Seasonal Changes in Tropical SST and the Surface ... - AMS Journals

Download PDF
7 downloads 465 Views 2MB Size Report
Comment
Aug 15, 2015 - cell in boreal winter (summer) (Seo et al. 2014). However, the ... et al. 2014). TABLE 1. List of the 31 CMIP5 models used in the present study.
15 AUGUST 2015

HUANG

6503

Seasonal Changes in Tropical SST and the Surface Energy Budget under Global Warming Projected by CMIP5 Models PING HUANG Center for Monsoon System Research, Institute of Atmospheric Physics, Chinese Academy of Sciences, and Joint Center for Global Change Studies, Beijing, China (Manuscript received 16 January 2015, in final form 12 April 2015) ABSTRACT The seasonal changes in tropical SST under global warming are investigated based on the representative concentration pathway 8.5 (RCP8.5) and historical runs in 31 models from phase 5 of CMIP (CMIP5). The tropical SST changes show three pronounced seasonal patterns: the peak locking to the equator throughout the year and the weaker equatorial changes and stronger hemispheric asymmetric changes (HACs) in boreal autumn. The magnitude of the seasonal patterns is comparable to the tropical-mean warming and the annual-mean patterns, implying great impacts on global climate changes. The peak locking to the equator is a result of the equatorial locking of the minimum damping of climatological latent heat flux and the ocean heat transport changes. Excluding the role of ocean heat transport suggested in previous studies, the weaker equatorial warming in boreal autumn is contributed by stronger evaporation damping as a result of stronger climatological evaporation and increased surface wind speed. The seasonal variations of the HAC are driven by the variations of the damping effect of climatological evaporation. In boreal summer, the damping effect of climatological evaporation, which is greater in the Southern Hemisphere, promotes the development of the HAC. Consequently, the HAC peaks in boreal autumn when the damping effect of climatological evaporation transforms to a reverse meridional pattern, which is greater in the Northern Hemisphere. The wind–evaporation–SST feedback, as the key process of the annual-mean HAC, amplifies the seasonal variations of the HAC in tropical SST.

1. Introduction The increases in atmospheric carbon dioxide (CO2) concentration are almost spatially and seasonally uniform, but it can induce nonuniform increases in sea surface temperature (SST) (Meehl et al. 2007; Xie et al. 2010; Christensen et al. 2013). Spatial patterns of surface warming often play a critical role for regional climate changes under global warming (Vecchi and Soden 2007b; Xie et al. 2010; Christensen et al. 2013; Huang et al. 2013; Ma and Xie 2013; Huang 2014). The annualmean pattern of changes in tropical SST, as one of the most important systems of regional climate changes, has been widely studied (Clement et al. 1996; Liu et al. 2005; Meehl et al. 2006; Vecchi and Soden 2007a; DiNezio et al. 2009; Xie et al. 2010).

Corresponding author address: Dr. Ping Huang, Center for Monsoon System Research, Institute of Atmospheric Physics, Chinese Academy of Sciences, Bei-Er-Tiao 6, Zhong-Guan-Cun, Beijing 100190, China. E-mail: [email protected] DOI: 10.1175/JCLI-D-15-0055.1 Ó 2015 American Meteorological Society

Annual-mean changes in tropical SST display two pronounced patterns: the equatorial peak and the hemispheric asymmetry with greater warming in the Northern Hemisphere. Seager and Murtugudde (1997) explained the equatorial peak of SST warming as a result of ocean dynamics, wherein the surface convergent flow gathers warmer surface water on the equator. Liu et al. (2005) emphasized the role of changes in surface heat flux on the equatorial peak, and Xie et al. (2010) further concluded that the equatorial peak of SST changes is a result of the climatological minimum of evaporation damping on the equator. The hemispheric asymmetric change (HAC) is another pronounced global warming pattern of SST changes with more warming in the Northern Hemisphere than in the Southern Hemisphere (Xie et al. 2010; Sobel and Camargo 2011; Friedman et al. 2013). The ultimate reason for the HAC should be the latitudinal asymmetry of land–sea distribution, which is also the ultimate reason of hemispheric asymmetric systems in the presentday climate (Xie and Philander 1994; Philander et al. 1996; Chiang and Friedman 2012). Processes such as

6504

JOURNAL OF CLIMATE

VOLUME 28

FIG. 1. (a) Seasonal variation of the zonal-mean tropical SST changes. (b) As in (a), but the annual mean removed. (c) Seasonal variation of the heat storage changes in the mixed layer.

