Effects of Southeastern Pacific Sea Surface Temperature on the

Download PDF
0 downloads 0 Views 4MB Size Report
Comment
Oct 15, 2016 - cipitation over the tropical Pacific is fundamentally de- termined by the ...... syndrome in coupled general circulation models: The role of large-scale vertical .... Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003:.
15 OCTOBER 2016

SONG AND ZHANG

7417

Effects of Southeastern Pacific Sea Surface Temperature on the Double-ITCZ Bias in NCAR CESM1 FENGFEI SONG AND GUANG J. ZHANG Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California (Manuscript received 1 December 2015, in final form 15 June 2016) ABSTRACT The double intertropical convergence zone (ITCZ) is a long-standing bias in the climatology of coupled general circulation models (CGCMs). The warm biases in southeastern Pacific (SEP) sea surface temperature (SST) are also evident in many CGCMs. In this study, the role of SEP SST in the double ITCZ is investigated by prescribing the observed SEP SST in the Community Earth System Model, version 1 (CESM1). Both the double ITCZ and dry equator problems are significantly improved with SEP SST prescribed. Both atmospheric and oceanic processes are involved in the improvements. The colder SST over the SEP decreases the precipitation, which enhances the southeasterly winds outside the prescribed SST region, cooling the ocean via increased evaporation. The enhanced descending motion over the SEP strengthens the Walker circulation. The easterly winds over the equatorial Pacific enhance upwelling and shoal the thermocline over the eastern Pacific. The changes of surface wind and wind curl lead to a weaker South Equatorial Countercurrent and stronger South Equatorial Current, preventing the warm water from expanding eastward, thereby improving both the double ITCZ and dry equator. The enhanced Walker circulation also increases the low-level wind convergence and reduces the wind speed in the tropical western Pacific, leading to warmer SST and stronger convection there. The stronger convection in turn leads to more cloud and reduces the incoming solar radiation, cooling the SST. These competing effects between radiative heat flux and latent heat flux make the atmospheric heat flux secondary to the ocean dynamics in the western Pacific warming.

1. Introduction Over the past three decades, coupled general circulation models (CGCMs) have been used widely in climate variability, change, and projection studies. However, the climatological biases in CGCMs limit their utility in these aspects and impair the creditability of model results (e.g., Cai and Cowan 2013; Zhou and Xie 2015; Tian 2015; Li et al. 2015). Among numerous CGCM climatological biases, the double intertropical convergence zone (ITCZ) stands out as the most prominent one (Mechoso et al. 1995; Davey et al. 2002; Meehl et al. 2005; Dai 2006; Lin 2007; Oueslati and Bellon 2015; Zhang et al. 2015). In an annual-mean sense, most CGCMs exhibit two parallel zonal rainfall bands straddling the equatorial Pacific, rather than a northern zonal ITCZ and a southern South

Corresponding author address: Guang J. Zhang, Climate, Atmospheric Sciences, and Physical Oceanography, Scripps Institution of Oceanography, UC San Diego, 9500 Gilman Drive #0221, La Jolla, CA 92093-0221. E-mail: [email protected] DOI: 10.1175/JCLI-D-15-0852.1 Ó 2016 American Meteorological Society

Pacific convergence zone (SPCZ) extending southeastward from the western Pacific in the observations. According to Mechoso et al. (1995) and De Szoeke and Xie (2008), the double ITCZ can be divided into two categories: one is the persistent double-ITCZ bias, in which precipitation persists too long in the Southern Hemisphere; and the other is the alternating ITCZ bias, in which the precipitation maximum migrates meridionally following the seasonal shift of insolation maximum. With the improvement in model development, the persistent double-ITCZ error has become less common, and the alternating double-ITCZ error dominates the majority of current CGCMs (Mechoso et al. 1995; De Szoeke and Xie 2008; Oueslati and Bellon 2015). Many efforts have been devoted to identifying the causes of the double-ITCZ problem, and both tropical and extratropical origins are suggested. Since ITCZ precipitation is of convective origin, convective parameterization schemes have often been blamed for the double-ITCZ problem. As such, significant efforts have been devoted to understanding the role of convection in the double ITCZ. For example, the double ITCZ is significantly improved (Zhang and Wang 2006; Song and

7418

JOURNAL OF CLIMATE

Zhang 2009; Zhang and Song 2010) when the convective parameterization closure (Zhang 2002) is changed in the Zhang–McFarlane convection scheme (Zhang and McFarlane 1995). Bacmeister et al. (2006) suggested that the strong rain reevaporation can also reduce the double ITCZ in the NASA Seasonal-to-Interannual Prediction Project, version 2 (NSIPP-2), since the evaporation cooling weakens the feedback between the convective heating and boundary layer convergence at high frequency (less than 15 days). The threshold sea surface temperature (SST) for convection is also found to be a critical parameter, with a higher threshold prone to a more obvious double-ITCZ pattern (Bellucci et al. 2010). Some studies also emphasized the importance of dynamical suppression of deep convection by the lateral entrainment of environmental dry air (Chikira 2010; Hirota et al. 2011; Oueslati and Bellon 2013). According to these studies, high lateral entrainment weakens the precipitation over the southern ITCZ and has a better representation of SPCZ. However, although the double ITCZ is mitigated by different modifications on convection schemes to some extent, it still plagues most CGCMs, suggesting that convection parameterization is not the only culprit. Deficiencies in midlatitude cloud simulation are also suggested to be possible causes of double ITCZ (Hwang and Frierson 2013; Li and Xie 2014). The negative biases in Southern Hemisphere midlatitude cloud amount lead to more incoming shortwave radiation and warm the Southern Ocean in the CGCM. In response to the interhemispheric heat contrast, the cross-equatorial atmospheric energy transport takes place, primarily by changes in Hadley circulation, which results in a meridional shift of ITCZ. The Southern Ocean warming shifts the ITCZ southward, in agreement with some theoretical studies that suggest that the ITCZ is drawn toward the warmer hemisphere (Kang et al. 2008, 2009). However, Kay et al. (2016) tested this argument by reducing the shortwave radiation bias over the Southern Ocean in the fully coupled National Center for Atmospheric Research (NCAR) Community Earth System Model, version 1 (CESM1), and found that the double ITCZ is not improved. They argued that the cross-equatorial energy transport corresponding to the cooled Southern Ocean occurs in the ocean instead of the atmosphere and thus has little effect on the double ITCZ. Hence, it is still unclear whether the Southern Ocean cloud simulation bias exerts any influence on the double ITCZ. The warm SST biases in the southeastern Pacific (SEP), which are common in CGCMs (Zheng et al. 2011), are regarded as another source for the double-ITCZ problem (Ma et al. 1996; Yu and Mechoso 1999; Dai et al. 2003, 2005). It is believed that the SEP warm bias is associated with the underestimated stratocumulus cloud, which results in more net

