The Effect of Greenhouse SSTs on ENSO Simulations ... - AMS Journals

1 downloads 0 Views 682KB Size Report
Rutherfurd 1994). Another method is to model the phenomenon under both present-day (1CO2) and possible future (2CO2) conditions. Since ENSO involves ...
342

JOURNAL OF CLIMATE

VOLUME 10

The Effect of Greenhouse SSTs on ENSO Simulations with an AGCM I. N. SMITH, M. DIX,

AND

R. J. ALLAN

CSIRO Division of Atmospheric Research, Mordialloc, Victoria, Australia (Manuscript received 6 November 1995, in final form 24 June 1996) ABSTRACT The potential for changes to the El Nin˜o–Southern Oscillation (ENSO) phenomenon under enhanced greenhouse conditions is addressed in a series of atmospheric general circulation model (AGCM) experiments with both 13CO2 and 23CO2 radiative forcing and prescribed sea surface temperatures. In the first series of experiments, climate anomalies associated with the 1987 ‘‘warm’’ event SSTs are simulated for present-day (13CO2) conditions. The results reveal realistic large-scale perturbations to mean sea level pressure, rainfall, evaporation, and low-level and upper-level winds compared to simulations using climatological SSTs. In the second series of experiments, the same warm anomalies are superimposed on the equilibrium SSTs calculated during the course of 23CO2 equilibrium experiments. The resultant SSTs provide an analog warm event for enhanced greenhouse conditions. The simulated climate anomalies for this greenhouse event are similar to those of the present-day event, but, although some changes are found in the western Pacific region, the results reveal little evidence of any large-scale intensification associated with warmer equilibrium SSTs. The results suggest that increased static stability may dampen the effects of any nonlinear increase in evaporation and that any changes in the behavior of ENSO due to predicted temperature increases in the Tropics may not be significant.

1. Introduction When dealing with greenhouse-related climate change, one area of concern is the possibility of changes to the El Nin˜o–Southern Oscillation (ENSO) phenomenon. At present, ENSO phases can lead to significant droughts, wet episodes, and temperature extremes over widespread regions of the globe (see, e.g., Ropelewski and Halpert 1987; Kiladis and Diaz 1989; Halpert and Ropelewski 1992) and several questions have been raised concerning possible changes in intensity, frequency, and duration of these events (Enfield 1988, 1989; Enfield and Cid 1991; Elliot and Angell 1988). One method of addressing these questions is to study the historical record as a means of determining past changes to ENSO that may have been related to changes in the climate system (see Allan 1993; Whetton and Rutherfurd 1994). Another method is to model the phenomenon under both present-day (13CO2) and possible future (23CO2) conditions. Since ENSO involves interactions between both the atmosphere and the oceans, this method usually involves coupled models. Coupled models can reproduce many significant features of the seasonal cycle but, at present, do not yet capture all the important features

Corresponding author and address: Ian Smith, CSIRO Division of Atmospheric Research, Private Bag No.1, Mordialloc 3195, Australia. E-mail: [email protected]

q 1997 American Meteorological Society

associated with interannual fluctuations (Mechoso et al. 1995). This may limit the interpretation of any simulated greenhouse-related changes of the ENSO phenomenon. Tett (1995) performed a 70-yr-long transient coupled model experiment in which CO2 was increased at a rate of 1% (compounded) per year. The results did not reveal any significant changes to the interannual variance of either SSTs or precipitation over the Pacific Ocean. Knutson and Manabe (1994, 1995) and Knutson et al. (1997) have also analyzed the results of a transient coupled model experiment with quadrupled CO2 values. They noted that the amplitude of ENSO-like SST anomalies decreased slightly, ENSO-like precipitation anomalies were enhanced, but the associated surface wind anomalies were not greatly affected. Knutson and Manabe (1996) also report finding no pronounced changes in simulated SST variability with a finer resolution model. Meehl et al. (1993) (MA) analyzed a coupled GCM experiment in which warm anomalies in the west Pacific, for a 13CO2 climate simulation, were interpreted as ENSO-like warm events. Neither the frequency nor strength of these anomalies altered significantly in a 23CO2 climate simulation. SST anomalies did not alter significantly, and in the Tropics, the patterns of anomalously wet and dry regions changed very little, although the anomalies were intensified. Similar results were obtained when they performed a simpler mixedlayer model greenhouse experiment with prescribed ENSO-like warm anomalies in the Pacific Ocean. In both experiments the intensification of the hydrological

FEBRUARY 1997

SMITH ET AL.

