Sensitivity of Mountain Hydroclimate Simulations in Variable-Resolution CESM to Microphysics and Horizontal Resolution
Using an ensemble of variable-resolution CESM simulations across four resolutions (i.e., 55km to 7km) and two versions of the Morrison and Gettelman microphysics scheme, the authors systematically explore the implications of these choices on simulated mountain precipitation, snow cover, snow water equivalent, and two-meter surface temperature in the California Sierra Nevada.Resolution alone does not improve the simulated mountain hydroclimatology. The use of an updated prognostic rainfall and snowfall microphysics parameterization (Morrison and Gettelman, 2015), in combination with high-resolution (i.e., ≤28km), is required to improve the temporal, spatial, and elevational characterization of mountain hydroclimate variables. However, a systemic mountain cold-bias in CESM was found, worsened with higher-resolution, that was also insensitive to the choice of microphysics.
Variable-resolution CESM is a promising cutting-edge tool for global-to-regional downscaling, however, it has not been systematically evaluated across resolution and microphysics. This paper adds value to the literature by showing how model bias arises in CESM with increasing resolution in mountainous regions. It further demonstrates how those biases shape important hydroclimate variables that, in turn, are used for water management decisions.
Mountains of the Western U.S. are important natural stores of freshwater through the capture and storage of atmospheric moisture. However, the ability of mountains to act like natural reservoirs has diminished over the historical record. This diminished capacity is largely due to the decline in mountain snowpack that will likely continue through the 21st century. Climate models are useful tools to understand and anticipate how mountain processes that shape this decline may act. This paper is a first attempt at uncovering how simulated mountain hydroclimatology in variable-resolution CESM responds to previously unattainable resolutions in global climate models (i.e., 28, 14 and 7km) and the choice of microphysics scheme.
Results of this study highlight that increasing model resolution to better capture topographical features is insufficient for improving the mountain hydroclimatology. However, with the use of a cutting-edge microphysics scheme (Morrison and Gettelman, 2015) and increased resolution, mountain windward/leeward distributions nearly matched observations, spatial correlations substantially improved, and, importantly, snow water equivalent bias diminished by 9.4x, 4.9x, and 3.5x from 55 to 7km. However, not all mountain processes improved, and hence further investigation is required. In particular, a mountain cold-bias in two-meter surface temperature persisted and worsened with increasing model resolution throughout much of the water year.