SST Variability and Its Mechanism in a Coupled Atmosphere–Mixed Layer Ocean Model

Akio Kitoh Meteorological Research Institute, Tsukuba, Ibaraki, Japan

Search for other papers by Akio Kitoh in
Current site
Google Scholar
PubMed
Close
,
Tatsuo Motoi Meteorological Research Institute, Tsukuba, Ibaraki, Japan

Search for other papers by Tatsuo Motoi in
Current site
Google Scholar
PubMed
Close
, and
Hiroshi Koide Meteorological Research Institute, Tsukuba, Ibaraki, Japan

Search for other papers by Hiroshi Koide in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Interannual SST variability in a coupled atmosphere–mixed layer ocean model is investigated. This model has no El Niño but shows a large interannual SST variability in the tropical Pacific. The basin-scale feature of SST variation has some common characteristics shared with that obtained by a global ocean–atmosphere coupled GCM and observational data in the subtropical to the midlatitude Pacific. Both the latent heat flux and shortwave radiation have their roles in producing the SST anomalies. There is no large contrast in the total heat flux between the eastern and the western Pacific. However, their main components, the shortwave radiation and the latent heat flux, have a remarkable contrast between the cold tongue in the east and the warm pool region in the west. In the east, the ocean is warmed by shortwave radiation and cooled by latent heat. This shortwave radiation is negatively correlated with low-level clouds. When the SST is warmer than normal in the eastern Pacific, there is less low-level stratus cloud cover and more shortwave radiation reaching the surface. In the western Pacific, the ocean is warmed by less evaporation due to weaker winds. When the ocean becomes warm, it is cooled by less shortwave radiation due to stronger activity in cumulus convection.

Corresponding author address: Dr. Akio Kitoh, Meteorological Research Institute, 1-1 Nagamine, Tsukuba, Ibaraki 305-0052, Japan.

Email: kitoh@mri-jma.go.jp

Abstract

Interannual SST variability in a coupled atmosphere–mixed layer ocean model is investigated. This model has no El Niño but shows a large interannual SST variability in the tropical Pacific. The basin-scale feature of SST variation has some common characteristics shared with that obtained by a global ocean–atmosphere coupled GCM and observational data in the subtropical to the midlatitude Pacific. Both the latent heat flux and shortwave radiation have their roles in producing the SST anomalies. There is no large contrast in the total heat flux between the eastern and the western Pacific. However, their main components, the shortwave radiation and the latent heat flux, have a remarkable contrast between the cold tongue in the east and the warm pool region in the west. In the east, the ocean is warmed by shortwave radiation and cooled by latent heat. This shortwave radiation is negatively correlated with low-level clouds. When the SST is warmer than normal in the eastern Pacific, there is less low-level stratus cloud cover and more shortwave radiation reaching the surface. In the western Pacific, the ocean is warmed by less evaporation due to weaker winds. When the ocean becomes warm, it is cooled by less shortwave radiation due to stronger activity in cumulus convection.

Corresponding author address: Dr. Akio Kitoh, Meteorological Research Institute, 1-1 Nagamine, Tsukuba, Ibaraki 305-0052, Japan.

Email: kitoh@mri-jma.go.jp

1. Introduction

Sea surface temperature (SST) anomalies have profound effects on the atmospheric general circulation and seasonal climate anomalies, and thus many analyses and experiments are performed to reveal their mechanism and predictability. Large-scale SST variability in the Pacific has a distinct spatial feature both in interannual and interdecadal timescales. It bears an opposite polarity between the Tropics and the North Pacific. Figure 1a shows a dominant spatial pattern of the observed SST anomalies based on the annual mean data for the period 1949–94. Here the monthly Global Sea Ice and Sea Surface Temperature dataset (GISST2, Rayner et al. 1995) is used. In the Pacific, a wedgelike pattern of positive SST anomalies extends from the central equatorial Pacific to off the coast of California. There extends a warm SST area along the western coast of North America from off the coast of California toward the Gulf of Alaska. Corresponding to positive SST anomalies in the tropical Pacific, there appears a region with cold SST anomalies in the central North Pacific. In the South Pacific, negative anomalies extend southeast from New Guinea to around 30°S, 120°W. The Tasman Sea is also covered by below normal SST.

Global ocean–atmosphere coupled general circulation models (GCM) can simulate some features of the observed interannual SST anomaly pattern. An example is shown in Fig. 1b, which is a leading EOF pattern of the interannual SST variability appearing in a long-term integration of the Meteorological Research Institute (MRI) global coupled GCM (Yukimoto et al. 1996). The coupled model reveals interannual variability in the tropical Pacific, which has several typical characteristics shared with the observed El Niño–Southern Oscillation (ENSO). The basin-scale feature of the North Pacific SST variation shows negative correlation in the central North Pacific with the tropical SST while showing positive correlation along the western coast of North America. It is noted that the coupled model shows a positive SST anomaly around 50°S, 120°W in conjunction with the warm equatorial SST anomaly. This is also the case for the real world. Many researchers have investigated the mechanisms of creating the SST anomaly pattern from various points of view including heat flux and oceanic heat transport, but in this paper we try to explain this by use of a mixed layer ocean model that excludes the effect of ocean dynamics in creating SST anomalies. This approach will distinguish the relative importance of in situ heat flux and ocean heat transport.

A coupled atmosphere–mixed layer ocean model is in the coupled model hierarchy, which allows SST to change as a result of atmosphere–ocean heat flux exchange but with implicitly prescribed climatological ocean heat transport. These coupled atmosphere–mixed layer ocean models are extensively used for equilibrium CO2 doubling experiments (Hansen et al. 1981; Wilson and Mitchell 1987; Wetherald and Manabe 1988) and paleoclimate simulations (Kutzbach and Gallimore 1988; Mitchell et al. 1988).

This type of mixed layer ocean model can produce significant interannual and interdecadal variability. Barnett et al. (1992) and Miller and Del Genio (1994) showed that the model’s variability can be attributed to a tropical cloud feedback mechanism. Davies and Hunt (1994) investigated a 500-yr run of a mixed layer ocean model and found decadal warming and cooling episodes in globally averaged annual mean surface temperature up to 0.7°C. Hunt and Davies (1997) showed that those episodes are associated with cloud variability in the tropical Pacific. They also showed the occurrence of pulses of 3–5 yr in duration. These pulses were associated with westerly wind bursts over the western Pacific that reduce wind speeds and reduce evaporation and thus increase SST. Manabe and Stouffer (1996) compared the variability of SST and surface air temperature in a 1000-yr integration of a coupled GCM and a mixed layer ocean model. They showed a similar geographical distribution and spectral power characteristics of the two models at interannual and interdecadal timescales. Although their coupled model underestimates the power spectrum regarding ENSO due to the coarse resolution, the mixed layer model also has relatively small variability of SST in the tropical Pacific.

Apparently the mechanism of the interannual and decadal natural climatic variability has not been fully resolved yet. In this paper we focus on the interannual variability and show that a coupled atmosphere–mixed layer ocean model can have a large interannual tropical SST variability. We begin with the model description in section 2, followed by the model climatology in section 3. Section 4 shows the dominant SST mode and associated atmospheric circulation changes. A mechanism of SST anomalies in the tropical Pacific is investigated in section 5. Discussion and summary are made in section 6.

2. The model

The coupled atmosphere–mixed layer ocean model used for this study is the same as that which Kitoh (1997) used to study the effect of mountain uplift on surface temperature changes. The model is based on the MRI global ocean–atmosphere coupled GCM (Tokioka et al. 1995). While the original coupled GCM has a fine horizontal resolution for the oceanic GCM, the horizontal resolution of the model in this paper is set as 4° lat × 5° long both in the atmosphere and in the ocean.