surface wind speed changes, Hadley circulation changes, and the Arctic amplification effect could contribute to the HAC in the subtropics and higher latitudes (Manabe et al. 1990; Holland and Bitz 2003; Sobel and Camargo 2011; Friedman et al. 2013). The wind–evaporation–SST (WES) feedback was suggested to be the dominant process forming the HAC in tropical SST (Xie et al. 2010). The process of the WES feedback can be understood as a meridional dipole of SST anomalies: positive (negative) anomalies north (south) of the equator drive southeasterly (southwesterly) anomalies south (north) of the equator, and then these wind anomalies increase (decrease) the background wind speed and intensify (weaken) the evaporation cooling south (north) of the equator; finally, the pattern of evaporation cooling amplifies the meridional dipole of SST anomalies (e.g., Xie and Philander 1994; Chiang and Vimont 2004). The tropical SST changes not only exhibit some spatial patterns in the annual mean but also show pronounced seasonal patterns. Figure 1a shows the seasonal variations of zonal-mean tropical SST changes under global warming. [The changes are defined by the differences between the representative concentration pathway 8.5 (RCP8.5) runs and historical run in the multimodel ensemble mean of 31 models from phase 5 of CMIP (CMIP5). The details of models and calculation are introduced in section 2.] The peak of SST changes is located on the equator throughout the year, but its magnitude is smaller in boreal autumn, which is attributed to the changes in ocean dynamics (Timmermann et al. 2004; Xie et al. 2010). The HAC in tropical SST is greater (smaller) in boreal autumn (spring).

The seasonal variation of SST, as the key variable in the air–sea interaction, dominates the seasonal cycle of tropical climate systems and also influences subtropical seasonal climate systems (Xie and Philander 1994; Chiang and Friedman 2012). The seasonal patterns of tropical SST changes contribute much to the seasonal patterns of tropical precipitation and circulation changes (Huang et al. 2013; Dwyer et al. 2014; Seo et al. 2014). For example, the maximum precipitation changes are closer to the equator than the maximum climatological precipitation as a result of the peak locking of SST changes to the equator (Huang et al. 2013). Moreover, the different meridional temperature gradients in two seasons play a crucial role in a robust (weak) weakening of the Hadley cell in boreal winter (summer) (Seo et al. 2014). However, the formation mechanisms of these seasonal patterns of tropical SST changes have not been studied as extensively as the annual-mean patterns. The present study will analyze the seasonal changes in surface energy budget under global warming to investigate the formation mechanisms of seasonal SST changes using outputs of 31 CMIP5 models. The models and surface heat flux decompositions are described in section 2. Results are presented in section 3, and conclusions are summarized in section 4.

2. Models and methods a. Models Outputs of 31 coupled general circulation models of CMIP5 are used. Change (denoted by D) under global

15 AUGUST 2015

6505

HUANG

TABLE 1. List of the 31 CMIP5 models used in the present study. (See http://cmip-pcmdi.llnl.gov/cmip5/availability.html for details.) Model

Institution

ACCESS1.0 ACCESS1.3 BCC_CSM1.1 BCC_CSM1.1(m) BNU-ESM CanESM2 CCSM4 CESM1(BGC) CESM1(CAM5) CMCC-CESM CMCC-CM CMCC-CMS CNRM-CM5 CSIRO Mk3.6.0 FGOALS-g2 GFDL CM3 GFDL-ESM2G GFDL-ESM2M HadGEM2-CC HadGEM2-ES IPSL-CM5A-LR IPSL-CM5A-MR IPSL-CM5B-LR MIROC5 MIROC-ESM MIROC-ESM-CHEM MPI-ESM-LR MPI-ESM-MR MRI-CGCM3 NorESM1-M NorESM1-ME

Commonwealth Scientific and Industrial Research Organization (CSIRO) and Bureau of Meteorology, Australia Beijing Climate Center, China Meteorological Administration, China College of Global Change and Earth System Science, Beijing Normal University, China Canadian Centre for Climate Modelling and Analysis, Canada National Center for Atmospheric Research (NCAR), United States National Science Foundation, U.S. Department of Energy, and NCAR, United States Centro Euro-Mediterraneo per I Cambiamenti Climatici, Italy

Centre National de Recherches Météorologiques, France CSIRO in collaboration with the Queensland Climate Change Centre of Excellence, Australia LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences, China Geophysical Fluid Dynamics Laboratory, United States

Met Office Hadley Centre, United Kingdom L’Institut Pierre-Simon Laplace, France

The University of Tokyo, National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology, Japan Max Planck Institute for Meteorology, Germany Meteorological Research Institute, Japan Norwegian Climate Centre, Norway

warming is defined by the difference in the long-term means between the RCP8.5 run for the period of 2061– 2100 and the historical run for 1961–2000 (Taylor et al. 2012). The 31 models are listed in Table 1. The variables of SST, surface longwave (QL) and shortwave (QS) radiation, surface sensible (QE) and latent heat flux (QH), surface vector wind velocity, and surface scalar wind speed are used. All outputs of the models are interpolated into a 2.58 3 2.58 grid before ensemble mean and other analyses. Analyses related to surface vector wind velocity are based on 26 of the 31 models, excluding CCSM4, FGOALS-g2, CESM1(BGC), NorESM1-ME, and CESM1(CAM5), in which the variables are unavailable. For the same reason, analyses related to surface scalar wind speed are based on all 31 models, excluding CCSM4, FGOALS-g2, CESM1(BGC), NorESM1-ME, and NorESM1-M. To remove the model uncertainty from global-mean temperature change, changes in each model are normalized by respective tropical- and subtropicalmean (608S–608N) SST warming into a uniform warming of 3 K, a typical warming in RCP8.5 runs relative to historical runs.

