1. Introduction
The El Niño–Southern Oscillation (ENSO) is the most significant interannual variability mode in the tropical Pacific climate system. Studies of ENSO have made great progress, reaching the stage where long-lead-time forecasts can be made routinely using various coupled ocean–atmosphere models (e.g., Cane and Zebiak 1985; Barnett et al. 1993; Latif et al. 1998; Ji et al. 1998; Barnston et al. 1999; Kirtman et al. 2002; DeWitt 2005; Zhang et al. 2005; see the summary of model ENSO forecasts at the International Research Institute for Climate and Society Web site: http://iri.columbia.edu/climate/ENSO/currentinfo/update.html).
However, systematic errors and large uncertainties still exist, with significant model tropical biases in simulation and prediction of ENSO (e.g., Chen et al. 2000; Large and Danabasoglu 2006; Wittenberg et al. 2006; Misra et al. 2008). Extensive efforts have been made to identify the potential sources for tropical biases in coupled ocean–atmosphere models. Previously, the roles of wind and heat flux forcing and feedback have been the main focus (e.g., Bjerknes 1969; Zebiak and Cane 1987; Jin and An 1999; Meehl et al. 2001). Here, we consider the extent to which freshwater flux (FWF), another important atmospheric forcing to the ocean, can play a role in interannual biases often seen in coupled ocean–atmosphere models of the tropical Pacific.
Recently, physical understanding and modeling of the roles of FWF in climate variability have been significantly advanced. It has been demonstrated that FWF forcing and its related salinity fields can play an active role in maintaining the Pacific climate and its low-frequency variability through its effect on the horizontal pressure gradients, the stratification, and the equatorial thermocline (e.g., Miller 1976; Lukas and Lindstrom 1991; Delcroix and Hénin 1991; Huang and Schmitt 1993; Sprintall and Tomczak 1992; Schneider and Barnett 1995; Vialard and Delecluse 1998; Kessler 1999; Maes 2000; Ballabrera-Poy et al. 2002; Lagerloef 2002; Levitus et al. 2005; Boyer et al. 2005; Huang et al. 2008; Cravatte et al. 2009). Previous modeling studies have been mostly conducted in forced ocean-only context. For example, sensitivity studies performed to examine the ocean response to a prescribed FWF perturbation have identified large effects on SSTs (e.g., Cooper 1988; Carton 1991; Reason 1992; Murtugudde and Busalacchi 1998; Yang et al. 1999; Huang and Mehta 2004, 2005; Fedorov et al. 2004; Huang and Mehta 2005). When the atmosphere and the ocean are allowed to couple, new additional processes come into play since the FWF forcing-induced changes in SST can further feedback to the atmosphere. More recently, Zhang and Busalacchi (2009) examine the role of anomalous FWF forcing in interannual variability using a hybrid-coupled model (HCM) of the tropical Pacific. To represent feedbacks between interannual variations in FWF and SST in an interactive way, a simple prognostic model for interannually anomalous FWF is constructed empirically using a singular value decomposition (SVD) analysis (e.g., Bretherton et al. 1992). When the covarying patterns of FWF and SST are incorporated in the HCM, a positive feedback is identified in the tropical Pacific. It has been demonstrated that FWF forcing can have a significant modulating effect on ENSO amplitude, acting to enhance interannual variability. Because of the positive feedback involved and the significant effects on interannual variability, FWF forcing needs to be taken into account adequately in diagnostic analyses and modeling studies for the tropical Pacific climate system.
At present, uncertainties and biases in estimating precipitation P and evaporation E make it difficult to accurately represent FWF forcing in models. Various approximations to FWF are often made in many previous modeling studies. For example, some neglect the contribution of E to FWF (e.g., Zhang et al. 2006a); some include interannually anomalous P only (e.g., Vialard and Delecluse 1998). As such, this forcing and related feedback can be misrepresented. In particular, FWF forcing has not been included in most simplified coupled ocean–atmosphere models used for ENSO simulation and prediction (e.g., Zebiak and Cane 1987; Barnett et al. 1993; Syu et al. 1995; Zhang et al. 2003, 2005). On the other hand, comprehensive coupled general circulation models (CGCMs) can take into account this forcing, but the so-called double ITCZ problem is still a prominent tropical bias in many CGCM simulations, indicating that P has not been captured realistically over the tropical Pacific. As such, the FWF forcing-induced feedback can be misrepresented in these CGCMs. Another common practice in ocean modeling is the use of relaxation-type boundary conditions for sea surface salinity (SSS) and SST. While heat flux acts to damp interannual SST variability in the tropical Pacific, FWF actually tends to amplify variability of SSS and SST (Zhang and Busalacchi 2009). This indicates that the relaxation-type boundary conditions are not adequate in representing FWF forcing. Also, FWF adjustments are adopted in some coupled models in order to reduce systematic biases in SSS simulations.