The atmospheric model has 15 vertical levels with its top at 1 hPa. Both the seasonal cycle and diurnal cycle of solar insolation are included. Shortwave radiation is calculated based on the parameterization of Lacis and Hansen (1974). Calculation of longwave radiation is based on the multiparameter random model by Shibata and Aoki (1989). Parameterization of cumulus convection is based on the scheme of Arakawa and Schubert (1974) but is modified by Tokioka et al. (1988). Five types of clouds are diagnostically determined. They are penetrative cumulus, midlevel convective, planetary boundary layer stratus, large-scale condensation, and cirrus anvil clouds. The latter three cloud types cover the entire grid box (cloud fraction = 1). Convective clouds are assumed to overlap randomly with a total cloud fraction of 0.3. Cirrus anvil cloud forms if cumulus convection penetrates to levels above 400 hPa. The performance of this atmospheric model is described in Kitoh et al. (1995).

The oceanic model was replaced with a constant 50-m depth mixed layer ocean (slab) model. The SST is calculated every 5 days from the heat balance of the 50-m depth slab using heat fluxes at the sea surface. A sea ice model, also included, calculates sea ice concentration and thickness (Tokioka et al. 1996).

To compensate for the absence of heat transport by ocean currents and upwelling, a heat flux adjustment term is added to the equation that calculates the SST. Monthly mean heat flux value is thus obtained by running the model for 3 yr by restoring the model SST to the observed SST. In this process, the seasonal march of the sea ice is used to obtain additional heat flux adjustment values in the polar oceans. Annual mean values of the prescribed heat flux (Q flux) are shown in Fig. 2. Heat is removed from the tropical oceans and added to the oceans in mid- and high latitudes. The model was then integrated for 60 model years.

3. Model climate

In this section we show the model climate as obtained with a long integration of the coupled atmosphere–mixed layer ocean model. Figure 3 shows the 60-yr averaged annual mean SST, precipitation, and surface wind stress. As we use the Q flux method, the model reproduces large-scale features of the observed SST distribution quite realistically with a warm water pool over the western tropical Pacific and Indian Oceans (Fig. 3a). The simulated SST over the Indian Ocean is slightly warmer than the observed SST. There appears to be a fine meridional structure east of the Philippines. This may be related to the fact that no horizontal diffusion process is working in the mixed layer ocean. Heat fluxes between the atmosphere and the ocean alone may not be sufficient to remove this fine structure in SST.

As the model well reproduces the SST, atmospheric circulation fields are very close to those in the atmospheric GCM forced with the observed SST (Kitoh et al. 1995). Simulated precipitation (Fig. 3b) captures salient observed characteristics with heavy precipitation maxima over the Asian–Australian, the American, and the African monsoon regions. There is a distinct ITCZ north of the equator in the central and the eastern Pacific. The South Pacific convergence zone extends eastward from New Guinea but has a tendency to become the double ITCZ in the eastern Pacific. This is a symptom found in other coupled GCMs (e.g., Mechoso et al. 1995). Dry regions cover the subtropical land and the subtropical eastern oceans. The anticyclonic circulation at the surface can be seen over the subtropical Pacific (Fig. 3c). Northeasterly trades with their maximum between Hawaii and the date line are simulated in the model.

4. SST variation and the associatedatmospheric circulation

Although a coupled atmosphere–mixed layer ocean model does not have the ocean dynamics that are believed to be responsible for causing the El Niño phenomenon (Philander 1990), it can have as large interannual variability as full coupled GCMs or the observation show. Actually, our mixed layer model has larger SST variability than either the coupled model version of the MRI GCM or the observation. Figure 4 compares the standard deviation of annual mean SST by the mixed layer model with the observation. The observed data show the maximum variability over the eastern equatorial Pacific. They also show local maxima over the central equatorial Pacific and off the coast of California. The simulated SST variability is very different from the observation. It is shown that the tropical Pacific is the region with the largest variability of the simulated SST. It exceeds 1°C along the equator with its maximum value at the central equatorial Pacific. Variability in the South Pacific extending southeastward from the central equatorial Pacific is also large. In the North Pacific, a region with large SST variability extends from east of the Philippines to around Hawaii. It is noteworthy that there are local minima of variability between the equatorial Pacific and the two elongated regions extending from the western equatorial Pacific to subtropical eastern Pacific. As both the observed and the coupled GCM have standard deviations in annual mean SST of about 0.5°C at the equatorial Pacific (Manabe and Stouffer 1996; Yukimoto et al. 1996), this mixed layer model is twice as energetic compared to the coupled model that includes ocean dynamics.

Figure 5 shows the geographical distribution of the lag 1 yr autocorrelation of the annual mean SST of the mixed layer model. It is greater than 0.4 in the tropical Pacific and in the central North Pacific. Lag 1 yr autocorrelation of the observed and the coupled model both tend to have larger values (greater than 0.4) in the central equatorial Pacific (not shown). Compared to other mixed layer model results, such as those shown by Manabe and Stouffer (1996), this model has larger persistence of the SST anomaly, particularly in the Tropics.

In order to investigate spatial patterns of the SST variations simulated by the mixed layer ocean model, we made an EOF analysis using the 60-yr annual mean SST anomalies. Figure 6 shows the first EOF and the corresponding spatial pattern. This first EOF explains 19.8% of the total variance. The correlation coefficients between the first principal component (Fig. 6b) and SST anomalies at all ocean grid points are plotted in Fig. 6a. As shown in Fig. 4, the model simulates rather large interannual variability of SST in the tropical Pacific. The peak-to-peak value of the SST anomalies in the equatorial Pacific (10°S–10°N, 120°E–90°W) reaches around 2°C. Being different from the observed El Niño that is characterized by a contrast between the eastern and the western equatorial Pacific SST anomalies, there is no distinct east–west contrast in the SST variation in this model. It tends to change signs in the entire tropical Pacific as far as the annual mean SST anomaly is concerned. This different behavior may be a manifestation that this model does not include ocean dynamics. As shown in Fig. 5, the persistence of simulated SST is large in the tropical Pacific and exceeds 0.5 in lag 1-yr autocorrelation in the equatorial strip. This is confirmed by calculating autocorrelation of the first principal component. They are 0.582, 0.065, and −0.184 at lag 1, 2, and 3 yr, respectively.

Except for the difference in the equatorial Pacific, which may be attributed to the absence of the ocean dynamics in the mixed layer ocean model, the spatial pattern of this first EOF has some similarity to that obtained by a global coupled ocean–atmosphere GCM and observational data in the Pacific, which is shown in Fig. 1. In the subtropical North Pacific between Hawaii and California, positive anomalies extend from southwest to northeast. There is a negative anomaly in the North Pacific east of Japan. In the South Pacific, the negative anomaly east of Australia extends from northwest to southeast. It is noteworthy that the two models share another similar feature with a positive anomaly in the South Pacific around 50°–60°S and a negative anomaly south of Australia.

Figure 7 shows the simultaneous regression coefficients between the first principal component of SST (Fig. 6b) and the 200-hPa geopotential height, sea level pressure, and precipitation based on the annual mean values. Regression coefficients are normalized so that the SST anomaly in the central equatorial Pacific (4°S–4°N, 160°E–150°W) corresponds to 1°C. Regions where regression coefficients are statistically significant at the 95% level are hatched.

The geopotential height anomaly at 200 hPa (Fig. 7a) is nearly symmetric about the equator with positive anomalies in the Tropics and negative anomalies in the midlatitudes when the tropical Pacific SST is warmer than normal. There are quasi-stationary Rossby wave trains in both hemispheres emanating from the central Pacific around the date line. This atmospheric response is similar to that obtained during the warm phase of ENSO (e.g., Horel and Wallace 1981).

The sea level pressure anomaly in the Tropics associated with the dominant SST variability is characterized by a negative sea level pressure anomaly in the Pacific and a positive anomaly in the Indian Ocean through the Atlantic Ocean (Fig. 7b). This feature is similar to the observed ENSO (Trenberth and Shea 1987). However, the region with negative sea level pressure anomalies extends more westward than the observed and occupies the western Pacific. This discrepancy comes from the fact that the entire tropical Pacific becomes warm in the model (Fig. 6a).

