Assessing the Sensitivity of Simulated East Coast Winter Storms to Horizontal Resolution Using Variable-Resolution CAM

Abstract:

East Coast winter storms can have a broad range of impacts on densely populated regions of eastern North America. These systems can pose threats to health and safety, significantly disrupt transportation infrastructure, and result in long-term national and global economic consequences. While considerable efforts have been made in the short-term forecasting of these storms, less investigation has detailed their representation in free-running climate simulations at seasonal to multi-year timescales, in part due to the relatively low resolution of traditional global general circulation models (GCMs). However, aspects of these systems determining local impacts include mesoscale storm dynamics, orography, and land surface properties, which require higher horizontal resolution for increased realism. Recently, variable-resolution GCMs have become more widely used as a method of targeting computing resources to regionally increase resolution while maintaining a global modeling framework. These techniques allow for the elimination of boundary conditions and the more physically consistent treatment of subgrid parameterizations.

In this presentation we assess the impact of model resolution on winter storm characteristics by using a hierarchy of variable-resolution Community Atmosphere Model version 5 (CAM5) simulations. Regional refinement is added within CAM over the eastern portions of North America. An automated detection scheme has been developed to objectively find and track storms in output data. In addition, implementation of precipitation type and snow liquid water equivalent algorithms allow for simulated snow, freezing rain, sleet, and mixed precipitation to be assessed in a climatological sense. We pay particular attention to changes in storm intensity and shifts in precipitation distribution as horizontal resolution is increased from typical GCM grid spacings (1°, ~110 km) to resolutions more typical of numerical weather prediction (0.125°, ~14km). Novel methods for analyzing discrete storm events will be discussed. We address the scale sensitivity of parameterizations important in the formation and evolution of these storms and suggest pathways forward for improving regional climate assessments using next-generation numerical techniques.