343

FIG. 1. The difference between 23CO2 and 13CO2 mean annual SSTs from coupled model simulations. Contour interval is 0.58C. The central equatorial Pacific (CEP) and eastern equatorial Pacific (EEP) regions are indicated.

cycle was attributed to higher mean SSTs that, near the rising branch of the Walker circulation, led to greater evaporation, low-level convergence, and therefore greater precipitation. Their finding that dry areas tended to be drier and wet areas to be wetter was related to an overall strengthening of the anomalous Walker circulation. Meehl et al. (1993) acknowledged that their results are somewhat uncertain and that more definite conclusions would rely on the development of higher resolution models and better representation of the critical coupling mechanisms. Their study suggested that an ‘‘analog’’ approach, in which SST anomalies are prescribed within the framework of a greenhouse equilibrium climate simulation, may also provide some insight. Here, we describe the results of an analog study that makes use of both observed and coupled model equilibrium SSTs. The term analog indicates that SST anomalies are prescribed rather than simulated—a relatively simple approach that avoids the costs and uncertainties associated with a fully coupled model. This study differs from the MA study in that it utilizes observed SST anomalies and makes use of a higher resolution atmospheric model than (R21 cf. R15). This appears to yield a more realistic atmospheric response to ENSO-like SST anomalies. 2. Model and experimental setup The model used in this study is CSIRO9—a ninelevel spectral atmospheric model with rhomboidal 21 (R21) resolution (McGregor et al. 1993). Results of long-term simulations with CSIRO4 (a four-level predecessor to CSIRO9) using observed SSTs are presented in Smith (1994, 1995), while Dix and Hunt (1995) present an analysis of AMIP (Atmospheric Model Intercomparison Project; Gates 1992) experiments with CSIRO9. These studies provide an example of the ability of relatively coarse resolution models to capture important features of ENSO-related phenomena. In addition to prescribed SST experiments, CSIRO9,

FIG. 2. Observed SST anomalies for (a) September–November 1987, (b) December 1987– February 1988. Shading corresponds to positive values.

when coupled to a primitive slab ocean model, has been used to simulate the equilibrium response of the climate system to a doubling of greenhouse gases (Watterson et al. 1995; Whetton et al. 1993). The increase in mean global surface temperature was found to be 4.8 K, which is at the higher end of estimates from a range of model studies (Gates et al. 1992). Figure 1 shows the differences in annual mean SSTs between equilibrium simulations of 1XCO2 and 2XCO2 conditions. Increases throughout most of the equatorial regions are about 2.5 K, but are greater and generally zonally uniform at higher latitudes. In the far eastern Pacific, particularly off the coast of South America, there is some evidence that the changes are less uniform. The reasons for this are not apparent, but since they occur in regions where upwelling is an important factor, may reflect the limitations of a simple slab model with no dynamics. Figure 1 also indicates two boxed regions. The first of these represents the eastern equatorial Pacific (EEP) and comprises 48 grid points covering the region 4.88S to 4.88N and 151.98W to 90.08W—a close approximation to the NINO3 region where average SST anomalies provide an index of observed ENSO events (Cane 1992). The second box represents the central equatorial Pacific (CEP) and comprises 66 grid points that cover the region 88S to 88N and 180.08W to 123.78W—a close approximation to the region used by MA for calculating area averages of various fields. 1987 is classified as a warm event year since warm SST anomalies persisted throughout the equatorial Pacific for all 12 months and into 1988. Figure 2 shows the mean SST anomalies for both the September—November (SON) and the December 1987—February 1988

344

JOURNAL OF CLIMATE

FIG. 3. The relationship between present-day mean conditions, observed anomalies, greenhouse mean conditions, and analog warm event anomalies: (a) Monthly mean SSTs in the EEP region. The four ¯ 1; observed curves correspond to present-day (13CO2) mean values A values (January 1987 to December 1988) A1; greenhouse (23CO2) ¯ 2; analog warm event values A2. (b) Meridionally avmean values A eraged (11.18S to 11.18N) surface temperatures (sea surface temperatures are prescribed, land temperatures are model values). The four curves correspond to present-day (13CO2) September to November ¯ 1, observed (and simulated) values September to Nomean values A ¯ 2, analog warm vember 1987 A1, greenhouse (23CO2) mean values A event values A2.