A comparison between the sea level pressure and the 200-hPa height anomalies reveals barotropic patterns in the mid- and high latitudes. There is a clear indication of large barotropic anomalies over the Southern Ocean:an anticyclonic circulation at 100°W and a cyclonic circulation at 100°E along the 60°S latitude circle. It is noted that positive and negative SST anomalies are located near the anticyclonic and cyclonic circulation anomalies, respectively (Fig. 6a). This sea level pressure anomaly in high southern latitudes is not well documented with the observed data due to data paucity in this region. However, limited data analysis reveals a positive sea level pressure anomaly in the Pacific sector and a negative anomaly in the Atlantic and Indian sectors over the Southern Ocean at the time of El Niño (Kitoh 1994).

The precipitation anomaly (Fig. 7c) generally corresponds to the SST anomalies, with heavier precipitation over the regions with warmer SST. Overall precipitation anomaly in the Tropics is an eastward shift of precipitation area and resembles the observed pattern associated with ENSO (Ropelewski and Halpert 1989). Over land, significantly negative precipitation is simulated over India when tropical Pacific SST is above normal. As the Indian precipitation anomaly mainly comes from the summer season, this relationship between the tropical Pacific SST and the Indian monsoon rainfall is consistent with the observed ENSO–monsoon relationship; that is, strong monsoon is associated with La Niña and vice versa.

5. Analysis of tropical SST variability

Because only the climatological ocean heat transport is included in this mixed layer ocean model in the context of the prescribed Q flux, any interannual variations of SST occur only through variations in air–sea heat exchanges. Besides the Q flux, the oceanic surface heat budget consists of shortwave radiation (SW), longwave radiation (LW), sensible heat flux (SH), and latent heat flux (LH). Throughout this paper we use the convention that a positive flux value implies that the heat flux acts to increase SST. Therefore, evaporation from the ocean is indicated as negative values.

a. Relative importance of heat flux components

Figure 8 shows relative roles of these four heat flux components on the interannual SST variations at each grid point. We calculated regression coefficients of each variable on the local SST tendency using the annual mean values. When all values are combined, it becomes 1. Negative value means that the flux acts to increase SST while the real SST tends to increase.

It is clear from Fig. 8 that LH is the most dominant term in contributing to the interannual SST variability in the mixed layer ocean model; that is, smaller evaporation means warmer SST and vice versa. This term exceeds the other terms in most regions. Over the tropical and the subtropical oceans the latent heat flux term explains more than 60% of the SST variability.

Contribution from SW is large over the tropical oceans. Actually, the shortwave radiation over the equatorial eastern Pacific and the equatorial Atlantic is larger than the latent heat flux and is the dominant term in regulating the local SST variability. This shortwave radiation component has a comparable contribution to the latent heat flux component over the tropical western Pacific. Variation of the shortwave flux has a good relationship with that of cloudiness, as will be shown later in Figs. 14c and 15d.

The role of SH is generally low in the low and middle latitudes but is comparable to that of the shortwave heat flux in the high latitudes. Particularly over the Southern Ocean, the sensible heat flux plays the most important role for the SST variations. This area corresponds to near the sea ice edge where variability of cold surface winds over the open ocean is responsible for the large sensible heat flux variations. The relative role of LW is generally small. It is negative in the Tropics and subtropics and mostly acts to dampen the SST anomalies.

In short, the changes in the shortwave radiation and the latent heat flux contribute to the interannual variability in the SST over the tropical Pacific. Over the North Pacific, change in the latent heat flux is the dominant player. This implies that the negative SST anomaly east of Japan is mainly produced by larger evaporative heat loss, which itself is associated with stronger surface winds (Tokioka et al. 1993). It should be stressed that the above analysis is based on the mixed layer ocean model that does not include the ocean dynamics that may play an important role at the equator. The role of the ocean dynamics will be discussed later.

b. Regressions in horizontal structure

Next, we investigate an evolution of the SST anomalies and the roles of the shortwave radiation and latent heat flux on it. Figures 9–13 show the lagged regression coefficients between the first principal component of SST (Fig. 6b) and (a) the SST, (b) the total heat flux, (c) the shortwave radiation, and (d) the latent heat flux and the surface wind stress with a lag −3 through +1 yr. Regression coefficients are normalized so that the maximum SST at no lag (Fig. 12a) equals 1°C. To change the ocean temperature of a 50-m slab by 1°C in one year, a heat flux of 6.3 W m−2 is needed. Therefore, heat flux values of 6.3 W m−2 in Figs. 12b–d correspond to 1°C SST change in one year.

At t = −3 yr, there is a negative SST anomaly in the equatorial Pacific (Fig. 9a). At this time the total heat flux anomaly is already positive in the eastern equatorial Pacific (Fig. 9b). The latent heat flux anomaly is responsible for this SST increase in the eastern Pacific (Fig. 9d). Smaller evaporation (positive value in the figure) is mainly due to the reduced wind speeds (weak trade winds). On the other hand, the shortwave radiation anomaly is negative in the tropical eastern Pacific and does not contribute to the SST increase. There is a good relationship between the shortwave radiation anomaly and the cloudiness anomaly; more cloudiness reduces the shortwave radiation into the ocean. From the annual mean values the simultaneous anomaly correlation coefficient exceeds −0.8 over most of the oceanic grid points. Correlation is relatively weaker over the equatorial central and western Pacific (around −0.4∼−0.6) but is still significant. The correlation coefficient between the latent heat flux and surface wind speeds is about 0.4; stronger surface wind speeds correspond to larger latent heat fluxes out of the ocean.

At t = −2 yr, the SST anomaly itself is nearly zero in the equatorial Pacific (Fig. 10a), while the total heat flux anomaly into the ocean is larger than that at t = −3 yr (Fig. 10b). Figure 10c shows a large change in the shortwave radiation compared to that in the previous year. The shortwave radiation anomaly is now positive in the tropical eastern Pacific. A positive SST anomaly off South America due to less evaporation (Fig. 9d) now contributes to reduced cloudiness and a large heating due to the shortwave radiation (Fig. 10c).

At t = −1 yr (Fig. 11a), the SST anomaly in the tropical Pacific becomes positive with the largest warming off the coast of South America that corresponds to the continued heating from the shortwave radiation anomaly from the previous year. The total heat flux anomaly becomes larger than at t = −2 yr. It is noted that the shortwave radiation is the dominant term at this stage for the warming in the central and the eastern Pacific. The latent heat flux now works to dampen the warm SST anomaly in the eastern Pacific (Fig. 11d). The role of this latent heat flux anomaly is opposite between the eastern Pacific and the western Pacific. It is negative in the eastern Pacific and positive in the western Pacific. The surface wind anomaly is easterly in the eastern Pacific and westerly in the western Pacific, corresponding to the increased and decreased wind speeds, respectively. This anomalous circulation can be interpreted as an atmospheric response to the El Niño–like SST anomaly. There appears to be a positive precipitation anomaly over the central Pacific. Both the shortwave radiation anomaly and the latent heat flux anomaly contribute to the SST warming in the western Pacific. The total heat flux anomaly (Fig. 11b) shows a clear contrast in the subtropical South Pacific with a zero line extending from northwest to southeast. This is reflected to a wedgelike feature of the SST variability (Fig. 6a).

At t = 0 yr, the overall pattern of the regression coefficients is similar to those at t = −1 yr, but the decrease in SST due to the latent heat flux anomaly becomes larger in the eastern Pacific (Fig. 12d) and surpasses the SST increase due to the shortwave radiation anomaly (Fig. 12c). The negative region of the total heat flux anomaly extends south of the equator in the eastern Pacific (Fig. 12b). At this stage the total heat flux anomaly in the western Pacific is still positive and the SST continues to warm up. Due to an east–west contrast of the total heat flux anomaly, the SST becomes cool in the eastern Pacific while it becomes warm in the western Pacific. This results in a westward movement of the SST anomaly along the equator. It is noted that, at this time, there appears to be a negative shortwave radiation anomaly in the western Pacific. The above normal SST in the western Pacific leads to increased cumulus activity and, thus, the increased cloudiness.