VOLUME 29

heat flux into the ocean. By artificially increasing the stratocumulus cloud amount (Ma et al. 1996; Yu and Mechoso 1999) or improving the low-level cloud simulation (Dai et al. 2003, 2005), it is found that both the SEP warm bias and double ITCZ are alleviated. However, the equatorial cold tongue is stronger and extends farther westward, leading to deficient precipitation over the western equatorial Pacific. Recently, Zheng et al. (2011) examined 19 IPCC AR4 CGCMs and found that most models have negative net heat flux biases into the SEP region and thus could not explain the warm SST bias there. Instead, they showed that biases in ocean heat transport through both upwelling and horizontal advection were most responsible. Yet there are other studies that suggest that the SEP warm SST bias may not only be due to a local process, but also from the tropical Atlantic bias (Wang et al. 2010, 2014; Zhang et al. 2014). Hence, solving the stratocumulus cloud problem only addresses part of the effect of SEP warm SST bias on the double ITCZ. Large and Danabasoglu (2006) investigated the effects of ocean eastern boundary biases along the Peruvian–Chilean coast by restoring model ocean temperature and salinity to observations in the upper ocean. They found that the coastal SST and salinity biases exert significant influences on the SEP SST and precipitation, though the mechanisms remain unknown. In this study, we investigate the effect of the SEP SST on the ITCZ simulation in the CESM1. The objectives are twofold: 1) to determine how much effect the SEP SST has on the ITCZ simulation in the CESM1 and 2) to understand the mechanism through which it contributes to the double ITCZ. The remainder of the paper is organized as follows: Section 2 describes the model experiments, observational datasets, and methods. The main results are presented in section 3, including the influences of SEP SST on the double ITCZ, physical mechanisms for the SST pattern maintenance, and atmospheric and oceanic roles in the SST change over other regions. Section 4 gives a summary and discussion.

2. Model experiments, observational datasets, and methods In this study, the latest model, CESM1.2.2, is used (refer to http://www.cesm.ucar.edu/models/cesm1.2/ for more details). The atmospheric component of the model is the Community Atmospheric Model, version 5.3 (CAM5.3), which includes some modifications to the spectral element dynamical core based on CAM5 (Neale et al. 2013). Its oceanic component is the Parallel Ocean Program, version 2 (POP2), and the land component is the Community Land Model, version 4.5 (CLM4.5). The f19_g16 configuration of the CESM1.2.2 is used, with approximately 28 resolution for the atmospheric model

15 OCTOBER 2016

SONG AND ZHANG

7419

FIG. 1. The annual-mean SST (8C) and 925-hPa wind in the (a) observations (SST from HadISST and 925-hPa wind from NCEP-1) and their differences between (b) CNTL and observations, (c) SEP and observations, and (d) the SEP and CNTL runs. The purple box in (b) shows the region in which observed SSTs are prescribed. The stippled regions in (d) indicate SST differences exceeding the 5% significance level.

and approximately 18 resolution for the ocean model. Two 12-yr experiments are performed. In the first experiment [referred to as the control (CNTL) run], the atmosphere and the ocean are freely coupled, with the external forcing (such as the greenhouse gas, aerosol, volcanic, and solar activity) fixed at the year-2000 level. The simulation is initialized in a ‘‘warm start’’ mode, with the default restart files provided by NCAR. The second experiment (referred to as the SEP run) is the same as the CNTL run except that the SST in the SEP region is prescribed with the seasonally varying climatological (1980– 94) SST from the Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST; http://www.metoffice. gov.uk/hadobs/hadisst/data/download.html; Rayner et al. 2003). Here, the SEP region is defined as 58–258S and from 1208W to the South American west coast (the rectangular region in Fig. 1). It includes part of the narrow coastal region (08–358S, 848–668W) in Large and Danabasoglu (2006). To avoid too sharp SST gradients at the boundaries of the region, a transition zone extending

58 outward is added along the boundaries in which SST is interpolated linearly from the observed SST to the model SST. The first 2 years are treated as the spinup, and the last 10 years are used for analysis. The differences of the two experiments measure the influences of the SEP SST on the ITCZ. The Student’s t test is used to calculate the statistical significance of the difference field. To compare with the model experiments, the following observational and reanalysis datasets are used: 1) precipitation from the Global Precipitation Climatology Project (GPCP; http://www.esrl.noaa.gov/psd/ data/gridded/data.gpcp.html; Adler et al. 2003) for 1995–2004; 2) atmospheric three-dimensional winds from the National Centers for Environmental Prediction (NCEP)–NCAR reanalysis project (NCEP-1; http://www.esrl.noaa.gov/psd/data/gridded/data.ncep. reanalysis.html; Kalnay et al. 1996) for 1995–2004; 3) SST from HadISST for 1995–2004; 4) atmospheric surface winds from Quick Scatterometer (QuikSCAT;

7420

JOURNAL OF CLIMATE

http://www.remss.com) satellite retrievals for 2000–08; and 5) ocean temperature, ocean zonal and vertical velocity, and sea surface height (SSH) from the Simple Ocean Data Assimilation (SODA) reanalysis (Carton and Giese 2008) for 1995–2004. The choice of climatological periods does not affect the results in this study.

3. Results a. The influences of SEP SST on the double ITCZ Figure 1 shows the annual-mean observed SST and 925-hPa wind, along with the differences between model simulations and observations. Compared to the observations, there is a cold bias along the equator over the western and central Pacific in the CNTL run, an indication that the equatorial cold tongue extends too much westward (Figs. 1a,b). The warm SST biases are evident across the Pacific basin in the Southern Hemisphere, with the maximum in the SEP. The cold (warm) SST bias over the equatorial Pacific (SEP) corresponds to the near-surface wind divergence (convergence) (Fig. 1b). The tropical North Atlantic cold bias and South Atlantic warm bias are also evident in the CNTL run. After prescribing the observed SST over the SEP region, the equatorial cold bias weakens and shifts eastward, an indication of improved cold tongue simulation (Fig. 1c). Compared to the CNTL run, the central and northeastern Pacific becomes warmer (Fig. 1d), while the subtropics and the far western Pacific become colder. The trade winds are enhanced east of 1608W south of the equator and reduced west of 1208W north of the equator, forming a near-surface wind convergence over the western Pacific and near the date line. The influence of SEP SST is mainly confined to the Pacific region, and the Atlantic biases are almost unchanged (Figs. 1b,c). The annual-mean precipitation in the observations and CNTL and SEP runs is shown in Fig. 2. The observations exhibit the ITCZ across the Pacific basin north of the equator and the SPCZ south of the equator extending southeastward from the western Pacific (Fig. 2a). The observed north–south asymmetry of precipitation over the tropical Pacific is fundamentally determined by the geometries of the American continents and influenced by many feedback processes (Philander et al. 1996; Xie and Philander 1994; Li 1997). In the CNTL run (Fig. 2c), the ITCZ north of the equator is stronger, while the SPCZ elongates zonally and extends farther into the eastern equatorial Pacific, forming the so-called double ITCZ. The overly dry equator in the CNTL run is also evident over the western Pacific. In