b. Surface energy budget decompositions The energy budget balance of the mixed layer ocean can be written as Qt 5 QL 1 QS 1 QE 1 QH 1 DO ,

(1)

where Qt 5 r0 Cp h›T/›t is the heat storage in the mixed layer, r0 and Cp are the density and specific heat of seawater, h is the thickness of the mixed layer, and DO is the ocean heat transport convergence. The sign of the heat fluxes (QL, QS, QE, QH, and DO) is defined such that a positive flux warms the ocean. A constant mixed layer (h) of 50 m is used in this analysis as in Dwyer et al. (2012), in which the results are insensitive to the seasonal variation of the mixed layer depth. Under the heat budget balance, the ocean heat transport convergence can be calculated by DO 5 QL 1 QS 1 QE 1 QH 2 Qt (e.g., Xie et al. 2010; Li and Xie 2012). This approximate diagnostic relationship is reasonable when SST changes are dominated by the mixed layer and upper-ocean processes at the end of the twenty-first century (Long et al. 2014).

6506

JOURNAL OF CLIMATE

In models, surface latent heat flux is calculated by a bulk formula as 0

QE 5 2ra LCE Wqs (T)(1 2 RHe2aT ) ,

(2)

where ra is the surface air density, L is the latent heat of evaporation, CE is the transfer coefficient, W is the surface scalar wind speed, RH is the surface relative humidity, T is the SST, and T 0 is the difference between SST and surface air temperature. The function qs(T) is the saturated specific humidity at temperature T following the Clausius–Clapeyron equation ›qs (T)/›T 5 aqs (T), where a 5 L/(Ry T 2 ) ; 0.06 K21, and Ry is the gas constant for water vapor. Equation (2) shows that change in latent heat flux could be contributed by change in SST, surface wind speed, sea–air temperature difference, and relative humidity. The latent heat flux change can be decomposed into several components following previous studies (e.g., Du and Xie 2008; Xie et al. 2010; Jia and Wu 2013): DQE 5 DQEO 1 DQEW 1 DQE-others ,

(3)

where DQEO represents the effect of ocean temperature change, DQEW the effect of surface wind speed change, and DQE-others the effect of surface relative humidity and surface stability change. The DQEO and DQEW can be obtained by linearizing the bulk formula into DQEO 5 aQE DSST and DQEW 5 QE DW/W, respectively. The DQEO is commonly known as the Newtonian cooling effect, while DQEW is known as the key term in the WES feedback (Xie and Philander 1994). This latent heat flux decomposition has been widely applied to investigate the interannual variability and change under global warming in tropical SST (Xie and Philander 1994; Du and Xie 2008; Jia and Wu 2013) and to evaluate the simulated seasonal cycle of tropical eastern Pacific climate in models (de Szoeke and Xie 2008). In DQEO 5 aQE DSST, the climatological latent heat flux QE and the response of SST change are nonlinearly mixed together. It is hard to discuss the role of QE with other heat flux changes on the spatial and seasonal patterns of DSST. In Xie et al. (2010), DQEO was supposed to be uniform in the tropics, and thus the distribution of SST response was concluded to be inversely proportional to the distribution of climatological latent heat flux QE. However, in the present discussion, the temporal variations of SST changes are not only in DQEO but also in DQt. To compare the contribution of QE with other processes, we decompose DQEO as

VOLUME 28

DQEO 5 ahQE ihDSSTi 1 ahQE iDSST0 1 aQ0E hDSSTi 1 aQ0E DSST0 ,

(4)

where h i denotes the regional and annual mean, and the prime denotes the deviation from the mean. Further, the spatial and seasonal deviation of DQ0EO can be represented by two linear components: DQ0EO 5 ahQE iDSST0 1 aQ0E hDSSTi 5 DQEO1 1 DQEO2 ,

(5)

where DQEO1 represents the nonuniform response of SST change, and DQEO2 represents the contribution of nonuniform climatological latent heat flux QE. The high-order term aQ0E DSST0 can be omitted. The effect of DQEO2 is independent of the seasonal and spatial deviation of DSST and can be understood as an external forcing on the seasonal and spatial variation of DSST. Using this decomposition, the effect of QE can be compared with the effects of other heat flux changes, and the seasonal variations of DSST-driven evaporation cooling can be compared with the variations of heat storage changes DQt.

3. Results a. Energy budget components The seasonal variations of the heat storage changes in the mixed layer DQt lead the variations of SST changes by around three months, or a quarter cycle (Figs. 1b,c). The heat storage changes in boreal summer are favorable for the decay in equatorial DSST and the development in the HAC, when the equatorial SST changes are smaller and the HAC in tropical SST is greater in boreal autumn (Fig. 1b). The seasonal changes in zonal-mean surface heat fluxes are shown in Fig. 2. Under global warming induced by increases in greenhouse gases, surface warming is mainly contributed by increases in downward longwave radiation (Fig. 2b), while surface evaporation plays a damping role suppressing surface warming (Fig. 2a). Sensible heat changes are relatively small and omitted in Fig. 2. The absolute values of these heat fluxes cannot represent their roles on the seasonal variations of SST changes. Thus, the respective annual mean is removed from the changes in each heat flux. The seasonal deviations of longwave radiation changes are small compared with the seasonal changes in latent heat flux, shortwave radiation, and ocean heat transport (Figs. 2e–h). (Seasonal change in heat flux is also defined as seasonal deviation for simplicity.)