(DJF) periods. During SON, the warm anomalies peaked near 128C in the central Pacific while cold anomalies occurred in the northern and southern Pacific. Three months later, the anomalies had weakened considerably with maximum values of about 118C just east of the date line. The evolution of the SST anomalies for the EEP region over the 2-yr period 1987–88 is demon¯ 1 represents clistrated in Fig. 3a. The curve labeled A matological SST values for this region while the curve labeled A1 represents the observed values over the 24 months. Although MA considered results for the ‘‘mature phase’’ season, namely DJF, it can be seen that the SST anomalies in this case study were near their maxima around SON 1987 and declined rapidly thereafter to almost zero by January 1988. Average surface temperatures for the SON period in the equatorial band between 11.18S and 11.18N are plotted as a function of longitude ¯ 1 represents climatological values in Fig. 3b. Again, A while A1 represents observed values. It can be seen that the warm anomalies were almost uniform around the globe, particularly east of the dateline. Furthermore, the

VOLUME 10

magnitude of the anomalies is similar to that of the ideal anomalies prescribed by MA in their mixed-layer experiment. The equilibrium climates for both present-day (13CO2) and greenhouse (23CO2) conditions were previously calculated with the coupled atmosphere/slabocean model. This model uses the Q-flux method whereby the oceanic heat fluxes for a 50-m mixed layer depth are deduced in an endeavor to constrain the simulated monthly SSTs for l3CO2 conditions near their presently observed long-term mean values. Despite this constraint, the equilibrium SSTs differ slightly from observed because of minor drift, which occurs at some points. The differences are greatest at polar latitudes but are of the order of only 10.28C in the Tropics and are ¯1 not considered significant here (the curves labeled A in Figs. 3a and 3b are, in fact, derived from the coupled model values and, in the case of Fig. 3b, include model values over land points). At this point the simulation was continued with the SSTs prescribed at their equilibrium values; that is, the ocean model was effectively ‘‘turned off.’’ The net result was an atmospheric modelonly simulation of present-day climate (involving cloud amounts, soil moisture, sea ice distributions, etc.) consistent with the SSTs as calculated by the coupled model. The next step involved simulating ENSO-related atmospheric anomalies for present-day conditions. Starting with 1 January initial conditions, global monthly SST anomalies based on those observed during 1987 and 1988 (Reynolds 1988) were imposed on the equilibrium SSTs, and the AGCM integration continued for ¯1 24 months. (In Figs. 3a and 3b, the curves labeled A are in fact based on adding the observed SST anomalies to the coupled model equilibrium SST values. As already noted, the differences between the prescribed and observed values are negligible in the Tropics.) This yielded a simulation of climatic anomalies associated with the warm anomalies for the period January 1987 to December 1988 (only the results for SON 1987 are presented here). This step was repeated a further four times, on each occasion starting with different 1 January initial conditions (from independent simulations). In all, a total of five ENSO-type simulations were completed in order to provide a reasonable ensemble sample from which to establish mean anomalies. The final steps involved prescribing the analog warm event for enhanced greenhouse conditions. The same procedure as described above was adopted except that the equilibrium SSTs were those calculated with the coupled model under 23CO2 conditions. Again, the SSTs were ‘‘frozen’’ so that an atmospheric model-only simulation of the equilibrium greenhouse climate was ¯ 2 repachieved. In Figs. 3a and 3b the curves labeled A resent the equilibrium SSTs from the coupled model 23CO2 simulation which, as Fig. 1 also illustrates, are about 2.58C uniformly warmer than present-day values. The same observed SST anomalies for 1987–88 were then imposed on these equilibrium SSTs and another

FEBRUARY 1997

SMITH ET AL.

345

FIG. 4. Simulated mean sea level pressure anomalies: (a) September to November 1987, present-day conditions. (b) September to November 1987, greenhouse conditions. (c) The difference (b) – (a). Shading corresponds to values greater than zero.

five independent 24-month AGCM simulations were completed with 23CO2 radiative forcing maintained in each case. The curves labeled A2 in Figs. 3a and 3b are based on the SSTs obtained by adding the observed anomalies to the greenhouse values. It can be seen that, because there is little change in the seasonal cycle in going from 13CO2 conditions to 23CO2 conditions, the resultant analog warm event is almost identical to the observed El Nin˜o event except for a shift of about 12.58C. It should be noted that the approach adopted here is different from simply simulating the effects of El Nin˜o SSTs under current conditions and then simulating the effects of adding a greenhouse warming to the SSTs. The difference lies in the fact that the radiative forcing and the equilibrium SSTs are consistent with each other. The relevance of this analog approach is based on the assumption that the SST anomalies observed under present-day conditions are likely to continue relatively unchanged under greenhouse conditions. There is some evidence from existing coupled model studies that anomalies may actually decrease and so the scenarios adopted here may involve SSTs that may be unrealistically warm. Figure 3a indicates that the analog warm event has SST values in the EEP region prescribed to peak near 318C (304 K), while Fig. 3b shows values exceeding 328C (305 K) in the region immediately east