Clearly, various approximations are often involved in representing FWF forcing in ocean and coupled ocean–atmosphere models, including the neglect of interannual variability of E and/or P. As demonstrated in Zhang and Busalacchi (2009), the FWF forcing-induced feedback presents a positive effect on interannual SST variability, in contrast to heat flux forcing, which indicates a negative feedback. It follows that if the ENSO-related FWF–SST feedback is not included adequately, a bias can be introduced in representing FWF forcing and related feedback, potentially leading to biases and errors in ocean simulations, which can be further amplified by air–sea coupling. However, the extent to which this atmospheric forcing causes biases in coupled models is not known. How does a misrepresentation in FWF forcing contribute to a bias in SST and SSS simulations in coupled ocean–atmosphere models of the tropical Pacific? How sensitive is the coupled ocean–atmosphere modeling of the region to correct interannual representations of FWF forcing? Are the biases induced significant in terms of ENSO simulations so that care should be taken adequately? What processes are involved in the FWF bias-induced effects?
In this article, we continue to investigate these remaining issues in more detail from a coupled ocean–atmosphere modeling perspective. Previously, we have constructed an HCM consisting of an ocean general circulation model (OGCM) and an empirical atmospheric model for interannual wind stress variability (Zhang et al. 2006a). As shown in Zhang et al. (2006a), this HCM can reasonably well simulate interannual variability in the tropical Pacific associated with ENSO. However, large errors and biases still exist. In particular, it significantly underestimates SST variability in the eastern equatorial Pacific and SSS variability in the central basin. These systematic biases can also be found in other model simulations (e.g., Syu et al. 1995; Barnett et al. 1993; Latif et al. 2001), and remain a major challenge to short-term ENSO-related prediction studies. To identify sources for these errors and to improve simulations of SST and SSS particularly, extensive efforts have been taken, including an embedding approach to optimally represent the entrainment temperature Te and salinity Se of subsurface waters into the mixed layer (e.g., Zhang et al. 2006a,b). Recently, based on understanding of the FWF forcing effects (e.g., Zhang and Busalacchi 2009), it has been realized that FWF forcing is misrepresented in our previous HCM (Zhang et al. 2006a). Here, as a test bed, we will use this HCM to illustrate the role of FWF forcing in tropical bias in interannual simulations of the tropical Pacific.
The paper is organized as follows. Section 2 describes the model and data used. Section 3 deals with model experiment design. A reference run is analyzed in section 4 to understand the role of anomalous FWF forcing in interannual variability, followed by sensitivity experiments in section 5 to demonstrate the links between interannual biases and the way anomalous FWF forcing is represented. The paper is concluded in section 6.
2. Model descriptions
We use a hybrid-coupled model of the tropical Pacific climate system (Fig. 1). The HCM consists of a layer OGCM and a simplified atmospheric representation of three forcing fields to the ocean. Empirical models for interannual wind stress and FWF variability are constructed using SVD analysis techniques from historical data; heat flux is estimated using an advective atmospheric mixed layer (AML) model (Seager et al. 1995). Thus, the HCM has three atmospheric forcings to the ocean: wind stress, heat flux, and freshwater flux. These are briefly described in this section, as well as observational and model-based datasets used to construct these empirical models. For more details, see Zhang et al. (2006a) and Zhang and Busalacchi (2009).
