1. Introduction
Although the double intertropical convergence zone (ITCZ) is an observed phenomenon in the tropical central-eastern Pacific during boreal spring (Hubert 1969; Waliser and Gautier 1993; Zhang 2001; Halpern and Hung 2001; Liu and Xie 2002), it is often overemphasized in general circulation models (GCMs) and is a common bias that has plagued GCMs for a few decades. Mechoso et al. (1995) categorized the double ITCZ bias into two types: a persistent double ITCZ feature throughout the year and a seasonally alternating ITCZ between two hemispheres. Most of the state-of-the-art GCMs, including models participating in phase 3 of the Coupled Model Intercomparison Project (CMIP3) and the Atmospheric Model Intercomparison Project (AMIP), suffer from one of the two types of biases (Lin 2007; de Szoeke and Xie 2008).
In GCMs, the double ITCZ bias is believed to be associated with local air–sea interactions. For example, Lin (2007) suggested that overly strong trade winds in the tropical northeastern Pacific may cause overestimated latent heat flux (LH), leading to a cold sea surface temperature anomaly (SSTA). This SSTA is then enhanced by the so-called Bjerknes feedback (Bjerknes 1969; Dijkstra and Neelin 1995) and creates the cold tongue bias. Over the southeastern Pacific, an insufficient amount of cloudiness allows more downward shortwave flux to warm the local SST, which reduces the static stability in the boundary layer and further decreases the amount of low clouds (Klein and Hartmann 1993). A warmer SST and a less stable lower atmosphere lead to unrealistic convection. The double ITCZ structure is then enhanced by these positive feedbacks in GCMs. However, the causes of the overly strong trade winds and insufficient cloudiness are not discussed in previous studies.
Recently, some studies suggested that the source of the double ITCZ bias is in the atmospheric component, particularly the deep convection scheme (Song and Zhang 2009; Hirota et al. 2011). Song and Zhang (2009) provided direct evidence that the double ITCZ bias is significantly reduced when the large-scale generation of convective available potential energy (CAPE) is used in the closure of the Zhang–McFarlane (ZM) scheme (Zhang and McFarlane 1995). Hirota et al. (2011) analyzed the tropical precipitation pattern, SST, and large-scale subsidence in the outputs of 19 CMIP3 models and the fifth version of the Model for Interdisciplinary Research on Climate (MIROC5). They found that if the dynamic suppression due to entrainment of environmental dry air is not strong enough over subsidence areas, the spatial distribution of deep convection will follow the SST field too closely. As a result, once the local SST warms up, the models tend to produce overly strong convection. Both studies pointed out that the main direction of improving the convection scheme is to decrease the dependence of precipitation on SST and to increase the influence of large-scale subsidence on convection.
The uncertainty of precipitation simulation in the Community Atmosphere Model version 5 (CAM5) using the ZM scheme with prescribed SST was assessed by Yang et al. (2013). By tuning nine parameters directly linked to subgrid vertical motions, CAPE, and cloud–rain conversion, they found that model precipitation is very sensitive to the CAPE consuming rate and the parcel fractional mass entrainment rate over oceans. The sensitivity and optimal values of parameters were identified through a stochastic approach. Their findings about the importance of the CAPE consuming rate and entrainment rate are in agreement with Song and Zhang (2009) and Hirota et al. (2011). Similar to Song and Zhang (2009), the ZM scheme with optimal parameters in Yang et al. (2013) reduces the double ITCZ bias but at the same time overestimates precipitation over the tropical western Pacific.
The double ITCZ bias has been an outstanding issue for the Community Earth System Model version 1 with CAM5 (CESM1/CAM5). Reasons for this bias are not yet fully understood. In this study, we revisit the issue by analyzing the double ITCZ bias in a set of simulations using CESM1/CAM5. Questions we want to answer include these: 1) What is the difference between the northern ITCZ and southern ITCZ in terms of the formation mechanism in CAM5? 2) How is this formation difference associated with model biases? 3) Through what process(es) are small model biases in AGCMs and OGCMs enhanced in coupled models? A brief description of CESM1, the set of simulations, and analysis data for comparison is given in section 2. The simulation of ITCZ and the associated dynamic and thermal fields are presented in section 3. Section 4 contains discussion of possible processes that enhance the biases and section 5 is concluding remarks.
2. Model, data, and method
a. CESM1 and experiments
CESM1 is a general circulation model composed of atmosphere (CAM5), land [Community Land Model (CLM)], ocean [Parallel Ocean Program version 2 (POP2)], and sea ice [Community Ice Code (CICE)] components. The difference between CESM1 and its previous version, the Community Climate System Model version 4 (CCSM4; Gent et al. 2011), is the substantially modified atmosphere component. CAM5 in CESM1 uses the same finite-volume dynamic core (Lin 2004) and deep convection scheme (Zhang and McFarlane 1995; Neale et al. 2008) as CAM4 in CCSM4, but with updated physical processes including shallow convection (Park and Bretherton 2009), moist boundary layer (Bretherton and Park 2009), cloud microphysics (Morrison and Gettelman 2008), radiative transfer (Iacono et al. 2008), and aerosol scheme involving the indirect effect (Liu et al. 2012).