of the date line. Although these are similar to the MA values, they may be unrealistically excessive since present-day observations reveal that only a small fraction of the tropical Pacific, Atlantic, and Indian Oceans is warmer than 29.58C (302.5 K). It has been argued (see Ramanathan and Collins 1993) that a ‘‘thermostat’’ effect operates, which prevents the oceans from exceeding this temperature for any length of time but the nature of this effect is debated (see Fu et al. 1992). Given that there are relatively few studies into this problem, it is considered that the use of (possibly) excessive values deserves attention. This should provide, at least, an indication of whether an intensification of the hydrological cycle could be expected in a warmer world. 3. Results In dealing with changes to climatic anomalies, we need to consider changes in the means and the relative anomalies—an important aspect discussed in some detail by MA. The labels attached to the SST curves shown in Figs. 3a and 3b can be used to refer to the behavior of any climatic variable (e.g., mean sea level pressure, precipitation, evaporation, etc.). We can refer to each ¯ 1, the mean for 13CO2 conditions; in turn as follows: A ¯ 2, the mean A1, the anomaly under 13CO2 conditions; A for 23CO2 conditions; and A2, the anomaly under

346

JOURNAL OF CLIMATE

VOLUME 10

FIG. 5. Simulated rainfall anomalies expressed as percentage differences. (a) September to November 1987, present-day conditions. (b) September to November 1987, greenhouse conditions. (c) The difference (b) – (a). Contours are 0, 620, 640, and 680%. Shading corresponds to values greater than zero.

¯ 22A ¯ 1 represents the 23CO2 conditions. The difference A shift in the means while A22A1 represents an absolute ¯ 1 represents a present-day anomdifference. Here, A12A ¯ aly while A22A2 represents a 23CO2 anomaly. The ¯ 2)2(A12A ¯ 1) represents a change, posquantity (A22A sibly an intensification, of the relative anomalies. Figure 4a shows the simulated anomalous mean sea level pressure (MSLP) field associated with SON 1987 ¯ 1) repunder present-day conditions. This pattern (A12A resents the difference between the mean MSLP, as calculated from the five experiments with warm SST anomalies, and those with climatological SSTs. The corresponding greenhouse pattern (when the same SST anomalies are superimposed on the greenhouse equilib¯ 2) is shown in Fig. 4b. Both patrium SSTs, i.e., A22A terns are similar, and represent a simulated Southern Oscillation (see, e.g., Deser and Wallace 1990) characterized by pressure decreases over the eastern equatorial Pacific and increases over the Indonesian, Australasian, and Indian Ocean regions. Most of these features are statistically significant according to t tests (Chervin and Schneider 1976). ¯ 2)2(A12A ¯ 1). Figure 4c shows the difference (A22A If an intensification of the anomalies occurred, this would manifest itself in a difference pattern similar to the original anomaly pattern. Comparing Fig. 4c with Fig. 4a reveals negligible changes to pressure anomalies

in the eastern equatorial Pacific. Most of the differences east of 1508W and across to tropical South America indicate a weakening of the original anomalies. There is some evidence of an intensification near the date line, but the pattern of positive anomalies in the Indonesian region is also weakened. Throughout a large region centered over the Indian Ocean there is the evidence of a strengthening of the positive anomalies although, with the exception of Western Australia, these are relatively weak. Other regions where intensification is apparent include Northern Africa (negative anomalies) and both North America and extratropical South America (positive anomalies). However, throughout much of the tropical Pacific region where SST anomalies are greatest, there is little evidence of intensification. Similar results are also found when comparing the results for either the following season (DJF) or the entire 12 months of 1987. Figure 5a shows the simulated present-day anomalous mean rainfall expressed as a percentage difference from the long-term average mean rainfall for SON. The same result for the greenhouse scenario is shown in Fig. 5b and a similar ENSO-like pattern is evident in both cases. This pattern is characterized by large increases over the equatorial Pacific, where the SST anomalies are greatest, and increases over extratropical North America and South America. Decreases occur near the central Americas and in the far western equatorial Pacific where they

FEBRUARY 1997

SMITH ET AL.

