1. Introduction
On the basis of observational and model studies (Bjerknes 1969; Manabe et al. 1974), sea surface temperature (SST) has been considered the most important factor controlling the convective activity over tropical oceans. SST affects the heat and moisture supplies from the sea surface to the lower troposphere that fuel convective activity. An empirical SST threshold for the onset of deep convection is found in the temperature range 26°–28°C (Gadgil et al. 1984; Waliser and Graham 1993). SST also affects convective activity through lower-level wind convergence driven by the SST gradient (Back and Bretherton 2009).
In addition to the SST’s control of convective activity, recent studies suggest that large-scale subsidence plays an important role in suppressing deep convection (Lau et al. 1997; Sherwood 1999; Yuan et al. 2008; Xie et al. 2010). Takayabu et al. (2010) analyzed the values of Q1 − QR (the apparent heat source minus radiative heating; Yanai et al. 1973) estimated by the spectral latent heating (SLH) algorithm (Shige et al. 2004, 2007, 2008, 2009) using satellite data from the Tropical Rainfall Measuring Mission Precipitation Radar (TRMM PR). They showed that two modes of Q1 − QR corresponding to congestus and deep convection are dominant in the tropics. They also noted that these two modes are significantly related to the sign of large-scale vertical velocity. The heating associated with congestus over the subsidence region is confined to the lower troposphere around 800 hPa, whereas that associated with deep convection over the ascending region has its maximum around 500 hPa. They suggested that since the mid to lower troposphere is very dry over the large-scale subsidence region, entrainment of environmental dry air to a convective parcel effectively reduces the parcel’s buoyancy, suppressing deep convection. Jensen and Del Genio (2006) also suggested the importance of the dry midtroposphere using a simple parcel model. They first estimated the entrainment rate by comparing the temperature and humidity profiles of an ascending parcel in a simple model to observational data at Nauru Island. They performed sensitivity tests of cloud-top heights against various factors including an environmental temperature profile, a humidity profile, and the latent heat of fusion. Their results indicate that midtropospheric drying is more likely responsible for limiting cloud-top heights than the temperature profile or latent heat of fusion.
As described above, both SST and large-scale subsidence play important roles in determining convective activity and hence precipitation distributions. Figure 1 shows the annual means of observed precipitation and SST averaged from 1998 to 2007. Very high SST accompanied by heavy precipitation is observed over the warm pool from the Indian Ocean to the western Pacific. From the warm pool, two convergence zones with heavy precipitation, the intertropical convergence zone (ITCZ) and the South Pacific convergence zone (SPCZ), extend eastward and southeastward, respectively. Roughly speaking, greater amounts of precipitation are observed over oceans with higher SSTs. Some exceptions are identified over subsidence regions such as the southeastern Pacific, where only light precipitation associated with shallow convection is observed with relatively high SST (Takayabu et al. 2010).
Annual mean precipitation (shading; units in the grayscale below are mm day−1; TRMM) and SST (contours; unit is °C; HadISST) averaged from 1998 to 2007.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
The precipitation over the eastern Pacific has long been a major question for researchers (reviewed by Xie 2004). Why are the ITCZ and corresponding high SST regions located only over the northern Pacific around 10°N, while the solar heating is symmetric about the equator? Numerical experiments suggested that a fundamental reason for the asymmetric distribution of SST is the northwest tilt of the American continent’s coastline (Philander et al. 1996). Ekman transport of ocean water driven by northeasterly surface winds in the Northern Hemisphere (NH) is directed northwestward and that driven by southeasterly winds in the Southern Hemisphere (SH) is directed southwestward. Therefore, the Ekman flow is nearly parallel to the coastline in the NH and perpendicular to the coastline in the SH. The resulting upwelling of cold water in the SH causes the asymmetric SST about the equator.
Once a north–south SST gradient across the equator is initiated, many positive feedback mechanisms can amplify the asymmetric anomalies. When the north–south SST gradient drives anomalous southerly winds across the equator, the Coriolis force acts to deflect the southerlies westward in the SH and eastward in the NH. Superposed on the background easterly trades, evaporation is enhanced in the SH and suppressed in the NH, thereby increasing the SST gradient (Xie and Philander 1994). The southerlies, which drive westward Ekman transport in the SH and eastward transport in the NH, further amplify the SST gradient through upwelling (Chang and Philander 1994). Moreover, radiation associated with clouds also works as a positive feedback. Klein and Hartmann (1993) reported that low-level stratus cloud cover is highly correlated with SST over the southeastern Pacific. Therefore, low-level cloud cover increases over the colder southeastern Pacific and reflects the solar heating back, thereby decreasing SST further (Philander et al. 1996). Xu et al. (2004) also pointed out that the downward motion of easterlies on the lee side of the Andes strengthens the inversion and increases the amount of stratocumulus cloud cover over the southeastern Pacific. Wang et al. (2005) identified the role of cloud-top radiative cooling associated with the boundary layer clouds over the southeastern Pacific in increasing the sea level pressure and cross-equatorial flow, and thus the convection in the ITCZ. This process can feed back to the boundary layer clouds in the subtropical southeast Pacific through enhancing the overturning subsidence, thereby maintaining the hemispheric asymmetries over the eastern Pacific.
