Abstract

Two suites of partial coupling experiments are devised with the upper-ocean dynamics version (UOM) of the CCSM3 to isolate the effects of the feedbacks from the change of the wind-driven ocean circulation and air–sea heat flux in the global climate response to the forcing of doubling CO2. The partial coupling is achieved by implementing a so-called overriding technique, which helps quantitatively partition the total response in the fully coupled model to the feedback component in question and the response to external forcing in the absence of the former. By overriding the wind stress seen by the ocean and the wind speed through the bulk formula for evaporation, the experiments help to reveal that (i) the wind–evaporation–SST (WES) feedback is the main formation mechanism for the tropical SST pattern under the CO2 forcing, verifying the hypothesis proposed by Xie et al.; (ii) the weakened tropical Pacific wind is shown in this UOM model not to be the cause for the enhanced equatorial Pacific warming, as one might expect from the thermocline and Bjerknes feedbacks; (iii) WES is also the leading mechanism for shaping the tropical precipitation response in the ocean; and (iv) both the wind-driven ocean dynamical feedback and the WES feedback act to increase the persistence of the southern annular mode (SAM) and the increased time scale of the SAM due to these feedbacks manifests itself in the response of the jet shift to an identical CO2 forcing, in a manner conforming to the fluctuation–dissipation theorem.

1. Introduction

Increasing the concentration of the heat-trapping greenhouse gases (GHGs) can have profound impacts on the climate dynamical system. Among the most robust are the patterns of the global atmospheric circulation, sea surface temperature, and global hydrological cycle. The overall strength of the tropical atmospheric overturning circulation is projected to be weakening despite intensification of the tropical hydrological cycle (e.g., Held and Soden 2006; Vecchi et al. 2006; Vecchi and Soden 2007a; Lu et al. 2007). In the subtropics and midlatitudes, the descending branch of the Hadley cell expands poleward and so does the subtropical dry zone controlled by the descending motion (Lu et al. 2007; Previdi and Liepert 2007; Lu et al. 2008); the midlatitude storm tracks and jet streams shift poleward, leading to an increase in the mean and extreme precipitation over most parts of the extratropics (Kushner et al. 2001; Yin 2005; Lorenz and DeWeaver 2007a,b; Chen et al. 2008; Wu et al. 2010). The underlying causes for these changes have been understood mostly from the atmospheric perspective in terms of basic dynamical constraints (e.g., angular momentum conservation for the case of Hadley cell expansion) and thermodynamical principles (e.g., the Clausius–Clapeyron relation for the hydrological cycle response). The robust features in the zonal mean atmospheric circulation have been successfully simulated with idealized settings without coupling to the ocean dynamics (such as slab mixed-layer models, e.g., O’Gorman and Schneider 2009) and even without sea surface boundary condition (such as in a dry dynamical core, e.g., Williams 2006). It remains largely elusive as to what role the ocean dynamical feedback plays in those robust climate features in responding to global warming. For example, what is the role of the ocean dynamic feedback in the midlatitude westerly jet and storm track response to the global warming forcing? Watterson (2003) investigated the role of dynamical ocean on climate sensitivity using an earlier version of the Commonwealth Scientific and Industrial Research Organisation (CSIRO) coupled climate model and identified ocean dynamical effects on the tropical warming and precipitation from the inferred oceanic heat transport. However, he did not consider the dynamical patterns of the climate response to global warming, which only emerged since the Fourth Assessment of the Intergovernmental Panel on Climate Change (IPCC) (Meehl et al. 2007).

For the ocean dynamics and air–sea interaction to impact the climate, they must work through the SST. The role of the ocean is especially important when regional detail of the SST response to global warming is the matter of concern. Xie et al. (2010) recently demonstrated that regional differences in SST can be as large as the tropical mean warming itself and that one cannot obtain correct regional rainfall response without considering the regional features of SST. In addition, the projection of the potential intensity of tropical cyclones under GHG warming is acutely sensitive to the details of the tropical SST pattern (Vecchi and Soden 2007b; Xie et al. 2010). Thus, it is of crucial importance to understand the formation mechanism of the tropical SST pattern under a relatively uniform GHG forcing.

The SST response to external climate forcing is a result of complex and poorly understood interplay among a myriad of factors involved in the surface energy budget (Hartmann and Michelsen 1993). The surface energy budget has been widely used to examine the mechanisms for the SST pattern formation under GHG forcing and several hypotheses have been proposed to account for the tropical Pacific SST pattern simulated by the coupled climate models (Knutson and Manabe 1995; Liu et al. 2006; DiNezio et al. 2009; Xie et al. 2010). Knutson and Manabe (1995) argued that the “evaporative damping” effect, due to the strong dependence of evaporation on SST through the Clausius–Clapeyron relation, is the leading cause for the reduced zonal SST gradient between the Pacific warm pool and the cold tongue in their model. Indeed, they found enhanced evaporative cooling over the warm pool relative to the cold tongue to back their argument. On the other hand, Clement et al. (1996) demonstrated that the “ocean thermostat,” resulted from the shallower thermocline over the eastern equatorial Pacific relative to the western Pacific, acts to enhance the west-to-east SST gradient through more efficient upwelling cooling over the eastern Pacific and the associated feedbacks. As a result of the cancellation between the two major opposing mechanisms, it is unclear whether the global warming signature over the tropical Pacific will be “El Niño–like” or “La Niña–like” (Vecchi et al. 2008). Liu et al. (2006) instead suggested that the enhanced equatorial warming relative to the subtropics, instead of the zonal SST gradient, is the more robust SST fingerprint of global warming. The basis for the enhanced equatorial warming is again the dependence of the evaporation on the wind speed. Upon a presumed uniform energy flux at the sea surface, the subtropical area with stronger trade winds can damp this energy surplus so as to stabilize the SST increase more efficiently than the equatorial area. A similar argument has been put forward to explain the lack of warming over the tropical North Atlantic in the future climate projection (Leloup and Clement 2009).

All the three propositions, so far, are “static” arguments, that is, not taking into consideration the effects of wind change and associated wind-driven ocean circulation change. The seminal work of G. Vecchi and collaborators discovered that the tropical atmospheric circulation, especially the Walker cell, will weaken under global warming, an assertion also backed by observational evidence (Vecchi et al. 2006; Vecchi and Soden 2007a). If this weaker Walker mechanism acted in isolation, one would expect some El Niño likeness in the equatorial Pacific response from the thermocline or Bjerknes feedbacks (Vecchi et al. 2008), with which we have familiarized ourselves from extensively studied ENSO dynamics. However, through careful diagnostics of air–sea energy flux and oceanic heat transport, DiNezio et al. (2009, 2010) dismissed El Niño as a useful analogy for understanding the equatorial SST warming under climate change by noting that the impact of oceanic dynamical response is to damp the warm signal over the eastern equatorial Pacific. On the other hand, by noting the association between the warming maximum (minimum) near the central equatorial Pacific (southeast subtropics) and weakening of the equatorial easterlies (intensification of the southeast trade winds), Xie et al. (2010) proposed that wind–evaporation–SST (WES) [to be explained later, see also Xie and Philander (1994)] may be the main feedback mechanism for the tropical SST pattern under GHG forcing. Which is right, then? While an energy budget can be helpful for identifying the major balance and forming hypotheses therefrom, it is ineffective when it comes to resolving issues of debate as such. Therefore, here we modify the code of a coupled climate model to turn off the wind-driven ocean dynamical feedback or/and the WES feedback to see how partially coupled versions of the model respond to an imposed CO2 perturbation. In so doing we may reach a “verdict” on the debate.