VOLUME 29

the SEP run (Fig. 2e), the northern ITCZ is much stronger than both the observations and CNTL run, but the spurious southern ITCZ is improved considerably to look more like the SPCZ. The dry equator is also improved. The bias patterns of CNTL and SEP runs are similar: there is more precipitation over the tropical Pacific, except for the western equatorial region, than in the observations; the precipitation over the Indian Ocean (Atlantic Ocean) shifts westward (southward) compared to the observations (Figs. 2b,d). However, compared to the CNTL run, precipitation is increased over the ITCZ and SPCZ but decreased over the southcentral and southeastern Pacific as well as the eastern equatorial region, leading to decreased biases in both the double ITCZ and dry equator in the SEP run (Fig. 2f). To ensure that 12-yr integrations are long enough for the model simulations to reach equilibrium so that the difference fields in Figs. 1 and 2 are statistically meaningful, the time evolution of SST and precipitation over the south-central Pacific (SCP; 258–58S, 1508–1208W), where the double-ITCZ bias is prominent, is shown in Fig. 3. The SST and precipitation time series in both the CNTL and SEP runs are stable and undergo regular annual cycles throughout the 12-yr integration period. Although there are some interannual variations, the SST and precipitation in the CNTL run are consistently higher than in the SEP run. The time evolutions of SST over the tropical western Pacific (TWP; 108S–108N, 1508E–1708W) and northeastern Pacific (NEP; 158– 308N, 1108–1608W) are also examined (not shown). Over the TWP region, SST warms up quickly in the first 2 years and remains warm afterward, although there are some fluctuations in years 7 through 10. Over the NEP region, SST warms up in the first three years but exhibits much larger variations than TWP and SCP. It indicates that, over the NEP, SST warming response to prescribed SEP SST is less robust than in SCP and TWP regions, and other factors have considerable influences. We also examined the subsurface ocean temperatures (not shown). They reach equilibrium in 3–4 years after the initiation of model integration. To exclude the possible influences of ENSO, the ENSO years in the CNTL run are identified using the Niño-3.4 index and removed from the climatology. The differences between the SEP and CNTL runs are recalculated. The results are similar to Figs. 1 and 2 (not shown). Thus, 12-yr integrations are adequate for examining the differences between the CNTL and SEP runs, especially in the tropical regions. The SST and precipitation averaged over three latitude bands (58–158N, 58N–58S, and 58–158S) over the Pacific are shown in Fig. 4. The zonal SST gradient in the

15 OCTOBER 2016

SONG AND ZHANG

7421

FIG. 2. The annual-mean precipitation (mm day21) in (a) GPCP observations, (c) CNTL run, (e) SEP run, and the differences between (b) CNTL run and GPCP, (d) SEP run and GPCP, and (f) SEP and CNTL runs. The stippled regions in (f) indicate that precipitation differences exceed the 5% significance level.

CNTL run is much smaller than in the observations but becomes closer in the SEP run, especially in the equatorial region and south of the equator (Figs. 4a–c). The SST and precipitation are higher than the observations west of 1108W north of the equator in the SEP run (Figs. 4a,d). Corresponding to the improvement of SST, precipitation in the SEP run is much closer to the observations in the equatorial region and south of the equator. Over the equator, precipitation is significantly less than the observations west of 1708W in the CNTL run but increases considerably in the SEP run (Fig. 4e), reducing the dry equator problem. South of the equator, the precipitation is much higher than the observations in the CNTL run across the entire Pacific. The positive biases are reduced by about 50% east of 1608W in the SEP run, while the biases are worsened in the western Pacific (Fig. 4f).

Many previous studies suggested that the doubleITCZ bias in the annual-mean sense could come from either persistent double-ITCZ bias year-round or alternating ITCZ bias with the season (Mechoso et al. 1995; De Szoeke and Xie 2008; Oueslati and Bellon 2015). It is important to examine which kind of bias this model has and whether the SEP SST has any effect on the seasonal variation of the double ITCZ. Figure 5 shows the seasonal cycle of precipitation and SST in the eastern Pacific averaged over 908–1408W, along with the meridional wind in the equatorial region. In the observations (Fig. 5a), SST greater than 278C appears north of the equator all year round, except January, and south of the equator in boreal spring (MAM). Precipitation follows SST closely, with double ITCZ in MAM and single ITCZ north of the equator in the rest of the year. The meridional wind is weak during boreal spring,

7422

JOURNAL OF CLIMATE

VOLUME 29

FIG. 3. The time evolution of (a) SST (8C) and (b) precipitation (mm day21) in the SEP run (red line) and CNTL run (blue line) over SCP (258–58S, 1508–1208W).

approximately corresponding to the period when the observations show a double ITCZ straddling the equator. From May to December, there is northward crossequatorial flow into the northern ITCZ. The annual variation of SST in the northern ITCZ region is similar

between the CNTL and SEP run: both are colder (warmer) than the observations in boreal spring (autumn). South of the equator, SST is much warmer from January to June in the CNTL run, whereas that in the SEP run is closer to the observations because a large

FIG. 4. The annual-mean (a)–(c) SST (8C) and (d)–(f) precipitation (mm day21) averaged over (left) 58–158N, (center) 58S–58N and (right) 58–158S in observations (black line), CNTL run (blue line), and SEP run (red line).

15 OCTOBER 2016

7423

SONG AND ZHANG

FIG. 5. The seasonal cycle of SST (shaded; 8C) and precipitation (contours; mm day21) and 925-hPa winds (only meridional wind shown in the equatorial region; vectors; m s21) averaged over 908–1408W in (a) observations, (b) CNTL run, and (c) SEP.

part of the region is prescribed with the observed SST. Precipitation in the northern ITCZ is largely disrupted in boreal spring when the southern ITCZ is overly strong in both the CNTL and SEP runs. Correspondingly, there exists northerly wind across the equator during this period. From May to December, there is southerly cross-equatorial wind toward the northern ITCZ, similar to the observations. Note that although the SST south of the equator is close to the observations in boreal spring, the precipitation there in the SEP run is still much higher than in the observations (Figs. 5a,c), suggesting that local SST is not the only factor responsible for the excessive precipitation south of the equator.

b. The physical mechanisms for the maintenance of SST difference pattern From the above results, the annual-mean precipitation change is closely related to SST change. Hence, investigating the maintenance of SST difference patterns between the CNTL and the SEP run will help us understand the role of SST in the SEP region in the double ITCZ. Outside the prescribed SST region, the evolution of SST difference between the SEP and CNTL runs ›(DT)/›t is determined by the net surface heat flux difference DQsfc and ocean dynamic heat transport difference DDo:

C

›(DT) 5 DQsfc 1 DDo , ›t

(1)

where C 5 Cpo ro H is the heat capacity of the ocean mixed layer, Cpo and ro are the specific heat at constant pressure [3850 J (kg 8C)21] and density of seawater (1025 kg m23), and H is the mixed layer depth. The D means the difference between the SEP and CNTL runs, and T is SST. The SST difference is in the range from 20.88 to 0.88C in the tropics for 10 years (Fig. 1d). This is equivalent to a magnitude of approximately 1 W m22 for a 100-m mixed layer on the left-hand side of Eq. (1). On the right-hand side, DQsfc can reach approximately 25 W m22 (see below), an order of magnitude larger than the SST difference term. In other words, the difference in ocean dynamic transport approximately balances that of the net surface heat flux. In the prescribed SST region, replacing model SST with observed SST is equivalent to nudging model SST with observed SST with a relaxation time equal to ocean model time step. Thus, the equation for the SST difference can be formally written as C