In this study, we performed a 150-yr preindustrial simulation using CESM1/CAM5 with all four components active at a horizontal resolution of 0.9° × 1.25° for the atmosphere and approximately 1° for the ocean. Following the method introduced by Subramanian et al. (2011), we selected a 30-yr period with the variance of Niño-3.4 index closest to the variance in the whole 150 yr to represent a general climate state in the tropics. In addition, a 12-yr CAM5 simulation with prescribed SST of year 1850 and a 40-yr POP2 simulation with prescribed atmospheric conditions of year 1850 (denoted as “CAM5 standalone” and “POP2 standalone,” respectively) were also performed to investigate the effect of air–sea interaction on double ITCZ. Both simulations are initialized in “warm start” mode, meaning using the restart files provided by the National Center for Atmospheric Research (NCAR). It appears that the CAM5 standalone and POP2 standalone simulations reach steady states around the third and the 25th year, respectively. Only the last 10 yr of both simulations are used for analysis. A 150-yr preindustrial simulation using CCSM4 is also carried out for the comparison of impacts from different schemes. We compared CAM5 standalone simulations with preindustrial and present-day conditions, and found that the differences in near-surface wind, precipitation, and surface fluxes are insignificantly small.
b. Observation and reanalysis data
The SST dataset is the Extended Reconstructed Sea Surface Temperature v3b (ERSST; Smith et al. 2008) from the National Oceanic and Atmospheric Administration (NOAA), which includes the available ship measurements and buoy data, but satellite data are not included due to residual biases. The spatial resolution is 2° × 2° globally and the temporal coverage is from 1854 to 2009. The precipitation data are from the Global Precipitation Climatology Project (GPCP; Adler et al. 2003; Huffman et al. 2009) at a 2.5° × 2.5° spatial resolution globally from 1979 to 2010. The data of ocean current and potential temperature are from the National Centers for Environmental Prediction (NCEP) Global Ocean Data Assimilation System (GODAS; Behringer and Xue 2004) with a 0.33° × 1.0° latitude/longitude global grid size from January 1980 to June 2013.
Moisture, wind, and surface heat fluxes are from the ERA-Interim dataset (Dee et al. 2011) at a 0.75° × 0.75° spatial resolution and temporal coverage from 1979 to 2009. The NCEP Climate Forecast System Reanalysis (CFSR; Saha et al. 2010) at a horizontal resolution of 0.5° × 0.5° and with a temporal coverage from 1979 to 2009 is also used for comparison. All available data were used in analysis.
c. Skill score
d. The apparent heat source
3. Current state of ITCZ simulation in CESM1
a. Mean state and seasonal cycle
The climatological annual means of SST and precipitation in CESM1 and their differences from observation are shown in Fig. 1. The observation fields (Fig. 1a) are the GPCP precipitation rate (color shading), ERSST (contour), and the contour of zero pressure velocity at 500 hPa (ω500) from ERA-Interim (red contour), which outlines the area of rising motion. In the CAM5 standalone simulation (Fig. 1b), the overall precipitation rate in the tropics is overestimated, especially over the South Pacific convergence zone (SPCZ) and the Indian Ocean. This precipitation bias has been noted in most of the GCM simulations, suggesting that the precipitation rate is too sensitive to the surface thermal condition (Hirota et al. 2011). The northern ITCZ over the central to eastern Pacific is significantly overestimated, which is slightly improved after coupling (Fig. 1c). The large-scale pattern of the precipitation rate in the CESM1/CAM5 simulation (Fig. 1c) shows some well-known biases. For example, the SPCZ is more zonally oriented than that of the observation (Fig. 1a) and extends to the southeastern Pacific, forming the so-called double ITCZ structure. Note that a weaker double ITCZ bias also exists in the CAM5 standalone simulation but is enhanced in the coupled simulation. The cold tongue in SST is stronger than observation and extends farther westward because the simulated tropical SST in CESM1/CAM5 exhibits an overall cold bias (Fig. 1d) as already noted in CCSM4 (Subramanian et al. 2011). To the south of the cold bias, there are warm SST biases over the southeastern Pacific and Atlantic Ocean, which may be due to a small bias in surface heat fluxes and will be discussed in section 4.