347

FIG. 6. Simulated evaporation rate anomalies expressed as percentage differences: (a) September to November 1987, present-day conditions. (b) September to November 1987, greenhouse conditions. (c) The difference (b) – (a). Contours are 0, 610, 620, and 640%. Shading corresponds to values greater than zero.

extend over the tropical Indian Ocean as far west as South Africa. These features are similar to those found in composite rainfall anomaly patterns observed during real events (see, e.g., Ropelewski and Halpert 1987; Lau and Sheu 1990). Rather than show the raw pattern of intensification ¯ 2)2(A12A ¯ 1), which is dominated by rainfall (A22A anomalies where the mean rainfall is highest, we instead show in Fig. 5c the differences between the previous two figures (i.e., simply the differences in the percentage differences). This does not distort the results since both are almost identical in qualitative terms. This figure reveals that some intensification occurs west of the date line including Indonesia/Australia and over the tropical Indian Ocean. East of the date line, the pattern of intensification does not greatly resemble the original anomalies but is characterized by weaker anomalies, particularly near 1508 W and farther east over the central Americas. Figure 6a shows the simulated present-day anomalous evaporation rate, again expressed as a percentage difference from the long-term average values for SON. Some relatively large anomalies (exceeding plus or minus 20%) are simulated in some places—particularly Australia and the western and eastern equatorial Pacific Ocean. These anomalies are also evident, and possibly

stronger, in the results from the greenhouse scenario simulations shown in Fig. 6b. The difference between the percentage anomalies, shown in Fig. 6c, confirms this latter observation for both Australia and the western Pacific Ocean. As with rainfall, this is almost identical in qualitative terms, but easier to display than the raw ¯ 2)2(A12A ¯ 1). Over Australia, the reducpattern (A22A tion in actual evaporation is associated with reduced soil moisture due to the reduction in rainfall. Over the western Pacific some intensification is apparent but Fig. 3b shows that this is associated with relatively small SST anomalies of the order of .58C or less. Factors such as increased wind speed or decreased stability may be associated with this response since there is no clear evidence that warmer SSTs east of the date line have resulted in an intensification of the evaporation anomalies. The results reveal either a weakening of the original anomalies or else very weak intensification, particularly for the eastern equatorial Pacific. To facilitate the comparisons, we show in Fig. 7 equatorial cross-section averages (for the band 11.18S– 11.18N) of the raw MSLP, rainfall, and evaporation anomalies associated with both present-day and greenhouse conditions. Figure 7a shows that positive MSLP anomalies intensify in the Indian Ocean region (408– 1208E), negative anomalies intensify in the western and

348

JOURNAL OF CLIMATE

VOLUME 10

alies evident in COADS surface wind data as analyzed by Nigam and Shen (1993) or Deser and Wallace (1990). The anomalies for the greenhouse event (not shown) are very similar and this is reflected in Fig. 8b, which shows the difference between the two patterns. This difference bears very little resemblance to the original anomalies and therefore reveals very little evidence of significant intensification. Most of the changes are small and, if anything, suggest a weakening of the westerly anomalies in the central Pacific. A similar result is found when considering the anomalous upper-level (approx 200 hPa) zonal winds. Figure 9a shows that, for present-day conditions, easterly anomalies over the equatorial Pacific accompany the warm SST anomalies—consistent with an anomalously weaker Walker circulation. Figure 9b shows the difference between the anomalous winds associated with the greenhouse event (not shown) and those in Fig. 9a. Weak positive differences throughout the equatorial Pacific correspond to weaker easterly anomalies—in contrast to the original anomalies. Both the low-level and upper-level results indicate a reduction in the strength of the anomalous Walker circulation. 4. Discussion and concluding remarks

FIG. 7. Meridionally averaged (11.18S to 11.18N) simulated anomalies for present-day conditions (solid curves) and greenhouse conditions (dashed curves): (a) mean sea level pressure, (b) rainfall, (c) evaporation.

central Pacific (1208–2008E), while over the remainder of the eastern Pacific the anomalies are weakened. Figure 7b shows that both negative rainfall anomalies in the vicinity of 1208E and positive anomalies in the vicinity of the date line intensify, while east of the date line the positive anomalies tend to diminish. A similar pattern is evident in the anomalous evaporation values shown in Fig. 7c. Again, any intensification is restricted to west of the date line, whereas the magnitude of the anomalies is relatively small when compared to the rainfall anomalies. These cross-section averages confirm that there is no significant large-scale signal associated with the warm SSTs east of the date line. The anomalous low-level (approx 990 hPa) zonal wind simulated for present-day conditions is shown in Fig. 8a. Throughout the western and central equatorial Pacific, the anomalous winds are westerly—reflecting a shift of the convergence zone during a warm event toward the region of warm SST anomalies. This response is very similar to observed tropical zonal wind anom-