15 AUGUST 2015

HUANG

6507

FIG. 2. Seasonal variations of changes in surface heat fluxes: (a) latent heat flux DQE, (b) net longwave radiation DQL, (c) net shortwave radiation DQS, and (d) oceanic transport DDO. (e)–(f) As in (a)–(d), but the respective annual mean removed.

Figure 3 shows the components of DQE decomposed in Eq. (3), and the components of DQEO decomposed in Eq. (5) are shown in Fig. 4. In Eq. (5), the seasonal variation of DQEO2 contributes most of the seasonal variation of DQEO (Figs. 3b and 4) because the ratio of the seasonal DSST0 to the annual-mean hDSSTi is much smaller than the ratio of the seasonal evaporation Q0E to its annual-mean hQE i. The effect of surface wind changes on latent heat (DQEW) can be influenced by multiple factors. Figure 5 first isolates DQEW into two components: the effects of seasonal variations of wind speed changes (Fig. 5a) and climatological latent heat flux (Fig. 5b). The result shows that seasonal variations of surface wind speed changes are the main contributor to DQEW. Figure 6 shows the climatology and changes in surface zonal and meridional winds and changes in surface scalar wind speed. It is noteworthy that the scalar wind speed changes are much smaller than the magnitude of surface vector wind velocity changes. The reason for this could be that scalar

wind speed, along with related variables such as evaporation and latent heat flux, is calculated at each time step in models. The monthly scalar wind speed is consistent with the monthly latent heat flux but not with the monthly mean of vector wind velocity. Thus, the damping effect of wind speed changes is calculated using the scalar wind speed directly from the monthly outputs, and vector wind velocity is used to illustrate the reason for scalar wind speed changes.

b. Peak locking of SST changes to the equator The peak of equatorial SST changes is located on the equator throughout the year (Fig. 1a). The peak locking of SST changes requires that the seasonal variations of surface energy budget changes near the equator do not exhibit pronounced hemispheric asymmetry (Fig. 1c) that would dramatically influence the annual-mean peak on the equator (Xie et al. 2010). Among the surface heat flux components, the ocean heat transport DDO, the shortwave radiation DQS, the effect of wind speed change

6508

JOURNAL OF CLIMATE

VOLUME 28

FIG. 3. Decomposition of the latent heat flux changes: (a) climatological latent heat flux, (b) Newtonian cooling component DQEO, (c) wind speed component DQEW, and (d) DQE-others the effects of air–sea temperature deviation and relative humidity changes.

DQEW, and the effect of climatological latent heat flux DQEO2 have considerable contributions on the equator. The DDO and DQEO2 are symmetrically located on the equator almost throughout the year, playing a positive role on the peak locking of SST changes, while the equatorial DQEW and DQS shift north and south across the equator. The variations of equatorial DDO could be attributed to the symmetry of the equatorial oceanic dynamics (Clement et al. 1996; Timmermann et al. 2004; Xie et al. 2010). The variations of DQEO2 can be explained by the minimum locking of climatological evaporation (Fig. 3a). The shift of equatorial DQEW and wind speed changes is mainly contributed by the meridional winds (Figs. 6a,b). The wind speed changes are decided by the magnitude of wind changes and the relative direction between wind changes and background winds. The equatorial peak of SST changes induces a low-level convergence near the equator, but the position of the convergence changes is not locked on the equator as the peak of SST changes (Huang et al. 2013). The low-level specific humidity changes can also contribute to the moist instability changes and then influence the circulation changes. Therefore, seasonally varying moisture changes influenced by the seasonally varying climatological humidity induce a weak north–south shift in the convergence changes (Huang et al. 2013). The southerly changes south of the convergence are stronger than the northerly changes north of the convergence because of the annual-mean southerly changes associated with greater northern warming (Fig. 1a). On the other hand, the direction of the background meridional winds on the equator also seasonally varies (Fig. 6b). As a result, the wind speed changes and the wind-induced evaporation

changes on the equator exhibit a pronounced asymmetry and seasonal shift (Figs. 5a and 6c–e). However, the asymmetric DQEW does not break the equatorial peak locking of SST changes because the asymmetry of DQEW is partly compensated by another asymmetric component DQS (Fig. 2g). The variations of DQS are dominated by the seasonal cycle of tropical precipitation changes due to the convective cloud– shortwave radiation feedback (Tan et al. 2008; Huang et al. 2013). Therefore, the total heat storage changes in the mixed layer do not show enough of a seasonal shift across the equator (Fig. 1c) to pronouncedly break the equatorial peak of the annual-mean SST changes (Xie et al. 2010).