Figure 2 shows the dependence of the precipitation rate and pressure velocity at 850 hPa (ω850) and at 500 hPa (ω500) on the SST, respectively, over the entire tropical ocean (30°S–30°N). Over most of the tropical oceans, when SST exceeds 26°C, the GPCP precipitation rate (Fig. 2a) is proportional to the SST, implying that it is closely related to the surface thermal condition. The GPCP precipitation rate exceeds 6 mm day−1 at 29°C and drops sharply when the SST is warmer than 29.5°C, meaning that the SST increases due to strong solar heating under clear-sky conditions over these regions. However, for the CAM5 standalone and CESM1/CAM5 simulations, the precipitation rate starts increasing when SST is over 25°C and reaches 6 mm day−1 at 28° and 27.5°C, respectively. Then, the precipitation rate saturates around 28.5°C for the CAM5 standalone and 29°C for CESM1/CAM5. The vertical motion shown in Figs. 2b and 2c indicates that a relatively shallower convection (ω850 is negative, and ω500 is positive) is developed in the CAM5 standalone simulation when the SST is around 27°C, and the simulated precipitation rate is around 4 mm day−1 (Fig. 2a). Once the SST exceeds 27.5°C, convection develops into deep convection (ω500 is negative in Fig. 2c), and the precipitation rate increases to 6 mm day−1. The threshold values for shallow and deep convection in CAM5 are both about 0.5°C lower than those in the observation. This result suggests that convection starts and develops into deep convection more easily in the model.
In the real world, the double ITCZ phenomenon appears in boreal spring (February–May) over the eastern Pacific (90°–120°W) as shown in Fig. 3a with a slightly stronger precipitation rate in its northern branch. Both the northern and southern precipitation bands are overestimated in the CAM5 standalone simulation (Fig. 3b) and the southern precipitation band persists for nearly half a year (January–June). After air–sea coupling, the northern precipitation band in boreal spring diminishes and the southern precipitation band enhances significantly (Fig. 3c), forming the so-called alternating ITCZ type of error (Mechoso et al. 1995).
The monthly values of precipitation skill score S for the CAM5 standalone and CESM1/CAM5 simulations with respect to GPCP over the entire tropical ocean (30°S–30°N) were computed and are shown in Fig. 4. It shows that S drops sharply during boreal spring in the CESM1/CAM5 simulation while it remains stable in the CAM5 standalone simulation (bottom panel). Note that S is significantly dominated by the spatial pattern of precipitation (top panel) rather than the ratio of spatial standard deviations (SDR) between the model and observation (middle panel). The SST skill score (not shown) shows no seasonality, but the ω500 skill score (not shown) has a sudden dip in boreal spring similar to the seasonal fluctuation in the precipitation skill score, implying that the precipitation bias is more directly related to biases in dynamical processes rather than in the surface thermal condition.
b. Meridional SST and sea level pressure gradients
The importance of SST gradient for the formation of tropical convergence zone was first proposed by Lindzen and Nigam (1987) and investigated in more detail by Tomas and Webster (1997). The Lindzen and Nigam mechanism suggests that surface wind is driven by the surface pressure field, which follows the underlying SST closely. As demonstrated by Tomas and Webster (1997), tropical convergence over some areas, such as the Indian Ocean, eastern Pacific, eastern Atlantic, and Africa, tends to be located away from the equator. These areas are characterized by large sea level pressure (SLP) gradient and the contour of zero absolute vorticity (ζa = 0) at the sea level is off the equator. This condition creates an area that is inertially unstable between the zero absolute vorticity line and the equator. The flow in this area is divergent due to inertial instability and creates a convergence-favorable region at the poleward side of ζa = 0. For the areas with weak meridional pressure gradient, the location of convergence is determined by thermal conditions, such as the SST, and the instability of the atmosphere.
Figures 5a and 5b show the monthly values of meridional SST gradient and SLP gradient, averaged over the equator to 10° latitude and 90°–120°W for the (top) Northern Hemisphere (NH) and (bottom) Southern Hemisphere (SH), respectively. The observed SST gradient in the NH is small during boreal spring and reaches its maximum in late summer while its SH counterpart is much smaller. While the CESM1/CAM5 simulates a realistic NH SST gradient seasonal cycle, it poorly simulates the seasonal cycle in the SH with a wrong phase and unrealistically large amplitude. As will be shown later, this bias is closely related to the double ITCZ bias.