Despite of the mechanisms described above, many coupled general circulation models (CGCMs) still suffer from the so-called double ITCZ problem (Mechoso et al. 1995; de Szoeke et al. 2006; de Szoeke and Xie 2008; Xie et al. 2007)—that is, the overestimation of precipitation over the southeastern Pacific parallel to the equator, corresponding to a fictitious ITCZ in the SH. Mechoso et al. (1995) and de Szoeke and Xie (2008) discussed how a positive SST bias appearing over the southeastern Pacific is associated with various factors such as surface southerly winds, upwelling, and low-level clouds.
Some recent studies, however, suggested that deep convection behaves differently even with the same SST owing to dynamical suppression, and the SST bias may not fully explain the double ITCZ problem. Bellucci et al. (2010) analyzed phase 3 of the World Climate Research Program’s (WCRP’s) Coupled Model Intercomparison Project (CMIP3) multimodel dataset and statistically obtained the SST threshold for the onset of deep convection in each CGCM. They showed that this SST threshold significantly varies among different CGCMs, and the double ITCZ appears in models with an SST above the threshold. Moreover, Zhang et al. (2007) and Chikira (2010) reported that the double ITCZ appears even in atmospheric general circulation models (AGCMs) with a prescribed SST. Song and Zhang (2009a,b) and Chikira (2010) showed that the double ITCZ problem is mitigated when the convective parameterization scheme is modified. This is consistent with the observational studies, suggesting the importance of dynamical suppression over the southeastern Pacific.
In this study, we first examine precipitation reproducibility in the twentieth-century runs with the CMIP3 models as well as with the current (fifth) version of the Model for Interdisciplinary Research on Climate (MIROC5; Watanabe et al. 2010) over the tropical oceans. We focus on the sensitivity of deep convection to SST and to dynamical suppression by large-scale subsidence. Then, we discuss the extent to which these sensitivities of convective activity explain the double ITCZ problem in CMIP3 models. Most previous researches on the double ITCZ focused on SST bias, but recent studies suggest that convection may behave differently even with the same SST, depending on the model’s sensitivities to dynamical suppression. This paper describes analyses of the September–November (SON) season except for section 6, which discusses the annual mean and seasonal cycles. SON is the season when precipitation over the southeastern Pacific is suppressed in the real world (e.g., Takayabu et al. 2010). This study examines why precipitation over the southeastern Pacific is not suppressed in those models showing a double ITCZ bias persistent through the whole annual cycle.
This paper is structured as follows. In section 2, we describe the data used in our analyses. In section 3, we explain the precipitation reproducibility and SST distributions of CGCMs. In section 4, we investigate the sensitivity of deep convection to SST and subsidence. In section 5, we discuss the double ITCZ problem. In section 6, we describe the seasonality of our results. Section 7 provides the summary and discussion.
2. Data and method
In this study, we analyze the output data from the twentieth-century experiments (20C3M) from CMIP3 and MIROC5 as well as the Japanese 25-yr Reanalysis (JRA-25; Onogi et al. 2007) and the 40-yr European Center for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40; Uppala et al. 2005) from 1979 to 2000 over the tropical oceans (30°S–30°N). The variables used in this study include precipitation, SST, surface pressure, temperature, zonal winds, meridional winds, and relative humidity. We used daily data to calculate Q1 as described below. These daily variables are available for 19 CMIP3 models from the Program for Climate Model Diagnosis and Intercomparison (PCMDI; http://www-pcmdi.llnl.gov; expansions of these model names are given in the appendix). The details of these CMIP3 models are given in Table 1. MIROC5 is a CGCM developed cooperatively by the Atmosphere and Ocean Research Institute, National Institute for Environmental Studies, and Research Institute for Global Change (AORI/NIES/RIGC). This model is a modified version for CMIP phase 5 (CMIP5) of MIROC3.2 for CMIP3. In this paper, MIROC3.2 is denoted as MIROC_M for a model resolution of about 2.8° and MIROC_H for a resolution of 1.1°. The primary modifications are those made in physical parameterization schemes for cumulus convection (Chikira and Sugiyama 2010), large-scale condensation (Watanabe et al. 2009), boundary layer turbulence (Nakanishi and Niino 2004), and radiation (Sekiguchi and Nakajima 2008). Chikira and Sugiyama (2010) used an entrainment rate that depends on the buoyancy of a convective parcel, whereas the entrainment rate was originally set as vertically constant for each type of cloud in the Arakawa and Schubert (1974) scheme. As a result, convection with intermediate height between very shallow boundary layer clouds and deep convection is increased in the new scheme, whereas only shallow clouds and deep convections are dominant in the old scheme. The height of convection is more sensitive to environmental moisture (SST) in the new (old) scheme. Watanabe et al. (2011) suggested the importance of entrainment for precipitation over the eastern Pacific. We also use precipitation data observed by TRMM PR2A25 and Hadley Centre Global Sea Ice and Sea Surface Temperature (HadISST) data as reference data.
Reanalyses and CGCMs analyzed in this study. Their convective schemes and horizontal resolutions of the atmospheric component (°) are also indicated. The asterisk (*) indicates models using flux adjustment.
We consider that Q1 − QR mainly represents the heating rate associated with convective activity. It is helpful to examine the vertical structure of this heating, which is closely related to the height of convection, when we consider the mechanism for suppressing deep convection. However, QR is also unavailable for the models, so we use the QR values estimated by L’Ecuyer and Stephens (2003, 2007) instead. They calculated the radiative transfer considering information from TRMM on high and low clouds and precipitation. Since this QR is common to all the models, when we are comparing Q1 − QR among the models, we are actually comparing only Q1. We also use the Q1 − QR values obtained from the SLH algorithm, which are produced using TRMM PR data and the Goddard Cumulus Ensemble (GCE; Tao and Simpson 1993; Tao et al. 2003).