Another point of interest is the role of ocean dynamic feedback (or “back interaction” termed by some researchers, e.g., Watterson 2001; Sen Gupta and England 2007) in the internal variability of the southern annular mode (SAM) and in the SAM-like response to GHG-induced global warming. Recent analysis of the data output from the Third Coupled Model Intercomparison Project (CMIP3) found that models with an equatorward bias in the position of the midlatitude eddy-driven westerlies in the Southern Hemisphere tend to have overly persistent SAM anomalies and overestimate the midlatitude circulation response to a given climate forcing (Kidston and Gerber 2010; Barnes et al. 2010), a behavior reminiscent of a fluctuation–dissipation (FD) system (Leith 1975; Gritsun and Branstator 2007; Gritsun et al. 2008). To the extent that the ocean’s feedback plays a part in shaping the intrinsic time scale of the SAM variability, ocean dynamics may also manifest itself in the SH wind response to climate change forcing if the SH climate behaves in any way like a FD system. In principle, two possible mechanisms may be involved in the back interaction due to coupling with the ocean. The first is related to local thermodynamical interaction, wherein the ocean response to an atmospheric anomaly modifies air–sea fluxes in such a way as to reduce the damping that the original atmospheric anomaly otherwise would experience in the absence of ocean feedback. This thermodynamical interaction is considered in both fully coupled atmosphere–ocean models as well as those coupled with a slab mixed layer ocean. The second is the indirect response to changes in the surface flux resulting from the wind-induced anomalous ocean circulation and heat transport, the effect of which may imprint back onto the original wind anomaly. By applying a SAM-like wind stress anomaly felt only by the ocean component of a fully coupled climate system, Sen Gupta and England (2007) demonstrated that ocean feedback can have a weak, but significant, impact on the temporal characteristics of the SAM—enhancing the persistence of atmospheric vacillations in the coupled integrations. As will be shown later in the present study, both the reduced damping mechanism (mainly through WES) and the indirect ocean dynamical feedback, indeed, serve as positive feedback mechanisms in the life cycle of the SAM. Additionally, these feedbacks act to reinforce the positive SAM-like response in the doubling CO2 experiment.

The primary goal of this study is to isolate and identify the effects of the ocean dynamical feedback and air–sea latent heat flux in many of the robust features of the tropical SST, global atmospheric circulation, and hydrological response under global-warming forcing. To this end, we utilize a partial-coupling technique with the upper-ocean dynamical model (UOM) of the Community Climate System Model, version 3 (CCSM3), a version in which temperature and salinity at depths below the nominal permanent pycnocline are prescribed as the climatology of the full depth ocean model of CCSM3. The UOM of CCSM3 (referred to as CCSM3-UOM hereafter) was developed by Zhao et al. (2011) for the purpose of climate sensitivity study. The basic configuration of the UOM and the associated advantage are briefly recapitulated in section 2. The partial coupling is realized through overriding the wind speed and/or wind stress from climate scenarios of interest. However, the wind-related surface heat and momentum fluxes are interactive in nature and wind overriding interferes with the physical consistence in the coupled processes: as a result, climate bias ensues. This bias can confound the feedback signal of interest in this study and should be addressed with care. In section 2b, we lay out the procedures for eliminating the bias caused by overriding and the design of the overriding experiments to isolate the WES feedback and the feedback from the wind-driven ocean circulation. The effects of these feedbacks on the global SST, global atmospheric circulation, and hydrological pattern are then delineated in section 3. The same sets of experiments can also help shed light on the role of oceanic dynamical feedback in the internal variability of the SAM and its response to CO2 forcing as well (section 4). The paper concludes with a summary and discussion.

2. Model and experimentation

a. CCSM3-UOM

The CCSM3-UOM is developed from a low-resolution version of the National Center for Atmospheric Research (NCAR) CCSM3 (Collins et al. 2006a) with T31 (approximately 3.75° grid cell) resolution for the atmospheric component [Community Atmosphere Model version 3 (CAM3) (Collins et al. (2006b)] and a nominal 3° horizontal resolution for the ocean component Parallel Ocean Program (POP). Yeager et al. (2006) documented some modifications specifically to this version to make it a viable alternative to a higher-resolution version of the CCSM3. For the UOM configuration, the temperature and salinity fields at levels below the permanent pycnocline are prescribed as monthly climatologies, which are obtained from a long integration of the CCSM3 with a full depth ocean (referred to as CCSM3-FDM hereafter). Similar treatment is applied to the baroclinic component of the horizontal velocity so that the thermal wind balance is satisfied. The permanent pycnocline in the CCSM3 is set at 500-m depth in the tropics, inclining to 1300 m near the poles. Configured as such, the deep ocean serves as a climatological reservoir of tracers and baroclinic velocity with infinite capacity. See Zhao et al. (2011) for further details about the construction of the CCSM3-UOM.

Since the thermal condition of the deep ocean is fixed, only the upper portion of the ocean of the CCSM3-UOM responds to a climate perturbation. This brings about a critical advantage over the conventional, full depth ocean general circulation model: a much faster adjustment to equilibrium. A typical adjustment time scale of the CCSM3-UOM is only 40–50 years (Danabasoglu and McWilliams 2000; Zhao et al. 2011) versus the thousand years or more for the CCSM3-FDM (Zhao et al. 2011). This affords us the computation of running multiple sets of simulations to equilibrium. The CCSM3-UOM so configured can reproduce a mean climate state and associated internal variability almost indistinguishable from that of the original CCSM3-FDM. As a consequence, the UOM also inherits most of the climate biases of the FDM.

It should also be noted that the equilibrium response of the CCSM3-UOM to CO2 doubling is not identical to that of the CCSM3-FDM. Instead, the equilibrium response of the CCSM3-UOM emulates the transient climate response of the CMIP models (Hegerl et al. 2007), which is the response after about 70 years, given a 1% CO2 doubling rate, or the response when the quick surface temperature response is in quasi equilibrium with the radiative forcing of CO2 while the sluggish deep ocean remains far from equilibrium (e.g., Meehl et al. 2006). Recently, Held et al. (2010) identified the time-scale separation between the transient and equilibrium response and suggested that the twenty-first century climate is characterized by a transient response as long as the CO2 doubles over a time period short enough (less than 100 years, for example) for the deep ocean to remain far from equilibrium. Thus, the response and feedbacks derived from the UOM experiments may be thought of as the fast response part of the FDM to doubling CO2 forcing.

To further justify use of the UOM, we compare the equilibrium surface temperature response to doubling CO2 among CCSM3-UOM (Fig. 1a); CCSM3-FDM (Fig. 1b, based on a 1500-yr control run and a 3000-yr run, in which the CO2 is instantaneously doubled, see Danabasoglu and Gent 2008); and the CCSM3 slab ocean model (Fig. 1c). One can see that the UOM is adequate in capturing the tropical temperature response in the FDM, while both differ significantly from the response in the slab ocean model. In particular, the features of interest in the current study, such as the enhancement of the equatorial SST warming and the lack thereof over the subtropics, can all be reproduced by the UOM. The effect of deep ocean adjustment, likely related to slowdown of the interhemispheric overturning circulation, is omitted in the UOM model. As a result, the UOM simulates a relatively weaker (larger) warming near the South (North) Pole than for the FDM. The role of deep ocean adjustment to climate change warrants further investigation, a topic we leave for future study.

Fig. 1.

Surface temperature response (K) to doubling CO2 in (a) CCSM3_UOM; (b) CCSM3_UOM; and (c) a slab ocean version of the CCSM3.

Fig. 1.

Surface temperature response (K) to doubling CO2 in (a) CCSM3_UOM; (b) CCSM3_UOM; and (c) a slab ocean version of the CCSM3.

b. Partial-coupling technique

The partially coupled UOM in this study is realized through overriding the full time series of a variable, or multiple variables, at the air–sea interface taken from a separate fully coupled realization. As a result, the feedbacks from the ocean through this (these) variable(s) are disabled. Specifically, the overriding is implemented through the following procedure. First, a baseline experiment (1x) and a 2xCO2 perturbation experiment (2x) with the fully coupled model are performed, each initialized from a long control CCSM3-FDM simulation; the overriding variables at the frequency of air–sea coupling are stored for the overriding experiments. If, for example, we are interested in the feedback from the modified wind-driven ocean circulation under 2xCO2 forcing, we then run the 2xCO2 experiment again but with the wind stress field prescribed (at every time step of coupling) to be that of the 1x experiment. We name this overriding run W1x_2x, meaning wind stress prescribed from 1x but with the CO2 level from 2x. The climatological mean wind stress of the 1x control and the difference 2x minus 1x are shown in Fig. 2. Had the climate bias (relative to the fully coupled run) due to overriding not been an issue, taking the difference between 2x and W1x_2x would give the feedback effect of the changing wind-driven ocean circulation. However, one should not compare the overriding run directly with the fully coupled run for the following reason. In the overriding run the ocean “feels” the perturbation in the atmospheric wind stress, which is not felt by the atmosphere; as such, the wind stress feedback loop is short-circuited. As a consequence, a climate bias results that can confound the actual climate change signal. To address the bias due to overriding, we need to rerun the 2x case but now with the time series of wind stress taken from the 2x experiment and perturbed only to interrupt the temporal coherence between the wind stress and the oceanic condition (it is hence named W2x_2x). This is achieved by simply shifting by one year the time of the wind stress field before applying it to the overriding rerun. Finally, the true feedback of the wind-driven ocean circulation can be obtained by subtracting W1x_2x from W2x_2x. An overriding rerun for the 1x case, that is, W1x_1x, has also been conducted to form an overriding baseline. Through this procedure, the total response to CO2 doubling in the fully coupled model can be linearly decomposed into two additive components: the response of the wind stress-driven oceanic feedback (W2x_2x–W1x_2x, or W2x_1x–W1x_1x) and the response of the wind stress overriding climate system without the feedback (W1x_2x–W1x_1x).