T 2 Tobs ›(DT) 5 DDo 1 DQsfc 2 C mod . ›t dt

(2)

The last term is the relaxation of model SST Tmod in the SEP run to observed SST Tobs, and dt is the ocean model

7424

JOURNAL OF CLIMATE

time step. The relaxation adjustment term can be interpreted as an artificial modification of either the surface heat flux into the ocean or the upper-ocean dynamic transport. In the analysis below, DQsfc is obtained directly from the model runs, and DDo is obtained as the residual in Eq. (1). Thus, in the prescribed SST region, DDo represents the sum of the relaxation adjustment term in Eq. (2) and the actual ocean dynamic transport, but we will call it loosely ocean dynamics. It represents the contribution that should be provided by the ocean in order to have the prescribed SSTs in the SEP region. The net surface heat flux difference DQsfc (positive downward) is composed of differences of radiative heat flux DQR, latent heat flux DQE, and sensible heat flux DQH. The radiative heat flux DQR in turn consists of shortwave radiation DQS and longwave radiation DQL. Following Zhang and Li (2014), DQL can be further divided into upward and downward longwave radiation, dn DQup L and DQL , respectively. According to the Stefan– Boltzmann law, Qup L can be written as 3 DQup L 5 24sT DT ,

(3)

VOLUME 29

where a ffi L/RyT2 ’ 0.067 K21, W is wind speed, L is the latent heat of evaporation, and Ry is the gas constant for water vapor. The derivation of Eqs. (5) and (6) is given in the appendix. By the sign convention, QE is negative. In the calculation of DQwE and DQoE , QE and W from the CNTL run are used. The atmospheric term DQaE is mainly contributed by that due to wind speed difference DW, while the Newtonian cooling term DQoE is related to SST difference DT. Hence, the atmospheric forcing on SST via surface heat flux (referred to as atmospheric heat flux term) DQatm consists of DQaR , DQH, and DQaE : DQatm 5 DQaR 1 DQH 1 DQaE .

(7)

We can combine the Newtonian cooling term DQoE and DQup L as the ocean damping term DQocn. To summarize, with the above decomposition of surface fluxes into various components, Eq. (1) can be written as C

  Q ›(DT) 5 DQaR 1 DQH 1 DQrE 1 E DW ›t W 1 (aQE DT 2 4sT 3 DT) 1 DDo .

(8)

where s is the Stefan–Boltzmann constant (5.67 3 1028 W m22 K24). The downward longwave radiation DQdn L can be estimated as the difference between DQL and DQup L . This decomposition considers the ocean damping effect through longwave radiation, which is thought to be important for the change of SST pattern under global warming (Zhang and Li 2014). Hence, DQR can be divided into atmospheric part DQaR (sum of DQS and DQdn L ) and oceanic damping part DQup L . Following Xie et al. (2010), DQE is also decomposed into the atmospheric part DQaE (because of differences in near-surface wind speed, humidity, and stability) and oceanic part DQoE (because of SST difference). The atmospheric contribution DQaE represents the atmospheric forcing on SST via latent heat flux, while the oceanic contribution DQoE represents the oceanic response to SST change (hereafter referred to as Newtonian cooling), which represents the ocean damping effects via latent heat flux. The atmospheric contribution DQaE can be further decomposed into that due to wind speed change DQwE and residuals due to relative humidity and stability changes DQrE :

Here, on the right-hand side, the first pair of the parentheses represents atmospheric forcing, the second one represents the ocean response or damping, and the last term is the ocean dynamics (including the relaxation term in the prescribed SST region). It should be noted that the ocean damping from longwave radiation is likely overestimated. A large part of the upward longwave radiation emitted from the ocean’s surface is radiated back to the ocean from the low-level atmosphere. In other words, the downward and upward longwave radiation has large cancellations, because downwelling longwave radiation is determined to a large degree by temperature and moisture of the boundary layer, which are strongly influenced by SST. As discussed above, the SST difference pattern is determined by DQatm, ocean dynamics DDo, and DQocn. These three components and each term of DQatm are shown in Fig. 6. The contributions to latent heat flux difference due to wind speed change DQwE , surface wind speed difference, and SST difference are shown in Fig. 7.

DQE 5 DQaE 1 DQoE 5 DQwE 1 DQrE 1 DQoE .

(4)

c. The atmospheric role in the SST maintenance over other regions

(5)

Over other regions, the energetic analysis suggests the contribution of each term to the SST change under the influence of prescribed SEP SST. Among the atmospheric heat fluxes (Fig. 6a), the radiative heat flux and latent heat flux are comparable (Figs. 6b,c), whereas the sensible heat flux is negligible (Fig. 6d). Among the

Here, ›QE Q DW 5 E DW and ›W W ›Q DQoE 5 E DT 5 aQE DT , ›T DQwE 5

(6)

15 OCTOBER 2016

SONG AND ZHANG

7425

FIG. 6. The differences between the SEP and CNTL runs (W m22) for (a) DQatm, (b) DQaR , (c) DQaE , (d) DQH, (e) DQocn, and (f) DDo. The stippled regions indicate the differences exceeding the 5% significance level.

atmospheric radiative heat flux, the shortwave radiation plays a dominant role, while the longwave radiation exerts a smaller and opposite effect. The upward longwave damping effects have a similar pattern to the Newtonian cooling term, but with smaller magnitude (figure not shown). The radiative heat flux contributes to the SST pattern negatively over the SCP and western Pacific through the negative shortwave radiation–SST feedback: warmer SST increases convection and precipitation, which increase the cloud amount and reduces the incoming shortwave radiation, leading to SST cooling, and vice versa. Over the NEP region, the positive stratus cloud–SST feedback makes radiative heat flux act as a positive contribution to the SST. The atmospheric latent heat flux contributes to the SST warming north of the equator and cooling over the SCP, while its contribution to the western Pacific warming south of the equator is small. The ocean dynamics term contributes much to the western Pacific warming and moderate cooling over the SCP. The ocean damping term resembles

the SST pattern with an opposite sign, indicating the ability of ocean to adjust the temperature itself through longwave radiation and latent heat flux. The contribution to latent heat flux difference from wind speed change can explain most of the atmospheric term of latent heat flux (Figs. 7a and 6c). The resemblance [with a pattern correlation of 20.73 for the tropical Pacific (208S–208N, 1208E–908W)] between the wind speed difference and SST difference patterns (Fig. 7b) indicates the importance of the wind– evaporation–SST (WES) feedback. A decrease in wind speed reduces evaporation, leading to warmer SST, and vice versa. Hence, it is crucial to understand how the wind field in the tropical Pacific changes when the observed SST is prescribed over the SEP region. Referring back to Figs. 1d and 2f, the prescribed cold SST over the SEP region suppresses convection and reduces precipitation locally. The decreased latent heat release in the troposphere induces the southeasterly winds crossing the equator and SCP (Fig. 1d), either as a

7426

JOURNAL OF CLIMATE

VOLUME 29

enhanced Walker circulation is related to the wind convergence over the western Pacific, reducing the wind speed and warming the SST. The above results suggest that the reduced SST in the SEP suppresses convection and precipitation. The negative change of atmospheric latent heat release induces the southeasterly wind outside the SEP region and strengthens the Walker circulation. The enhanced ascending motion over the western Pacific causes the wind convergence and reduces the wind speed, which warms the SST through the WES feedback. However, the enhanced ascending motion also increases the cloud amount and reduces the incoming shortwave radiation (Fig. 6b), which cools the SST over the western Pacific. Hence, the net atmospheric flux over the western Pacific south of the equator is small, while the ocean dynamics plays a more important role.

d. The oceanic role in the SST maintenance over other regions

FIG. 7. The difference of (a) wind term of latent heat flux (W m22) and (b) SST (shaded; 8C) and wind speed (contours; contour interval is 0.3 m s21). In (b), the dashed (solid) line represents the negative (positive) wind speed, with zero as the thick line. The stippled regions indicate the differences exceeding the 5% significance level.