a. An ocean general circulation model
The ocean GCM used is based on a reduced gravity, primitive equation, sigma coordinate model of Gent and Cane (1989), which was developed specifically for studying the coupling between the dynamics and the thermodynamics of the upper tropical ocean. Several related efforts have improved this ocean model significantly (e.g., Chen et al. 1994; Murtugudde and Busalacchi 1998; Murtugudde et al. 2002). The details of the OGCM used in this work can be found in Murtugudde and Busalacchi (1998), Hackert et al. (2001), Murtugudde et al. (2002), Zhang et al. (2006a), and Zhang and Busalacchi (2009).
b. Atmospheric models for interannual variability
The atmospheric forcing fields to the ocean consist of three components (Fig. 1): wind stress, heat flux, and freshwater flux. As mentioned above, heat flux is calculated using the AML model (Seager et al. 1995). This heat flux parameterization allows for a realistic representation of the feedbacks between mixed layer depths (MLDs), SSTs, and the heat fluxes (e.g., Murtugudde et al. 1996; Murtugudde and Busalacchi 1998). Two empirical feedback models for interannual variability of wind stress and FWF are constructed from historical data.
Various observational and model-based data are used for constructing the empirical models and validate model simulations. The observed SST data are from Reynolds et al. (2002); observed SSS data are from the in situ measurements in the tropical Pacific provided by the Institut de Recherche pour le Developpement (IRD)/ECOP, France (e.g., Delcroix and Hénin 1991; Maes 2000; Ballabrera-Poy et al. 2002). Data of wind stress, evaporation, and precipitation used to construct empirical models are from the ensemble mean of a 24-member ECHAM 4.5 atmospheric general circulation model (AGCM) simulations during the period 1950–99, forced by observed SST anomalies. The ECHAM4.5 AGCM is developed by the Max Planck Institute for Meteorology (MPI) and the European Centre for Medium-Range Weather Forecasts; see Roeckner et al. (1996) for details. Using the ensemble mean data from the ECHM4.5 AGCM is an attempt to enhance SST forced signals by reducing atmospheric noise. In addition, the preference of using the simulated P and E data from the ECHAM4.5 AGCM to satellite data is that the former has long time series consistently for the period 1950–99, while the latter can be strongly product-dependent (e.g., Xie and Arkin 1995). In particular, satellite-based P analysis data are only available since about 1979.
1) A wind stress (τ) model
The atmospheric wind stress model adopted in this work is statistical, specifically relating interannual wind stress variability τinter to large-scale SST anomalies SSTinter. The τinter model is constructed from an SVD of the covariance matrix that is calculated from time series of monthly mean SST and wind stress fields (e.g., Syu et al. 1995; Chang et al. 2001); the more details of SVD analyses can be found in Bretherton et al. (1992). In this work, we perform a combined SVD analysis for anomalies of SST, zonal and meridional wind stress components. To construct seasonally dependent models for τinter, the SVD analyses are performed separately for each calendar month (e.g., Zhang and Busalacchi 2005), and so the τinter model consists of 12 different submodels, one for each calendar month. To achieve reasonable amplitudes, the first five SVD modes are retained in estimating τinter fields from SST anomalies. This wind stress anomaly model has been extensively used for coupled ocean–atmosphere modeling studies in the tropical Pacific by Zhang and Busalacchi (2005) and Zhang et al. (2003, 2005, 2006a).
2) Dominant variability modes for E, P, and (P − E)
The total freshwater flux considered in the HCM can be separated into two parts: FWF = (P − E)clim + FWFinter. The first is its climatological part, which is represented by the differences between climatological P (Pclim) and E (Eclim) fields, in which Pclim is prescribed from observations (Xie and Arkin 1995) and Eclim is calculated using the AML model from simulated SST and other specified atmospheric fields. The second part (FWFinter) is concerned with interannual FWF variability associated with ENSO.
In the tropical Pacific, large interannual variations in E, P, and (P − E) have been observed during ENSO cycles, with their coherent relationships to SST (e.g., Xie and Arkin 1995; Yu and Weller 2007; Zhang and Busalacchi 2009). To determine their statistically dominant modes, SVD analyses of these respective fields with SST are performed. Figure 2 demonstrates the spatial patterns of the first SVD mode derived for SST, and for (P − E), E, and P. The SVD analyses are all performed during the periods 1963–96 (a total of 34 years of data) using SST data from Reynolds et al. (2002), and anomaly data for (P − E), E, and P from the ECHAM4.5 ensemble simulations. Note that the spatial pattern of (P − E) does not look like (P) − (E), which can be caused by the fact that the corresponding SVD decomposition is made based on the covariance matrix that is calculated from normalized time series of monthly mean P, E, and (P − E).