Fig. 2.

Wind stress and its magnitude (N m−2) in (a) 1x coupled control simulation and (b) difference between 2x and 1x.

Fig. 2.

Wind stress and its magnitude (N m−2) in (a) 1x coupled control simulation and (b) difference between 2x and 1x.

c. Rationale for experiment design

Let us start with a thought experiment wherein the energy balance at the air–sea interface is perturbed by a constant anomaly (, positive downward), which mimicks the forcing of doubling CO2. Ignoring the change of oceanic heat transport temporarily and assuming this surplus is balanced solely by the evaporative cooling, we have

 
formula

where the overbar denotes climatological mean, is latent heat flux, SST change, and wind speed change. In (1) only the parameter changes deemed to be of leading-order importance are considered. In comparison to Eq. (3) in Xie et al. (2010), the contribution due to changes in surface static stability and relative humidity have been ignored. Alternatively, (1) can be written as

 
formula

Since the radiative forcing of CO2 is relatively uniform, (2) implies that the gradient of the tropical SST response is dominated by that of near-surface wind speed, the strengthening of which cools the SST through enhanced evaporation, an effect referred to as WES feedback within the context of this paper. Indeed, Xie et al. (2010) recognized an intriguing resemblance between the tropical SST pattern and wind speed pattern and proposed WES to be the main mechanism for minimum warming of the subtropical South Pacific. However, even in the absence of WES, the denominator, which measures the sensitivity of the evaporative cooling to SST (denoted as ), can exert a strong influence on the pattern of the SST. According to the bulk formula, it can be expresssed as

 
formula

where the notation is conventional and the subtle difference between the SST and the near-surface air temperature is neglected. It measures the resistance to the SST warming through evaporative cooling and is directly proportional to surface wind speed and saturation specific humidity, the latter being a quasi-exponential function of SST following the Clausius–Clapeyron relation. We will refer to as evaporation–SST denominator in the context of the present study.

Adding back the change of heat convergence related to ocean dynamics , which comprises the change due to the advection/diffusion of anomalous temperature by the mean ocean circulation and that due to the change of wind-driven circulation , the SST response now becomes

 
formula

This is the leading-order energy balance that maintains the tropical SST anomalies under climate perturbations: we will use it to guide our experimental design. Herein we leave the radiative/cloud feedbacks out of the picture, only focusing on the processes at the air–sea interface. To test the weaker Walker hypothesis (related to the weakened equatorial wind stress) and the WES hypothesis, we conduct two sets of overriding experiments: one with wind stress and the other with both wind stress and wind speed through the evaporation simultaneously. The former is intended to isolate the feedback from the wind-stress-driven ocean dynamics change, that is, the effect of (referred to as wind stress feedback hereafter), and the latter includes both wind stress feedback and WES feedback (). The former set will be referred to as W overriding category and the latter WE overriding category (Table 1). Assuming linearity, the effect of the WES feedback can be inferred via (WE2x_2x–WE1x_2x) – (W2x_2x–W1x_2x). For the linearity to hold, the climate perturbation should be within the range wherein the climate response scales linearly with the external forcing. This is the rationale behind the choice of doubling CO2 concentration, which is believed to reside safely within the linear response regime (as per our experience with the climate response of the CCSM3 under different climate change scenarios). The downside for using a moderate forcing is that long simulations are needed to obtain a robust signal. Therefore, for each overriding experiment a 240-yr-long integration is conducted, and the analysis is conducted using the output of the last 190 years. For each of the experiments discussed in this study, the CO2 concentration is specified as a constant in both time and space throughout the integration. All of the experiments used for the purpose of this study are listed in Table 1; for most cases annual mean results are presented except where explicitly stated.

Table 1.

List of experiments used in the current study.

List of experiments used in the current study.
List of experiments used in the current study.

3. Results

a. SST

1) Decomposition of the SST response

The full response of the tropical Pacific SST to doubling CO2 is neither “El Niño–like” nor “La Niña–like” in this model, but is characteristic of an equatorial warming centered at 150°W (Fig. 3a), epitomizing the most robust aspect of the tropical Pacific response amongst the climate models (Liu et al. 2006; Collins et al. 2005). This feature, together with other extratropical features can be faithfully replicated by the overriding runs (see Figs. 3c and 4c). The wind-stress-overriding replication of the SST response to doubling CO2 (W2x_2x–W1x_1x) can then be further decomposed into wind stress feedback (W2x_2x-W1x_2x, Fig. 3d) and the response to CO2 in the absence thereof (W1x_2x–W1x_1x, Fig. 3b). The wind stress feedback turns out to be a cooling over the central and eastern equatorial Pacific, in stark contrast to what one might envisage from the weakening of the tropical Walker circulation alone: the relaxation of the equatorial easterlies deepens the thermocline, acting to warm the central and eastern Pacific through the thermocline and Bjerknes feedbacks. It is likely that other processes, such as the Sverdrup response to the near-equatorial wind stress curl or the “discharge” component of the Recharge–Discharge Oscillator (Jin 1997), may also be involved in the cooling response. We defer discussion of the cause of the cooling until the next subsection.

Fig. 3.

Partitioning of the SST response (K) using wind stress overriding experiments: (a) full response, 2x–1x; (b) response to 2xCO2 in the absence of wind stress feedback, W1x_2x–W1x_1x; (c) replication of the full response with wind stress overriding simulations, W2x_2x–W1x_1x; and (d) wind stress feedback, W2x_2x–W1x_2x.

Fig. 3.

Partitioning of the SST response (K) using wind stress overriding experiments: (a) full response, 2x–1x; (b) response to 2xCO2 in the absence of wind stress feedback, W1x_2x–W1x_1x; (c) replication of the full response with wind stress overriding simulations, W2x_2x–W1x_1x; and (d) wind stress feedback, W2x_2x–W1x_2x.

Fig. 4.

Partitioning of the SST response (K) using wind stress and wind speed overriding experiments: (a) full response, 2x–1x, duplicated from Fig. 3a; (b) response to 2xCO2 in the absence of both wind stress and WES feedbacks, WE1x_2x–W1x_1x; (c) replication of the full response with overriding simulations, WE2x_2x–WE1x_1x; (d) wind stress plus WES feedbacks, WE2x_2x–WE1x_2x; and (e) WES feedback, (WE2x_2x–WE1x_2x)–(W2x_2x–W1x_2x).

Fig. 4.

Partitioning of the SST response (K) using wind stress and wind speed overriding experiments: (a) full response, 2x–1x, duplicated from Fig. 3a; (b) response to 2xCO2 in the absence of both wind stress and WES feedbacks, WE1x_2x–W1x_1x; (c) replication of the full response with overriding simulations, WE2x_2x–WE1x_1x; (d) wind stress plus WES feedbacks, WE2x_2x–WE1x_2x; and (e) WES feedback, (WE2x_2x–WE1x_2x)–(W2x_2x–W1x_2x).

With the aid of the WE overriding experiments, the wind-related feedback effects (wind stress feedback plus WES) can be isolated by subtracting WE1x_2x from WE2x_2x. We first notice that the gradients in SST over the tropical and subtropical Pacific and Indian Ocean largely diminish when the wind feedbacks are disabled (WE1x_2x-WE1x_1x, Fig. 4b). By corollary, the pattern of the tropical Pacific SST should mainly be attributed to these wind-related feedbacks (Fig. 4d).