If we consider just mean sea level pressure, rainfall, and evaporation, then the results reveal some intensification of ENSO-like anomalies in the western equatorial Pacific. These indicate that in the Indonesian/Australian region it may be possible that pressure anomalies and rainfall deficits may become more severe during warm events under enhanced greenhouse conditions. However, where some intensification can be identified, it is not characterized by significant SST anomalies and is unlikely that the nonlinear increase of evaporation with temperature can totally explain this response. In central and eastern equatorial Pacific where the SST anomalies are greatest, no such intensification is found. This can be seen in more detail by considering the absolute values of these quantities averaged over the CEP region (see Fig. 1) as defined by MA. Meehl et al.’s tabulated results showed two sets of composite results: the first set based on their dynamical oceanic model, in which relatively weak, warm SST anomalies were detected, and the second set based on a mixed-layer model in which stronger SST anomalies were prescribed. Figure 10 compares, for the CEP region, the CSIRO9 results with the MA mixed-layer model results. SST, rainfall, and evaporation are all presented in terms ¯ 2, ¯ 1, A1, A of their climatological and anomalous values A and A2. The main difference is that the model results refer to the September–November period, whereas the MA results refer to the December–February period. This is not regarded as a significant difference since the SON period involves SST values very similar to those prescribed by MA and is also a time of year when ENSOrelated anomalies are observed to be strong. Further-

FEBRUARY 1997

SMITH ET AL.

FIG. 8. Simulated low-level zonal wind anomalies: (a) September to November 1987, presentday conditions. (b) The difference between the anomalies simulated for greenhouse conditions (not shown) and (a). Shading corresponds to values greater than 1.0 m s21.

FIG. 9. Same as Fig. 8 except for upper-level zonal wind anomalies.

349

350

JOURNAL OF CLIMATE

VOLUME 10

were found to be statistically significant. However, the changes can be seen to be relatively small. For example, the difference between the anomalous rainfall found by ¯ 22A2)2(A ¯ 12A1)] amounts to 0.58 mm MA [i.e., (A day21 compared to a mean rainfall rate of between 7 and 8 mm day21. Expressed in terms of percentages, the MA anomalous rainfall for present-day conditions ¯ 12A1) is 125.2%, while that for greenhouse condi(A ¯ 22A2) is 128.5%—a net increase of only 3.3%. tions (A The equivalent results reported here are 150.0% and 148.0%—a net decrease of 2.0%. There may be several reasons for these differences, particularly as the experimental framework in each case is different, but it is worth noting that CSIRO9 yields a stronger, more realistic, ENSO-like response than that shown by MA. This may be due to the imposition of more realistic ENSO-like SST anomalies but is more likely due to the use of a higher resolution atmospheric model (R21 vs R15). Meehl et al. noted that evaporation increases over the CEP region were proportionally larger in their 23CO2 warm events due mainly to the nonlinear relationship with SST. Although we find that evaporation does increase with the higher SSTs, the relationship is more linear such that similar anomalies are generated in either the 13CO2 or 23CO2 scenarios. The expression for evaporation over an oceanic grid point (as implemented in CSIRO9) is E 5 C V F(Rib) [qs (SST) 2 q1],

FIG. 10. Box average values for SST, rainfall, and evaporation for the CEP region from this study (denoted by CSIRO9) compared with those of Meehl et a1. (1993) (denoted by MA). CSIRO9 values refer to SON while MA values refer to DJF.

more, there is very little qualitative variation in the results for either 3-month period. The simulated anomalies for the 3-month period December 1987 to February 1988 reflect the weaker SST anomalies and are much smaller than either the SON anomalies or the MA anomalies. The first point to note is that although the CSIRO9 equilibrium greenhouse simulation tends to yield relatively warm SSTs, in this comparison the MA SSTs are more than 0.58C higher. The second point to note is that the CSIRO9-simulated ENSO rainfall anomalies ¯ 12A1) are much greater than that found by MA. In (A absolute terms, the increases are about twice as great while, in percentage terms, the increases are of the order of 140% compared to 125%. Despite this relatively large difference in rainfall response, the ENSO-related evaporation anomalies are relatively small (of the order of 14%) and about half that found by MA. Meehl et al. interpret their results as evidence of intensification, particularly as some of the differences