(i) Evaporation (E)
As has been shown in the SVD analysis (Fig. 2c), E exhibits large interannual variations over the tropical Pacific in association with ENSO. A positive correlation exists with SST (Fig. 2a). During El Niño, a warm SST anomaly in the central and eastern tropical Pacific is accompanied by a positive E anomaly in the Central Basin. This pattern largely reflects an increased latent heat flux out of the ocean due to the ENSO-induced warming (e.g., Yu and Weller 2007). During La Niña, a cold SST anomaly is associated with a negative E anomaly.
(ii) Precipitation (P)
Interannual variations in P are well documented in association with ENSO (e.g., Xie and Arkin 1995). For example, during the development of El Niño events, deep convection and heavy precipitation are observed to shift eastward along the equator. In the Central Basin, P increases significantly (Fig. 2d), but reduces moderately over the far western region (Fig. 2c). La Niña conditions show a reversed pattern, including a deficit of P in the central equatorial Pacific.
(iii) Precipitation minus evaporation (P − E)
In the tropical Pacific, a coherent correlation exists between interannual variations in SST and FWF, as represented by P minus E. As indicated in their SVD analyses (Fig. 2), interannual variations in SST and (P − E) show coherent covariability spatial patterns (e.g., Zhang and Busalacchi 2009). Because of the dominance of Pinter over Einter, anomalous (P − E)inter patterns are mainly attributed to those of P, with a small offset of E. For example, during ENSO cycles, (P − E)inter has been observed to undergo large interannual variations, with a coherent positive correlation with SST. During El Niño when SSTs are high in the central and eastern regions, both Einter and Pinter are anomalously positive. The differences between Pinter and Einter are positive (a freshwater gain into the ocean). When the tropical Pacific is in a cold condition during La Niña, Pinter is reduced, accompanied with a negative Einter anomaly. The resultant (P − E)inter is anomalously negative but with a reduced amplitude (a freshwater loss from the ocean).
(iv) The SVD-based empirical models for E, P, and (P − E)
As detailed in Zhang and Busalacchi (2009), using the derived spatial patterns of the SVD modes (Fig. 2), anomaly feedback models can be constructed to statistically relate (P − E)inter, Pinter, and Einter to large-scale SST anomalies. From the consideration of the sequence of the singular values and the reconstruction test of these anomaly fields from SST anomalies, the first two SVD modes are retained. Then, given an SST anomaly, variations in (P − E)inter, Pinter, and Einter can be calculated correspondingly. Note that all these FWF-related empirical models are constructed from the SVD analyses that are performed during the period 1963–96 on all time series data irrespective of season; as a result, the constructed models are season-independent, one model for all seasons. Results indicate that the empirical models can well capture interannual (P − E) variability when the two modes are retained, which account for about 70% variance in the central and eastern tropical Pacific.
One obvious assumption with this empirical approach is that interannual FWF variability in the tropical Pacific is determined nonlocally by SST anomalies, which are dominated by ENSO signals. As illustrated from observations (e.g., Xie and Arkin 1995; Yu and Weller 2007), interannual variations in SST are well correlated with P, E, and (P − E) over the tropical Pacific, with a dominant control of the former on the latter. Furthermore, as large SST anomalies are produced by ENSO, a response of FWF is quick and almost simultaneous. In particular, the temporal expansion coefficients of the SVD analyses clearly indicate that variations in FWF closely follow those in SST during ENSO evolution (Zhang and Busalacchi 2009). As a good indicator for changes in FWF induced by ENSO, the SST field can be used to construct an empirical model for interannual FWF variability. The other assumption is that, since the atmospheric fields, including P, E, and (P − E), are highly coupled with SST in the region, most of their variance can be explained by that associated with SST variations (i.e., the portion of variance not captured by coupled modes is relatively small). These arguments provide a physical basis for adopting empirical methods by which ENSO-induced interannual FWF variability can be reasonably well captured using a feedback model presenting its response to SSTs in the tropical Pacific.
c. Coupling procedure
As shown in Fig. 1, the OGCM is coupled to two empirical submodels for interannual variability of wind stress and FWF, while the advective AML model is used to estimate heat flux. In the context of the HCM, the climatological fields are specified from observations, including wind stress τclim, Pclim, solar radiation, cloud, and wind speed. The Eclim is estimated using the AML model from simulated climatological SST of the OGCM and other specified atmospheric fields.