At t = +1 yr (Fig. 13a), the maximum SST region shifts westward and its magnitude decreases compared to that at t = 0 yr. A strong easterly wind anomaly contributes to the large evaporation over the entire tropical Pacific (Fig. 13d). The largest negative latent heat flux anomaly appears at the eastern equatorial Pacific. A negative heat flux anomaly due to shortwave radiation, which was found at t = 0 yr in the western Pacific, intensifies. The total heat flux anomaly is negative in the entire tropical Pacific, although the eastern half is mainly due to the anomalous evaporation and the western half is due to the solar radiation anomaly. This reduces positive SST anomaly and results in turning to negative SST anomaly.

c. Time-sequence analysis of lagged regressions by monthly mean data

In the previous section we explored the time evolution of the SST anomaly and the role of the heat flux components by using the annual mean data. In this section, in order to further investigate the time sequences of the shortwave radiation and the latent heat flux anomalies and their contrasting behavior between the eastern and the western Pacific, we investigate the lagged regression analysis using the monthly mean data. Figures 14 and 15 show the longitude–time diagrams showing the lagged regression coefficients of selected variables at the equator on the SST anomaly at 0°, 120°W. Calculations are made for lags from −4 to +4 yr. Values shown in Fig. 14 are the SST, the total heat flux, the shortwave flux, and the latent heat flux, and those in Fig. 15 are the zonal wind stress, the convective precipitation, the stratiform precipitation, and the total cloudiness. Heat fluxes shown are positive downward.

Figure 14a shows the SST history itself. This figure shows that there is a large negative SST anomaly at t = −4∼−3 yr, and the SST anomaly changes its sign from negative to positive around t = −2 yr, reaching the peak value at t = 0 by definition and again changes its sign around t = +2∼+2.5 yr. It becomes a large negative anomaly again around t = +4 yr. It also shows that the SST anomaly moves from the eastern equatorial Pacific to the western Pacific in about 1 yr (∼100° yr−1). These westward propagating modes are shown to result from wind–evaporation feedbacks by Xie (1996) and Miller and Jiang (1996).

Figure 14b shows the regression time series of the total heat flux, and Figs. 14c and 14d are contributions by the shortwave radiation and the latent heat flux, respectively. Maps for longwave radiation and sensible heat flux are not shown as they are of secondary importance (see Fig. 8). The total heat flux anomaly is positive from t = −4 yr to t = 0 and is negative from t = 0 to t = +4 yr. A maximum value appears at t = −4 months and a minimum appears at t = 4 month at the eastern equatorial Pacific.

Compared to the total heat flux, component fluxes, that is, the shortwave radiation and the latent heat flux, have two distinct characteristics. First, both the shortwave radiation and the latent heat flux show a contrast between the eastern and the western Pacific, although the total heat flux has little contrast in the zonal direction. Second, while the total heat flux changes its sign at t = 0 by definition of the lagged regressions, both the shortwave radiation anomaly and the latent heat flux anomaly tend to change their signs at around t = −2 yr and t = 2 yr in the eastern Pacific. Therefore, the total heat flux anomaly is a result of cancellation of these two dominant terms both in a temporal and a spatial sense.

There is a good relationship between the variation in the shortwave radiation and that in cloudiness. A comparison between Fig. 14c (shortwave radiation) and Fig. 15d (total cloudiness) reveals a large negative correlation, particularly in the eastern Pacific where stratiform clouds dominate. A 1% change of total cloudiness in the eastern Pacific roughly corresponds to a 2 W m−2 change of the surface shortwave flux in the model. In the western Pacific, the relationship between the shortwave radiation and the total cloudiness becomes unclear due to cancellation of convective and stratiform clouds as will be shown below.

Changes in convective precipitation and those in large-scale precipitation are shown in Figs. 15b and 15c, respectively. They show that changes in convective precipitation dominate in the western Pacific while changes in large-scale precipitation dominate in the eastern Pacific. It is also shown that when SST anomaly is positive there is a positive convective precipitation anomaly and a negative large-scale precipitation anomaly. Although convective precipitation is positively correlated to the underlying SST, its magnitude is larger in the western Pacific and less in the eastern Pacific. This may be explained by a difference in the climatological SST values between the two regions. In the western Pacific where SST is high, an SST anomaly can easily result in changes in convective precipitation. On the other hand in the eastern Pacific where SST is low, convective precipitation hardly occurs even when there is a positive SST anomaly. With a similar argument, the magnitude of large-scale precipitation is larger in the eastern Pacific where the climatologically low SST is a favorable condition for large-scale precipitation and resultant low-level stratiform clouds.

In order to investigate more closely different behaviors of convective and stratiform clouds between the eastern and the western Pacific, we now look at vertical distribution of cloudiness. Figure 16a shows the height–longitude cross section of the simulated climatological annual mean cloudiness at the equatorial Pacific region. There are two regions of large cloudiness. One is over the warm pool region in the western Pacific, where the mid- and upper troposphere are covered by tall clouds. They are associated with deep cumulus convection. The other maximum in cloudiness is found at the model’s lowest layer in the cold tongue region in the eastern Pacific. These clouds are due to large-scale condensation and nonprecipitating stratus clouds.

Figure 16b shows the vertical distribution of regression coefficients in cloudiness anomaly at t = 0 when the SST is the warmest. It is shown that convective clouds generally increase when SST is above normal. It dominates in the western Pacific, where the maximum increase in cloudiness appears around the midtroposphere with a slight compensating decrease in the highest clouds. The total cloudiness change is little in the western Pacific due to cancellation between increase in convective clouds and decrease in stratiform clouds. However, the shortwave radiation at the surface significantly decreases (Fig. 14c). A comparison of Fig. 14c (shortwave radiation) and Fig. 15b (convective precipitation) reveals a negative correlation in the western Pacific. This is because cloud optical thickness becomes due more to thick clouds in the western Pacific.

Figure 16b also shows that an overall decrease in low clouds is clearly seen associated with warmer surface temperatures. A negative correlation between the SST and the large-scale precipitation comes from the fact that the low-level stratiform clouds develop more when the underlying SST is cold as often observed in the eastern Pacific. The largest anomaly is located in the eastern Pacific.

This cloud–radiation feedback and its dominance in the eastern Pacific for the total heat flux term can be seen in Fig. 14c. When SST is low (t = −4∼−2 yr), there are more low-level clouds with a negative shortwave radiation anomaly. As SST warms by positive latent heat flux anomalies, low-level clouds decrease and the shortwave radiation anomaly becomes positive. Warmer SST further reduces clouds, which results in increased shortwave radiation. This SST tendency is reversed when the cooling by latent heat flux anomaly (Fig. 14d) exceeds the warming by shortwave radiation.

The latent heat flux (Fig. 14d) has two effects on the total heat flux in the eastern Pacific. One is the contribution on positive total heat flux anomaly of less evaporation. This can be seen at the beginning phase of SST increase (t = −4∼−2 yr) when the SST anomaly is negative. This anomalous heat input into the ocean by smaller latent heat flux is caused by weaker surface wind stress, which is associated with a westerly wind anomaly (Fig. 15a). The second role is related to a change of the SST tendency. When SST becomes warm, the shortwave radiation dominates the total heat flux as already shown (Fig. 14c). When the SST anomaly becomes positive, the anomalous east–west circulation with an upward motion around 150°W is produced. This results in the surface easterly wind anomaly in the eastern Pacific and the westerly wind anomaly in the western and the central Pacific (Fig. 15a). This easterly wind anomaly in the eastern Pacific corresponds to stronger surface winds and, thus, larger evaporation (Fig. 14d). Therefore, negative latent heat flux anomaly becomes larger as SST becomes warm and eventually exceeds the shortwave radiation, which occurs by definition at t = 0. This results in negative total heat flux and causes a beginning of SST decrease. As SST becomes cool in the eastern Pacific, low-level clouds increase and the shortwave radiation anomaly begins to decrease in turn.