Matsuno–Gill response (Matsuno 1966; Gill 1980) or as a Lindzen–Nigam response (Lindzen and Nigam 1987). Although the Matsuno–Gill model and the Lindzen–Nigam model view the tropical circulation from different perspectives, Neelin (1989) unified interpretation of the two models, at least to the first order. Superposing on the climatological southeasterly wind, the wind speed is increased north of the equator and SCP (Fig. 7b). Hence, the SST cooling is extended outside the SEP and decreases precipitation there. The suppressed convection over the eastern Pacific enhances the Walker circulation, as shown in Fig. 8. Compared to the observations, the Walker circulation is much weaker in the CNTL run, with weaker easterly (westerly) divergent wind in the low level (upper level) (Figs. 8a–c). The low-level upward motion branch also shifts eastward, corresponding to the double ITCZ. In the SEP run, the magnitude of the Walker circulation is much closer to the observations, although with westward shift (Figs. 8d,e). Both the ascending motion over the western Pacific and descending motion over the eastern Pacific are enhanced compared to the CNTL run (Fig. 8f). The

The above analysis shows the importance of atmospheric processes in the improvement of double ITCZ. The surface wind changes also influence the upperocean current, which in turn influences SST via the ocean dynamical transport. Before we delve into the ocean’s role, the SEP region warrants some special remarks, because SST there is prescribed, and thus the energetic analysis over the region is slightly different from that over other regions. Over the SEP region, in observations colder SST means greater gain of heat from the atmosphere by the ocean. In the SEP run, model SST is replaced by the much colder observed SST. From the energetic point of view, there must be a heat loss to achieve this. The colder the observed SST is relative to the model SST, the more heat loss. This heat loss is achieved through nudging (note that replacing model SST with observed SST is equivalent to nudging with relaxation time equal to one model time step). Since the colder (compared to the CNTL run) SST is prescribed in the SEP run, it favors the development of stratus cloud (figure not shown), resulting in less shortwave radiation at the ocean surface (Fig. 6b). The colder SST also suppresses the convection and induces low-level divergence, which enhances the surface wind speed and latent heat flux out of the ocean (Fig. 6c). Together these lead to negative atmospheric heat flux change into the ocean (Fig. 6a). Averaged over the SEP region, the difference in atmospheric heat flux is 28.95 W m22. The atmospheric heat flux comes from both radiative heat flux and latent heat flux about equally, suggesting the importance of reasonable near-surface wind and stratus cloud simulation. The difference in ocean dynamic transport (which includes the nudging correction)

15 OCTOBER 2016

SONG AND ZHANG

7427

FIG. 8. The annual-mean vertical velocity-divergent zonal wind (vectors), the vertical velocity (contours; contour interval of 24 3 1023 Pa s21), and divergent zonal wind speed (shaded; m s21) averaged over 58–108S in (a) NCEP-1, (c) CNTL run, and (e) SEP run, and differences between (b) the CNTL run and observations, (d) SEP run and observations, and (f) SEP and CNTL runs. The dashed (solid) line represents the negative (positive) wind speed, with zero as the thick line. The stippled areas in (b),(d),(f) indicate the differences of zonal wind speed exceeding the 5% significance level.

is 213.22 W m22. Since the relaxation term is included in ocean dynamic transport over the SEP region, at least part of this difference is as a result of the nudging correction, which is negative. Nevertheless, the coastal upwelling is increased in the SEP run (Fig. 9b), which should contribute to the negative ocean dynamics difference between SEP and CNTL runs. Off the Peruvian–Chilean coast, the upwelling bias exists in the low-resolution ocean model, which contributes to the large cooling via ocean dynamics (Fig. 6f). A successful simulation of coastal upwelling requires a correct simulation of the coastal winds and even ocean mesoscale eddies (Mechoso et al. 2014).

The difference of surface and upper-ocean states between the SEP and CNTL runs is shown in Fig. 9. Consistent with the low-level wind difference (Fig. 1d), the surface wind stress is increased in and northwest of the SEP region and decreased north of the equator (Fig. 9a). The enhanced southeasterly wind leads to negative surface wind stress curl along the Peruvian coast and enhances the upwelling there (Fig. 9b). The divergence of wind stress change leads to positive wind stress curl over the SCP, corresponding to downwelling. Because of the reduced wind stress north of the equator, the North Equatorial Countercurrent (NECC) is enhanced (Fig. 9c). This contributes to the warmer SST

7428

JOURNAL OF CLIMATE

VOLUME 29

FIG. 9. The annual-mean differences of (a) surface wind stress (vectors; 1022 N m22) and its curl (shaded; 1027 N m23), (b) ocean vertical velocity at 50 m (1024 m s21), (c) ocean surface zonal current (1022 m s21), and (d) sea surface height (1022 m) between the SEP and CNTL runs. The stippled areas indicate that the differences exceed the 5% significance level.

across the Pacific basin north of the equator, as shown in Fig. 4a, which can also be inferred from the ocean dynamics contribution to the SST (Fig. 6f). Because of the enhanced southeasterly wind stress over the SEP, the South Equatorial Current (SEC) is enhanced along 108S. The enhanced ocean current prevents the warm water in the western Pacific from expanding eastward, contributing to the western Pacific warming and central Pacific cooling (Fig. 6e). The enhanced SEC also leads to higher SSH over the western Pacific. North of the equator, the negative wind stress curl over the western Pacific induces downwelling and higher SSH there. Meanwhile, upwelling leads to lower SSH over the eastern Pacific. Hence, the SSH change exhibits a west high–east low seesaw pattern. This gives an SSH distribution in the SEP run in closer agreement with the observations (figure not shown). The meridional structure of surface zonal wind stress and wind stress curl averaged over 1208–1608W is shown in Fig. 10. In contrast to the SODA and SEP run, the easterly wind stress decreases from 208 to 88S, then slightly increases to the equator in the CNTL run. The minimum easterly wind stress near 88S is due to the spurious convection, corresponding to the wind convergence. It leads to a negative wind curl along 208–88S and positive wind curl equatorward, corresponding to

downwelling and upwelling, respectively. The related change in SSH causes stronger South Equatorial Countercurrent (SECC), which transports warm water eastward in the CNTL run. The warmer SCP leads to more convection and wind convergence, forming a positive feedback. This feedback is also suggested by Zhang et al. (2007) and Zhang and Song (2010). In the SEP run, this positive feedback is broken because of the enhanced southeasterly wind. On the one hand, the enhanced southeasterly strengthens the SEC; on the other hand, the change of wind curl weakens the SECC, both contributing to the improvement. The surface wind change also leads to changes in upper-ocean current and temperature, as shown in Fig. 11. In the SODA and SEP run, the SECC is only confined to west of 1708E and the date line, respectively, in the upper 150 m, while east of the date line is dominated by the westward SEC (Figs. 11a,c). On the other hand, the SECC in the CNTL run extends eastward to as far as 1508W and reaches more than 250 m, while the SEC is only confined to east of 1608W in the upper 200 m (Fig. 11b). The stronger SECC in the CNTL run transports the warm water farther eastward, leading to a shallower and flatter thermocline across the Pacific (Fig. 11b). Compared to SODA, there is positive zonal current bias in the upper 250 m across the Pacific in the

