Evaluation of Leading Modes of North American Summer Heatwave Variability in CMIP6 Models

Abstract

This study evaluates 33 CMIP6 GCMs for their ability to simulate the dominant modes of heatwave variability (monthly maximum 3-day temperatures) during extended summers (June–September 1961–2010) across North America (NA) and associated teleconnections with the North Pacific and regional processes affecting formation. By applying common Empirical Orthogonal Function (EOF) analysis to three global reanalysis data sets, combined with Common Basis Function (CBF) approach for the GCMs, a unified framework for evaluating performance is established. Two distinct leading modes of monthly summer heatwave variability over NA–dipole and tripole patterns–are identified using the common EOF. The GCMs reproduce these modes, showing agreement in the centers of positive and negative anomalies with the reanalysis data sets, particularly when evaluated using the CBF approach. The GCMs capture large-scale North Pacific to North American wave train patterns associated with the two modes of heatwave variability over NA, showing a phase shift between them. They reproduce atmospheric moisture conditions, such as total column water vapor and precipitation; however, performance in capturing these anomalies is lower due to the complexities of moisture transport and convection. The GCMs also effectively reproduce regional-scale surface radiation and turbulent heat flux anomalies related to heatwave variability. Ensemble means of the GCMs generally outperform individual models, highlighting the advantages of multi-model ensembles in reducing uncertainty and improving overall accuracy. Higher-resolution models outperform lower-resolution counterparts in capturing the intricate details of heatwave variability and associated processes, underscoring the importance of resolution in achieving accurate simulations of these extremes.