(1)

where C can be regarded as a constant, V is the surface wind speed, F(Rib) is a stability-dependent function involving the Richardson number, qs (SST) is the saturation mixing ratio corresponding to the SST, and q1 is the mixing ratio at some level above the surface. A nonlinear relationship between evaporation and SST holds only if all else is equal. In the model results presented here, variations in the other terms obviously act to dampen the nonlinear effects. The magnitude of these factors has been estimated by Boer (1993), who analyzed the surface and energy budgets simulated with the Canadian Climate Centre model. One of these, identified by Knutson and Manabe (1994, 1995) and Knutson et al. (1997), is a change in static stability. This is illustrated in Fig. 11, which compares, for both present-day and greenhouse climatological mean conditions, the temperature difference between a midlevel of the atmosphere (approx 500 hPa) and the surface. This difference is less for greenhouse conditions and corresponds to a significant increase in static stability, particularly over the eastern Pacific. This can counteract what would otherwise be relatively large evaporation increases by suppressing the exchange of latent heat between the surface and the atmosphere. This may occur via a reduction in F(Rib) and/or an increase in q1 if vertical mixing is suppressed. The absence of any large-scale intensification is reflected in the results for anomalous low-level and upperlevel winds. A similar lack of a dynamic response is

FEBRUARY 1997

SMITH ET AL.

FIG. 11. Meridionally averaged (11.18S–11.18N) temperature differences between the surface and the middle level (approx 500 hPa) of the atmosphere as simulated for present-day conditions (solid curve) and greenhouse conditions (dashed curve).

reported by Knutson et al. (1997), who performed a detailed analysis of a coupled model experiment involving the simulation of the response of ENSO to enhanced (43CO2) greenhouse conditions. Although they found evidence of intensification of the precipitation anomalies on the large scale, they found no similar intensification of the surface wind anomalies. This is attributed to compensating processes, which determine the tropospheric heat balance—the increase in static stability playing an important part in this compensation. The effect on surface evaporation is not explicitly identified and so it is difficult to judge whether their results also include a less-than-expected increase in this term. Either way, their results support the conjecture that increased static stability in the Tropics due to enhanced greenhouse conditions may play a part in dampening one aspect of the atmospheric response to ENSO-like SST anomalies. Both the pattern of greenhouse induced warming and future ENSO-related SST anomalies will possibly differ from the simple scenarios assumed in this study. There also exists a possibility that, even if these scenarios are broadly correct, the frequency of ENSO events may change. Further results from improved fully coupled models are expected to provide the best answers to these problems. Acknowledgments. The authors would like to thank Barrie Hunt, Barrie Pittock, Linda Waterman, Peter Whetton, and Stephen Wilson for many helpful discussions and two anonymous referees whose comments helped improve the manuscript. We also thank Harvey Davies for providing software support and Xingren Wu for assistance with graphics. REFERENCES Allan, R. J., 1993: Historical fluctuations in ENSO and teleconnection structure since 1879: Near-global patterns. Quat. Australasia, 11, 17–27.