15 OCTOBER 2016

SONG AND ZHANG

7429

through ocean dynamic transport (Fig. 6e), further enhancing the east–west SST gradient and forming a positive feedback.

4. Conclusions and discussion

FIG. 10. The annual-mean (a) zonal wind stress (N m22 ) and (b) wind stress curl (1027 N m23) averaged over 1608–1208W in SODA (black line), the CNTL run (blue line), and the SEP run (red line).

southern ITCZ region. The eastern Pacific is warmer in the top 50 m and the western Pacific is much colder at the subsurface (Fig. 11d). The subsurface cold bias is greatly improved in the SEP run, and the surface warming bias is shifted to the western Pacific (Figs. 11e,f). The zonal current bias is also reduced significantly. This is because of the improved westward SEC current in the SEP run. Over the equatorial region, the SEC reaches more than 100 m in depth in the SODA and SEP run, but only about 60 m in the CNTL run as a result of the weaker easterly wind stress in the central Pacific (Figs. 11g–i). Compared to the CNTL run, the thermocline is also deepened over the equatorial western Pacific in the SEP run (Fig. 11l). The thermocline and wind stress changes suggest that the Bjerknes feedback is operating: the prescribed SEP cooling and extended equatorial cooling because of the WES feedback enhance the east–west SST gradient, which induces the easterly wind over the equatorial Pacific (Fig. 9a). The easterly wind induces the upwelling over the eastern Pacific (Figs. 9a,b), resulting in an enhanced east–west thermocline gradient (Figs. 11c,i). The enhanced east–west thermocline contributes to the SST warming over the western Pacific

In this study, the influences of the SEP SST on the double ITCZ are investigated by prescribing the observed SST over the SEP region in CESM1.2.2. By comparing the simulations with and without the prescribed SST (referred to as the SEP run and CNTL run, respectively), it is found that both the double ITCZ and dry equator are significantly improved in the SEP run. However, the northern ITCZ precipitation becomes more excessive. Compared to the CNTL run, the southern ITCZ in the eastern equatorial Pacific is reduced in the SEP run, making it more closely resemble the observed SPCZ. The improvement of double ITCZ and dry equator is closely related to SST changes. Over the equatorial region, the precipitation is increased west of 1708W because of the higher SST in the SEP run, thereby improving the dry equator. South of the equator, because of the colder SST in the SEP run, the precipitation is decreased significantly east of 1708W. However, the seasonal evolution of ITCZ shows no obvious improvement in the SEP run. Although the SST is similar to the observations in boreal spring over the eastern Pacific, the alternating ITCZ bias is still evident, accompanied by the cross-equatorial northerly wind. This suggests that other factors may be in play in the alternating ITCZ bias. Because of the importance of SST in the improvements of precipitation, the SST difference pattern is investigated following the approach of Zhang and Li (2014) and Xie et al. (2010). The main contribution terms include the atmospheric heat fluxes, ocean dynamics, and ocean damping effects through the longwave radiation and latent heat flux. Among the atmospheric heat fluxes, the radiative heat flux and latent heat flux are comparable, whereas the sensible heat flux is negligible. The latent heat flux is mainly determined by wind speed through WES feedback: The decreased (increased) wind speed corresponds to positive (negative) SST change. The radiative heat flux, dominated by shortwave radiative heat flux, plays an opposite role in the maintenance of the SST pattern over the SCP and western Pacific through negative shortwave radiation–SST feedback. The prescribed colder SST over the SEP region suppresses convection and precipitation locally. The resulting negative latent heat release induces the southeasterly crossing the equator and SCP, further cooling SST there through the WES feedback. The Walker circulation is strengthened

7430

JOURNAL OF CLIMATE

VOLUME 29

FIG. 11. The annual-mean upper-level ocean temperature (shaded; 8C) and zonal current (contour; cm s21) averaged over (a)–(f) 108– 58S and (g)–(l) 58S–58N for (a),(g) SODA; (b),(h) the CNTL run; and (c),(i) the SEP run, as well as the differences between (d),(j) CNTL and SODA; (e),(k) SEP and SODA; (f),(l) SEP and CNTL. The thick contour is zero, and solid (dashed) lines represent positive (negative) values; the contour interval is 3 cm s21 for mean state and 2 cm s21 for the differences. The stippled areas indicate the differences of zonal current exceeding the 5% significance level.

as a result of the enhanced descending motion over the SEP, and ascending motion occupies much of the region over the western Pacific. The resultant wind convergence decreases the wind speed and increases the SST over the western Pacific. The increased SST in turn leads to more clouds and reduces the incoming solar radiation, which cancels out a large part of the effects of latent heat flux. Eventually, ocean dynamics

plays a more important role in the SST warming over the western Pacific. The ocean circulation is affected by the surface wind changes between the SEP and CNTL runs, which further changes the SST and atmosphere circulation. The Bjerknes feedback plays an important role in the equatorial SST pattern maintenance: the prescribed SEP cooling and equatorial eastern Pacific cooling due

15 OCTOBER 2016

7431

SONG AND ZHANG

to the WES feedback enhance the east–west SST gradient, inducing the easterly wind over the equatorial Pacific. The easterly wind leads to upwelling over the eastern Pacific and the thermocline shoals. The enhanced east–west thermocline contributes to the western Pacific warming through the ocean dynamics, further enhancing the east– west SST gradient. South of the equator, there also exists another positive feedback in the CNTL run: the spurious convection corresponds to wind convergence. It leads to negative (positive) wind curl north (south) of 88S, corresponding to upwelling (downwelling). The resultant SSH change leads to strong SECC, which transports more warm water eastward and favors further convection. This positive feedback is broken in the SEP run because of the enhanced southeasterly wind. The increased southeasterly wind enhances the SEC, and, at the same time, the wind curl change eliminates the spurious SECC. The double ITCZ is mostly a coupled model problem, although it also appears in atmosphere-only models. It is closely related to the SST biases in the coupled model. In fact, the double ITCZ and SST biases mutually reinforce each other. Recognizing that there is a warm SST bias over the SEP, this study did not address its causes, but rather sought to investigate what effect it may have on double ITCZ if it were improved (be it through improvement in atmospheric processes, such as clouds, or oceanic processes, such as upwelling). By prescribing SST over the SEP, we aim to understand the mechanisms through which it affects the SST and precipitation in the rest of the tropical Pacific. The double-ITCZ problem is extremely challenging, as indicated by the lack of substantial progress in the last three decades. Deficiencies in both atmosphere and ocean models contribute to the SST biases and establish that process-level studies are needed to improve the double-ITCZ problem. At the process level, some previous studies (Song and Zhang 2009; Zhang and Song 2010) investigated how atmospheric convection affects double ITCZ and SST. Likewise, one could improve ocean process parameterization to investigate its effect on SST simulation. Besides, the atmosphere and ocean model resolution will likely have important effects as well. For example, a reasonable depiction of narrow and deep Andes and Central American mountains are found to be important for the ITCZ simulation (Xu et al. 2004, 2005). These are beyond the scope of this study and will be the subjects of future research.

