Regional climates are strongly influenced by modes of climate variability (see Box 14.1 for definitions of mode, regime and teleconnection). This chapter assesses major modes such as El Niño-Southern Oscillation (ENSO, Section 14.4), the North Atlantic Oscillation/Northern Annular Mode (NAO/NAM) and Southern Annular Mode (SAM) in the extratropics (Section 14.5) and various other well-known modes such as the Pacific North American (PNA) pattern, Pacific Decadal Oscillation (PDO), Atlantic Multi-decadal Oscillation (AMO), etc. (Section 14.7). Many of these modes are described in previous IPCC reports (e.g., Section 3.6 of AR4 WG1). Chapter 2 gives operational definitions of mode indices (Box 2.5, Table 1) and an assessment of observed historical behaviour (Section 2.7.8). Climate models are generally able to simulate the gross features of many of the modes of variability (Section 9.5), and so provide useful tools for understanding how modes might change in the future (Müller and Roeckner, 2008; Handorf and Dethloff, 2009).

Modes and regimes provide a simplified description of variations in the climate system. In the simplest paradigm, variations in climate variables are described by linear projection onto a set of mode indices (Baldwin et al., 2009; Baldwin and Thompson, 2009; Hurrell and Deser, 2009). For example, a large fraction of interannual variance in Northern Hemisphere (NH) sea level pressure is accounted for by linear combinations of the NAM and the PNA modes (Quadrelli and Wallace, 2004). Alternatively, the nonlinear regime paradigm considers the probability distribution of local climate variables to be a multi-modal mixture of distributions related to a discrete set of regimes/types (Palmer, 1999; Cassou and Terray, 2001; Monahan et al., 2001). There is ongoing debate on the relevance of the different paradigms (Stephenson et al., 2004; Christiansen, 2005; Ambaum, 2008; Fereday et al., 2008), and care is required when interpreting these constructs (Monahan et al., 2009; Takahashi et al., 2011).

Modes of climate variability may respond to climate change in one or more of the following ways:

  • No change—the modes will continue to behave as they have done in the recent past.
  • Index changes—the probability distributions of the mode indices may change (e.g., shifts in the mean and/or variance, or more complex changes in shape such as changes in local probability density, e.g., frequency of regimes).
  • Spatial changes—the climate patterns associated with the modes may change spatially (e.g., new flavours of ENSO; see Section 14.4 and Supplementary Material) or the local amplitudes of the climate patterns may change (e.g., enhanced precipitation for a given change in index (Bulic and Kucharski, 2012)).
  • Structural changes—the types and number of modes and their mutual dependencies may change; completely new modes could in principle emerge.

An assessment of changes in modes of variability can be problematic for several reasons. First, interpretation depends on how one separates modes of variability from forced changes in the time mean or variations in the annual cycle (Pezzulli et al., 2005; Compo and Sardeshmukh, 2010). Modes of variability are generally defined using indices based on either detrended anomalies (Deser et al., 2010b) or anomalies obtained by removing the time mean over a historical reference period (see Box 2.5). The mode index in the latter approach will include changes in the mean, whereas by definition there is no trend in a mode index when it is based on detrended anomalies. Second, it can be difficult to separate natural variations from forced responses, for example, warming trends in the N. Atlantic during the 20th century that may be due to trends in aerosol and other forcings rather than natural internal variability (see Sections 14.6.2 and 14.7.1). Finally, modes of climate variability are nonlinearly related to one another (Hsieh et al., 2006) and this relationship can change in time (e.g., trends in correlation between ENSO and NAO indices).

Even when the change in a mode of variability index does not contribute greatly to mean regional climate change, a climate mode may still play an important role in regional climate variability and extremes. Natural variations, such as those due to modes of variability, are a major source of uncertainty in future projections of mean regional climate (Deser et al., 2012). Furthermore, changes in the extremes of regional climate are likely to be sensitive to small changes in variance or shape of the distribution of the mode indices or the mode spatial patterns (Coppola et al., 2005; Scaife et al., 2008).