Characterizing Large-Scale Meteorological Patterns and Associated Temperature and Precipitation Extremes over the Northwestern United States using Self Organizing Maps

Ongoing and projected anthropogenic climate change is driving demand for climate information at scales relevant to societal and environmental impacts. Projecting climate at such scales presents modeling and computational challenges, especially in regions of complex topography and high climatic heterogeneity like the northwestern United States (US). This challenge is amplified when focusing on extreme events, which are often associated with the most severe climate impacts. Here we present the first steps toward employing large-scale meteorological patterns (LSMPs) associated with local temperature and precipitation extremes to obtain climate change information at societally relevant scales, using existing observational and climate model data. LSMPs are defined as patterns of atmospheric circulation in the lower, mid, and upper troposphere. First, using the method of self-organizing maps (SOMs), we demonstrate the ability to characterize the LSMPs associated with extremes in temperature and precipitation across the northwestern US for both the winter and summer seasons over the last 35 years. In particular, we use SOMs to identify characteristic LSMPs (the nodes or outputs of the SOMs) for the Northwest and from the assignment of individual days to nodes, we relate extreme event occurrence to LSMPs. Our results suggest that the SOMs effectively capture the range of circulation patterns and associated temperature and precipitation extremes for plausible physical reasons, especially in winter and in areas most dominated by large-scale meteorology. With this observational foundation, we then apply SOMs to evaluate the fidelity of CMIP5 historical simulations of LSMP nodes, to assess the degree to which climate models capture key circulation patterns related to local-scale extreme events in the region. Moving forward, we aim to develop and apply SOMs-based evaluation tools to constrain uncertainties in extremes associated with projected changes in LSMPs under anthropogenic warming.