Quantifying the predictability of the timing of jokulhlaups from Merzbacher Lake, Kyrgyzstan

Glacier-dammed lakes can yield subglacial outburst floods (jökulhlaups) repeatedly. Predicting flood timing is crucial for hazard mitigation, but incomplete understanding of flood-initiation physics makes this challenging. Here we examine the predictability of the timing of jökulhlaups from Merzbacher Lake, Kyrgyzstan, using five flood-date prediction models of varying complexity. The simplest model, which offers a benchmark against which the other models are compared, assumes that floods occur on the same date each year. The other four models predict flood dates using a flood-initiation threshold approach and incorporate weather forcing (approximated by the output of two climate reanalyses) behind the meltwater input to the lake; the most complex of these models accounts for a moving subglacial water divide beneath the glacier that dams the lake. Each model is optimized against recorded flood dates to maximize its prediction ability. In terms of their flood prediction ability, our two best models are those that assume a variable outburst threshold governed by the rate of meltwater input to the lake and the rate of lake-level rise. They excel over the simplest and most complex models and correctly predict flood dates to within ±20 days 57.4% of the time. We also quantify the impact of weather uncertainty on prediction success. Our findings can inform practical flood-forecasting schemes and future investigations of flood-initiation physics.