In the absence of the wind stress and WES feedbacks, the flat pattern of SST response is determined jointly by the limiting effect of evaporative cooling on SST warming (Knutson and Manabe 1995; Liu et al. 2006) and the regulating effect of heat transport/diffusion by the mean ocean circulation (Seager and Murtugudde 1997; Clement and Seager 1999). In other words, even in the absence of these wind-related feedbacks, the evaporative cooling mechanism is inadequate to account for the SST response in WE1x_2x-WE1x_1x, as to be elaborated next. Assuming linearity, the pure WES can be inferred from subtracting the wind stress feedback (W2x_2x-W1x_2x) from the total wind feedbacks (WE2x_2x-WE1x_2x), and the result is presented in Fig. 4e. The pattern correlation between the WES-induced SST anomalies and the SST pattern of the full response over the tropical oceans is 0.82, indicative of the dominant role of the WES mechanism in the formation of the tropical SST pattern. This result provides a direct verification for the WES hypothesis proposed by Xie et al. (2010).

The wind stress feedback and WES also contribute to the warming near the Kuroshio and its extension, in accord with the finding of poleward expansion of the North Pacific subtropical gyre (e.g., Sato et al. 2006). However, the bulk of the North Pacific warming persists even in the absence of the wind feedbacks, resulted likely from other climate feedback processes (such as albedo and cloud feedbacks) and the advection of warm anomalies by the climatological subpolar currents. In a doubling CO2 simulation with the slab version of the model wherein all oceanic dynamical feedback is eliminated, the warming in the North Pacific interior is somewhat suppressed and the warming is relatively confined in the western Pacific (Fig. 1c). It is of interest to note that in the absence of wind stress and WES feedbacks, the zonal band of minimum warming near 50°–55°S in the Southern Ocean (see Figs. 3a and 4a)—another robust feature in the SST response to global warming—also disappears (Fig. 4b). This underscores the importance of the wind-driven ocean circulation and WES feedback in shaping the Southern Ocean SST patterns. As will be shown later in section 3d, both wind-related feedbacks play a constructive role in maintaining the wind anomalies of the internal variability of the SAM and in the SAM-like response to increasing CO2 forcing.

2) Interpreting the SST response via a partially coupled hierarchy

This set of overriding experiments in combination with the slab ocean model constitute a modeling hierarchy for interpreting the fully coupled response to greenhouse gas warming, from the simple to complex: slab ocean coupled model, partially coupled without wind stress and WES feedbacks, partially coupled without wind stress feedback, and fully coupled model. We now utilize this hierarchy to assess the relative importance of the aforementioned mechanisms for the full SST response under CO2 forcing.

The most fundamental of the mechanisms is the evaporation–SST denominator effect, which operates in all members of the hierarchy. Following Leloup and Clement (2009), we calculate according to Eq. (3) based on the mean wind and SST from the WE baseline experiment (WE1x_1x); the negative of it is displayed in Fig. 5a to facilitate comparison with the actual SST response. The tropical Atlantic sector can be compared with Fig. 3b of Leloup and Clement (2009). What is appealing about the evaporation–SST denominator is that it is a function of the climatological mean values and, if it is the dominant mechanism in balancing the GHG-induced radiative forcing, the SST response can be predicted simply from the climatological mean values of the surface quantities. The regulating effect of the evaporation–SST denominator on the subtropical SST is discernable in all 2xCO2 response patterns in the coupling hierarchy (Fig. 1c for slab, Figs. 3a for UOM, Fig. 1b for FDM, Fig. 3b for the W overriding, Fig. 4b for WE overriding). However, it is obviously inadequate to account for the patterns quantitatively, even for the response in the slab model.1 Of particular interest among the hierarchy is the case of WE overriding since the evaporation–SST denominator effect is expected to work the best in the absence of the wind related feedbacks. On the contrary, the resemblance to the pattern of the SST response, shown in Fig. 4b, is arguably the least amongst the model hierarchy. Further energy budget analysis (see Fig. A1 for the case of WE1x_2x–WE1x_1x) indicates that the oceanic heat convergence acts to cool (warm) the equatorial Pacific and Indian oceans (subtropical Pacific and North Atlantic) so as to compensate the evaporation–SST denominator effect. In fact, the sensitivity of the oceanic heat divergence on SST, denoted symbolically as , can be thought of the “ocean dynamical thermostat” owing to the advective and diffusive cooling by the mean ocean circulation. The term ocean dynamical thermostat was first dubbed by Clement et al. (1996), but for a dynamically coupled tropical system. The exact meaning of the thermostat depends on the model configuration. In this current study, we refer to as the thermostat effect due to the mean ocean dynamical advection and diffusion only, without the feedback through the wind change or any radiative feedbacks. While the exact estimation of requires further carefully designed experiments, its gross feature may be inferred roughly from the surface energy balance:

 
formula

The opposing effect on SST between the evaporation–SST denominator and ocean dynamical thermostat is clear in the contrast between the structure of (Fig. 5a) and (Fig. 5b), despite the contamination in the latter by the radiative feedbacks. It is thus conceivable that the flat SST response over the tropics in experiments WE1x_2x – WE1x_1x is the result of cancelation between these two opposing mechanisms. The implication for the minimum warming in the tropical North Atlantic is that it cannot be simply explained by the evaporation–SST denominator effect, that is, the influence of the climatological mean wind speed on the efficiency of latent heat flux when coupling to ocean dynamics is involved. Owing to the strong compensation from the oceanic thermostat effect, the evaporation–SST denominator effect only contributes secondarily to the SST pattern formation under GHG forcing.

Fig. 5.

(a) Evaporation–SST denominator effect () and (b) ocean dynamical thermostat ().

Fig. 5.

(a) Evaporation–SST denominator effect () and (b) ocean dynamical thermostat ().

The dominant factor for the tropical SST pattern formation is the WES feedback. The direct WES effect can be estimated as

 
formula

and the result is presented in Fig. 6. Its resemblance to actual WES response [(WE2x-2x–WE1x_2x) – (W2x-2x–W1x_2x), Fig. 4e] is conspicuous. This represents the immediate SST tendency that one might expect when the sea surface suddenly feels the impact of the changed wind. However, when we calculate the energy budget explicitly at equilibration (see Fig. A2), it turns out that the evaporative cooling tends to increase over the region of weakened wind speed and it is, instead, the oceanic heat convergence that maintains the warm SST anomalies. A similar role of ocean heat transport in the central equatorial Pacific warming was also noticed in earlier studies [Watterson (2009), see the increase of ocean heat convergence west of the date line in their Fig. 20]. That the evaporation term in the equilibrium energy budget might assume an opposite sign to the actual WES SST response points to the limitation of surface energy budget for understanding the SST formation (e.g., DiNezio et al. 2009). This exemplifies the challenge of understanding the SST pattern in an atmosphere–ocean coupled dynamical system.

Fig. 6.

The direct WES effect estimated based on the climatologically mean surface quantities: .

Fig. 6.

The direct WES effect estimated based on the climatologically mean surface quantities: .

It is worth noting that the features of minimum warming over the tropical North and southwest Atlantic reemerge in the fully coupled response (Fig. 4a) as well as for the overriding response whenever the WES feedback is operating (see Fig. 4c for the WE overriding replication and Fig. 3b for case of CO2 response without wind stress feedback). The WES cooling due to the enhanced wind speed there (it is still a cooling despite that the total change of evaporation is a warming effect) is arguably the leading cause for the lack of warming over the main development region in this model when ocean dynamics is involved. This assertion may be generalized to the lack of warming in all subtropical oceans through our experimentatal hierarchy: the balance between the evaporation–SST effect and the advection/diffusion by the mean ocean circulation instigates a flat SST warming pattern with a broad subtropical suppression and a wind pattern that further shapes the former into the final form through the WES feedback.