351

Boer, G. J., 1993: Climate change and the regulation of the surface moisture and energy budgets. Climate Dyn., 8, 225–239. Cane, M. A., 1992: Tropical Pacific ENSO models: ENSO as a mode of the coupled system. Climate System Modelling, K. E. Trenberth, Ed., Cambridge University Press, 583–614. Chervin, R. M., and S. H. Schneider, 1976: On determining the statistical significance of climate experiments with general circulation models. J. Atmos. Sci., 33, 405. Deser, C., and J. M. Wallace, 1990: Large-scale atmospheric circulation features of warm and cold episodes in the tropical Pacific. J. Climate, 3, 1254–1281. Dix, M. R., and B. G. Hunt, 1995: Chaotic influences and the problem of deterministic seasonal predictions. Int. J. Climatol., 15, 729– 752. Elliott, W. P., and J. K. Angell, 1988: Evidence for changes in Southern Oscillation relationships during the last 100 years. J. Climate, 1, 729–737. Enfield, D. B., 1988: Is El Nin˜o becoming more common? Oceanogr. Mag., 1, 23–37. , 1989: El Nin˜o, past and present. Rev. Geophys., 27, 159–187. , and L. S. Cid, 1991: Low-frequency changes in El Nin˜o– Southern Oscillation. J. Climate, 4, 1137–1146. Fu, R., W. T. Liu, A. D. Del Genio, and W. B. Rossow, 1992: Cirruscloud thermostat for tropical sea surface temperatures tested using satellite data. Nature, 358, 394–397. Gates, W. L., 1992: AMIP—The Atmospheric Model Intercomparison Project. Bull. Amer. Meteor. Soc., 73, 1962–1970. , J. F. B. Mitchell, G. J. Boer, U. Cubasch, and V. P. Meleshko, 1992: Climate modelling, climate prediction, and model validation. Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment. J. T Houghton, B. A. Callendar, and S. K. Varney, Eds., Cambridge University Press, 97–134. Halpert, M. S., and C. F. Ropelewski, 1992: Surface temperature patterns associated with the Southern Oscillation. J. Climate, 5, 577–593. Kiladis, G. N., and H. F. Diaz, 1989: Global climatic anomalies associated with extremes in the Southern Oscillation. J. Climate, 2, 1069–1090. Knutson, T. R., and S. Manabe, 1994: Impact of increasing CO2 on simulated ENSO-like phenomena. Geophys. Res. Lett., 21, 2295– 2298. , and , 1995: Time-mean response over the tropical Pacific to increased CO2 in a coupled ocean–atmosphere model. J.Climate, 8, 2181–2199. , , and D. Gu, 1997: Simulated ENSO in a global coupled ocean–atmosphere model: Multidecadal amplitude modulation and CO2 sensitivity J. Climate, 10, 138–161. Lau, K.-M., and P. J. Sheu, 1990: Teleconnections in global rainfall anomalies: Seasonal to inter-decadal time scales. Teleconnections Linking Worldwide Climate Anomalies, M. Glantz, W. Katz, and N. Nicholls, Eds., Cambridge University Press, 227–256. McGregor, J. L., H. B. Gordon, I. G. Watterson, M. R. Dix, and L. D. Rotstayn, 1993: The CSIRO 9-level atmospheric general circulation model. CSIRO-Division of Atmospheric Res. Tech. Paper 26, 89 pp. [Available from CSIRO DAR, PBN1, Aspendale, 3195, Australia.] Mechoso, C. R., and Coauthors, 1995: The seasonal cycle over the tropical Pacific in coupled ocean–atmosphere general circulation models. Mon. Wea. Rev., 123, 2825–2838. Meehl, G. A., and W. M. Washington, 1993: South Asian summer monsoon variability in a model with doubled atmospheric carbon dioxide concentration. Science, 260, 1101–1104. , G. W. Branstator, and W. M. Washington, 1993: Tropical Pacific interannual variability and CO2 climate change. J. Climate, 6, 42–63. Nigam, S., and H.-S. Shen, 1993: Structure of oceanic and atmospheric low-frequency variability over the tropical Pacific and Indian Oceans. Part I: COADS observations. J.Climate, 6, 657– 676.

352

JOURNAL OF CLIMATE

Ramanathan, V., and W. Collins, 1993: A thermostat in the tropics? Nature, 361, 410–411. Reynolds, R. W., 1988: A real-time global sea surface temperature analysis. J. Climate, 1, 75–86. Ropelewski, C. F., and M. S. Halpert, 1987: Global and regional scale precipitation patterns associated with the El Nin˜o/Southern Oscillation. Mon. Wea. Rev., 115, 1606–1626. Smith, I. N., 1994: A GCM simulation of global climate trends: 1950– 88. J. Climate, 7, 732–744. , 1995: A GCM simulation of global climate interannual variability 1950–88. J. Climate, 8, 709–718. Tett, S. F. B., 1995: Simulation of El Nin˜o–Southern Oscillation-like

VOLUME 10

variability in a global AOGCM and its response to CO2 increase. J. Climate, 8, 1473–1502. Watterson, I. G., M. R. Dix, H. B. Gordon, and J. L. McGregor, 1995: The CSIRO nine-level atmospheric general circulation model and its equilibrium present and doubled CO2 climates. Aust. Meteor. Mag., 44, 111–125. Whetton, P. H., and I. Rutherfurd, 1994: Historical ENSO teleconnections in the eastern hemisphere. Climate Change, 28, 221– 253. , A. M. Fowler, M. R. Haylock, and A. B. Pittock, 1993: Implications of climate change due to the enhanced greenhouse effect on floods and droughts in Australia. Climate Change, 25, 289–317.