The coupling between the atmospheric components and the OGCM is as follows (Fig. 1). At each time step, the OGCM calculates SSTs, which are averaged to obtain daily mean fields. The corresponding large-scale SST anomalies (SSTinter) are obtained relative to its SSTclim fields that are predetermined from the spinup OGCM run forced by observed τclim fields (monthly mean climatology SST fields are obtained but are temporally interpolated to daily fields for use in the coupling). The interannual anomaly fields (τinter and FWFinter components) are calculated using the corresponding empirical models from the SST anomalies. The FWFinter and τinter fields are updated every day from the corresponding large-scale SSTinter fields; heat flux is calculated using AML model from model SST and is updated every time step. Interannual variations in solar radiation, cloudiness, and wind speed are not taken into account in the heat flux computation. Note that two methods can be used to calculate anomalous E (Einter) one using the AML model as an evaporative component of the heat flux (Einter = E − Eclim), and other using the SVD-based empirical Einter model.
3. Model experiment designs
Previously, various approximations are often involved in representing FWF forcing in ocean and coupled ocean–atmosphere models, including the neglect of interannual variability of E and/or P (e.g., Zhang et al. 2006a; Vialard and Delecluse 1998). The relationships between biases represented in FWF and those induced in SST and SSS simulations have not been clearly demonstrated in a coupled ocean–atmosphere modeling context of the tropical Pacific. In this article, four sensitivity experiments are performed with different FWFinter representations (Table 1). In a case referred to as the interannual (P − E) forcing, interannual (P − E) variability is calculated using the empirical model (i.e., anomalous P and E are both taken into account in the FWFinter calculation) and is explicitly incorporated in the HCM. This SST-dependent, prognostic representation of FWFinter allows for an interactive feedback between FWF and SST during ENSO cycles. But in the other three cases, FWFinter is misrepresented somehow, giving rise to a bias in the FWF calculation. For example, the interannual E forcing is referred to as a run in which only Einter is taken into account in the FWFinter calculation; the interannual P forcing is referred as a run in which only anomalous P (Pinter) is taken into account; the climatological (P − E) forcing is for a run in which both anomalous P and E are not taken into account. When these approximations are taken, biases are induced in the representations of FWF forcing. For example, excluding an Einter contribution to FWFinter leads to the calculated FWFinter variability that can be exaggerated since the offset by E is not taken into account. On the other hand, neglecting a Pinter contribution to FWFinter tends to reverse the sign of the calculated FWFinter patterns (i.e., FWFinter = −Einter).
4. A simulation with interannual (P − E) forcing
A simulation is performed using the HCM in which the total FWF into the ocean is fully taken into account [i.e., FWFTotal = (P − E)clim + (P − E)inter]. Its anomalous part, (P − E)inter, is calculated using the SVD-based empirical model from large-scale SST anomalies. Thus, the evolving anomalous FWF patterns associated with ENSO (e.g., Fig. 2b) are explicitly incorporated in the HCM. This interannual (P − E) forcing run represents the most realistic case in all HCM simulations we perform in this article.
As shown in Zhang and Busalacchi (2009), the HCM can well simulate interannual oscillations with about a 4-year period. Examples are shown in Figs. 3 –5 for simulated SSS and SST fields. SSS has largest variability around the eastern edge of the warm pool near the date line (Figs. 3 –4) where the interannual FWF variability is also largest (Fig. 2b). Associated with ENSO events, the SSS front moves back and forth along the equator. For instance, during El Niño, the fresh zone in the west extends eastward to the central basin (Fig. 3a).