Another puzzle is the tropical Pacific cooling induced by the wind stress feedback. To shed light on the possible cause of it, we examine the components that constitute the energy budget balance at the sea surface: vertically integrated oceanic heat convergence , net radiative flux , sensible heat flux , and latent heat flux . The sign convention for each term is chosen in such a way that positive means warming the SST. The result is displayed in the  appendix (see Fig. A3). In accordance with our physical intuition, the weakened wind over the equatorial Pacific gives rise to a convergence of ocean heat flux there, but sandwiched by divergences off the equator. In the meantime, both radiative flux (Fig. A3b) and latent heat flux (Fig. A3c) counter the dynamical warming near the equator, resulting in an overall cooling as a result. This serves as another example challenging our understanding the SST response when coupling with ocean dynamics is considered. In this case, even though the response as a whole is, by design, initiated by a wind-driven ocean circulation anomaly, the final equilibrium response of the SST can still take an opposite sign to what the ocean dynamics intends it to be. Given the coarse resolution of the ocean component and the likely model dependence of this result, we defer more careful analysis of the ocean processes until the availability of similar experiments with state-of-the-art models.

b. Precipitation

The dominance of WES in the tropical Indo-Pacific SST pattern formation has important implications for tropical precipitation. As shown in Fig. 7g (WE2x_2x–WE1x_2x), the regional distribution of the tropical precipitation response (with the exception of African and South American continents) to CO2 doubling is predominantly controlled by the WES feedback! At first glance, it is somewhat surprising to note that the relatively uniform tropical SST warming in the absence of the WES feedback (as shown in Fig. 4b) only contributes secondarily to the hydrological pattern over the tropical oceans (Fig. 7f), despite its larger absolute values. The sensitivity of convective precipitation to the pattern of tropical SST is consistent with what one might expect from the close covariability between the SST threshold for convection and the tropical mean SST (as found in Johnson and Xie 2010): the effect of the increase of the tropical mean SST is muffled by the attendant increase in the convection threshold and local changes in convective precipitation is thus dictated by deviation between the local and tropical mean SST change, rather than the local SST change alone (Xie et al. 2010). It is also interesting to note that the land precipitation response over tropical Africa and South America is dominated by warming in the absence of wind stress and WES feedbacks. The wind stress feedback seems to play a secondary, but appreciable, role in shaping the patterns of the tropical precipitation. In the tropical Pacific, the wind-stress-induced oceanic feedback to the hydrological field (Fig. 7d) runs counter to the hydrological response in the absence of wind stress feedback (Fig. 7c). Were the tropical cooling induced by the wind stress feedback real, by corollary, it is conceivable that the slab mixed-layer models might have overestimated the precipitation response over the equatorial oceans compared to the corresponding models with ocean dynamics.

Fig. 7.

The precipitation response (mm day−1) in (a) 2x–1x; (b) W2x_2x–W1x_1x; (c) W1x_2x–W1x_1x; (d) W2x_2x–W1x_2x; (e) WE2x_2x–WE1x_1x; (f) WE1x_2x–WE1x_1x; and (g) WE2x_2x–WE1x_2x . The meaning of each of these experiments is labeled accordingly.

Fig. 7.

The precipitation response (mm day−1) in (a) 2x–1x; (b) W2x_2x–W1x_1x; (c) W1x_2x–W1x_1x; (d) W2x_2x–W1x_2x; (e) WE2x_2x–WE1x_1x; (f) WE1x_2x–WE1x_1x; and (g) WE2x_2x–WE1x_2x . The meaning of each of these experiments is labeled accordingly.

As far as the extratropical precipitation response is concerned, neither WES nor wind stress feedback amounts to the same dominance as WES does in the tropical precipitation. The bulk of the extratropical precipitation response can be accounted for by climate warming in the absence of wind-related feedbacks. This might be the underlying reason for the fact that many features in the extratropical climate response can be understood in terms of basic atmospheric dynamical and thermodynamical principles such as those put forward in recent studies (e.g., Lu et al. 2007, 2008, 2010; Chen et al. 2008; Lorenz and DeWeaver 2007a; Riviere 2011). On the other hand, over the Euro–Atlantic sector, the total wind feedbacks (Fig. 7g) can have important modulating effect on the south European and Mediterranean climate. For example, without the WES feedback, mediterranean Europe might not be as dry as it is in the fully coupled climate system.

c. Atmospheric circulation

The most prominent feature in the atmospheric circulation response, as delineated by sea level pressure (Fig. 8), is the poleward expansion of the subtropical high in both hemispheres and both Pacific and Atlantic basins. As the circulation in the atmosphere must be dynamically consistent with the SLP pattern, the meridional mass overturning circulation also exhibits an expansion in both the Hadley and Ferrel cell (no shown). Consistent with the impression gained from the precipitation response, the total sea level pressure pattern in the extratropics and expansion of the meridional cells are dominated by the response to 2xCO2 in the absence of the wind stress and WES feedbacks. To the extent that the main features in the sub- to extratropical circulation can be accounted for without considering these feedbacks, the key to understanding the extratropical climate response is held in the intrinsic dynamics of the atmosphere.

Fig. 8.

Sea level pressure response (Pa) in (a) 2x–1x; (b) W2x_2x–W2x_1x; (c) W2x_2x–W1x_2x; (d) WE1x_2x–WE1x_1x; and (e) WE2x_2x–WE1x_2x. The contours indicate the total SLP field with respect to which the difference is taken. Values less than the global mean are displayed as dashed contours.

Fig. 8.

Sea level pressure response (Pa) in (a) 2x–1x; (b) W2x_2x–W2x_1x; (c) W2x_2x–W1x_2x; (d) WE1x_2x–WE1x_1x; and (e) WE2x_2x–WE1x_2x. The contours indicate the total SLP field with respect to which the difference is taken. Values less than the global mean are displayed as dashed contours.

The zonal-mean zonal wind response and the decomposition thereof are presented in Fig. 9. The total response (2x–1x) is characterized by a rise and an intensification of the subtropical jet and a poleward shift of the barotropic eddy-driven jet, epitomizing the response of the zonal wind in typical coupled climate models. This total response can be reasonably reproduced by both the wind stress overriding experiments and the total wind overriding experiments (not shown). Thus, the total response can be further partitioned into the effect of warming and the feedbacks due to changing wind stress (for the W group) or changing both wind stress and wind speed (for the WE group). Again, as for the case of precipitation and SLP, the zonal wind response in midlatitudes is also dominated by the effect of 2xCO2 warming in the absence of these feedbacks. However, as will be elaborated later, both wind stress and WES feedbacks play a significant, though weak, reinforcing role in the Southern hemispheric wind response. Whereas, the same may not be said for the Northern Hemispheric zonal wind response. Further inspection of the three-dimensional wind structures in experiment WE2x_2x–WE1x_2x suggests that weakening of the zonal wind in the Northern Hemisphere comes mainly from the midlatitude ocean sectors, possibly resulted from a weakened total SST gradient on the equatorward side of the midlatitude SST warming (see Fig. 4d).

Fig. 9.

As in Fig. 8 but for the response of zonal-mean zonal wind (m s−1). The wind field for comparison in each panel is also overlaid.

Fig. 9.

As in Fig. 8 but for the response of zonal-mean zonal wind (m s−1). The wind field for comparison in each panel is also overlaid.