In the western Pacific, the roles of the shortwave radiation and the latent heat flux are different from that in the eastern Pacific. When SST is low, there is less convection and shortwave radiation anomaly reaching the surface is positive. The period when the SST anomaly is negative is characterized by the time of a stronger Walker circulation with lower-tropospheric divergence located in the central Pacific (Fig. 15a). Thus the evaporation is larger due to stronger trade winds. When the SST anomaly becomes positive, the western Pacific is now covered by the westerly surface wind anomaly as the anomalous east–west circulation is established with lower-tropospheric convergence around 150°W. As shown in Figs. 14c and 14d, it is the latent heat flux, not the shortwave radiation, that contributes to heat up the ocean from t = −2 to t = 0 yr in the western Pacific. The shortwave radiation here is negatively correlated with cumulus convection, that is, the larger cumulus tower prevents the shortwave radiation from reaching the surface. As SST approaches its peak value in the western Pacific (t = +1 yr), a center of anomalous convection moves farther westward. The easterly wind anomaly also extends westward, and the latent heat flux anomaly becomes negative and reduces SST further. It is the shortwave radiation due to the reduced cumulus convection that is responsible for changing the SST tendency to be positive in the western Pacific region.

In summary, both the latent heat flux and the shortwave radiation are important in regulating the tropical SST variability in this mixed layer ocean model. It is found that their effects are opposite between the eastern and the western Pacific, and they also differ in different phases of the anomaly cycle. The eastern Pacific is warmed by the shortwave radiation and cooled by the latent heat flux, while the western Pacific is warmed by the latent heat flux and cooled by the shortwave radiation.

6. Discussion and conclusions

We have investigated the interannual variability of a mixed layer ocean coupled with an atmospheric GCM. This model uses a constant 50-m depth mixed layer and thus has no explicit ocean dynamics, which is believed to be essential for El Niño, but reveals a large interannual SST variability in the tropical Pacific.

The leading mode of this model’s SST variability has some similarity to that obtained by a global coupled ocean–atmosphere GCM and observational data in its spatial pattern in the extratropical oceans including the Southern Ocean. This implies that the interannual SST variability in the extratropical oceans is mainly driven by the heat flux anomaly, particularly the latent heat flux anomaly, which itself is driven by the tropical SST anomaly through atmospheric circulation changes, that is, teleconnections (e.g., Lau and Nath 1996; Kitoh 1991; Tokioka et al. 1993).

A careful comparison of Figs. 1b and 6a reveals a difference of the structure of the North Pacific negative SST anomaly between the two models. The anomaly produced by the global coupled ocean–atmosphere GCM has its maximum around the date line in the North Pacific while that in the mixed layer ocean model has its maximum in the western Pacific. This difference may be attributed to the Ekman-drift advection that is neglected in the mixed layer ocean model. Tokioka et al. (1993) analyzed the atmosphere–ocean coupled mode in the North Pacific in a coupled GCM where the deepening of the Aleutian low and the cooling of midlatitude SST appeared when model El Niño occurred. They calculated all components that produce SST changes and concluded that the evaporation and the Ekman-drift advection are the main controlling factors in creating the SST anomaly in the North Pacific. In the central North Pacific the anomaly induced by the Ekman drift is as large as or greater than that induced by the evaporation.

It is noted that our mixed layer model shows the positive SST anomaly in the southern South Pacific around 60°S when the tropical Pacific is at the time of warm phase. This feature is also noted in the coupled GCM and in the observed data. Recently, interannual variability of the ocean–atmosphere system over the Southern Ocean has been investigated by many researchers (e.g., White and Peterson 1996) and this coupled system is called the Antarctic circumpolar wave (ACW). Considering that this oceanic area corresponds to where an ACW activity is the largest in the coupled GCM (Motoi et al. 1998), there is a possibility that the SST variability in the Tropics may be responsible for the ACW.

In the model tropical Pacific, the solar radiation and the evaporation are the two dominant players in regulating interannual SST variations. However, their role is different between the eastern and the western Pacific and in different phases of the anomaly.

This is illustrated schematically in Fig. 17. When there is a cold SST anomaly (Fig. 17a), the dense low-level stratus clouds prevail in the eastern Pacific, which prevents solar radiation into the ocean. On the other hand, there is a diverging surface wind anomaly from the central part of the ocean and convection is less active there. As this wind anomaly superimposes on the climatological easterly trade wind, the wind speed decreases in the eastern Pacific and contributes to warming the SST. As the evaporation effect dominates the solar radiation effect, the SST increases. When the SST anomaly becomes positive (Figs. 17b,c), the stratus decreases and the resultant increased solar radiation further increases the SST. At this time of the positive SST anomaly, the convection anomaly appears in the central Pacific with the easterly surface wind anomaly to the east and the westerly wind anomaly to the west. This results in the cooling and the warming of SST due to the latent heat flux in the eastern and the western Pacific, respectively. As the SST anomaly grows (Fig. 17d), the Walker circulation anomaly intensifies and eventually the evaporative cooling dominates the warming by solar radiation, thus the SST begins to decrease (Fig. 17e). As the SST decreases the stratus increases and that contributes to lower the SST (Fig. 17f). In the western Pacific, it is the solar radiation by suppressed convection due to low SST that is responsible for increasing the SST (Figs. 17a,b). But when the SST anomaly becomes positive, warming due to less evaporation by the westerly wind anomaly converging into the central Pacific dominates (Fig. 17c). As SST becomes warmer, the convection increases and the resultant decreased solar radiation becomes responsible for a negative SST tendency (Fig. 17e). When the convection center moves westward, the evaporative cooling also acts to further decrease the SST in the western Pacific (Fig. 17f).

Therefore, the roles of the shortwave radiation and the latent heat flux are opposite between the eastern Pacific and the western Pacific; the eastern Pacific is warmed by shortwave radiation and cooled by latent heat flux, while the western Pacific is warmed by latent heat flux and cooled by shortwave radiation. The wind-evaporation feedback in the surface heat flux has recently been investigated to shape the climatological asymmetric SST pattern in the Tropics in the meridional direction (Xie and Philander 1994) or in the zonal direction (Dijkstra and Neelin 1995). Xie (1996) and Miller and Jiang (1996) show that wind-evaporation feedbacks can create westward-propagating atmosphere–ocean coupled modes in a region of mean easterlies. The westward propagating mode in this model shown in Fig. 14a can be explained by the wind-evaporation feedback.

Importance of low-level stratus clouds in regulating the tropical Pacific climate is shown by Ma et al. (1996) and Philander et al. (1996). In this paper we demonstrated by a coupled model that both the wind-evaporation feedback and the cloud feedback can be responsible for the interannual SST variability in the tropical Pacific. Their contrasting behavior between the eastern and the western Pacific comes from the fact that the Walker circulation anomaly has the opposing wind direction between the two regions and that the low-level stratus clouds are dominant in the eastern Pacific while the convective clouds dominate in the western Pacific.

Regression maps in Figs. 14 and 15 show that the timescale of the atmosphere–ocean coupled mode in our model is about 8–9 yr. The timescale may depend on the strength of wind-evaporation and cloud-radiation feedbacks and on the heat content of the ocean mixed layer (50 m in this case). However, further sensitivity experiments will be needed to investigate mechanisms to determine the timescale.

Several results have been reported concerning the interannual and decadal variability with the atmosphere–mixed layer ocean coupled model (e.g., Hunt and Davies 1997; Manabe and Stouffer 1996). Our results show much larger SST variability in the tropical Pacific compared to these previous models. As has been shown, the positive feedback among the low-level stratus clouds, the solar radiation, and the SST is the essence of the mechanism of the variability of our air–sea coupled mode. Therefore, how the stratus clouds are simulated in each model may be responsible for the different magnitude of the interannual variability among the models.

In our experiment the atmospheric GCM is coupled to a mixed layer ocean where there is no ocean dynamics, thus excluding the ENSO and the thermohaline circulation. In the coupled GCM the ocean current would effectively transport the SST anomaly westward and this may act to weaken the above mechanism. How the stratus–solar-radiation–SST feedback and the evaporation–wind–SST feedback are working in the coupled GCM and in the real tropical atmosphere–ocean system needs further investigation.

Acknowledgments

The authors are grateful to two anonymous reviewers whose comments were very useful in revising the original manuscript. Discussion with S.-P. Xie was invaluable. The authors thank the members of the Climate Research Department of the MRI for continuous discussion for this study.

REFERENCES

  • Arakawa, A., and W. H. Schubert, 1974: Interactions of a cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci.,31, 674–701.

  • Barnett, T. P., A. D. Del Genio, and R. A. Ruedy, 1992: Unforced decadal fluctuations in a coupled model of the atmosphere and ocean mixed layer. J. Geophys. Res.,97, 7341–7354.