All the data described above are linearly interpolated into 2.5° × 2.5° horizontal grids for equality of comparison. In this study, climatological averages are defined as the average of 1979–2000 for the models and that of 1998–2007 for TRMM.
3. Precipitation reproducibility and SST distributions
Skill scores (Taylor 2001) of precipitation distributions over the tropical oceans averaged from 1979 to 2000 for the reanalyses, the CMIP3 models, and MIROC5 in the SON season. The reference is the precipitation observed by the TRMM PR during 1998–2007. An asterisk (*) indicates models using flux adjustment. Dashed line indicates the score of MIROC5.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
The climatological precipitation in SON for the TRMM PR, the reanalyses, the CMIP3 models, and MIROC5 are shown in Fig. 3. Blue dashed lines in the figure indicate contours for ω = 0 hPa s−1 at 500 hPa. The panels for the CMIP3 models are ordered by the skill score. The observation shows heavy precipitation over ascending regions such as the Indo-western Pacific warm pool, over the ITCZ, and over the SPCZ. These characteristics are roughly reproduced in the reanalyses, most of the CMIP3 models, and MIROC5, but the detailed structures differ with the skill scores. To investigate factors causing differences in the model performance, composites of precipitation and SST from the five lowest-scoring models (LSMs) and the five highest-scoring models (HSMs) are shown in Figs. 4a and 4b, respectively, and the difference between the LSMs and HSMs (LSM − HSM) is shown in Fig. 5. In this grouping of the models, we excluded models using flux adjustment to make sure that the differences do not arise from the adjustment. The statistical significance in Fig. 5 confirms that the grouping by skill score successfully extracted systematic differences between the LSMs and the HSMs. The SPCZ in the LSM composite extends eastward rather than southeastward, forming a double ITCZ structure. The corresponding positive precipitation anomaly over the southeastern Pacific is statistically significant at a level of 95%. Significant positive anomalies are also found over the subtropics in the south Indian Ocean and the northeastern Pacific. Negative anomalies appear over the western Pacific, the central equatorial Pacific, and the central south Pacific, but only a small number of them are statistically significant. The low significance level probably occurs because the number of samples is very small. Note that the positive anomalies are located mostly in large-scale subsidence regions, such as those over the southeastern Pacific and the south Indian oceans, whereas the negative anomalies are located in ascending regions. This systematic anomaly pattern implies that the double ITCZ problem is not only a local problem depending on the SST distribution over the southeastern Pacific, but also a problem with the precipitation schemes that control precipitation over the entire tropics.
Precipitation (shading; mm day−1) and SST (black contours; °C) in the SON season for (a) the observations (TRMM and HadISST), (b) JRA, (c) ERA, (d)–(v) the CMIP3 models, and (w) MIROC5. Blue dashed lines are contours of ω500 = 0 hPa s−1, indicating the boundaries of the large-scale subsidence regions and ascending regions. CMIP3 models are shown in order of decreasing skill score.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
As in Fig. 3, but for the (a) HSM composite and (b) LSM composite.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
Difference in precipitation between LSM and HSM composites (contours; mm day−1) in the SON season for 1979–2000. Contour intervals are 0.12 (±0.6, ±0.18, ±0.3.). Negative values are indicated by dashed lines. Light and dark shading denote significance levels of 90% and 95%, respectively.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
The climatological SST distributions for the LSMs and HSMs are also indicated in Fig. 4. Very high SSTs (above 28°C) are located over the Indo-western Pacific Ocean and extend zonally around 10°N and 10°S to the eastern Pacific in both the LSMs and the HSMs. The areal average of the SST over the entire tropical oceans (30°S–30°N) is 25.9°C for the LSMs and 24.8°C for the HSMs. The effect of the SST on the convective behavior will be examined in the next section.
The LSMs and HSMs seem to differ in the relationship between precipitation and SST. As described above, the SPCZ in the LSM composite extends eastward instead of southeastward from the western Pacific (Fig. 4b). This is where the SST is relatively high compared to the surrounding areas. On the other hand, the SPCZs in the HSM composite (Fig. 4a) as well as in the observation (Fig. 3a) extend southeastward, crossing SST contours. The patterns of precipitation and SST look more similar in the LSMs than in the HSMs or the observation. We speculate that the precipitation in the LSMs follows SST too closely. This speculation will be further examined quantitatively in the next section.
4. Sensitivity of deep convection to SST and subsidence
In this section, we examine the sensitivity of deep convection to SST and subsidence over the tropical oceans (30°S–30°N). These two environmental factors are considered to be essential in controlling convective activity (see section 1).
As described in section 3, the precipitation distribution in the LSM seems to follow the SST distribution very closely. We quantify the degree of relationship between the climatological precipitation and SST by a pattern correlation over the tropical oceans (30°S–30°N). Then, we plot the precipitation skill scores against this precipitation–SST pattern correlation, as shown in Fig. 6. A significant intermodel relationship among the CMIP3 models appears between the skill scores and the precipitation–SST pattern correlation, with a correlation coefficient of −0.58. In other words, models with higher significance in relationship between precipitation and SST tend to have lower precipitation skill scores. Interestingly, the reanalyses are the exceptions. Their skill scores are very high even when the relationship between precipitation and SST is relatively significant compared to the observation and the HSMs. These higher pattern correlations of the reanalyses possibly represent the characteristics of the models rather than the assimilation of the observational data. In general, the degree of dependence on the model in a reanalysis is considered to be larger for variables with less accurate observations, such as humidity (Uppala et al. 2005).