The SLP gradient in the NH shows a seasonal evolution that varies in opposite phase compared to the variation in SST gradient, indicating that the two fields are closely related and consistent with the Lindzen and Nigam’s mechanism. The SLP gradients of ERA-Interim and CFSR are included for comparison. It shows that the simulation of SLP in CAM5 is good in the NH but the gradient is oversimulated in the SH. The difference between ERA-Interim and CFSR in the SH points out that the southeastern Pacific is particularly difficult to deal with. In spite of the difference between CAM5 standalone and the reanalysis data, the discussion below remains valid. In the SH, both SST and SLP gradients are negative. During boreal summer, the SLP gradients in both hemispheres are negative, creating a northward force. This force moves air parcels with negative absolute vorticity from the SH to the NH, shifting the ζa = 0 contour to the north of the equator. Therefore, the convergence zone is located at the northern edge of the inertially unstable area, which is between the ζa = 0 contour and the equator. From January to April, the SLP gradient force becomes southward in the NH but remains northward in the SH, with the latter slightly stronger. Because the net northward SLP gradient force is the smallest during boreal spring, the location of the ζa = 0 contour would be very sensitive to model biases and so would the convergence zone. In other words, the model biases always exist but have no significant impact during boreal summer because of the strong net SLP gradient force. When the SLP gradient force weakens, the influence of model biases gradually increases and eventually produces the double ITCZ problem.
c. Meridional circulation and relative humidity
Takayabu et al. (2010) have pointed out that the entrainment of middle to low tropospheric dry air is important to suppress convection in the subsidence area. Zhang et al. (2004) and Nolan et al. (2007) both suggested that the local meridional circulation may redistribute moisture in the lower to middle troposphere. The meridional circulation and the vertical distribution of relative humidity (RH) over the eastern Pacific are diagnosed to understand the moisture distribution and transport in the CAM5 standalone and CESM1/CAM5 simulations. Two reanalysis datasets are utilized for comparison. April is selected to show as an example (Fig. 6) since the double ITCZ bias is most significant in boreal spring.
The circulation in the region between 5°N and 15°S is characterized by a shallow meridional circulation (Zhang et al. 2004, 2008), which includes a southward surface return flow (SRF) above the top of the planetary boundary layer (PBL) and a northward boundary layer inflow (BLI) driven by the surface temperature gradient (Zhang et al. 2008; Nolan et al. 2010). The moisture detrained from the top of shallow cumuli can be carried by the SRF and influences the moisture distribution over the southeastern Pacific. The BLI in both reanalysis datasets exhibits a major convergence around 5°N and a secondary convergence around 3°S (Figs. 6a,b). The BLI slows down more significantly south of the equator in ERA-Interim (Fig. 6b) than in the CFSR (Fig. 6a).
The CFSR RH field has a local maximum around 750 hPa near 5°N accompanied with a clear SRF around 700 hPa, while the maximal RH in ERA-Interim is located much closer to the surface at 5°N with weaker SRF. The differences in the meridional circulation appear to be partly responsible for the difference in the vertical distribution of RH. The consensuses between the two datasets are the two shallow meridional counterclockwise circulations at 10°–5°S and 0°–5°N wrapped in a cross-equatorial shallow meridional circulation and thick moist air in the lower troposphere.
The situation in the CAM5 standalone simulation (Fig. 6c) is very different from both reanalysis datasets. The BLI converges around 4°S, rather than blowing across the equator, causing too much surface moisture convergence and upward transport south of the equator. A rather deep return flow is seen around 750 to 400 hPa in the SH (Fig. 6c). However, this flow originates not from the northern ITCZ, but from the southern branch. The RH pattern clearly shows that the strong surface convergence results in too much moisture transport up to the middle troposphere in the SH, forming two peaks in the RH field with almost equal strength straddling the equator. This surface wind and vertical circulation bias, which can be identified starting from January to April (not shown), is exacerbated greatly in the coupled simulation (Fig. 6d), switching the major precipitation band to the SH with weak upward motion in the NH.
d. Surface wind anomaly and the influence on latent heat flux
The surface wind anomaly presented in the previous session has critical impacts on the surface LH. Figure 7 shows the LH and areas with a precipitation rate higher than 6 mm day−1 for the standalone and coupled model simulations. It is found that the areas with a high precipitation rate are coincident with the areas of large LH gradient. Since the LH is mostly controlled by the surface wind speed, the change in LH indicates that the corresponding surface wind slows down at the precipitation band.
Figure 8 shows the near-surface wind (975 hPa, vector) and wind speed (shading) from ERA-Interim (Fig. 8a), the CAM5 standalone simulation (Fig. 8b), and the differences between these two (Fig. 8c) over the tropical eastern Pacific (90°–150°W). The equatorial easterly between 90° and 130°W in CAM5 standalone (Fig. 8b) has an erroneous southward component in the SH (Fig. 8c), and the trade wind over the equatorial eastern Pacific and most of the south eastern Pacific is stronger (positive values in Fig. 8c) than in ERA-Interim. Two narrow bands of negative anomaly (5°N and 4°S) indicate weaker winds at where the double ITCZ structure is located in CAM5 standalone. The anomalous deceleration near 4°S reflects the surface convergence seen in Fig. 6c. Weaker winds may cause weaker LH and increase local SST. Since the SST is prescribed in the CAM5 standalone simulation, the wind bias presented in Fig. 8c is not caused by the SST, but from errors in the atmosphere component. As will be shown later, this bias is exacerbated through the air–sea interaction and becomes much more prominent in the coupled model.