  • Davies, H. L., and B. G. Hunt, 1994: The problem of detecting climatic change in the presence of climatic variability. J. Meteor. Soc. Japan,72, 765–771.

  • Dijkstra, H. A., and J. D. Neelin, 1995: Ocean–atmosphere interaction and the tropical climatology. Part II: Why the Pacific cold tongue is in the east. J. Climate,8, 1343–1359.

  • Hansen, J., D. Johnson, A. Lacis, S. Lebedeff, P. Lee, D. Rind, and G. Russel, 1981: Climate impact of increasing atmospheric carbon dioxide. Science,213, 957–966.

  • Horel, J. D., and J. M. Wallace, 1981: Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Hunt, B. G., and H. L. Davies, 1997: Mechanism of multi-decadal climatic variability in a global climatic model. Int. J. Climatol.,17, 565–580.

  • Kitoh, A., 1991: Interannual variability in an atmospheric GCM forced by the 1970–1989 SST. Part II: Low-frequency variability of the wintertime Northern Hemisphere extratropics. J. Meteor. Soc. Japan,69, 271–291.

  • ——, 1994: Tropical influence on the South Pacific double jet variability. Proc. NIPR Symp. Polar Meteor. Glaciol.,8, 34–45.

  • ——, 1997: Mountain uplift and surface temperature changes. Geophys. Res. Lett.,24, 185–188.

  • ——, A. Noda, Y. Nikaidou, T. Ose, and T. Tokioka, 1995: AMIP simulations of the MRI GCM. Pap. Meteor. Geophys.,45, 121–148.

  • Kutzbach, J. E., and R. G. Gallimore, 1988: Sensitivity of a coupled atmosphere/mixed layer ocean model to changes in orbital forcing at 9000 years B.P. J. Geophys. Res.,93, 803–821.

  • Lacis, A. A., and J. E. Hansen, 1974: A parameterization for the absorption of solar radiation in the earth’s atmosphere. J. Atmos. Sci.,31, 118–133.

  • Lau, N.-C., and M. J. Nath, 1996: The role of the “atmospheric bridge” in linking tropical Pacific ENSO events to extratropical SST anomalies. J. Climate,9, 2036–2057.

  • Ma, C.-C., C. R. Mechoso, A. W. Robertson, and A. Arakawa, 1996:Peruvian stratus clouds and the tropical Pacific circulation: A coupled ocean–atmosphere GCM study. J. Climate,9, 1635–1645.

  • Manabe, S., and R. J. Stouffer, 1996: Low-frequency variability of surface air temperature in a 1000-year integration of a coupled atmosphere–ocean–land surface model. J. Climate,9, 376–393.

  • 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.

  • Miller, R. L., and A. D. Del Genio, 1994: Tropical cloud feedbacks and natural variability of climate. J. Climate,7, 1388–1402.

  • ——, and X. Jiang, 1996: Surface energy fluxes and coupled variability in the Tropics of a coupled general circulation model. J. Climate,9, 1599–1620.

  • Mitchell, J. F. B., N. S. Grahame, and K. J. Needham, 1988: Climate simulations for 9000 years before present: Seasonal variations and effect of the Laurentide ice sheet. J. Geophys. Res.,93, 8283–8303.

  • Motoi, T., A. Kitoh, and H. Koide, 1998: Antarctic circumpolar wave in a coupled ocean–atmosphere model. Ann. Glaciol., in press.

  • Philander, S. G. H., 1990: El Niño, La Niña, and the Southern Oscillation. Academic Press, 289 pp.

  • ——, D. Gu, D. Halpern, G. Lambert, N.-C. Lau, T. Li, and R. C. Pacanowski, 1996: Why the ITCZ is mostly north of the equator. J. Climate,9, 2958–2972.

  • Rayner, N. A., C. K. Folland, D. E. Parker, and E. B. Horton, 1995:A new global sea-ice and sea surface temperature (GISST) data set for 1903–1994 for forcing climate models. Hadley Centre Internal Note 69, 9 pp. [Available from Hadley Centre for Climate Prediction and Research, Meteorological Office, London Road, Bracknell, Berkshire RG12 2SY, United Kingdom.].

  • Ropelewski, C. F., and M. S. Halpert, 1989: Precipitation patterns associated with the high index phase of the Southern Oscillation. J. Climate,2, 268–284.

  • Shibata, K., and T. Aoki, 1989: An infrared radiative scheme for the numerical models of weather and climate. J. Geophys. Res.,94, 14 923–14 943.

  • Tokioka, T., K. Yamazaki, A. Kitoh, and T. Ose, 1988: The equatorial 30–60 day oscillation and the Arakawa–Schubert penetrative cumulus parameterization. J. Meteor. Soc. Japan,66, 883–901.

  • ——, A. Kitoh, and S. Nakagawa, 1993: Interactions between lower atmosphere and the ocean realized in a coupled atmosphere–ocean general circulation model. Extended Abstracts, Int. WCRP Symp. on Clouds and Ocean in Climate, Nagoya, Japan, Nagoya University, 1.5–1.8.

  • ——, A. Noda, A. Kitoh, Y. Nikaidou, S. Nakagawa, T. Motoi, S. Yukimoto, and K. Takata, 1995: A transient CO2 experiment with the MRI CGCM—Quick report. J. Meteor. Soc. Japan,73, 817–826.

  • ——, ——, ——, ——, ——, ——, ——, and ——, 1996: A transient CO2 experiment with the MRI CGCM—Annual mean response. CGER’s Supercomputer Monograph Report, Vol. 2, CGER/NIES, Tsukuba, 86 pp. [Available from National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba, Ibaraki 305-0053, Japan.].

  • Trenberth, K. E., and D. J. Shea, 1987: On the evolution of the Southern Oscillation. Mon. Wea. Rev.,115, 3078–3096.

  • Wetherald, R. T., and S. Manabe, 1988: Cloud feedback processes in a general circulation model. J. Atmos. Sci.,45, 1397–1415.

  • White, W. B., and R. G. Peterson, 1996: An Antarctic circumpolar wave in surface pressure, wind, temperature and sea-ice extent. Nature,380, 699–702.

  • Wilson, C. A., and J. F. B. Mitchell, 1987: A doubled CO2 climate sensitivity experiment with a global climate model including a simple ocean. J. Geophys. Res.,92, 13 315–13 343.

  • Xie, S.-P., 1996: Westward propagation of latitudinal asymmetry in a coupled ocean–atmosphere model. J. Atmos. Sci.,53, 3236–3250.

  • ——, and S. G. H. Philander, 1994: A coupled ocean–atmosphere model of relevance to the ITCZ in the eastern Pacific. Tellus,46A, 340–350.

  • Yukimoto, S., M. Endoh, Y. Kitamura, A. Kitoh, T. Motoi, A. Noda, and T. Tokioka, 1996: Interannual and decadal variabilities in the Pacific in an MRI coupled GCM. Climate Dyn.,12, 667–683.

Fig. 1.
Fig. 1.

(a) Spatial pattern of the first EOF based on the observed annual mean SST. The GISST2 data (Rayner et al. 1995) from 1949 to 1994 are used. Correlation coefficients of local SST on the time series of the first EOF are plotted. The contour interval is 0.1 and negative values are hatched. (b) Same as (a) except the 71-yr annual mean SST of the MRI CGCM (Yukimoto et al. 1996) is used.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 2.
Fig. 2.

Annual mean prescribed heat flux. The contour interval is 20 W m−2. Negative values are hatched.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 3.
Fig. 3.

Annual mean 60-yr averaged model climates with the coupled atmosphere–mixed layer ocean model. (a) SST. The contour interval is 2°C. The 29°C contour is also drawn. Values larger than 28°C are hatched. (b) Precipitation. Contours are 1, 2, 5, and 9 mm day−1. (c) Surface wind stress. The reference vector is 0.1 N m−2. Values larger than 0.1 N m−2 are hatched.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 4.
Fig. 4.

