Sensitivity of Mountain Hydroclimate Simulations in Variable‐Resolution CESM to Microphysics and Horizontal Resolution

Abstract

Mountains are natural dams that impede atmospheric moisture transport and water towers that cool, condense, and store precipitation. They are essential in the western United States where precipitation is seasonal, and snowpack is needed to meet water demand. With anthropogenic climate change increasingly threatening mountain snowpack, there is a pressing need to better understand the driving climatological processes. However, the coarse resolution typical of modern global climate models renders them largely insufficient for this task, and signals a need for an advanced strategy. This paper continues the assessment of variable‐resolution in the Community Earth System Model (VR‐CESM) in modeling mountain hydroclimatology to understand the role of grid‐spacing at 55, 28, 14, and 7 km and microphysics, specifically the Morrison and Gettelman (, MG1, https://doi.org/10.1175/2008JCLI2105.1) scheme versus the Gettelman and Morrison (, MG2, https://doi.org/10.1175/JCLI-D-14-00102.1) scheme. Eight VR‐CESM simulations were performed from 1999 to 2015 with the F_AMIP_CAM5 component set, which couples the atmosphere‐land models and prescribes ocean data. Refining horizontal grid‐spacing from 28 to 7 km with the MG1 scheme did not improve the simulated mountain hydroclimatology. Substantial improvements occurred with the use of MG2 at grid‐spacings ≤28 km compared to MG1 as shown with subsequent statistics. Average SWE bias diminished by 9.4X, 4.9X, and 3.5X from 55 to 7 km. The range in minimum (maximum) DJF spatial correlations increased by 0.1–0.2 in both precipitation and SWE. Mountain windward/leeward distributions and elevation profiles improved across hydroclimate variables, however not always with model resolution alone. Disconcertingly, all VR‐CESM simulations exhibited a systemic mountain cold bias that worsened with elevation and will require further examination.

Impact of Atmospheric Rivers on Surface Hydrological Processes in Western U.S. Watersheds

“…Also, this lack of prognostic rainfall/snowfall could lead to an overestimation bias on the windward sides of the mountain ranges, particularly snowfall (even at 6‐km grids). Advection of droplets and snowflakes may have important effects on the fine‐scale distribution of snowpack for very high resolution simulations (Rhoades et al, ). Since our analysis is done at an aggregated HUC8 watershed scale, our results may not be very sensitive to cross‐grid advection of precipitation.…”

The Changing Character of the California Sierra Nevada as a Natural Reservoir

“…Although the NA‐CORDEX ensemble only provides simulation data at 25‐ and 50‐km resolutions, coarser than generally preferred for mountain snowpack products (Ikeda et al, ; Letcher & Minder, ; Pavelsky et al, ; Wrzesien et al, ; Wrzesien et al, ), it is nonetheless a significant improvement over other multimodel ensembles such as the Coupled Model Intercomparison Project phase 5 (CMIP5) GCM ensemble. Further, it is at a sufficiently high resolution to still provide value for snowpack assessment when factoring in the important trade‐offs between model resolution, subgrid‐scale parameterizations, and global forcing data set (Rhoades, Ullrich, Zarzycki, et al, ; Xu et al, ).…”

Exploring a Variable‐Resolution Approach for Simulating Regional Climate Over the Tibetan Plateau Using VR‐CESM

“…This is an inappropriate assumption over areas for which the timescale of hydrometeor fallout is longer than the timescale for horizontal transport of hydrometeors across the model grid. Rhoades et al () found that precipitation biases over the California Mountainous Region are significantly improved in their VR‐CESM simulations with prognostic rain and snow treatment (Gettelman & Morrison, ), especially at model resolutions finer than 0.25°.…”

“…Other studies have found their way around the resolution-computation conundrum by implementing variable resolution (VR) models to simulate climate phenomena (e.g., Fox-Rabinovitz et al, 2006;Harris & Lin, 2012, 2014Hourdin et al, 2006;McGregor, 1996;Sabin et al, 2013;Sakaguchi et al, 2015;Zhou & Li, 2002). Specifically, VR versions of CESM have been used to study tropical cyclones (Zarzycki et al, 2013;Zarzycki & Jablonowski, 2014), snowpack, and hydroclimate in the western United States Rhoades et al, 2016Rhoades et al, , 2018Wu et al, 2017Wu et al, , 2018Zarzycki et al, 2015) and to examine their capability for regional climate simulations (Gettelman et al, 2017). This process generally makes use of a "regional refinement" of the horizontal grid to very high horizontal resolutions of a domain of interest, which coarsens to large horizontal grid spacings away from the domain of interest.…”