Quantitative model-data comparison of mid-Holocene lake-level change in the central Rocky Mountains

Abstract

Recently-developed Holocene lake-level reconstructions from the Rocky Mountains offer a quantitative target for testing the skill of state-of-the-art climate system models in simulating hydroclimate change. Here, we use a combination of hydrologic models of catchment streamflow, lake energy balance, and lake water balance to simulate lake level at Little Windy Hill Pond (LWH) in the Medicine Bow Range of Wyoming for a period of severe drought during the mid-Holocene (MH; approximately 6000 years ago). Using Coupled Model Intercomparison Project (CMIP5) output to drive our hydrologic models, we find that none of our simulations reproduce the significantly lowered lake levels at LWH during the MH. Rather, simulated hydroclimate changes for the MH are modest (< 10% reductions in precipitation and streamflow and generally 10–30% increases in lake evaporation), and LWH lake-level changes are buffered by the large volume of snowmelt runoff that the lake receives. Only when winter precipitation is approximately halved in sensitivity experiments do water inputs to the lake become small enough that lake level can be significantly drawn down by year-over-year negative water balances. Possible explanations for the model-data mismatch could lie in the realism of our hydrological modeling framework or in the accuracy of the CMIP5 output, the latter having important implications for projections of future drying in western North America.