Intermodel relationship between the skill score (ordinate) shown in Fig. 2 and the pattern correlation between precipitation and SST (abscissa) over tropical oceans. Parentheses [()] show the observations (TRMM, HadISST), JRA, and ERA. Solid triangle indicates MIROC5. Variables are averages in the SON season of 1998–2007 for the observations and 1979–2000 for the models. Intermodel correlation among CMIP3 models is −0.58.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
Similarly, the effect of large-scale subsidence on the precipitation distribution is examined. Figure 7 is the same as Fig. 6 but for the scatter against the pattern correlation calculated between the climatological precipitation and vertical velocity at 500 hPa (ω500). In this case, models with a higher correlation between precipitation and ω500 are likely to have higher skill scores. The intermodel correlation between the skill scores and the pattern correlations is −0.81.
(a) As in Fig. 6, but the abscissa is the pattern correlation between precipitation and ω500. Intermodel correlation is −0.81. (b) An enlarged figure of the box in (a).
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
We consider that the pattern correlation between precipitation and ω500 represents the effects of “dynamical suppression.” The correlation suggests that a decrease in the buoyancy of convective parcels owing to the entrainment of environmental dry air is mainly responsible for suppressing deep convection over subsidence regions (see section 1).
To investigate the effect of midlevel dry layer on the height of convection, the Q1 − QR profiles over the tropical oceans (30°S–30°N) for the LSMs and HSMs are stratified against relative humidity at 600 hPa (RH600), as shown in Figs. 8a and 8b, respectively. Similar results can be obtained if the relative humidity is used at 850 hPa or 400 hPa; hence, the results are not very sensitive to the selection of levels, as long as they are in the lower-to-mid free troposphere. Note that QR is based on TRMM observations (see section 1), so the QR profiles are stratified using RH600 information taken not from the models but from the JRA reanalysis data. The Q1 − QR profiles in both the HSMs and LSMs show a clear bimodality with a deep mode for RH600 > 50% and a shallow mode for RH600 < 50%. This is consistent with Takayabu et al. (2010), who analyzed the TRMM data and showed bimodal Q1 − QR profiles stratified against ω500. However, a difference appears in the shallow heating mode, as the maximum heating is located at the lowest level in the climate models but at around 800 hPa in the TRMM data. The models probably underestimate heating associated with congestus at 800 hPa. This is reminiscent of the fewer convection with intermediate height in the old version MIROC3.2, which problem is largely mitigated in the new version MIROC5 (Chikira and Sugiyama 2010). On the other hand, the TRMM data possibly lack a shallower mode at the lowest level because the sensitivity of the TRMM PR sensor is limited to 15 dBZ and scarcely detects the weak precipitation associated with trade cumulus. Therefore, the shallow heating observed by TRMM PR in Takayabu et al. (2010) is mainly associated with congestus clouds, whereas that in the models is also associated with very shallow boundary layer clouds.
Q1 − QR (contour, K day−1) profiles stratified against relative humidity at 600 hPa (horizontal axis, %) for (a) the HSM composite, (b) the LSM composite, and (C) LSM − HSM over the tropical oceans for 1979–2000. Negative values are indicated by dashed lines. Contour intervals are 0.6 in (a) and (b) and 0.2 in (c). Light and dark shadings in (c) denote significance levels of 90% and 95%, respectively.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
Although the bimodal structure is common to the HSMs and LSMs, the degree of its dependency on RH600 differs significantly. The heating difference LSM − HSM shown in Fig. 8c is positive for RH600 < 50% and negative for RH600 > 50%. The positive and negative heating anomalies correspond to the positive precipitation anomalies over subsidence regions and negative anomalies over ascending regions, respectively (Fig. 5). Therefore, the RH600 dependency of the bimodal heating is significantly weaker in the LSMs than in the HSMs.
In addition to the RH600 dependency, the effect of SST on Q1 − QR profiles is examined. Figures 9a–c are the same as Fig. 8c, but stratified for limited SST ranges of 21°–23°C, 24°–26°C, and 27°–29°C, respectively. Overall, they are similar in that they show positive anomalies with small RH600 and negative anomalies with large RH600, indicating that the RH600 dependency is weaker in the LSMs than in the HSMs. The largest positive anomaly is found over very dry areas with very high SST (Fig. 9c). Even when the mid to lower troposphere is very dry (RH600 < 10%), deep convection is active in the LSMs compared to the HSMs over oceans with very high SST (>27°C). These results suggest that dynamical suppression in the LSMs is very weak, and deep convection follows SST very closely. These disruptions of the heating dependencies on RH600 and SST provide us with physical interpretations for why the precipitation in the HSMs has a weaker correlation with SST and a stronger correlation with ω500, as shown in Figs. 6 and 7, respectively.
As in Fig. 8c, but over the tropical oceans with SST ranges of (a) 21°–23°C, (b) 24°–26°C, and (c) 27°–29°C.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
The weak sensitivity of deep convection to environmental humidity in the LSMs compared to the HSMs is consistent with the positive and negative precipitation anomalies over subsidence and ascending region, respectively. On the other hand, the reduced convective overturning due to negative precipitation anomalies over ascending regions may contribute to positive anomalies over subsidence regions. However, it is difficult to examine such effects diagnostically. Further investigations including model experiments are necessary.