PNNL Contract DOE/PNNL 190110. The computational support for this work was provided by the NCAR Computational and Information Systems Laboratory.

Acknowledgments. This material is based upon work supported by the U.S. Department of Energy, Office of Science, Biological and Environmental Research Program (BER), under Award DE-SC0008880, the National Science Foundation Grant AGS-1549259, and the

DQE 5

APPENDIX The Components of Latent Heat Flux The latent heat flux QE (negative upward) can be written as QE 5 2Lra CE W[qs (T) 2 qa ],

(A1)

where L is latent heat of evaporation, ra is surface air density, CE is aerodynamic transfer coefficient, W is wind speed at 10 m, and qs(T) is the saturation specific humidity at sea surface temperature. The specific humidity of air at 10 m qa can be approximated by the following equation: qa 5 RHqs (Ta ) 5 RHqs (T 2 dT)   qs (T) dT 5 RH qs (T) 2 T ffi RHqs (T)(1 2 adT) ,

(A2)

where RH is relative humidity, Ta is the surface air temperature, dT 5 T 2 Ta is the surface air–sea temperature difference (referred to as the surface stability parameter), and a5

1 ›qs (T) L ffi ’ 0:067 K21 qs (T) ›T Ry T 2

is from the Clausius–Clapeyron equation. Therefore, from Eq. (A1) and Eq. (A2), we can obtain QE 5 2Lra CE Wqs (T)[1 2 RH(1 2 adT)] 5 2Wqs (T)R ,

(A3)

where R 5 LraCE[1 2 RH(1 2 adT)] is the combination of relative humidity and surface stability. Taking the logarithm of Eq. (A3) and differentiating gives DQE DW Dqs (T) DR 1 . 1 5 W R qs (T) QE

(A4)

Hence, the differences of latent heat flux DQE can be obtained as

5

QE Q ›q Q DW 1 E s DT 1 E DR W qs ›T R QE DW 1 aQE DT 1 DQrE 5 DQwE 1 DQoE 1 DQrE . W (A5)

7432

JOURNAL OF CLIMATE REFERENCES

Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 1147–1167, doi:10.1175/1525-7541(2003)004,1147:TVGPCP.2.0.CO;2. Bacmeister, J. T., M. J. Suarez, and F. R. Robertson, 2006: Rain reevaporation, boundary layer–convection interactions, and Pacific rainfall patterns in an AGCM. J. Atmos. Sci., 63, 3383– 3403, doi:10.1175/JAS3791.1. Bellucci, A., S. Gualdi, and A. Navarra, 2010: The double-ITCZ syndrome in coupled general circulation models: The role of large-scale vertical circulation regimes. J. Climate, 23, 1127– 1145, doi:10.1175/2009JCLI3002.1. Cai, W., and T. Cowan, 2013: Why is the amplitude of the Indian Ocean dipole overly large in CMIP3 and CMIP5 climate models? Geophys. Res. Lett., 40, 1200–1205, doi:10.1002/ grl.50208. Carton, J. A., and B. S. Giese, 2008: A reanalysis of ocean climate using Simple Ocean Data Assimilation (SODA). Mon. Wea. Rev., 136, 2999–3017, doi:10.1175/2007MWR1978.1. Chikira, M., 2010: A cumulus parameterization with statedependent entrainment rate. Part II: Impact on climatology in a general circulation model. J. Atmos. Sci., 67, 2194–2211, doi:10.1175/2010JAS3317.1. Dai, A. G., 2006: Precipitation characteristics in eighteen coupled climate models. J. Climate, 19, 4605–4630, doi:10.1175/ JCLI3884.1. Dai, F., R. Yu, X. Zhang, Y. Yu, and J. Li, 2003: The impact of lowlevel cloud over the eastern subtropical Pacific on the ‘‘double ITCZ’’ in LASG FGCM-0. Adv. Atmos. Sci., 20, 461–474, doi:10.1007/BF02690804. ——, ——, ——, and ——, 2005: Impacts of an improved low-level cloud scheme on the eastern Pacific ITCZ–cold tongue complex. Adv. Atmos. Sci., 22, 559–574, doi:10.1007/BF02918488. Davey, M. K., and Coauthors, 2002: STOIC: A study of coupled model climatology and variability in tropical ocean regions. Climate Dyn., 18, 403–420, doi:10.1007/s00382-001-0188-6. De Szoeke, S., 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. Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447–462, doi:10.1002/qj.49710644905. Hirota, N., Y. N. Takayabu, M. Watanabe, and M. Kimoto, 2011: Precipitation reproducibility over tropical oceans and its relationship to the double ITCZ problem in CMIP3 and MIROC5 climate models. J. Climate, 24, 4859–4873, doi:10.1175/ 2011JCLI4156.1. Hwang, Y.-T., and D. M. M. Frierson, 2013: Link between the double-Intertropical Convergence Zone problem and cloud biases over the Southern Ocean. Proc. Natl. Acad. Sci. USA, 110, 4935–4940, doi:10.1073/pnas.1213302110. Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077,0437:TNYRP.2.0.CO;2. Kang, S. M., I. M. Held, D. M. M. Frierson, and M. Zhao, 2008: The response of the ITCZ to extratropical thermal forcing: Idealized slab-ocean experiment with a GCM. J. Climate, 21, 3521– 3532, doi:10.1175/2007JCLI2146.1. ——, D. M. M. Frierson, and I. M. Held, 2009: The tropical response of extratropical thermal forcing in an idealized GCM:

VOLUME 29

The importance of radiative feedbacks and convective parameterization. J. Atmos. Sci., 66, 2812–2827, doi:10.1175/ 2009JAS2924.1. Kay, J. E., V. Yettella, B. Medeiros, C. Hannay, and P. Cadwell, 2016: Global climate impacts of fixing the Southern Ocean shortwave radiation bias in the Community Earth System Model (CESM). J. Climate, 29, 4617–4636, doi:10.1175/ JCLI-D-15-0358.1. Large, W. G., and G. Danabasoglu, 2006: Attribution and impacts of upper-ocean biases in CCSM3. J. Climate, 19, 2325–2346, doi:10.1175/JCLI3740.1. Li, G., and S.-P. Xie, 2014: Tropical biases in CMIP5 multimodel ensemble: The excessive equatorial Pacific cold tongue and double ITCZ problems. J. Climate, 27, 1765–1780, doi:10.1175/ JCLI-D-13-00337.1. ——, ——, and Y. Du, 2015: Climate model errors over the South Indian Ocean thermocline dome and their effect on the basin mode of interannual variability. J. Climate, 28, 3093–3098, doi:10.1175/JCLI-D-14-00810.1. Li, T., 1997: Air–sea interactions of relevance to the ITCZ: The analysis of coupled instabilities and experiments in a hybrid coupled GCM. J. Atmos. Sci., 54, 134–147, doi:10.1175/ 1520-0469(1997)054,0134:ASIORT.2.0.CO;2. Lin, J.-L., 2007: The double-ITCZ problem in IPCC AR4 coupled GCMs: Ocean–atmosphere feedback analysis. J. Climate, 20, 4497–4525, doi:10.1175/JCLI4272.1. Lindzen, R. S., and S. Nigam, 1987: On the role of sea surface temperature gradients in forcing low-level winds and convergence in the tropics. J. Atmos. Sci., 44, 2418–2436, doi:10.1175/ 1520-0469(1987)044,2418:OTROSS.2.0.CO;2. 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, doi:10.1175/1520-0442(1996)009,1635: PSCATT.2.0.CO;2. Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44, 25–43. 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, doi:10.1175/ 1520-0493(1995)123,2825:TSCOTT.2.0.CO;2. ——, and Coauthors, 2014: Ocean–cloud–atmosphere–land interactions in the southeastern Pacific: The VOCALS program. Bull. Amer. Meteor. Soc., 95, 357–375, doi:10.1175/ BAMS-D-11-00246.1. Meehl, G. A., C. Covey, B. McAvaney, M. Latif, and R. J. Stouffer, 2005: Overview of the Coupled Model Intercomparison Project. Bull. Amer. Meteor. Soc., 86, 89–93, doi:10.1175/BAMS-86-1-89. Neale, R. B., and Coauthors, 2013: Description of the NCAR Community Atmosphere Model (CAM 5.0). NCAR Tech. Note NCAR/TN-4861STR, 268 pp. [Available online at http://www.cesm.ucar.edu/models/cesm1.0/cam/docs/description/ cam5_desc.pdf.] Neelin, J. D., 1989: On the interpretation of the Gill model. J. Atmos. Sci., 46, 2466–2468, doi:10.1175/1520-0469(1989)046,2466: OTIOTG.2.0.CO;2. Oueslati, B., and G. Bellon, 2013: Convective entrainment and largescale organization of tropical precipitation: Sensitivity of the CNRM-CM5 hierarchy of models. J. Climate, 26, 2931–2946, doi:10.1175/JCLI-D-12-00314.1. ——, and ——, 2015: The double ITCZ bias in CMIP5 models: Interaction between SST, large-scale circulation and

15 OCTOBER 2016

SONG AND ZHANG

precipitation. Climate Dyn., 44, 585–607, doi:10.1007/ s00382-015-2468-6. Philander, S. G. H., D. Gu, D. Halpern, G. Lambert, N.-C. Lau, T. Li, 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. Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670. Song, X., and G. Zhang, 2009: Convection parameterization, tropical Pacific double ITCZ, and upper-ocean biases in the NCAR CCSM3. Part I: Climatology and atmospheric feedback. J. Climate, 22, 4299–4315, doi:10.1175/2009JCLI2642.1. Tian, B., 2015: Spread of model climate sensitivity linked to double-Intertropical Convergence Zone bias. Geophys. Res. Lett., 42, 4133–4141, doi:10.1002/2015GL064119. Wang, C., S.-K. Lee, and C. R. Mechoso, 2010: Interhemispheric influence of the Atlantic warm pool on the southeastern Pacific. J. Climate, 23, 404–418, doi:10.1175/2009JCLI3127.1. ——, L. Zhang, S.-K. Lee, L. Wu, and C. R. Mechoso, 2014: A global perspective on CMIP5 climate model biases. Nat. Climate Change, 4, 201–205, doi:10.1038/nclimate2118. 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. Xu, H., Y. Wang, and S.-P. Xie, 2004: Effects of the Andes on eastern Pacific climate: A regional atmospheric model study. J. Climate, 17, 589–602, doi:10.1175/1520-0442(2004)017,0589: EOTAOE.2.0.CO;2. ——, S.-P. Xie, Y. Wang, and R. J. Small, 2005: Effects of Central American mountains on the eastern Pacific winter ITCZ and moisture transport. J. Climate, 18, 3856–3873, doi:10.1175/ JCLI3497.1. Yu, J.-Y., and C. R. Mechoso, 1999: Links between annual variations of Peruvian stratocumulus clouds and of SST in the eastern

7433

equatorial Pacific. J. Climate, 12, 3305–3318, doi:10.1175/ 1520-0442(1999)012,3305:LBAVOP.2.0.CO;2. Zhang, G. J., 2002: Convective quasi-equilibrium in midlatitude continental environment and its effect on convective parameterization. J. Geophys. Res., 107, ACL 12-1–ACL 12-16, doi:10.1029/2001JD001005. ——, and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.– Ocean, 33, 407–446, doi:10.1080/07055900.1995.9649539. ——, and H. Wang, 2006: Toward mitigating the double ITCZ problem in NCAR CCSM3. Geophys. Res. Lett., 33, L06709, doi:10.1029/2005GL025229. ——, and X. Song, 2010: Convection parameterization, tropical Pacific double ITCZ, and upper-ocean biases in the NCAR CCSM3. Part II: Coupled feedback and the role of ocean heat transport. J. Climate, 23, 800–812, doi:10.1175/ 2009JCLI3109.1. Zhang, L., and T. Li, 2014: A simple analytical model for understanding the formation of sea surface temperature patterns under global warming. J. Climate, 27, 8413–8421, doi:10.1175/ JCLI-D-14-00346.1. ——, C. Wang, Z. Song, and S.-K. Lee, 2014: Remote effect of the model cold bias in the tropical North Atlantic on the warm bias in the tropical southeastern Pacific. J. Adv. Model. Earth Syst., 6, 1016–1026, doi:10.1002/2014MS000338. Zhang, X., W. Lin, and M. Zhang, 2007: Toward understanding the double Intertropical Convergence Zone pathology in coupled ocean–atmosphere general circulation models. J. Geophys. Res., 112, D12102, doi:10.1029/2006JD007878. ——, H. Liu, and M. Zhang, 2015: Double ITCZ in coupled ocean– atmosphere models: From CMIP3 to CMIP5. Geophys. Res. Lett., 42, 8651–8659, doi:10.1002/2015GL065973. Zheng, Y., T. Shinoda, J.-L. Lin, and G. N. Kiladis, 2011: Sea surface temperature biases under the stratus cloud deck in the southeast Pacific Ocean in 19 IPCC AR4 coupled general circulation models. J. Climate, 24, 4139–4164, doi:10.1175/ 2011JCLI4172.1. Zhou, Z., and S. Xie, 2015: Effects of climatological model biases on the projection of tropical climate change. J. Climate, 28, 9909–9917, doi:10.1175/JCLI-D-15-0243.1.