c. Seasonal variations of equatorial SST changes The equatorial peak of SST changes is stronger in February–July than in August–November (Fig. 1a), although its location does not shift across the equator. This seasonal variation requires the heat storage changes in the mixed layer to be positive in September– May and negative in June–August (Fig. 1c). Of the heat flux components, the ocean heat transport DDO, the effect of wind speed change DQEW, and the effect of climatological evaporation DQEO2 have similar seasonal variation with DQt (Figs. 1c, 2h, 3b, and 3c)—positive (negative) in the first (second) half of the year. These three components are favorable for the formation of the seasonal variations of equatorial SST changes. The equatorial-mean (58S–58N) changes in these components are shown in Fig. 7a. The variations of another two considerable components, DQS and DQE-others, are opposite to that of DQt. This result indicates that DQS and DQE-others are negative factors to the variations of DSST.

15 AUGUST 2015

HUANG

6509

FIG. 4. Decomposition of the Newtonian cooling component of the latent heat flux changes DQEO contributed by (a) DQEO1 the seasonal SST changes and (b) DQEO2 the seasonal variations of climatological latent heat flux.

The variations of DDO could be attributed to the annual cycle in mean upwelling (Timmermann et al. 2004). The stronger climatological upwelling during August– October enhances the upwelling damping mechanism and reduces the SST changes (Clement et al. 1996; Xie et al. 2010). The variations of DQEO2 are directly decided by the climatological evaporation. Stronger climatological evaporation on the equator in May–August induces stronger damping cooling. As discussed in the last section, the seasonal variations of changes and climatology in the meridional winds jointly form the variations of surface wind speed changes and DQEW (Fig. 6b). The equatorial southerly changes decrease the speed of background northerly winds with positive DQEW in December–May, while the southerly changes enhance the background southerly winds with negative DQEW in June–November (Figs. 5a, 6b, and 6d). This indicates that the same southerly wind changes will induce different wind speed changes because of the direction transform of background winds. The cross-equatorial wind changes are associated with the interhemispheric temperature difference (Xie and Philander 1994; Chiang and Vimont 2004; Chiang and Friedman 2012). Thus, the HAC in SST and cross-equatorial wind speed changes

can play a direct role in the equatorial SST changes by modifying the evaporation damping and can indirectly influence ocean heat transport, as suggested in Timmermann et al. (2004). Because of the direction transform of background winds, the strength of southerly changes can influence the seasonal difference in wind speed changes even though the southerly changes have no seasonal variations. For example, stronger southerly changes can decrease the wind speed more in December–May and increase the wind speed more in June–November. Therefore, a stronger annual-mean off-equatorial HAC could induce stronger annual-mean southerly changes and enlarge the seasonal difference in the equatorial warming. This conclusion is verified by the significant correlation between the seasonal difference in equatorial SST changes and the strength of the annual-mean off-equatorial HAC in 31 models (Fig. 8). The seasonal difference in equatorial SST changes is defined by the difference of equatorial-mean (58S–58N) SST changes between January–June and July–December, and the strength of the off-equatorial HAC is defined by the difference in annual-mean SST changes between the Northern (58–308N) and Southern (58–308S) Hemispheres. The correlation of these two indexes in 31 models is up

6510

JOURNAL OF CLIMATE

VOLUME 28

FIG. 5. Seasonal variation of heat flux changes induced by (a) the seasonal variations of wind speed changes and (b) the seasonal variations of climatological latent heat flux.

to 0.62 at 99% confidence level based on the Student’s t test. The DDO, DQEW, and DQEO2 have an identical positive/ negative transform in May, the transform month of climatological tropical climate. This transform month is consistent with the transform month of DQt, which explains the maximum warming month of May for the equatorial SST changes. Another positive/negative transform month of DQt is September, corresponding to the month of the minimum SST changes. Of the three components, only DQEO2 has the same positive/negative transform in September. The transform month of DDO and DQEW is November, lagging two months to DQt. The lag of DDO and DQEW is compensated by the changes in shortwave radiation DQS and the effect of surface relative humidity and surface stability on latent heat changes DQE-others. The more complicated mechanism of minimum warming in September implies that it could be not as robust as the month of maximum warming.

d. Off-equatorial hemispheric asymmetric change As shown in Fig. 1, the HAC in SST shows a pronounced seasonal cycle on around 258 in two hemispheres. The off-equatorial HAC in July–October is around 3 times of that in January–April. These seasonal

variations of HAC require that DQt be positive (negative) in March–September (October–February) in the Southern Hemisphere, and vice versa in the Northern Hemisphere (Figs. 1c, 7b, and 7c). The variations of DQt are consistent with the variations of climatological evaporation damping (Figs. 7b,c). The other two considerable factors, DDO and DQEW, both show some phase mismatch with DQt. The variations of ocean heat transport DDO are almost out of phase with those of DQEW and DQt (Figs. 7b,c). The DQEW, representing the role of the positive WES feedback, is the key process in the formation of the annual-mean HAC (Xie et al. 2010). The seasonal variations of HAC are also consistent with the variations of damping effect induced by surface wind speed changes off the equator. In Fig. 6, the easterly changes with enhanced damping develop in the Southern Hemisphere (158–308S) in May and peak in September, and the westerly changes with suppressed damping develop in the Northern Hemisphere (58–258S) in June and peak in September. A typical WES feedback pattern can be observed from the horizontal structures of surface wind changes and SST changes in August–October (ASO) and February–April (FMA) (Fig. 9), and the WES pattern is more pronounced in ASO.