4. Time scale and the response of the SAM

Several studies have investigated the possible role of ocean dynamics in the internal variability of the SAM and a positive oceanic feedback to the SAM has been suggested (e.g., Sen Gupta and England 2007; Watterson 2001). However, few have studied the possible feedback of the wind-driven ocean circulation to the SAM-like wind or sea level pressure response to increasing greenhouse gas forcing. As the wind stress feedback has been successfully isolated through the wind stress override experiment, we can now interpret the total December–February (DJF) SAM-like response (Fig. 10a) as the linear addition of the direct response to 2xCO2 forcing (Fig. 10b) and the wind stress feedback to the former (Fig. 10d), the latter projecting positively onto the SAM-like pattern, indicative of a positive feedback from the modified wind-driven ocean circulation. We chose to show DJF because this is the season in which the SAM is well defined and, hence, is most extensively studied. Inspection of the annual-mean sea level pressure response shows a qualitatively similar partition as for DJF. The SAM mainly reflects the meridional shift of the eddy-driven jet and an alternative way to measure the response of, and feedback to, the SAM is the shift of the location of the maximum surface westerly wind. Figure 11 depicts the positions of the surface wind against the corresponding lag-1 correlation coefficient of the monthly time series of the SAM under different coupling and forcing conditions. The time series of SAM is calculated as the first principle component of the deseasonalized monthly SLP poleward of 20°S. The lag-1 correlation may be the best that we can come up with to estimate persistence of the internal variability of the SAM using monthly data. The W overriding counterparts of the 1x and 2x experiments both show reduced persistence in the monthly annular mode variability, suggesting that coupling through the wind stress plays a positive role in maintaining the anomalous SAM phase in its life cycle. Although the wind stress feedback augments the lag-1 correlation by only 0.07 (from 0.44 to 0.51 for the 1x case and from 0.38 to 0.45 for the 2x case), the increase is still significant at a 10% significance level according to the Student’s t test, thanks to the large degree of freedom (190 for our cases). What is remarkable about the diagram of Fig. 11 is that the total shift of the surface westerly from the fully coupled experiments (2x-1x) can be decomposed into two additive components: the shift due to the CO2 forcing alone (W1x_2x–W1x_1x) plus the wind stress feedback (W2x_2x–W1x_2x). If one defines the feedback strength as the ratio of the feedback to the imposed wind anomalies (following Delworth and Zeng 2008), the wind stress feedback strength can be quantified as 0.36/1.56 ≈ 23%. However, the same quantification of feedback may not be extended to the WES feedback since the linearity does not hold well for the decomposition through the WE experiments. Instead, the wind speed override leads to a conspicuous bias in the eddy-driven westerly and a drastic reduction in persistence of the SAM. The latter suggests that the reduced thermal damping through the WES mechanism also serves as an important positive feedback to intraseasonal SAM variability. Through these two suites of overriding experiments, we demonstrate that both feedbacks from the wind-driven ocean circulation and the WES can play a positive role in reddening the spectrum of SAM variability, as do synoptic eddies in the atmosphere.

Fig. 10.

Southern Hemisphere DJF sea level pressure difference (Pa) in (a) 2x–1x; (b) W1x_2x–W1x_1x; (c) W2x_2x–W1x_1x; and (d) W2x_2x–W1x_2x.

Fig. 10.

Southern Hemisphere DJF sea level pressure difference (Pa) in (a) 2x–1x; (b) W1x_2x–W1x_1x; (c) W2x_2x–W1x_1x; and (d) W2x_2x–W1x_2x.

Fig. 11.

Relationship between the location of the mean surface westerly maximum and the lag-1 correlation of the monthly time series of the SAM. The red frame is added to assist in gauging linearity of the wind stress overriding results.

Fig. 11.

Relationship between the location of the mean surface westerly maximum and the lag-1 correlation of the monthly time series of the SAM. The red frame is added to assist in gauging linearity of the wind stress overriding results.

Recent analyses of the CMIP3 coupled models (Kidston and Gerber 2010; Barnes et al. 2010) showed that the magnitude of a forced SAM-like response scales with the time scale of natural SAM variability in a manner reminiscent of a fluctuation–dissipation system (Leith 1975; Gritsun and Branstator 2007; Gritsun et al. 2008; Ring and Plumb 2008). Will the enhanced persistence in the SAM due to the coupling with the ocean manifest itself in the response to increased CO2 forcing? To answer this question, here we treat all of the simulations with 1xCO2 concentration as control climates (four of them: one fully coupled control plus three overriding ones, see the legend of Fig. 12) and for each compare the surface westerly shift in response to CO2 doubling against the lag-1 correlation coefficient of the SAM index for the corresponding control climate. The result, though only consisting of four cases, indeed corroborates the FD thinking above; that is, a control climate with a more persistent SAM tends to respond with a greater poleward shift in the surface westerly wind. Yet, it remains to be seen if this result will still hold for the CCSM3-FDM as multiple equilibrium simulations with the FDM remain formidably expensive even for such a coarse resolution.

Fig. 12.

Relationship between the lag-1 correlation of the monthly time series of the SAM in the control climates and the corresponding response of the surface westerly shift to 2xCO2 forcing. The four purported control climates are indicated in the legend, each subjected to 2xCO2 forcing.

Fig. 12.

Relationship between the lag-1 correlation of the monthly time series of the SAM in the control climates and the corresponding response of the surface westerly shift to 2xCO2 forcing. The four purported control climates are indicated in the legend, each subjected to 2xCO2 forcing.

5. Summary and discussion

The climate feedbacks due to the changing wind-driven ocean circulation and air–sea latent heat flux in the equilibrium climate response to greenhouse gas forcing are investigated through overriding the wind stress and/or wind speed at the air–sea interface. The model used for this purpose is an upper-ocean dynamics version of the NCAR CCSM3 climate model at a rather coarse resolution; this effort should serve as proof of the concept for further more comprehensive investigation by applying similar techniques to more contemporary models. Notwithstanding the obvious caveat of the coarse resolution, this approach proves to be fruitful and many findings can be made from the two sets of overriding runs with wind stress and total wind, respectively. On the technical aspect, we find that the override can cause unintentional bias in the model climate state, as one reasonably suspects from the disruption of the physical consistency. While the bias due to the wind stress override is relatively benign compared to that due to the wind speed override, both must be carefully accounted for in order to quantitatively estimate these wind-related feedbacks. The success of implementating the override technique helps test antecedent hypotheses regarding formation of the tropical SST pattern and unveils the feedback effects from the wind-driven ocean dynamics and latent heat flux in the response of the climate system to 2xCO2 forcing. The main conclusions include the following.

  • The wind–evaporation–SST feedback is the main formation mechanism for the pattern of the tropical SST response to CO2 forcing, supporting the WES hypothesis proposed by Xie et al. (2010). By corollary, the weakened equatorial easterlies and related upper ocean feedback are not responsible for the accentuated equatorial SST warming over the central equatorial Pacific.

  • The warming minimum over the “main development region” of the tropical cyclones in the Atlantic is predominately controlled by the WES feedback, as the evaporation–SST denominator effect is largely offset by the oceanic dynamical thermostat effect.

  • The wind-related feedbacks also play an essential role in the minimum SST warming near 55°S.

  • WES is also the dominant mechanism for shaping the tropical precipitation response in the ocean, while both WES and wind stress feedback only play a secondary role in the pattern of the midlatitude hydrological cycle.

  • Wind overriding, especially the wind speed overriding, reduces substantially the persistence time scale of the SAM, implicating the supportive roles of air–sea coupling in sustaining the life cycle of the SAM.

  • The change in wind-driven ocean circulation is found to feed back positively to the SAM-like wind response. The poleward shift in the Southern Hemispheric surface westerly wind scales with the time scale of the intrinsic variability of the SAM under different control climate conditions in a manner conforming to the fluctuation–dissipation theorem.

We caution that some of the above results might be model and resolution dependent. For example, the tropical Pacific cooling caused by the wind stress feedback remains to be verified with state-of-the-art coupled climate models at finer resolution. As far as the oceanic feedback in the midlatitudes is concerned, the quantitative details are very likely to differ depending on the different models, resolutions, UOM or FDM, and eddy-permitting or not. For example, Farneti et al. (2010) compared the response between a coarse-resolution version and a fine resolution, eddy-permitting version of a same climate model, and found that active eddy dynamics buffers the oceanic response and thereby substantially reduces the oceanic feedback to the atmospheric perturbation. As such, the wind stress feedback investigated here is very likely overestimated with a coarse-resolution ocean component. Further experiments in an intermodal comparison format are necessary to advance our understanding of the effects of the ocean dynamical feedback and air–sea interaction in the coupled response to GHG forcing, especially when details in the response of the ocean are concerned.

Meanwhile, these partial coupling experiments also provide a useful framework for isolating the sources of many robust features in the climate response to greenhouse gas forcing. By partitioning the total response into a component that represents the direct response to CO2 forcing plus those due to feedbacks, we may interpret the latter as the consequence of the former. As such, the complex processes that lead up to the final response may be dissected into much simpler subprocesses, each of those is more readily to be understood in terms of basic physical and dynamical principles. Within this hierarchy comprising a slab ocean, partially coupled, and fully coupled models, one interesting example is the case of the WE overriding in which, since the WES feedback and wind stress feedback are suppressed, the SST response is potentially predictable through the information of the evaporation–SST denominator and the ocean dynamical thermostat , both being inferable from the climatological state of the surface variables and ocean circulation. The WES and wind stress feedbacks can then be further deduced as feedbacks to the response under WE overriding. Moreover, since the response under WE overriding is dependent on the model climatology, the repercussions of model bias (especially that of the ocean component) are more readily to be diagnosed and corrected in this partially coupled system than the fully coupled climate model.