Standard deviation of annual mean SST for (a) the observation and (b) the coupled atmosphere–mixed layer ocean model. The contour interval is 0.2°C. Values larger than 0.6 are hatched.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 5.
Fig. 5.

Lag 1-yr autocorrelation of the simulated annual mean SST. The contour interval is 0.1. Values larger than 0.4 are hatched.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 6.
Fig. 6.

(a) Spatial pattern of the first EOF based on the 60-yr annual mean SST of the coupled atmosphere–mixed layer ocean model. Correlation coefficients of local SST on the time series of the first EOF are plotted. The contour interval is 0.1 and negative values are hatched. (b) Time series of the first EOF of SST.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 7.
Fig. 7.

Regression coefficient maps between the first EOF of SST and (a) 200-hPa geopotential height, (b) mean sea level pressure, and (c) precipitation. Regression coefficients are normalized so that the SST anomaly in the central equatorial Pacific (4°S–4°N, 160°E–150°W) corresponds to 1°C. Contour intervals are (a) 5 m, (b) 0.5 hPa, and (c) 0.2 mm day−1, respectively. Regions where regressions are statistically significant at 95% level are hatched.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 8.
Fig. 8.

Relative contributions of (a) LH, (b) SW, (c) SH, and (d) LW fluxes on local SST interannual variability. The contour interval is 0.2. Positive values are those fluxes that act to increase SST.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 9.
Fig. 9.

Lagged regression coefficients of (a) SST, (b) total heat flux, (c) shortwave radiation, and (d) latent heat flux and surface wind stress at 3 yr before on the time series of the first EOF of SST (Fig. 6b). Contour intervals are (a) 0.2°C, (b)–(d) 1 W m−2. The reference vector in (d) is 0.01 N m−2. Positive values are those fluxes that increase SST. Values are normalized as the maximum SST at no lag (Fig. 12a) equals 1°C. Heat flux of 6.3 W m−2 corresponds to 1°C SST change. Spatial nine-point smoothing is applied before plotting.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 10.
Fig. 10.

Same as Fig. 9 except 2 yr before.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 11.
Fig. 11.

Same as Fig. 9 except 1 yr before.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 12.
Fig. 12.

Same as Fig. 9 except simultaneous regression (no time lag).

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 13.
Fig. 13.

Same as Fig. 9 except 1 yr later.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 14.
Fig. 14.

Time–longitude plots of regression coefficients of (a) SST, (b) total heat flux, (c) shortwave radiation flux, (d) latent heat flux on the SST anomaly at 0°, 120°W. Contour intervals are (a) 0.1°C and (b)–(d) 1 W m−2. Monthly mean data are used and lagged regression coefficients from t = −4 yr to t = +4 yr are shown.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 15.
Fig. 15.

Same as Fig. 14 except for (a) zonal wind stress, (b) convective precipitation, (c) stratiform precipitation, and (d) total cloudiness. Contour intervals are (a) 1 × 10−3 N m−2, (b) 0.3 mm day−1, (c) 0.03 mm day−1, and (d) 1%.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 16.
Fig. 16.

(a) Height–longitude plots of climatological distribution of simulated cloudiness at the equator. The contour interval is 10%. There are eight model levels in the troposphere and the values in the axis of ordinates correspond to pressure levels when sea level pressure is 1000 hPa. (b) Same as (a) except for regression coefficients for cloudiness on the SST anomaly at 0°, 120°W at t = 0. The contour interval is 1% and negative values are hatched.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Fig. 17.
Fig. 17.

Schematic diagram illustrating tropical Pacific SST changes. (a)–(g) roughly correspond to t = −4, −2, −1, 0, +1, +2, +4 yr, respectively. Vertical arrows with LH and SW denote SST changes by latent heat flux and shortwave radiation, respectively. Arrows out of the ocean indicate SST decrease and those into the ocean indicate SST increase. Horizontal arrows denote surface zonal wind anomaly. Length of arrows shows relative magnitude. Plus and minus signs in the ocean denote positive and negative SST anomalies, respectively.

Citation: Journal of Climate 12, 5; 10.1175/1520-0442(1999)012<1221:SVAIMI>2.0.CO;2

Save
  • Arakawa, A., and W. H. Schubert, 1974: Interactions of a cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci.,31, 674–701.

  • Barnett, T. P., A. D. Del Genio, and R. A. Ruedy, 1992: Unforced decadal fluctuations in a coupled model of the atmosphere and ocean mixed layer. J. Geophys. Res.,97, 7341–7354.

  • Davies, H. L., and B. G. Hunt, 1994: The problem of detecting climatic change in the presence of climatic variability. J. Meteor. Soc. Japan,72, 765–771.

  • Dijkstra, H. A., and J. D. Neelin, 1995: Ocean–atmosphere interaction and the tropical climatology. Part II: Why the Pacific cold tongue is in the east. J. Climate,8, 1343–1359.

  • Hansen, J., D. Johnson, A. Lacis, S. Lebedeff, P. Lee, D. Rind, and G. Russel, 1981: Climate impact of increasing atmospheric carbon dioxide. Science,213, 957–966.

  • Horel, J. D., and J. M. Wallace, 1981: Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Hunt, B. G., and H. L. Davies, 1997: Mechanism of multi-decadal climatic variability in a global climatic model. Int. J. Climatol.,17, 565–580.

  • Kitoh, A., 1991: Interannual variability in an atmospheric GCM forced by the 1970–1989 SST. Part II: Low-frequency variability of the wintertime Northern Hemisphere extratropics. J. Meteor. Soc. Japan,69, 271–291.

  • ——, 1994: Tropical influence on the South Pacific double jet variability. Proc. NIPR Symp. Polar Meteor. Glaciol.,8, 34–45.

  • ——, 1997: Mountain uplift and surface temperature changes. Geophys. Res. Lett.,24, 185–188.

  • ——, A. Noda, Y. Nikaidou, T. Ose, and T. Tokioka, 1995: AMIP simulations of the MRI GCM. Pap. Meteor. Geophys.,45, 121–148.

  • Kutzbach, J. E., and R. G. Gallimore, 1988: Sensitivity of a coupled atmosphere/mixed layer ocean model to changes in orbital forcing at 9000 years B.P. J. Geophys. Res.,93, 803–821.

  • Lacis, A. A., and J. E. Hansen, 1974: A parameterization for the absorption of solar radiation in the earth’s atmosphere. J. Atmos. Sci.,31, 118–133.

  • Lau, N.-C., and M. J. Nath, 1996: The role of the “atmospheric bridge” in linking tropical Pacific ENSO events to extratropical SST anomalies. J. Climate,9, 2036–2057.

  • Ma, C.-C., C. R. Mechoso, A. W. Robertson, and A. Arakawa, 1996:Peruvian stratus clouds and the tropical Pacific circulation: A coupled ocean–atmosphere GCM study. J. Climate,9, 1635–1645.

  • Manabe, S., and R. J. Stouffer, 1996: Low-frequency variability of surface air temperature in a 1000-year integration of a coupled atmosphere–ocean–land surface model. J. Climate,9, 376–393.

  • 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.

  • Miller, R. L., and A. D. Del Genio, 1994: Tropical cloud feedbacks and natural variability of climate. J. Climate,7, 1388–1402.

  • ——, and X. Jiang, 1996: Surface energy fluxes and coupled variability in the Tropics of a coupled general circulation model. J. Climate,9, 1599–1620.

  • Mitchell, J. F. B., N. S. Grahame, and K. J. Needham, 1988: Climate simulations for 9000 years before present: Seasonal variations and effect of the Laurentide ice sheet. J. Geophys. Res.,93, 8283–8303.

  • Motoi, T., A. Kitoh, and H. Koide, 1998: Antarctic circumpolar wave in a coupled ocean–atmosphere model. Ann. Glaciol., in press.

  • Philander, S. G. H., 1990: El Niño, La Niña, and the Southern Oscillation. Academic Press, 289 pp.

  • ——, D. Gu, D. Halpern, G. Lambert, N.-C. Lau, T. Li, and R. C. Pacanowski, 1996: Why the ITCZ is mostly north of the equator. J. Climate,9, 2958–2972.