15 AUGUST 2015

HUANG

6511

FIG. 6. Change (shaded) and climatology (contours; m s21) in surface (a) zonal and (b) meridional wind. (c) The changes in surface scalar wind speed, (d) the change percentages of surface wind speed relative to the climatological surface wind speed, and (e) the annualmean-removed change percentages of surface wind speed.

The consistency between the seasonal DQEW and the HAC (Figs. 1a and 6d) indicates that the atmospheric wind speed and the wind-induced latent heat quickly respond to the SST pattern in the WES feedback. However, the DQEW induced by the wind speed changes is just part of DQt driving the evolution of the HAC. As shown in Figs. 7b and 7c, the phase of DQEW is not very consistent with that of DQt. This result implies that there exists an external heat flux factor

driving the variations of WES feedback associated with DQEW and HAC. A heat flux favoring warming (cooling) in the Northern (Southern) Hemisphere will enhance the annual-mean WES feedback, and vice versa. Among the considerable heat flux changes, the seasonal variation of DQEO2 is consistent with that of DQt. From March to September, the weaker climatological latent heat flux in the north (Fig. 3a) induces weaker damping in the north than in the south (Figs. 3b, 4b, 5b, and 5c),

6512

JOURNAL OF CLIMATE

VOLUME 28

FIG. 7. Seasonal variation of the changes in heat storage of the mixed layer, the effect of climatological evaporation, the effect of surface wind speed, and oceanic heat transport of (a) the equatorial mean (58S–58N), (b) the off-equatorial mean of the Southern Hemisphere (308–108S), and (c) the off-equatorial mean of the Northern Hemisphere (108–308N).

whereas the climatological latent heat flux with its induced damping reverses its meridional pattern during October–February. Under the forcing of DQEO2, the HAC develops in April and decays in November. Therefore, it can be concluded that the seasonal variations of the damping effect of climatological evaporation are the dominant factor driving the seasonal variations of the HAC. The WES feedback can enlarge the variations of the HAC, but it plays only a delaying and amplifying role, although it is the dominant factor for the annual-mean HAC. The role of evaporation damping changes DQE on the seasonal variation of DSST can also be demonstrated by their consistence in regional distribution. Figure 10 shows the seasonal-mean DQE and DQt in two representative seasons, November–January (NDJ) and May–July (MJJ). The regional consistence between the seasonal DQE and DQt exhibits that the consistence of the zonal-mean seasonal evolution of DQE and DQt in Fig. 5 is not coincidental. The latent heat changes including the driving role of climatological evaporation and the amplifying role of the WES feedback are the dominant mechanism of the seasonal variations of the HAC in tropical SST.

mechanisms of these seasonal patterns are discussed based on the surface energy budget analyses. The minimum locking to the equator of climatological latent heat flux and the ocean heat transport favor the peak locking of SST changes to the equator. The other two north–south asymmetric heat flux changes, the shortwave radiation changes and the wind-induced latent heat changes, partially cancel out each other. The weak north–south asymmetry of seasonal heat flux changes cannot pronouncedly break the equatorial peak

4. Summary In the present study, the seasonal variations of tropical SST changes under global warming are investigated using outputs of RCP8.5 and historical runs in 31 CMIP5 models. The tropical SST changes have pronounced seasonal patterns: the peak locking to the equator and the weaker equatorial changes and stronger hemispheric asymmetric changes in boreal autumn. The formation

FIG. 8. Relationship between the hemispheric difference of the off-equatorial SST changes and the seasonal difference of the equatorial SST changes in the 31 CMIP5 models.

15 AUGUST 2015

HUANG

6513

FIG. 9. Seasonal-mean changes in tropical SST (shaded), surface vector wind (vectors; m s21), and the change percentages of surface wind speed [contours; contour interval (CI) is 3%; and negative contours are dashed] in (a) ASO and (b) FMA.

pattern of the annual-mean SST changes induced by the equatorial minimum in climatological latent heat flux (Xie et al. 2010). Consequently, the equatorial SST changes exhibit a seasonal peak-locking pattern. For the weaker equatorial warming in the boreal autumn, the analysis shows that the wind-induced latent heat changes and the damping role of climatological evaporation are as important as the ocean heat transport changes suggested in previous studies (Timmermann et al. 2004; Xie et al. 2010). The stronger climatological evaporation in May–August induces stronger evaporation damping changes on the equator. The crossequatorial southerly changes associated with the HAC in SST speed up the background southerly winds on the

equator in June–November but weaken the background northerly winds in December–May. Because of the connection of the cross-equatorial southerly changes, the seasonal difference of equatorial warming is significantly correlated to the annual-mean hemispheric difference in off-equatorial SST changes in the 31 models. The wind-induced latent heat changes, the damping role of climatological evaporation, and the ocean heat transport changes collectively make the equatorial SST changes stronger in the first half of the year. The seasonal tendency of the off-equatorial HAC in SST is dominated by the variations of evaporation damping changes induced by climatological evaporation. In March–September, the greater evaporation damping

FIG. 10. Seasonal-mean DQt (shaded) and DQE (contours; CI is 2 W m22; negative contours are dashed; and zero lines are in red) in (a) NDJ and (b) MJJ.

