Sources of spatial and intermodel variability in CMIP5 precipitation projections for the western and northwestern USA

Tuesday, 16 December 2014
David Earl Rupp1, John T Abatzoglou2 and Philip Mote1, (1)Oregon State University, Corvallis, OR, United States, (2)University of Idaho, Department of Geography, Moscow, ID, United States
Anthropogenically forced projected changes in seasonal precipitation by the latter 21st century vary in magnitude and sign across the western US. Within a subregion, projections also vary widely among global climate models (GCMs), where only a fraction of this variability can be attributed to internal unforced variability. To aid in determining the cause(s) of this variability, we decompose CMIP5 projections of precipitation change (as change in latent heat release) into changes in short-wave and long radiative fluxes, sensible heat flux, air flux and stability (the latter two quantified in terms of the divergence of dry static energy, H). In both winter and summer, change in H is the primary control on the spatial pattern of precipitation change across the western US, though in summer the change in sensible heat flux modulates the pattern. Within the northwestern US, changes in H also explain most of the variability in precipitation changes among GCMs, again with sensible heat playing a secondary role in summer. Moreover, the dynamic component of the change in H, proxied by the change in 500hPa vertical velocity (omega, in units of pressure/time) appears to be a substantially larger control than the thermodynamic component of the change in H. These results imply that a GCM's ability to accurately reproduce regional wind patterns and regional convergence/divergence may be a useful metric for determining which GCMs are providing more plausible projections of regional precipitation change.