Acknowledgments

The experiment design of this paper benefited greatly from the conversation with William Large and Clara Deser of NCAR. The authors are grateful to Gokhan Danabasoglu for providing the data of the long integrations with the full CCSM3-FDM. Gang Chen suggested the analysis of the lag-correlation time scale of the SAM and its relationship with the shift of the midlatitude westerlies. Comments from Shang-Ping Xie on the early version of manuscript are also gratefully acknowledged. Comments from three anonymous reviewers helped improve the manuscript substantially. Discussion with Edwin Schneider and Ben Kirtman helped us understand the bias caused by the wind overriding. V. Krishnamurthy and Abraham Solomon provided editorial assistance during the revision of the manuscript. This research is supported by the COLA omnibus fund from NSF Grant 830068, NOAA Grant NA09OAR3210058, and NASA Grant NNX09AN50G. J. Lu is also partially supported by the NSF Grant AGS-1064045. The experiments were conducted by B. Zhao when he was a postdoctoral research scientist at Los Alamos National Laboratory, supported by IGPP Award 1544 from the Los Alamos branch of the Institute of Geophysics and Planetary Physics.

APPENDIX

Energy Budget Analysis

To shed light on the roles of ocean heat transport and air–sea fluxes in maintaining the tropical SST patterns, we performed a simple energy budget analysis for the sea surface for three overriding experiments: WE1x_2x–WE1x_1x, WE2x_2x–WE1x_2x, and W2x_2x–W1x_2x. For each of these experiments, since both the upper ocean and atmosphere are equilibrated, the surface energy budget is balanced between the change in net downward fluxes of energy (comprised of net radiative fluxes , sensible heat , and latent heat ) and divergence of the vertically integrated oceanic heat transport . It can be written as follows:

 
formula

The results are presented in Figs. A1A3 for the aforementioned overriding experiments, respectively. Note that the convergence of oceanic heat transport , which is the negative of , is shown so that the positive value in each panel indicates warming to the SST. The sensible heat is at least an order smaller than the latent heat over the tropical oceans and, hence, is ignored in the discussion of the tropical SST formation in the main text. Generally speaking, the radiative term is smaller than the other two terms, and thus the leading-order balance is between the change of latent heat flux and that of the oceanic heat divergence as it is in the climatological balance. Lastly, adding Fig. A1 to Fig. A2, one can recover accurately the four-term budget for the fully coupled case 2x -1x (not shown).

Fig. A1.

Terms involved in the surface energy budget for the case WE1x_2x-WE1x_1x (response to 2 × CO2 in the absence of wind stress and WES feedbacks): (a) vertically integrated oceanic energy convergence ; (b) latent heat flux ; (c) sensible heat flux ; and (d) net downward radiative flux . Unit is W m−2.

Fig. A1.

Terms involved in the surface energy budget for the case WE1x_2x-WE1x_1x (response to 2 × CO2 in the absence of wind stress and WES feedbacks): (a) vertically integrated oceanic energy convergence ; (b) latent heat flux ; (c) sensible heat flux ; and (d) net downward radiative flux . Unit is W m−2.

Fig. A2.

As in Fig. A1 but for the case WE2x_2x-WE1x_2x (wind stress and WES feedbacks).

Fig. A2.

As in Fig. A1 but for the case WE2x_2x-WE1x_2x (wind stress and WES feedbacks).

Fig. A3.

As in Fig. A1 but for the case W2x_2x-W1x_2x (wind stress feedback).

Fig. A3.

As in Fig. A1 but for the case W2x_2x-W1x_2x (wind stress feedback).