6514

JOURNAL OF CLIMATE

in the Southern Hemisphere induced by the climatological evaporation promotes the development of the HAC, whereas the meridional pattern of the evaporation damping changes suppresses the HAC in October– February. The WES feedback, the dominant process of the annual-mean HAC in tropical SST, synchronously varies along with the HAC but does not decide the tendency of the HAC. The WES feedback plays only an amplifying role to increase the HAC in March–September. The seasonal variation of equatorial peak and HAC in SST indicates that the meridional patterns of SST changes mainly show an equatorial peak pattern in the first half of the year and a HAC pattern in the second half. The crucial role of different meridional patterns of SST changes on circulation and precipitation changes in two seasons has been noticed in several recent studies (Huang et al. 2013; Dwyer et al. 2014; Seo et al. 2014). More attention should be paid to studying the seasonal climate changes under global warming in the future. The present study is based on the multimodel ensemble mean of 31 CMIP5 models, and the intermodel uncertainty in the seasonal SST changes is not discussed. In CMIP5 models, the intermodel uncertainty in the annual-mean SST changes is quite large and induces great uncertainty in projections of circulation and precipitation changes under global warming (DiNezio et al. 2009; Huang et al. 2013; Ma and Xie 2013; Seo et al. 2014; Huang and Ying 2015). The uncertainty in SST changes could seasonally vary as the SST changes. It will influence the robustness of projected seasonal climate changes correlated to the SST change patterns. Acknowledgments. This work was supported by the National Basic Research Program of China (2014CB953903 and 2012CB955604) and the National Natural Science Foundation of China (Grant 41461164005). I acknowledge the World Climate Research Programme’s Working Group on Coupled Modeling, which is responsible for CMIP5, and the climate modeling groups (listed in Table 1) for producing and making available their model output. I would also like to thank the three anonymous reviewers for their constructive suggestions.

REFERENCES Chiang, J. C. H., and D. J. Vimont, 2004: Analogous Pacific and Atlantic meridional modes of tropical atmosphere–ocean variability. J. Climate, 17, 4143–4158, doi:10.1175/JCLI4953.1. ——, and A. R. Friedman, 2012: Extratropical cooling, interhemispheric thermal gradients, and tropical climate change. Annu. Rev. Earth Planet. Sci., 40, 383–412, doi:10.1146/ annurev-earth-042711-105545. Christensen, J. H., and Coauthors, 2013: Climate phenomena and their relevance for future regional climate change. Climate

VOLUME 28

Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1217–1308. [Available online at http://www.ipcc.ch/pdf/assessment-report/ar5/wg1/ WG1AR5_Chapter14_FINAL.pdf.] Clement, A. C., R. Seager, M. A. Cane, and S. E. Zebiak, 1996: An ocean dynamical thermostat. J. Climate, 9, 2190–2196, doi:10.1175/ 1520-0442(1996)009,2190:AODT.2.0.CO;2. de Szoeke, S. P., and S.-P. Xie, 2008: The tropical eastern Pacific seasonal cycle: Assessment of errors and mechanisms in IPCC AR4 coupled ocean–atmosphere general circulation models. J. Climate, 21, 2573–2590, doi:10.1175/ 2007JCLI1975.1. DiNezio, P. N., A. C. Clement, G. A. Vecchi, B. J. Soden, B. P. Kirtman, and S.-K. Lee, 2009: Climate response of the equatorial Pacific to global warming. J. Climate, 22, 4873–4892, doi:10.1175/2009JCLI2982.1. Du, Y., and S.-P. Xie, 2008: Role of atmospheric adjustments in the tropical Indian Ocean warming during the 20th century in climate models. Geophys. Res. Lett., 35, L08712, doi:10.1029/ 2008GL033631. Dwyer, J. G., M. Biasutti, and A. H. Sobel, 2012: Projected changes in the seasonal cycle of surface temperature. J. Climate, 25, 6359–6374, doi:10.1175/JCLI-D-11-00741.1. ——, ——, and ——, 2014: The effect of greenhouse gas–induced changes in SST on the annual cycle of zonal mean tropical precipitation. J. Climate, 27, 4544–4565, doi:10.1175/ JCLI-D-13-00216.1. Friedman, A. R., Y.-T. Hwang, J. C. H. Chiang, and D. M. W. Frierson, 2013: Interhemispheric temperature asymmetry over the twentieth century and in future projections. J. Climate, 26, 5419–5433, doi:10.1175/JCLI-D-12-00525.1. Holland, M. M., and C. M. Bitz, 2003: Polar amplification of climate change in coupled models. Climate Dyn., 21, 221–232, doi:10.1007/s00382-003-0332-6. Huang, P., 2014: Regional response of annual-mean tropical rainfall to global warming. Atmos. Sci. Lett., 15, 103–109, doi:10.1002/asl2.475. ——, and J. Ying, 2015: A multimodel ensemble pattern regression method to correct the tropical Pacific SST change patterns under global warming. J. Climate, 28, 4706–4723, doi:10.1175/ JCLI-D-14-00833.1. ——, S.-P. Xie, K. Hu, G. Huang, and R. Huang, 2013: Patterns of the seasonal response of tropical rainfall to global warming. Nat. Geosci., 6, 357–361, doi:10.1038/ngeo1792. Jia, F., and L. Wu, 2013: A study of response of the equatorial Pacific SST to doubled-CO2 forcing in the coupled CAM–1.5layer reduced-gravity ocean model. J. Phys. Oceanogr., 43, 1288–1300, doi:10.1175/JPO-D-12-0144.1. Li, G., and S.-P. Xie, 2012: Origins of tropical-wide SST biases in CMIP multi-model ensembles. Geophys. Res. Lett., 39, L22703, doi:10.1029/2012GL053777. Liu, Z., S. Vavrus, F. He, N. Wen, and Y. Zhong, 2005: Rethinking tropical ocean response to global warming: The enhanced equatorial warming. J. Climate, 18, 4684–4700, doi:10.1175/ JCLI3579.1. Long, S.-M., S.-P. Xie, X.-T. Zheng, and Q. Liu, 2014: Fast and slow responses to global warming: Sea surface temperature and precipitation patterns. J. Climate, 27, 285–299, doi:10.1175/ JCLI-D-13-00297.1. Ma, J., and S.-P. Xie, 2013: Regional patterns of sea surface temperature change: A source of uncertainty in future projections of precipitation and atmospheric circulation. J. Climate, 26, 2482–2501, doi:10.1175/JCLI-D-12-00283.1.

