Influences of North Pacific Ocean Domain Extent on the Western US Winter Hydroclimatology in Variable-Resolution CESM
Few studies have assessed how the choice in the location and extent of the refinement domain in variable-resolution enabled Earth systems models may influence model fidelity in representing the hydrologic cycle, particularly in the western United States. We address this gap by developing a series of variable-resolution grids with fixed refinement resolution (28km) and latitudinal extent, yet different longitudinal extents over the North Pacific and assess the influence of refinement domain size on both upstream (e.g., atmospheric river genesis and transport) and downstream (e.g., topographic resolution implications for precipitation magnitude and phase) hydrologic cycle processes of the western United States.
Isolating Earth system model fidelity in representing the hydrologic cycle across resolutions has important implications for questions surrounding the scale-awareness of sub-grid-scale physics parameterizations (e.g., convection and microphysics) and implications for the interpretability of future changes in the hydrologic cycle for water resource management (e.g., changes in storm behavior such as precipitation magnitude and phase). More practically, this work also shows that a core-hour saving of ~30% can occur when running the Earth system model with a refinement domain that is smaller yet comparable in historical hydroclimate representation skill than a much larger refinement domain.
Variable-resolution global climate models (VRGCMs) are a dynamical downscaling method that can reach spatiotemporal scales needed for regional climate assessments. Over the years, several users of VRGCMs have assumed where the location and extent of the refinement domain should be based on knowledge of the prevailing storm tracks and resolution dependence of important regional climate processes (e.g., atmospheric rivers [ARs] and orographic uplift), but the effect of high resolution domain size and extent on the simulation of downstream hydroclimatic phenomena has not been systematically evaluated. Here, we use variable-resolution in the Community Earth System Model (VR-CESM) to perform such a test. To do this, three VR-CESM grids were generated that span the entire, two-thirds, and one-third of the North Pacific and evaluated for a 30-year climatology using Atmospheric Model Intercomparison Project protocols. Simulations are compared with reanalysis products offshore (ERA5) and onshore (Livneh, 2015 and Parameter-elevation Regressions on Independent Slopes Model [PRISM]) of the western US. The westward expansion of refinement domain influenced integrated vapor transport (IVT), which was generally high-biased, but minimally impacted AR characteristics. Due to slight differences in landfalling AR counts in the western US, California winter precipitation generally improved with westward expansion of the refinement domain. Western US mountain snowpack and surface temperatures were insensitive to refinement domain size and were more influenced by changes in topographic resolution and/or land-surface model version. Given minimal dependence of simulated western US hydroclimate on refinement domain size over the North Pacific we advise future VR-CESM studies to focus grid resolution on better resolving land-surface heterogeneity.