Finally, Fig. 10 shows the RH600 dependence of the Q1 − QR profiles for the difference of LSM − MIROC5. Since this difference is similar to that of LSM − HSM, convective activity in MIROC5 is similar to that in the HSMs. The RH600 dependence of the bimodal heating in MIROC5 is stronger than that in the LSMs, and deep convection is effectively suppressed at low midtropospheric relative humidity.
As in Fig. 8c, but for LSM − MIROC5.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
5. The double ITCZ problem
In this section, we focus on the double ITCZ problem over the southeastern Pacific in the SON seasons. Our analyses in this paper so far are based on the precipitation skill score over the entire tropical oceans (30°S–30°N). Bellucci et al. (2010) defined an index for the double ITCZ using precipitation averaged over the southeastern Pacific (20°S–0°, 100°–150°W, hereafter referred to as the SEP region). The correlation coefficient between the precipitation skill score and the double ITCZ index is −0.68, as shown in Fig. 11. Models with lower reproducibility of precipitation over the entire tropical oceans suffer from the double ITCZ problem over the southeastern Pacific. This relationship of the entire tropical oceans and the southeastern Pacific supports the suggestion in section 3 that the double ITCZ problem can be attributed to a problem with precipitation schemes.
(a) Intermodel relationship between the skill score (ordinate) and average precipitation over the southeastern Pacific (20°S–0°, 100°–150°W) (abscissa). Intermodel correlation among CMIP3 models is −0.68. (b) An enlarged figure of the box in (a).
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
The sensitivities of deep convection to the SST and large-scale subsidence are also investigated just over the SEP region. Recall that in the results for the entire tropical oceans shown in Figs. 6 and 7, the skill score is high if the precipitation is weakly correlated with SST and significantly correlated with ω500. The next question is to what extent the above sensitivities of deep convection explain the double ITCZ problem over the SEP region, where the mean SST is 25°C (HadISST) and the mean RH600 is 16% (JRA-25 and ERA-40). As shown in Fig. 9b, convective activity is higher in the LSMs than in the HSMs when the environmental conditions are about SST = 25°C and RH600 = 16%. The corresponding precipitation difference is about 0.89 mm day−1. This explains about half of the total precipitation difference (1.5 mm day−1) for LSM − HSM over the SEP region. The other half is probably due to the following difference in the environmental conditions. Although the SST in the LSMs is about 25°C, similar to the observation, RH600 is 33% and a very humid bias exists over the SEP region. As shown in Fig. 8a, the convective activity significantly depends on RH600. The difference in RH600 (16% vs 35% in the HSMs) accounts for 0.56 mm day−1 of precipitation. Since the actual convective activity and environmental conditions interact significantly, this is merely a rough estimate, and verification by sensitivity experiments using a climate model may clarify this point.
6. Annual mean and seasonal cycle
The results of the same analyses using an annual mean are almost the same as that using the SON mean. The precipitation skill scores for the annual mean and that for the SON mean have an intermodel correlation of 0.89; thus, a model with a higher SON score has a higher score for the annual mean as well.
The difference between the LSMs and HSMs in the relationship between precipitation and SST is best illustrated by the seasonal cycle at 150°W, as shown in Fig. 12. Note that the grouping of the LSMs and the HSMs is based on the precipitation skill score of the SON seasons for the consistency with the results in the previous sections. Although the seasonal SST cycle around 10°S is very similar in the LSMs and HSMs, precipitation is suppressed from July to September in the HSMs, whereas precipitation persists during this period in the LSMs, following the maximum SST. Furthermore, seasonal variations in precipitation in the LSMs are more dependent on the SST and less dependent on ω500 than in the HSMs (not shown). These results confirm the previous conclusion that in the LSMs, dynamical suppression is too weak and the precipitation distribution follows the SST distribution too closely.
Seasonal cycle of precipitation (shading; mm day−1) and SST (contours; °C) for the (a) HSM composite and (b) LSM composite at 150°W for 1979–2000.
Citation: Journal of Climate 24, 18; 10.1175/2011JCLI4156.1
In the February–April (FMA) seasons, the precipitation overestimate over the SEP region is identified not only in the LSMs but also in the HSMs. The reason for the overestimate in the FMA seems somewhat different from that in the SON. FMA is the season when the SPCZ in the real world extends farther eastward along the equator compared to other seasons. In climate models, the eastward extension is more exaggerated. Both in the HSMs and LSMs, the precipitation over the SEP region is associated with the large-scale ascending motion. In this case, the air–sea interactions suggested in previous studies (see section 1) may play a more important role in the double ITCZ problem. Note that even in the FMA seasons, the deep convection over large-scale subsidence regions, such as farther southeast of the SEP region, around 20°–10°S, 110°–80°W, is suppressed in the HSMs but not in the LSMs.
