaeoflood hydrology of Central Asia. Copyright (C) 2017 John Wiley & Sons, Ltd.