Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection

9.3.5.2 Interannual variability ENSO

ENSO is associated with some of the most pronounced year-to-year variability in climate features in many parts of the world (Chapters 2 and 7). Since global climate models simulate some aspects of ENSO-like phenomena (Chapter 8), there have been a number of studies that have attempted to use climate models to assess the changes that might occur in ENSO in connection with future climate warming and in particular, those aspects of ENSO that may affect future climate extremes.

Firstly, will the long-term mean Pacific SSTs shift toward a more El Niño-like or La Niña-like regime? Since 1995, the analyses of several global climate models indicate that as global temperatures increase due to increased greenhouse gases, the Pacific climate will tend to resemble a more El Niño-like state (Knutson and Manabe, 1995; Mitchell et al., 1995; Meehl and Washington, 1996; Timmermann et al., 1999; Boer et al., 2000b). However, the reasons for such a response are varied, and could depend on the model representation of cloud feedback (Senior, 1999; Meehl et al., 2000b); the quality of the unperturbed El Niño state in the models (Chapter 8) or the stronger evaporative damping of the warming in the warm pool region, relative to the eastern Pacific due to the non-linear Clausius-Clapeyron relationship between temperature and saturation mixing ratios (e.g., Knutson and Manabe, 1995). Additionally, a different coupled model (Noda et al., 1999b) shows a La Niña-like response and yet another model shows an initial La Niña-like pattern which becomes an El Niño-like pattern due to subducted warmed extra-tropical water that penetrates through the sub-tropics into the tropics (Cai and Whetton, 2000). A possible reason for the La Niña-like response has been suggested in a simple coupled model study where the dominant role of ocean dynamics in the heat balance over the tropical Pacific is seen for a specified uniform positive forcing across the Pacific basin (Cane et al., 1997).


Figure 9.26: Standard deviations of Niño-3 SST anomalies (Unit: °C) as a function of time during transient greenhouse warming simulations (black line) from 1860 to 2100 and for the same period of the control run (green line). Minimum and maximum standard deviations derived from the control run are denoted by the dashed green lines. A low-pass filter in the form of a sliding window of 10 years width was used to compute the standard deviations. (a) ECHAM4/OPYC model. Also shown is the time evolution of the standard deviation of the observed from 1860 to 1990 (red line). Both the simulated and observed SST anomalies exhibit trends towards stronger interannual variability, with pronounced inter-decadal variability superimposed, (reproduced from Timmermann et al., 1999), (b) HadCM3 (Collins, 2000b).

Secondly, will El Niño variability (the amplitude and/or the frequency of temperature swings in the equatorial Pacific) increase or decrease? Attempts to address this question using climate models have again shown conflicting results, varying from slight decreases or little change in amplitude (Tett 1995; Knutson et al., 1997; Noda et al., 1999b; Collins, 2000b; Washington et al., 2001; Figure 9.26b) to a small increase in amplitude (Timmermann et al., 1999; Collins, 2000a; Figure 9.26a), which has been attributed to an increase in the intensity of the thermocline in the tropical Pacific. Knutson et al. (1997) and Hu et al. (2001) find that the largest changes in the amplitude of ENSO occur on decadal time-scales with increased multi-decadal modulation of the ENSO amplitude. Several authors have also found changes in other statistics of variability related to ENSO. Timmermann et al. (1999) find that the interannual variability of their model becomes more skewed towards strong cold (La Niña type) events relative to the warmer mean climate. Collins (2000a) finds an increased frequency of ENSO events and a shift in the seasonal cycle, so that the maximum occurs between August and October rather than around January as in the unperturbed model and the observations. Some recent coupled models have achieved a stable climate without the use of flux adjustments and an important question to ask is what is the effect of flux adjustment on changes in variability. Collins (2000b) finds different responses in ENSO in two models, one of which has been run without the use of flux-adjustments. However, he concludes that differences in response are most likely to be due to differences in the response of the meridional temperature gradient in the two models arising from different cloud feedbacks (Williams et al., 2001) rather than due to the presence or absence of flux adjustment.