7. Summary and discussion
The precipitation reproducibility over the tropical oceans (30°S–30°N) of CMIP3 models and MIROC5 is examined. The skill score defined by Taylor (2001) is used to quantify the precipitation reproducibility of the models against TRMM observations. This score examines the distribution and amplitude of a spatial pattern. Precipitation distributions of five HSMs and LSMs determined by the skill scores are compared. Positive precipitation anomalies of LSM − HSM are located over large-scale subsidence regions, whereas negative anomalies are identified over ascending regions. An anomaly over the southeastern Pacific corresponds to the double ITCZ precipitation bias. The reproducibility of precipitation distributions over the entire tropics corresponds well to the location of the double ITCZ precipitation bias, suggesting that the double ITCZ is associated not only with the local SST but also with the precipitation schemes that control precipitation over the entire tropical ocean.
Next, the sensitivity of deep convection to SST and large-scale subsidence is investigated. The results indicate that precipitation in LSMs has a stronger correlation with SST and a weaker correlation with ω500 than that in HSMs. These intermodel relationships are shown to be significant over the entire tropical oceans as well as over the southeastern Pacific.
Regarding the interpretation of these statistically significant relationships, we emphasize the importance of dynamical suppression of deep convection owing to the entrainment of environmental dry air over subsidence regions. The vertical profiles of Q1 − QR stratified against RH600 show bimodal convection characteristics, with deep and shallow heating depending on the midtropospheric humidity. The deep and shallow modes are predominant where RH600 is larger and smaller than ~50%, respectively. Although a similar RH600 dependency of convective heating appears in both LSMs and HSMs, the dependency is significantly weaker in the former. In other words, dynamical suppression in the LSMs seems to be too weak, and deep convection follows the SST too closely. In particular, even when the mid to lower troposphere is very dry (RH600 < 10%), deep convection is still active in the LSMs compared to the HSMs for very high SST (SST > 27°C). A rough estimate suggests that the difference in the dynamical suppression of deep convection explains about half of the double ITCZ precipitation bias. We consider that this also provides a physical interpretation of the SST threshold problem discussed in Bellucci et al. (2010). They showed that the statistically obtained SST threshold for deep convection onset varies greatly among different CGCMs. We speculate that in the models with an unrealistic double ITCZ, dynamical suppression is not effective enough, so the SST thresholds for deep convection tend to be lower than those in the models without the double ITCZ.
The current version (MIROC5) of CGCM has significantly better precipitation reproducibility than its older versions, MIROC_M and MIROC_H. The modifications to the convective scheme and the boundary layer turbulence in MIROC5 are discussed in Chikira and Sugiyama (2010) and Nakanishi and Niino (2004), respectively. Chikira and Sugiyama (2010) used an entrainment rate that depends on the buoyancy of a convective parcel, whereas the entrainment rate was originally set vertically constant for each type of clouds in the Arakawa and Schubert (1974) scheme. As a result, the entrainment rate in the lower troposphere increases, resulting in stronger dynamical suppression of deep convection over dry subsidence regions and mitigation of the double ITCZ problem. It is noteworthy that models with convective schemes that consider the effects of environmental moisture also have relatively higher skill scores (Table 1; Fig. 2). INGV_ECHAM4, MIUB, and MPI_ECHAM5 used a scheme based on the Tiedtke (1989) cumulus parameterization scheme modified by Nordeng (1994) with a CAPE closure in which deep convection requires large-scale moisture convergence over the depth of the clouds. GFDL_0, GFDL_1, and MRI introduced a minimum entrainment rate of environmental air based on Tokioka et al. (1988). MIROC_M and MIROC_H introduced an empirical suppression condition in the cloud-layer relative humidity for convective activity (Emori et al. 2001). These models attain higher skill scores.
In this study, we emphasized that the conditions for the occurrence of deep convection significantly affect the existence of the double ITCZ bias among the CMIP3 models. On the other hand, many previous studies suggested that air–sea interaction is the fundamental factor for the north–south asymmetry in SST over the eastern Pacific (see section 1). Although we agree that air–sea interaction is a fundamental reason for the asymmetric SST, we think that it could not fully explain the difference in the double ITCZ precipitation among the CMIP3 models. In fact, a double ITCZ also appears in some AGCM experiments with a prescribed SST (Zhang et al. 2007; Chikira 2010).
This study is based on the output data from coupled models rather than atmospheric models because air–sea interaction is known to amplify the SST asymmetry associated with the double ITCZ precipitation. While data from the coupled models enable us to discuss realistic phenomena, including many feedback processes, separating the processes becomes rather difficult. To further clarify the contributions of each process, sensitivity experiments using a coupled model and an atmospheric model are necessary.
Acknowledgments
The authors appreciate two anonymous reviewers for their helpful comments and suggestions to improve the manuscript. This study was supported by the Global Environment Research Fund (S-5-2) of the Ministry of the Environment, Japan. The WCRP CMIP3 multimodel dataset is made available by the modeling groups, PCMDI, and the WCRP’s Working Group on Coupled Modelling (WGCM), and support of this dataset is provided by the Office of Science, U.S. Department of Energy. The authors also acknowledge the “Data Integration and Analysis System” Fund for the National Key Technology and the Innovative Program of Climate Change Projection for the 21st Century (“Kakushin” program) from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The dataset for the convective latent heating is produced under the support of the Japan Aerospace Exploration Agency, and the radiative heating rate is provided by Tristan S. L’Ecuyer. The Grid Analysis and Display System (GrADS) was used to plot the figures.