REFERENCES

REFERENCES
Barnes
,
E. A.
,
D. L.
Hartmann
,
D. M.
Frierson
, and
J.
Kidston
,
2010
:
The effect of latitude on the persistence of eddy-driven jets
.
Geophys. Res. Lett.
,
L11804
,
doi:10.1029/2010GL043199
.
Chen
,
G.
,
J.
Lu
, and
D. M. W.
Frierson
,
2008
:
Phase speed spectra and the latitude of surface westerlies: Interannual variability and global warming trend
.
J. Climate
,
21
,
5942
5959
.
Clement
,
A. C.
, and
R.
Seager
,
1999
:
Climate and tropical oceans
.
J. Climate
,
12
,
3393
3401
.
Clement
,
A. C.
,
R.
Seager
,
M. A.
Cane
, and
S. E.
Zebiak
,
1996
:
An ocean dynamical thermostat
.
J. Climate
,
9
,
2190
2196
.
Collins
,
M.
, and
Coauthors
,
2005
:
El Niño- or La Niña-like climate change?
Climate Dyn.
,
24
,
89
104
.
Collins
,
W. D.
, and
Coauthors
,
2006a
:
The Community Climate System Model version 3 (CCSM3)
.
J. Climate
,
19
,
2122
2143
.
Collins
,
W. D.
, and
Coauthors
,
2006b
:
The formulation and atmospheric simulation of the Community Atmosphere Model Version 3 (CAM3)
.
J. Climate
,
19
,
2144
2161
.
Danabasoglu
,
G.
, and
J. C.
McWilliams
,
2000
:
An upper-ocean model for short-term climate variability
.
J. Climate
,
13
,
3380
3411
.
Danabasoglu
,
G.
, and
P. R.
Gent
,
2008
:
Equilibrium climate sensitivity: Is it accurate to use a slab ocean model?
J. Climate
,
22
,
2494
2499
.
Delworth
,
T. L.
, and
F.
Zeng
,
2008
:
Simulated impact of altered Southern Hemisphere winds on the Atlantic meridional overturning circulation
.
Geophys. Res. Lett.
,
35
,
L20708
,
doi:10.1029/2008GL035166
.
DiNezio
,
P. N.
,
A. C.
Clement
,
G. A.
Vecchi
,
B. J.
Soden
,
B. P.
Kirtman
, and
S.-K.
Lee
,
2009
:
Climate response of the equatorial Pacific to global warming
.
J. Climate
,
22
,
4873
4892
.
DiNezio
,
P. N.
,
A. C.
Clement
, and
G. A.
Vecchi
,
2010
:
Reconciling differing views of tropical Pacific climate change
.
Eos, Trans. Amer. Geophys. Union
,
91
,
141
142
.
Farneti
,
R.
,
T. L.
Delworth
,
A. J.
Rosati
,
S. M.
Griffies
, and
F.
Zeng
,
2010
:
The role of mesoscale eddies in the rectification of the Southern Ocean response to climate change
.
J. Phys. Oceanogr.
,
40
,
1539
1557
.
Gritsun
,
A.
, and
G.
Branstator
,
2007
:
Climate response using a three-dimensional operator based on the fluctuation–dissipation theorem
.
J. Atmos. Sci.
,
64
,
2558
2575
.
Gritsun
,
A.
,
G.
Branstator
, and
A.
Majda
,
2008
:
Climate response of linear and quadratic functionals using the fluctuation–dissipation theorem
.
J. Atmos. Sci.
,
65
,
2824
2841
.
Hartmann
,
D. L.
, and
M. L.
Michelsen
,
1993
:
Large-scale effects on the regulation of tropical sea surface temperature
.
J. Climate
,
6
,
2049
2062
.
Hegerl
,
G. C.
, and
Coauthors
,
2007
: Understanding and attributing climate change. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 663–745.
Held
,
I. M.
, and
B. J.
Soden
,
2006
:
Robust response of the hydrological cycle to global warming
.
J. Climate
,
19
,
5686
5699
.
Held
,
I. M.
,
M.
Winton
,
K.
Takahashi
,
T. L.
Delworth
,
F.
Zeng
, and
G. K.
Vallis
,
2010
:
Probing the fast and slow components of global warming by returning abruptly to preindustrial forcing
.
J. Climate
,
23
,
2418
2427
.
Jin
,
F.-F.
,
1997
:
An equatorial ocean recharge paradigm for ENSO. Part II: A stripped-down coupled model
.
J. Atmos. Sci.
,
54
,
830
847
.
Johnson
,
N. C.
, and
S.-P.
Xie
,
2010
:
Changes in the sea surface temperature threshold for tropical convection
.
Nat. Geosci.
,
3
,
842
845
.
Kidston
,
J.
, and
E. P.
Gerber
,
2010
:
Intermodel variability of the poleward shift of the austral jet stream in the CMIP3 integrations linked to biases in 20th century climatology
.
Geophys. Res. Lett.
,
37
,
L09708
,
doi:10.1029/2010GL042873
.
Knutson
,
T. R.
, and
S.
Manabe
,
1995
:
Time-mean response over the tropical Pacific to increased CO2 in a coupled ocean–atmosphere model
.
J. Climate
,
8
,
2181
2199
.
Kushner
,
P. J.
,
I. M.
Held
, and
T. L.
Delworth
,
2001
:
Southern Hemisphere atmospheric circulation response to global warming
.
J. Climate
,
14
,
2238
2249
.
Leith
,
C. E.
,
1975
:
Climate response and fluctuation dissipation
.
J. Atmos. Sci.
,
32
,
2022
2026
.
Leloup
,
J.
, and
A.
Clement
,
2009
:
Why is there a minimum in projected warming in the tropical North Atlantic Ocean?
Geophys. Res. Lett.
,
36
,
L14802
,
doi:10.1029/2009GL038609
.
Liu
,
Z.
,
S. J.
Vavrus
,
F.
He
,
N.
Wen
, and
Y.
Zhang
,
2006
:
Rethinking tropical ocean response to global warming: The enhanced equatorial warming
.
J. Climate
,
18
,
4684
4700
.
Lorenz
,
D. J.
, and
E. T.
DeWeaver
,
2007a
:
The tropopause height and the zonal wind response to global warming in the IPCC scenario integrations
.
J. Geophys. Res.
,
112
,
D10119
,
doi:10.1029/2006JD008087
.
Lorenz
,
D. J.
, and
E. T.
DeWeaver
,
2007b
:
The response of the extratropical hydrological cycle to global warming
.
J. Climate
,
20
,
3470
3484
.
Lu
,
J.
,
G.
Vecchi
, and
T.
Reichler
,
2007
:
Expansion of the Hadley cell under global warming
.
Geophys. Res. Lett.
,
34
,
L06805
,
doi:10.1029/2006GL028443
.
Lu
,
J.
,
G.
Chen
, and
D.
Frierson
,
2008
:
Response of the zonal mean atmospheric circulation to El Niño versus global warming
.
J. Climate
,
21
,
5835
5851
.
Lu
,
J.
,
G.
Chen
, and
D. M. W.
Frierson
,
2010
:
The position of the midlatitude storm track and eddy-driven westerlies in aquaplanet AGCMs
.
J. Atmos. Sci.
,
17
,
3984
4000
.
Meehl
,
G. A.
, and
Coauthors
,
2006
:
Climate change projections for the twenty-first century and climate change commitment in the CCSM3
.
J. Climate
,
19
,
2597
2616
.
Meehl
,
G. A.
, and
Coauthors
,
2007
: Global climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 747–845.
O’Gorman
,
P. A.
, and
T.
Schneider
,
2009
:
Scaling of precipitation extremes over a wide range of climates simulated with an idealized GCM
.
J. Climate
,
22
,
5676
5685
.
Previdi
,
M.
, and
B. G.
Liepert
,
2007
:
Annular modes and Hadley cell expansion under global warming
.
Geophys. Res. Lett.
,
34
,
L22701
,
doi:10.1029/2007GL031243
.
Ring
,
M. J.
, and
R. A.
Plumb
,
2008
:
The response of a simplified GCM to axisymmetric forcings: Application of the fluctuation–dissipation theorem
.
J. Atmos. Sci.
,
65
,
3880
3898
.
Riviere
,
G.
,
2011
:
A dynamical interpretation of the poleward shift of the jet streams in global warming scenarios
.
J. Atmos. Sci.
,
68
,
1253
1272
.
Sato
,
Y.
,
S.
Yukimoto
,
H.
Tsujino
,
H.
Ishizaki
, and
A.
Noda
,
2006
:
Response of North Pacific ocean circulation in a Kuroshio-resolving ocean model to an Arctic Oscillation (AO)-like change in Northern Hemisphere atmospheric circulation due to greenhouse-gas forcing
.
J. Meteor. Soc. Japan
,
84
,
295
309
.
Seager
,
R.
, and
R. G.
Murtugudde
,
1997
:
Ocean dynamics, thermocline adjustment, and regulation of tropical SST
.
J. Climate
,
10
,
521
534
.
Sen Gupta
,
A.
, and
M. H.
England
,
2007
:
Coupled ocean–atmosphere feedback in the southern annular mode
.
J. Climate
,
20
,
3677
3692
.
Vecchi
,
G. A.
, and
B. J.
Soden
,
2007a
:
Global warming and the weakening of the tropical circulation
.
J. Climate
,
20
,
4316
4340
.
Vecchi
,
G. A.
, and
B. J.
Soden
,
2007b
:
Increased tropical Atlantic wind shear in model projections of global warming
.
Geophys. Res. Lett.
,
34
,
doi:10.1029/2006GL028905
.
Vecchi
,
G. A.
,
B. J.
Soden
,
A. T.
Wittenberg
,
I. M.
Held
,
A.
Leetmaa
, and
M. J.
Harrison
,
2006
:
Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing
.
Nature
,
441
,
doi:10.1038/nature04744
.
Vecchi
,
G. A.
,
A.
Clement
, and
B. J.
Soden
,
2008
:
Examining the tropical Pacific’s response to global warming
.
Eos, Trans. Amer. Geophys. Union
,
89
,
9
,
doi:10.1029/2008EO090002
.
Watterson
,
I. G.
,
2001
:
Zonal wind vacillation and its interaction with the ocean: Implication for interannual variability and predictability
.
J. Geophys. Res.
,
106
(
D20
),
23 965
23 975
.
Watterson
,
I. G.
,
2003
:
Effects of a dynamic ocean on simulated climate sensitivity to greenhouse gases
.
Climate Dyn.
,
21
,
197
209
,
doi:10.1007/s00382-003-0326-4
.
Watterson
,
I. G.
,
2009
:
Components of rainfall and temperature anomalies and change associated with modes of the Southern Hemisphere
.
Int. J. Climatol.
,
29
,
809
826
,
doi:10.1002/joc.1772
.
Williams
,
G. P.
,
2006
:
Circulation sensitivity to tropopause height
.
J. Atmos. Sci.
,
63
,
1954
1961
.
Wu
,
Y.
,
M.
Ting
,
R.
Seager
,
H.-P.
Huang
, and
M. A.
Cane
,
2010
:
Changes in storm tracks and energy transport in a warmer climate simulated by the GFDL CM2.1 model
.
Climate Dyn.
,
37
,
53
72
,
doi:10.1007/s00382-010-0776-4
.
Xie
,
S.-P.
, and
G.
Philander
,
1994
:
A coupled ocean-atmosphere model of relevance to the ITCZ in the eastern Pacific
.
Tellus
,
46A
,
340
350
.
Xie
,
S.-P.
,
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
.
Yeager
,
S. G.
,
C. A.
Shields
,
W. G.
Large
, and
J. J.
Hack
,
2006
:
The low-resolution CCSM3
.
J. Climate
,
19
,
2545
2566
.
Yin
,
J. H.
,
2005
:
A consistent poleward shift of the storm tracks in simulations of 21st century climate
.
Geophys. Res. Lett.
,
32
,
L18701
,
doi:10/1029/2005GL023684
.
Zhao
,
B.
,
J. C.
McWilliams
, and
P. W.
Jones
,
2011
: The Los Alamos Upper Ocean Model. Los Alamos National Laboratory Tech. Rep. LA-UR-11-11622, 44 pp. [Available online at http://oceans11.lanl.gov/twiki/pub/Cosim/CosimPublications/UpperOceanModelZhao.pdf.]

Footnotes

1

A calculation of similar to that in Fig. 4 of Leloup and Clement (2009) was repeated for the control simulation with the slab verion of the CCSM3. However, the agreement with the actual SST response deteriorates when the actual latent heat flux is used to replace the hypothetical, constant , as is also evident in their Fig. 4b. We argue that this is caused primarily by neglecting the WES term.