4. Discussion
A few studies have documented that the tropical precipitation over the southeastern Pacific is overestimated in most of the GCMs (Hirota et al. 2011; Hirota and Takayabu 2013) and suggested several possible reasons, such as the high humidity in the middle of troposphere, the bias of entrainment process in the convection scheme, or weak subsidence from large-scale dynamics. Some of these issues exist in the CESM1/CAM5 simulation as well. For example, the midlevel RH is overestimated and the large-scale subsidence is too weak in the SH. However, through what processes these errors may produce a systematic precipitation error has not been investigated in details. Here, some further analyses are discussed and a possible mechanism that may explain the double ITCZ bias in CESM1/CAM5 is suggested.
a. Inertial instability and ITCZ formation
The cross-equator flow, which is induced by the tropical meridional SLP gradient, advects air parcels from one hemisphere to the other. The resulting inertial instability can create convergence off the equator (Tomas and Webster 1997). This mechanism can be diagnosed by the location of the ζa = 0 contour (red line in Fig. 9). In the CAM5 standalone simulation and in observation, the location of the ζa = 0 contour over the eastern Pacific is mostly north of the equator during boreal summer and early fall (Fig. 9a), corresponding to the northward SLP gradient force (Fig. 5b). The precipitation band (contour in Fig. 9a) in the NH is located right on the poleward side of the ζa = 0 contour, indicating that the northern ITCZ forms due to inertial instability. In the central to western tropical Pacific, the contour of ζa = 0 is almost on the equator. The precipitation bands straddle on both sides of the equator, implying that the convergence is associated with the warmer SST (shading in Fig. 9a), but not inertial instability.
In April (Fig. 9b), the northern ITCZ over the eastern Pacific is located right on the poleward side of the ζa = 0 contour while the southern ITCZ is located over a relatively warmer SST band around 90°–150°W. As shown in Fig. 5b, during boreal spring, the net northward SLP gradient force is only slightly greater than the southward force. Therefore, the location of the northern convergence zone can be affected easily by small model biases. After coupling (Fig. 9c), the eastern part of the northern ITCZ disappears and the southern ITCZ enhances, representing the alternative ITCZ characteristics. The moving of ζa = 0 contour to the SH indicates that the enhancement of the southern ITCZ is partly contributed by inertial instability. On the other hand, the northern ITCZ becomes thermally forced and eventually disappears east of 120°W due to cold SST.
b. Vertical heating modes
Because the meridional circulation presented in section 3c is tied closely to the vertical heating profile, the daily diabatic heating profile [i.e., the apparent heat Q1 defined by Yanai et al. (1973)] over the tropical oceans is analyzed using rotated empirical orthogonal functions (REOF). The most intriguing result is that there are three dominant modes in the CAM5 standalone and CESM1/CAM5 simulations. Figure 10 shows the results of REOF analysis, which was applied to the daily heating profiles at each grid point in the southeastern Pacific region (0°–10°S, 90°–120°W) in boreal spring (March–May). ERA-Interim clearly shows a unique structure in the eastern Pacific that has a major heating mode with a maximum around 850 hPa (black in Fig. 10a) and a secondary mode with a maximum around 500 hPa (blue in Fig. 10a). The two modes explain about 75% variance together. A secondary shallow mode (REOF3; 8.8%) is drawn for later comparison. CFSR shows similar structure to that in ERA-Interim and therefore it is not shown here. CCSM4 (Fig. 10b) does not have a sharp shallow heating mode as does REOF1 in ERA-Interim, but it simulates the bimodal heating structure documented by Zhang and Hagos (2009) and Hagos et al. (2010) using data over the entire tropical ocean. The difference between CCSM4 and CESM1/CAM5 shows the effects of using new shallow convection scheme and boundary layer scheme in CAM5.
Both the CAM5 standalone and the CESM1/CAM5 simulations have a very different heating structure from the CCSM4. The CAM5 standalone simulation (Fig. 10c) did produce a sharp shallow mode with the maximum around 850 hPa (REOF3; 15.3%) that is similar to REOF3 in ERA-Interim in terms of the vertical structure. The bottom heavy heating mode (REOF1 48.7%), which was not found in observation, has developed too deep in the vertical and explains too much variance. The middle-heavy deep heating mode (REOF2; 22.7%) corresponds to REOF2 in ERA-Interim. After air–sea coupling, the structure of these three modes is adjusted to be closer to those of ERA-Interim, having two shallow modes with the maxima at 850 and 800 hPa, respectively, and a deep mode. However, the variance explained by the deep mode is too small and one by the shallow mode is too large. Although the vertical heating structure in CESM1/CAM5 is improved, the double ITCZ bias is worse than that in CCSM4. The double ITCZ indices (the average precipitation rate in the area 0°–20°S, 100°–150°W) for CESM1/CAM5, CCSM4, and GPCP are 3.3, 2.9, and 1.1, respectively.