APPENDIX
Expansions of Model Names
BCCR Bjerknes Centre for Climate Research
CCCMA Canadian Centre for Climate Modelling and Analysis
CNRM-CM3 Centre National de Recherches Météorologiques Coupled Global Climate Model, version 3
CSIRO Commonwealth Scientific and Industrial Research Organisation
ERA ECMWF Re-Analysis
GFDL Geophysical Fluid Dynamics Laboratory
GISS_AOM Goddard Institute for Space Studies Atmosphere–Ocean Model
GISS_E_H Goddard Institute for Space Studies Model E-H
GISS_E_R Goddard Institute for Space Studies Model E-R
IAP Institute of Atmospheric Physics
INGV Istituto Nazionale di Geofisica e Vulcanologia
INMCM3 Institute of Numerical Mathematics Coupled Model, version 3.0
JRA-25 Japanese 25-yr Reanalysis
MIROC5 Model for Interdisciplinary Research on Climate, version 5
MIROC_H MIROC3.2 with a model resolution of about 1.1°
MIROC_M MIROC3.2 with a model resolution of about 2.8°
MIUB Meteorological Institute, University of Bonn
MPI_ECHAM5 Max Planck Institute ECHAM5
MRI Meteorological Research Institute
REFERENCES
Arakawa, A., and W. H. Schubert, 1974: Interaction of cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci., 31, 674–701.
Back, L. E., and C. S. Bretherton, 2009: On the relationship between SST gradients, boundary layer winds, and convergence over the tropical oceans. J. Climate, 22, 4182–4196.
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.
Betts, A. K., 1986: A new convective adjustment scheme. Part I. Observational and theoretical basis. Quart. J. Roy. Meteor. Soc., 112, 677–691.
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, 163–172.
Bougeault, P., 1985: A simple parameterization of the large-scale effects of cumulus convection. Mon. Wea. Rev., 113, 2108–2121.
Chang, P., and S. G. H. Philander, 1994: A coupled ocean–atmosphere instability of relevance to the seasonal cycle. J. Atmos. Sci., 51, 3627–3648.
Chikira, M., 2010: A cumulus parameterization with state-dependent entrainment rate. Part II: Impact on climatology in a general circulation model. J. Atmos. Sci., 67, 2194–2211.
Chikira, M., and M. Sugiyama, 2010: A cumulus parameterization with state-dependent entrainment rate. Part I: Description and sensitivity to temperature and humidity profiles. J. Atmos. Sci., 67, 2171–2193.
Del Genio, A. D., and M.-S. Yao, 1993: Efficient cumulus parameterization for long-term climate studies: The GISS scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 181–184.
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.
de Szoeke, S. P., Y. Wang, S.-P. Xie, and T. Miyama, 2006: Effect of shallow convection on the eastern Pacific climate in a coupled model. Geophys. Res. Lett., 33, L17713, doi:10.1029/2006GL026715.
Emori, S., T. Nozawa, A. Numaguti, and I. Uno, 2001: Importance of cumulus parameterization for precipitation simulation over East Asia in June. J. Meteor. Soc. Japan, 79, 939–947.
Gadgil, S., P. V. Joseph, and N. V. Joshi, 1984: Ocean–atmosphere coupling over monsoon regions. Nature, 312, 141–143.
Gregory, D., and P. R. Rowntree, 1990: A mass flux convection scheme with representation of cloud ensembles characteristics and stability-dependent closure. Mon. Wea. Rev., 118, 1483–1506.
Gregory, D., J.-J. Morcrette, C. Jakob, A. C. M. Beljaars, and T. Stockdale, 2000: Revision of convection, radiation and cloud schemes in the ECMWF Integrated Forecasting System. Quart. J. Roy. Meteor. Soc., 126, 1685–1710.
Jensen, M. P., and A. D. Del Genio, 2006: Factors limiting convective cloud-top height at the ARM Nauru Island climate research facility. J. Climate, 19, 2105–2117.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587–1606.
Lau, N.-C., H.-T. Wu, and S. Bony, 1997: The role of large-scale atmospheric circulation in the relationship between tropical convection and sea surface temperature. J. Climate, 10, 381–392.
L’Ecuyer, T., and G. L. Stephens, 2003: The tropical oceanic energy budget from the TRMM perspective. Part I: Algorithm and uncertainties. J. Climate, 16, 1967–1985.
L’Ecuyer, T., and G. L. Stephens, 2007: The tropical atmospheric energy budget from the TRMM perspective. Part II: Evaluating GCM representations of the sensitivity of regional energy and water cycles to the 1998–99 ENSO cycle. J. Climate, 20, 4548–4571.
Manabe, S., D. G. Hahn, and J. L. Holloway, 1974: The seasonal variation of the tropical circulation as simulated by a global model of the atmosphere. J. Atmos. Sci., 31, 43–83.
Mechoso, C. R., and Coauthors, 1995: The seasonal cycle over tropical Pacific in coupled ocean–atmosphere general circulation models. Mon. Wea. Rev., 123, 2825–2838.
Moorthi, S., and M. J. Suarez, 1992: Relaxed Arakawa-Schubert: A parameterization of moist convection for general circulation models. Mon. Wea. Rev., 120, 978–1002.
Nakanishi, M., and H. Niino, 2004: An improved Mellor–Yamada level-3 model with condensation physics: Its design and verification. Bound.-Layer Meteor., 112, 1–31.
Nordeng, T. E., 1994: Extended versions of the convective parameterization scheme at ECMWF and their impact on the mean and transient activity of the model in the tropics. ECMWF Tech. Memo. 206, 41 pp.