  • Rayner, N. A., C. K. Folland, D. E. Parker, and E. B. Horton, 1995:A new global sea-ice and sea surface temperature (GISST) data set for 1903–1994 for forcing climate models. Hadley Centre Internal Note 69, 9 pp. [Available from Hadley Centre for Climate Prediction and Research, Meteorological Office, London Road, Bracknell, Berkshire RG12 2SY, United Kingdom.].

  • Ropelewski, C. F., and M. S. Halpert, 1989: Precipitation patterns associated with the high index phase of the Southern Oscillation. J. Climate,2, 268–284.

  • Shibata, K., and T. Aoki, 1989: An infrared radiative scheme for the numerical models of weather and climate. J. Geophys. Res.,94, 14 923–14 943.

  • Tokioka, T., K. Yamazaki, A. Kitoh, and T. Ose, 1988: The equatorial 30–60 day oscillation and the Arakawa–Schubert penetrative cumulus parameterization. J. Meteor. Soc. Japan,66, 883–901.

  • ——, A. Kitoh, and S. Nakagawa, 1993: Interactions between lower atmosphere and the ocean realized in a coupled atmosphere–ocean general circulation model. Extended Abstracts, Int. WCRP Symp. on Clouds and Ocean in Climate, Nagoya, Japan, Nagoya University, 1.5–1.8.

  • ——, A. Noda, A. Kitoh, Y. Nikaidou, S. Nakagawa, T. Motoi, S. Yukimoto, and K. Takata, 1995: A transient CO2 experiment with the MRI CGCM—Quick report. J. Meteor. Soc. Japan,73, 817–826.

  • ——, ——, ——, ——, ——, ——, ——, and ——, 1996: A transient CO2 experiment with the MRI CGCM—Annual mean response. CGER’s Supercomputer Monograph Report, Vol. 2, CGER/NIES, Tsukuba, 86 pp. [Available from National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba, Ibaraki 305-0053, Japan.].

  • Trenberth, K. E., and D. J. Shea, 1987: On the evolution of the Southern Oscillation. Mon. Wea. Rev.,115, 3078–3096.

  • Wetherald, R. T., and S. Manabe, 1988: Cloud feedback processes in a general circulation model. J. Atmos. Sci.,45, 1397–1415.

  • White, W. B., and R. G. Peterson, 1996: An Antarctic circumpolar wave in surface pressure, wind, temperature and sea-ice extent. Nature,380, 699–702.

  • Wilson, C. A., and J. F. B. Mitchell, 1987: A doubled CO2 climate sensitivity experiment with a global climate model including a simple ocean. J. Geophys. Res.,92, 13 315–13 343.

  • Xie, S.-P., 1996: Westward propagation of latitudinal asymmetry in a coupled ocean–atmosphere model. J. Atmos. Sci.,53, 3236–3250.

  • ——, and S. G. H. Philander, 1994: A coupled ocean–atmosphere model of relevance to the ITCZ in the eastern Pacific. Tellus,46A, 340–350.

  • Yukimoto, S., M. Endoh, Y. Kitamura, A. Kitoh, T. Motoi, A. Noda, and T. Tokioka, 1996: Interannual and decadal variabilities in the Pacific in an MRI coupled GCM. Climate Dyn.,12, 667–683.

  • Fig. 1.

    (a) Spatial pattern of the first EOF based on the observed annual mean SST. The GISST2 data (Rayner et al. 1995) from 1949 to 1994 are used. Correlation coefficients of local SST on the time series of the first EOF are plotted. The contour interval is 0.1 and negative values are hatched. (b) Same as (a) except the 71-yr annual mean SST of the MRI CGCM (Yukimoto et al. 1996) is used.

  • Fig. 2.

    Annual mean prescribed heat flux. The contour interval is 20 W m−2. Negative values are hatched.

  • Fig. 3.

    Annual mean 60-yr averaged model climates with the coupled atmosphere–mixed layer ocean model. (a) SST. The contour interval is 2°C. The 29°C contour is also drawn. Values larger than 28°C are hatched. (b) Precipitation. Contours are 1, 2, 5, and 9 mm day−1. (c) Surface wind stress. The reference vector is 0.1 N m−2. Values larger than 0.1 N m−2 are hatched.

  • Fig. 4.

    Standard deviation of annual mean SST for (a) the observation and (b) the coupled atmosphere–mixed layer ocean model. The contour interval is 0.2°C. Values larger than 0.6 are hatched.

  • Fig. 5.

    Lag 1-yr autocorrelation of the simulated annual mean SST. The contour interval is 0.1. Values larger than 0.4 are hatched.

  • Fig. 6.

    (a) Spatial pattern of the first EOF based on the 60-yr annual mean SST of the coupled atmosphere–mixed layer ocean model. Correlation coefficients of local SST on the time series of the first EOF are plotted. The contour interval is 0.1 and negative values are hatched. (b) Time series of the first EOF of SST.

  • Fig. 7.

    Regression coefficient maps between the first EOF of SST and (a) 200-hPa geopotential height, (b) mean sea level pressure, and (c) precipitation. Regression coefficients are normalized so that the SST anomaly in the central equatorial Pacific (4°S–4°N, 160°E–150°W) corresponds to 1°C. Contour intervals are (a) 5 m, (b) 0.5 hPa, and (c) 0.2 mm day−1, respectively. Regions where regressions are statistically significant at 95% level are hatched.

  • Fig. 8.

    Relative contributions of (a) LH, (b) SW, (c) SH, and (d) LW fluxes on local SST interannual variability. The contour interval is 0.2. Positive values are those fluxes that act to increase SST.

  • Fig. 9.

    Lagged regression coefficients of (a) SST, (b) total heat flux, (c) shortwave radiation, and (d) latent heat flux and surface wind stress at 3 yr before on the time series of the first EOF of SST (Fig. 6b). Contour intervals are (a) 0.2°C, (b)–(d) 1 W m−2. The reference vector in (d) is 0.01 N m−2. Positive values are those fluxes that increase SST. Values are normalized as the maximum SST at no lag (Fig. 12a) equals 1°C. Heat flux of 6.3 W m−2 corresponds to 1°C SST change. Spatial nine-point smoothing is applied before plotting.

  • Fig. 10.

    Same as Fig. 9 except 2 yr before.

  • Fig. 11.

    Same as Fig. 9 except 1 yr before.

  • Fig. 12.

    Same as Fig. 9 except simultaneous regression (no time lag).

  • Fig. 13.

    Same as Fig. 9 except 1 yr later.

  • Fig. 14.

    Time–longitude plots of regression coefficients of (a) SST, (b) total heat flux, (c) shortwave radiation flux, (d) latent heat flux on the SST anomaly at 0°, 120°W. Contour intervals are (a) 0.1°C and (b)–(d) 1 W m−2. Monthly mean data are used and lagged regression coefficients from t = −4 yr to t = +4 yr are shown.

  • Fig. 15.

    Same as Fig. 14 except for (a) zonal wind stress, (b) convective precipitation, (c) stratiform precipitation, and (d) total cloudiness. Contour intervals are (a) 1 × 10−3 N m−2, (b) 0.3 mm day−1, (c) 0.03 mm day−1, and (d) 1%.

  • Fig. 16.

    (a) Height–longitude plots of climatological distribution of simulated cloudiness at the equator. The contour interval is 10%. There are eight model levels in the troposphere and the values in the axis of ordinates correspond to pressure levels when sea level pressure is 1000 hPa. (b) Same as (a) except for regression coefficients for cloudiness on the SST anomaly at 0°, 120°W at t = 0. The contour interval is 1% and negative values are hatched.

  • Fig. 17.

    Schematic diagram illustrating tropical Pacific SST changes. (a)–(g) roughly correspond to t = −4, −2, −1, 0, +1, +2, +4 yr, respectively. Vertical arrows with LH and SW denote SST changes by latent heat flux and shortwave radiation, respectively. Arrows out of the ocean indicate SST decrease and those into the ocean indicate SST increase. Horizontal arrows denote surface zonal wind anomaly. Length of arrows shows relative magnitude. Plus and minus signs in the ocean denote positive and negative SST anomalies, respectively.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 247 27 2
PDF Downloads 76 11 1