A COMPARISON OF THE EFFECTIVENESS OF CMIP5 MODELS AT SIMULATING LAKE LEVEL CHANGES IN THE WESTERN U.S. DURING THE MID-HOLOCENE

Climate models project a greater likelihood of drought across much of the American southwest in the future due to increased greenhouse gas concentrations, though there is large uncertainty about the possible severity of these extremes. One way to assess the skill of these models is to test their ability to simulate past known hydroclimate change. Middle to late Holocene droughts in the western US are a particularly good target because they were more severe than any observed in the short instrumental record and because they are documented by quantitative paleoclimate records. We chose Lake Tahoe for this comparison given its well-studied current condition and quantitative indicators of lake level changes during the middle Holocene (circa 6,000 years ago). We used data output from eight different CMIP5 (Coupled Model Intercomparison Project) models as input into a lake forward model to quantify lake level changes. The lake forward model has two components: an energy balance model that calculates lake evaporation and a water balance model that uses lake evaporation in conjunction with precipitation, runoff and other hydrologic quantities to calculate lake level. We find that the lake forward model significantly underestimates the severity of drought as indicated by proxy records.