15 AUGUST 2015

HUANG

Manabe, S., K. Bryan, and M. J. Spelman, 1990: Transient response of a global ocean–atmosphere model to a doubling of atmospheric carbon dioxide. J. Phys. Oceanogr., 20, 722–749, doi:10.1175/1520-0485(1990)020,0722:TROAGO.2.0.CO;2. Meehl, G. A., and Coauthors, 2006: Climate change projections for the twenty-first century and climate change commitment in the CCSM3. J. Climate, 19, 2597–2616, doi:10.1175/ JCLI3746.1. ——, and Coauthors, 2007: Global climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 747–846. [Available online at http://www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1chapter10.pdf.] Philander, S., D. Gu, G. Lambert, T. Li, D. Halpern, N. Lau, and R. Pacanowski, 1996: Why the ITCZ is mostly north of the equator. J. Climate, 9, 2958–2972, doi:10.1175/1520-0442(1996)009,2958: WTIIMN.2.0.CO;2. Seager, R., and R. Murtugudde, 1997: Ocean dynamics, thermocline adjustment, and regulation of tropical SST. J. Climate, 10, 521–534, doi:10.1175/1520-0442(1997)010,0521: ODTAAR.2.0.CO;2. Seo, K. H., D. M. W. Frierson, and J. H. Son, 2014: A mechanism for future changes in Hadley circulation strength in CMIP5 climate change simulations. Geophys. Res. Lett., 41, 5251– 5258, doi:10.1002/2014GL060868.

6515

Sobel, A. H., and S. J. Camargo, 2011: Projected future seasonal changes in tropical summer climate. J. Climate, 24, 473–487, doi:10.1175/2010JCLI3748.1. Tan, P.-H., C. Chou, and J.-Y. Tu, 2008: Mechanisms of global warming impacts on robustness of tropical precipitation asymmetry. J. Climate, 21, 5585–5602, doi:10.1175/2008JCLI2154.1. Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, doi:10.1175/BAMS-D-11-00094.1. Timmermann, A., F.-F. Jin, and M. Collins, 2004: Intensification of the annual cycle in the tropical Pacific due to greenhouse warming. Geophys. Res. Lett., 31, L12208, doi:10.1029/2004GL019442. Vecchi, G. A., and B. J. Soden, 2007a: Global warming and the weakening of the tropical circulation. J. Climate, 20, 4316– 4340, doi:10.1175/JCLI4258.1. ——, and ——, 2007b: Effect of remote sea surface temperature change on tropical cyclone potential intensity. Nature, 450, 1066–1070, doi:10.1038/nature06423. Xie, S.-P., and S. G. H. Philander, 1994: A coupled ocean–atmosphere model of relevance to the ITCZ in the eastern Pacific. Tellus, 46A, 340–350, doi:10.1034/j.1600-0870.1994.t01-1-00001.x. ——, C. Deser, G. A. Vecchi, J. Ma, H. Teng, and A. T. Wittenberg, 2010: Global warming pattern formation: Sea surface temperature and rainfall. J. Climate, 23, 966–986, doi:10.1175/2009JCLI3329.1.