Onogi, K., and Coauthors, 2007: The JRA-25 reanalysis. J. Meteor. Soc. Japan, 85, 369–432.
Pan, D.-M., and D. A. Randall, 1998: A cumulus parameterization with a prognostic closure. Quart. J. Roy. Meteor. Soc., 124, 949–981.
Philander, S. G. H., D. Gu, D. Halpern, G. Lambert, N.-C. Lau, T. Li, and R. C. Pacanowski, 1996: Why the ITCZ is mostly north of the equator. J. Climate, 9, 2958–2972.
Russell, G. L., J. R. Miller, and D. Rind, 1995: A coupled atmosphere–ocean model for transient climate change studies. Atmos.–Ocean, 33, 683–730.
Sekiguchi, M., and T. Nakajima, 2008: A k-distribution-based radiation code and its computational optimization for an atmospheric general circulation model. J. Quant. Spectrosc. Radiat. Transfer, 109, 2779–2793.
Sherwood, S. C., 1999: Convective precursors and predictability in the tropical western Pacific. Mon. Wea. Rev., 127, 2977–2991.
Shige, S., Y. N. Takayabu, W.-K. Tao, and D. E. Johnson, 2004: Spectral retrieval of latent-heating profiles from TRMM PR data. Part I: Development of a model-based algorithm. J. Appl. Meteor., 43, 1095–1113.
Shige, S., Y. N. Takayabu, W.-K. Tao, and C.-L. Shie, 2007: Spectral retrieval of latent heating profiles from TRMM PR data. Part II: Algorithm improvement and heating estimates over tropical ocean regions. J. Appl. Meteor. Climatol., 46, 1098–1124.
Shige, S., Y. N. Takayabu, and W.-K. Tao, 2008: Spectral retrieval of latent heating profiles from TRMM PR data. Part III: Estimating apparent moisture sink profiles over tropical oceans. J. Appl. Meteor. Climatol., 47, 620–640.
Shige, S., Y. N. Takayabu, S. Kida, W.-K. Tao, X. Zeng, C. Yokoyama, and T. L’Ecuyer, 2009: Spectral retrieval of latent heating profiles from TRMM PR data. Part IV: Comparisons of lookup tables from two- and three-dimensional cloud-resolving model simulations. J. Climate, 22, 5577–5594.
Song, X., and G. Zhang, 2009a: 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.
Song, X., and G. Zhang, 2009b: Coupling between sea surface temperature and low-level winds in mesoscale numerical models. J. Climate, 22, 146–164.
Takayabu, Y. N., S. Shige, W.-K. Tao, and N. Hirota, 2010: Shallow and deep latent heating modes over tropical oceans observed with TRMM PR spectral latent heating data. J. Climate, 23, 2030–2046.
Tao, W.-K., and J. Simpson, 1993: Goddard Cumulus Ensemble model. Part I: Model description. Terr. Atmos. Oceanic Sci., 4, 35–72.
Tao, W.-K., and Coauthors, 2003: Microphysics, radiation and surface processes in the Goddard Cumulus Ensemble (GCE) model. Meteor. Atmos. Phys., 82, 97–137.
Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106 (D7), 7183–7192.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800.
Tokioka, T., K. Yamazaki, A. Kitoh, and T. Ose, 1988: The equatorial 3060-day oscillation and the Arakawa-Schubert penetrative cumulus parameterization. J. Meteor. Soc. Japan, 66, 883–901.
Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 2961–3012.
Waliser, D. E., and N. E. Graham, 1993: Convective cloud systems and warm-pool surface temperatures: Coupled interactions and self-regulation. J. Geophys. Res., 98, 12 881–12 893.
Wang, Y., S.-P. Xie, B. Wang, and H. Xu, 2005: Large-scale atmospheric forcing by southeast Pacific boundary layer clouds: A regional model study. J. Climate, 18, 934–951.
Watanabe, M., S. Emori, M. Satoh, and H. Miura, 2009: A PDF-based hybrid prognostic cloud scheme for general circulation models. Climate Dyn., 33, 795–816.
Watanabe, M., and Coauthors, 2010: Improved climate simulation by MIROC5: Mean states, variability, and climate sensitivity. J. Climate, 23, 6312–6335.
Watanabe, M., M. Chikira, Y. Imada, and M. Kimoto, 2011: Convective control of ENSO simulated in MIROC. J. Climate, 24, 543–562.
Xie, S., T. Hume, C. Jakob, S. A. Klein, R. B. McCoy, and M. Zhang, 2010: Observed large-scale structures and diabatic heating and drying profiles during TWP-ICE. J. Climate, 23, 57–79.
Xie, S.-P., 2004: The shape of continents, air–sea interaction, and the rising branch of the Hadley circulation. The Hadley Circulation: Past, Present, and Future, Kluwer Academic, 121–152.
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.
Xie, S.-P., and Coauthors, 2007: A regional ocean–atmosphere model for eastern Pacific climate: Towards reducing tropical biases. J. Climate, 20, 1504–1522.
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.
Yanai, M., S. Esbensen, and J.-H. Chu, 1973: Determination of bulk properties of tropical cloud clusters from large-scale heat and moisture budgets. J. Atmos. Sci., 30, 611–627.
Yuan, J., D. L. Hartmann, and R. Wood, 2008: Dynamic effects on the tropical cloud radiative forcing and radiation budget. J. Climate, 21, 2337–2351.
Zhang, G. J., 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, 3, 407–446.
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.