Finally, how will ENSO’s impact on weather in the Pacific Basin and other parts of the world change? Meehl et al. (1993) and Meehl and Washington (1996) indicate that future seasonal precipitation extremes associated with a given ENSO event are likely to be more intense due to the warmer, more El Niño-like, mean base state in a future climate. That is, for the tropical Pacific and Indian Ocean regions, anomalously wet areas could become wetter and anomalously dry areas become drier during future ENSO events. Also, in association with changes in the extra-tropical base state in a future warmer climate, the teleconnections to mid-latitudes, particularly over North America, may shift somewhat with an associated shift of precipitation and drought conditions in future ENSO events (Meehl et al., 1993).

When assessing changes in ENSO, it must be recognised that an “El Niño-like” pattern can apparently occur at a variety of time-scales ranging from interannual to inter-decadal (Zhang et al., 1997), either without any change in forcing or as a response to external forcings such as increased CO2 (Meehl and Washington, 1996; Knutson and Manabe, 1998; Noda et al., 1999a,b; Boer et al., 2000b; Meehl et al., 2000b). Making conclusions about “changes” in future ENSO events will be complicated by these factors. Additionally, since substantial internally generated variability of ENSO statistics on multi-decadal to century time-scales occurs in long unforced climate model simulations (Knutson et al., 1997), the attribution of past and future changes in ENSO amplitude and frequency to external forcing may be quite difficult, perhaps requiring extensive use of ensemble climate experiments or long experiments with stabilised forcing (e.g., Knutson et al., 1997).

Although there are now better ENSO simulations in global coupled climate models (Chapter 8), further model improvements are needed to simulate a more realistic Pacific climatology and seasonal cycle as well as more realistic ENSO variability (e.g., Noda et al., 1999b). It is likely that such things as increased ocean resolution, atmospheric physics and possibly flux correction can have an important effect on the response of the ENSO in models. Improvements in these areas will be necessary to gain further confidence in climate model projections.

Monsoon
One of the most significant aspects of regional interannual variability is the Asian Monsoon. Several recent studies (Kitoh et al., 1997; Hu et al., 2000a; Lal et al., 2000) have corroborated earlier results (Mitchell et al., 1990; Kattenberg et al., 1996) of an increase in the interannual variability of daily precipitation in the Asian summer monsoon with increased greenhouse gases. Lal et al. (2000) find that there is also an increase in intra-seasonal precipitation variability and that both intra-seasonal and inter-annual increases are associated with increased intra-seasonal convective activity during the summer. Less well studied is the Asian winter monsoon, although Hu et al. (2000b) find reductions in its intensity with a systematic weakening of the north-easterlies along the Pacific coast of the Eurasian continent. However, they find no change in the interannual or inter-decadal variability.

The effect of sulphate aerosols on Indian summer monsoon precipitation is to dampen the strength of the monsoon compared to that seen with greenhouse gases only (Lal et al., 1995; Cubasch et al., 1996; Meehl et al., 1996; Mitchell and Johns 1997; Roeckner et al., 1999), reinforcing preliminary findings in the SAR. The pattern of response to the combined forcing is at least partly dependent on the land-sea distribution of the aerosol forcing, which in turn may depend upon the relative size of the direct and indirect effects (e.g., Meehl et al., 1996; Roeckner et al., 1999). There is still considerable uncertainty in these forcings (Chapter 6). To date, the effect of aerosol forcing (direct and indirect) on the variability of the monsoon has not been investigated.

In summary, an intensification of the Asian summer monsoon and an enhancement of summer monsoon precipitation variability with increased greenhouse gases that was reported in the SAR has been corroborated by new studies. The effect of sulphate aerosols is to weaken the intensification of the mean precipitation found with increases in greenhouse gases, but the magnitude of the change depends on the size and distribution of the forcing.



Other reports in this collection