It is known that the model boundary layer tends to be too humid (Park and Bretherton 2009), and this bias is seen in this study as well. Results shown above also indicate that the low-to-middle troposphere is too humid in both CAM5 and CESM1/CAM5. It is possible that the extra heating mode enhances the shallow circulation, which then transports too much moisture upward from the PBL and creates an overly humid middle troposphere (Fig. 6c). This overly humid environment may provide excessive moisture for precipitation and weakens the large-scale subsidence. The convection in the lower troposphere may also influence surface wind and contribute to the wind bias over the southeastern Pacific. It is still not clear how the extra heating mode is produced in CAM5 and CESM1/CAM5, and it will require further in-depth numerical investigation.
c. Effects of air–sea coupling
So far, we have demonstrated that the bias associated with double ITCZ exists in the near-surface wind field and the alternating ITCZ error is exacerbated after the air–sea coupling. Figure 11 shows the zonal mean SST in April over the eastern Pacific (90°–120°W) for CESM1/CAM5 (solid) and CAM5 standalone (dashed; i.e., observed SST) simulations. The SST simulated by CESM1/CAM5 in the NH dramatically drops while it increases in the SH, which would change the SST gradient and consequently the SLP gradient. The next question is through what processes the small surface wind bias is enhanced by air–sea coupling, especially in April.
The surface wind anomaly could influence SST thermally through evaporation and dynamically through changing the ocean surface currents and vertical mixing. As shown in section 3d, this wind bias may cause weaker LH over the narrow bands at 5°N and 4°S, leaving more energy stored in the ocean and increasing the local SST. The situation is opposite outside these two narrow bands and results in a stronger and narrower cold tongue structure.
Over the tropical eastern Pacific (90°–120°W), the wind-driven ocean circulation has a complex structure and unique dynamic processes that have significant influences on SST. The upper ocean conditions in April are shown in Fig. 12. In Fig. 12a, the observed North Equatorial Current (NEC; negative contour) is located around 10°N. The positive contour south of the NEC is the north equatorial countercurrent (NECC; around 5°N). The South Equatorial Current (SEC) between 2°N and 10°S is much wider than NEC and NECC. The strong eastward current under SEC is the Equatorial Undercurrent (EUC). In the POP2 standalone simulation (Fig. 12c), it can be found that the SEC is much narrower and disappears near the equator. Both the NEC and NECC move equatorward and the equatorial upwelling, represented by the ocean temperature (shading in Fig. 12), is also weaker. It reveals that POP2 cannot well simulate the complex current structure in the eastern Pacific.
When CAM5 is coupled with POP2, the combination of biases from the atmosphere and ocean further deteriorates the model performance in simulating ocean currents. In the coupled simulation (Fig. 12b), the ocean current around 5°S erroneously becomes eastward because of the too weak surface easterly wind. Consequently, the LH reduction at the surface and eastward warm ocean advection result in the warm SST bias in the SH. Recall that once the local SST exceeds 27.5°C, deep convection develops and the surface wind speed is further decreased at this new convergence zone, causing a positive feedback through air–sea coupling. At the same time, because the NECC is erroneously located closer to the equator, ocean temperature at 5°–10°N significantly decreases due to less warm advection. The colder SST reduces convection in the NH. Eventually, convection in the SH is stronger than that in the NH, leading to a southward cross-equator wind (Fig. 6d), and the ζa = 0 contour is moved to a new balanced location in the SH. Now, both thermal and dynamical conditions favor convection in the SH, causing the alternating ITCZ bias.
5. Concluding remarks
This study analyzes the double ITCZ phenomenon in the eastern Pacific to diagnose the performance of CESM1/CAM5 in the tropics. In the prescribed SST simulation with CAM5, the precipitation rate of the southern branch of ITCZ is erroneously stronger than the northern branch in boreal spring. In the coupled simulation, the southern branch becomes even stronger while the northern branch almost disappears.
The vertical distribution of RH, the meridional circulation, and the heating profile over the tropical eastern Pacific in boreal spring are analyzed to investigate the possible causes of the convection bias associated with the double ITCZ. The simulated BLI slows down too quickly over the south edge of the cold tongue, and the return flow is too deep in the vertical. The vertical distribution of RH indicates that the PBL and the lower troposphere are too humid, probably due to the moisture transported by shallow convection that is too strong and too thick. The ROEF analysis for vertical diabatic heating profile reveals that shallow convection is too deep, too thick, and too strong in CAM5. The excessive shallow convection may influence the BLI, SRF, and surface wind field which then cause further moisture bias in the PBL and influence surface heat fluxes and SST. The overly humid mid-to-lower troposphere is consistent with an environment that is favorable for the development of deep convection in the southeastern Pacific.
In the prescribed-SST CAM5 simulation and in observation, the southern branch of the double ITCZ is formed due to surface thermal condition (i.e., warmer SST) while the northern branch is due to dynamical process (i.e., inertial instability). After coupled with POP2, the double ITCZ bias further deteriorates. The initial ocean temperature bias over the southeastern Pacific in POP2 is produced by weaker and narrower SEC. The wind bias in CAM5 provides additional error to amplify the SST bias through the wind–evaporation–SST feedback. At the same time, the SST reduction in the NH results from less warm advection due to the wrong location of the NECC. In the end, the amplification of the SST and SST gradient biases over the eastern Pacific enhance the convection bias and produce the alternating ITCZ error.
Some possible biases associated with the double ITCZ phenomenon have been identified in both the atmospheric and oceanic components. We will carry out further numerical study to examine these biases, especially the connection between the extra vertical diabatic heating mode in the lower troposphere and convection schemes in CAM5.
Acknowledgments
We thank Dr. Hua-Lu Pan from NCEP for his valuable discussions and suggestions. We also thank two anonymous reviewers for their helpful comments and suggestions to improve the manuscript. We are grateful to the National Center for High-Performance Computing for computer time and facilities. This research is supported by the National Science Council, Taiwan, through Grants NSC 102-2111-M-034-004, NSC 100-2119-M-001-029-MY5, and NSC 102-2111-M-001-009.
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.
Behringer, D. W., and Y. Xue, 2004: Evaluation of the global ocean data assimilation system at NCEP: The Pacific Ocean. Eighth Symp. on Integrated Observing and Assimilation Systems for Atmosphere, Oceans, and Land Surface, Seattle, WA, Amer. Meteor. Soc., 11–15. [Available online at https://ams.confex.com/ams/pdfpapers/70720.pdf.]
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97, 163–172, doi:10.1175/1520-0493(1969)097<0163:ATFTEP>2.3.CO;2.
Bretherton, C. S., and S. Park, 2009: A new moist turbulence parameterization in the Community Atmosphere Model. J. Climate, 22, 3422–3448, doi:10.1175/2008JCLI2556.1.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, doi:10.1002/qj.828.
de Szoeke, S. P., and S.-P. Xie, 2008: The tropical eastern Pacific seasonal cycle: Assessment of errors and mechanisms in IPCC AR4 coupled ocean–atmosphere general circulation models. J. Climate, 21, 2573–2590, doi:10.1175/2007JCLI1975.1.
Dijkstra, H. A., and J. D. Neelin, 1995: Ocean–atmosphere interaction and the tropical climatology. Part II: Why the Pacific cold tongue is in the east. J. Climate, 8, 1343–1359, doi:10.1175/1520-0442(1995)008<1343:OAIATT>2.0.CO;2.
Gent, P. R., and Coauthors, 2011: The Community Climate System Model version 4. J. Climate, 24, 4973–4991, doi:10.1175/2011JCLI4083.1.
Hagos, S., and Coauthors, 2010: Estimates of tropical diabatic heating profiles: Commonalities and uncertainties. J. Climate, 23, 542–558, doi:10.1175/2009JCLI3025.1.
Halpern, D., and C.-W. Hung, 2001: Satellite observations of the southeast Pacific intertropical convergence zone during 1993–1998. J. Geophys. Res., 106, 28 107–28 112, doi:10.1029/2000JD000056.
Hirota, N., and Y. N. Takayabu, 2013: Reproducibility of precipitation distribution over the tropical oceans in CMIP5 multi-climate models compared to CMIP3. Climate Dyn., 41, 2909–2920, doi:10.1007/s00382-013-1839-0.
Hirota, N., Y. N. Takayabu, M. Watanabe, and M. Kimoto, 2011: Precipitation reproducibility over tropical ocean and its relationship to the double ITCZ problem in CMIP3 and MIROC5 climate models. J. Climate, 24, 4859–4873, doi:10.1175/2011JCLI4156.1.
Hubert, L. F., A. F. Krueger, and J. S. Winston, 1969: The double intertropical convergence zone—Fact or fiction? J. Atmos. Sci., 26, 771–773, doi:10.1175/1520-0469(1969)026<0771:TDICZF>2.0.CO;2.
Huffman, G. J., R. F. Adler, D. T. Bolvin, and G. Gu, 2009: Improving the global precipitation record: GPCP version 2.1. Geophys. Res. Lett., 36, L17808, doi:10.1029/2009GL040000.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, doi:10.1029/2008JD009944.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587–1606, doi:10.1175/1520-0442(1993)006<1587:TSCOLS>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.
Lin, S. J., 2004: A “vertically Lagrangian” finite-volume dynamical core for global models. Mon. Wea. Rev., 132, 2293–2307, doi:10.1175/1520-0493(2004)132<2293:AVLFDC>2.0.CO;2.
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.
Liu, W. T., and X. Xie, 2002: Double intertropical convergence zones—A new look using scatterometer. Geophys. Res. Lett., 29, 2072, doi:10.1029/2002GL015431.
Liu, X., and Coauthors, 2012: Toward a minimal representation of aerosols in climate models: Description and evaluation in the Community Atmosphere Model CAM5. Geosci. Model Dev., 5, 709–735, doi:10.5194/gmd-5-709-2012.
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.
Morrison, H., and A. Gettelman, 2008: A new two-moment bulk stratiform cloud microphysics scheme in the Community Atmosphere Model, version 3 (CAM3). Part I: Description and numerical tests. J. Climate, 21, 3642–3659, doi:10.1175/2008JCLI2105.1.
Neale, R. B., J. H. Richter, and M. Jochum, 2008: The impact of convection on ENSO: From a delayed oscillator to a series of events. J. Climate, 21, 5904–5924, doi:10.1175/2008JCLI2244.1.
Nolan, D. S., C. Zhang, and S.-H. Chen, 2007: Dynamics of the shallow circulation around ITCZ regions. J. Atmos. Sci., 64, 2262–2285, doi:10.1175/JAS3964.1.
Nolan, D. S., S. W. Powell, C. Zhang, and B. E. Mapes, 2010: Idealized simulations of the intertropical convergence zone and its multilevel flows. J. Atmos. Sci., 67, 4028–4053, doi:10.1175/2010JAS3417.1.
Park, S., and C. S. Bretherton, 2009: The University of Washington shallow convection and moist turbulence schemes and their impact on climate simulations with the Community Atmosphere Model. J. Climate, 22, 3449–3469, doi:10.1175/2008JCLI2557.1.
Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1057, doi:10.1175/2010BAMS3001.1.
Smith, T. M., R. W. Reynolds, T. C. Peterson, and J. Lawrimore, 2008: Improvements to NOAA’s historical merged land–ocean surface temperature analysis (1880–2006). J. Climate, 21, 2283–2296, doi:10.1175/2007JCLI2100.1.
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.
Subramanian, A. C., M. Jochum, A. J. Miller, R. Murtugudde, R. B. Neale, and D. Waliser, 2011: The Madden–Julian oscillation in CCSM4. J. Climate, 24, 6261–6282, doi:10.1175/JCLI-D-11-00031.1.
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, doi:10.1175/2009JCLI3110.1.
Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 7183–7192, doi:10.1029/2000JD900719.
Tomas, R. A., and P. J. Webster, 1997: The role of inertial instability in determining the location and strength of near-equatorial convection. Quart. J. Roy. Meteor. Soc., 123, 1445–1482, doi:10.1002/qj.49712354202.
Waliser, D. E., and C. Gautier, 1993: A satellite-derived climatology of the ITCZ. J. Climate, 6, 2162–2174, doi:10.1175/1520-0442(1993)006<2162:ASDCOT>2.0.CO;2.
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, doi:10.1175/1520-0469(1973)030<0611:DOBPOT>2.0.CO;2.
Yang, B., and Coauthors, 2013: Uncertainty quantification and parameter tuning in the CAM5 Zhang-McFarlane convection scheme and impact of improved convection on the global circulation and climate. J. Geophys. Res. Atmos., 118, 395–415, doi:10.1029/2012JD018213.
Zhang, C., 2001: Double ITCZs. J. Geophys. Res., 106, 11 785–11 792, doi:10.1029/2001JD900046.
Zhang, C., and S. M. Hagos, 2009: Bi-modal structure and variability of large-scale diabatic heating in the tropics. J. Atmos. Sci., 66, 3621–3640, doi:10.1175/2009JAS3089.1.
Zhang, C., M. McGauley, and N. A. Bond, 2004: Shallow meridional circulation in the tropical eastern Pacific. J. Climate, 17, 133–139, doi:10.1175/1520-0442(2004)017<0133:SMCITT>2.0.CO;2.
Zhang, C., D. S. Nolan, C. Thorncroft, and H. Nguyen, 2008: Shallow meridional circulations in the tropical atmosphere. J. Climate, 21, 3453–3470, doi:10.1175/2007JCLI1870.1.
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, 33, 407–446, doi:10.1080/07055900.1995.9649539.