The El Niño conditions (Fig. 7) are characterized by a positive QS anomaly (into the ocean; Fig. 7b) and a negative SSS anomaly (a freshening; Fig. 7e) in the central and western regions, accompanied with a shallow mixed layer in the western and central regions (Fig. 7f), and negative QT (Fig. 7d) and QB (Fig. 7c) anomalies. The FWF forcing tends to modulate these El Niño–produced anomaly patterns. The positive FWF anomaly (Fig. 7b) has direct effects on SSS and QB (Fig. 6). The former acts to freshen the mixed layer in the central basin, which leads to a stable mixed layer. The latter, QB, has direct effects on MLDs and the entrainment. As shown and described in Zhang and Busalacchi (2009), interannual variations in QB are positively correlated with QT, but negatively with QS (Fig. 7). Since interannual variations in QT and QS are negatively correlated during ENSO cycles, their effects on QB tend to be compensated for each other. In the eastern equatorial Pacific, QT is a dominant contributor to QB; but in the central and western regions, QS can also make a significant contribution to QB (about 20%). Thus, as part of QB, the positive FWF anomaly (Fig. 7b) acts to compensate for the negative QT anomalies (Fig. 7d), leading to a reduced negative QB anomaly (Fig. 7c). The decreased negative QB anomalies (less negative because of the contribution of the positive QS anomaly to QB; Fig. 7c) tend to reduce MLDs, giving rise to less entrainment of subsurface water into the mixed layer. Thus the positive QS anomaly directly acts to stabilize the upper layer and reduce the strength of the mixing and entrainment at the base of the mixed layer. These positive FWF anomaly-induced oceanic processes tend to warm SSTs in the central basin, leading to an enhanced warming during El Niño. The opposite anomaly patterns and involved processes can be seen during La Niña phase (Fig. 8). The relationships between changes in SST and FWF indicate a positive feedback induced by FWF forcing during ENSO cycles. The amplifying effect of anomalous FWF forcing on SST is contrasted to the damping effect of heat flux on interannual time scales.
5. Model biases induced by misrepresented anomalous FWF forcing
As represented by a net difference, both anomalous P and E contribute to FWF variability. Because of their direct forcing effects on SSS and the compensating effects on QB, these fields of P, E, QT, and QS need to be taken into account coherently and consistently in coupled ocean–atmosphere models. It follows that if interannual variations in P and E are not represented simultaneously, a bias is introduced in the FWF calculation, which can modulate ocean processes that affect SSTs and further influence the atmosphere. In this section, sensitivity experiments will be performed using the same HCM with the only differences being in the specifications of anomalous FWF (Fig. 1; Table 1).
a. The interannual E forcing
When only Einter is taken into account in the FWFinter calculation (while P is kept to its climatology), the total FWF into the ocean can be approximated as FWFTotal = (P − E)clim − Einter. This run is referred to as the interannual E forcing run or the Einter run.
Two methods can be adopted to calculate Einter. One is to use the AML model (Einter = E − Eclim), in which Eclim is estimated from simulated SSTclim and other specified atmospheric climatological fields. This approximation to FWF has been adopted in our previous modeling studies (e.g., Zhang et al. 2006a; Ballabrera-Poy et al. 2007).
In this case, when the contribution of Pinter is not taken into account, FWFinter is determined solely by anomalous E (i.e., FWFinter = −Einter); a bias is introduced in calculating FWFinter. For example, during El Niño when P and E anomalies are both positive, their net differences should be positive in the central basin because of the dominance of P over E (i.e., an FWF gain from the ocean; Fig. 7b). When taking Einter only for FWFinter while excluding the Pinter contribution, the calculated FWFinter is artificially biased to be negative during El Niño (i.e., an FWF loss in the ocean instead). Thus, the sign of the FWFinter polarity is incorrectly reversed to be negative. Similarly, during La Niña, P and E anomalies are both negative, the resultant FWFinter should be negative (i.e., an FWF loss from the ocean; Fig. 8b). When the contribution of a negative Pinter anomaly to FWFinter is excluded, the calculated FWFinter is incorrectly reversed to be positive (i.e., a net FWF gain in the ocean). Therefore, this Einter forcing approximation to FWFinter acts to reverse the sign of FWFinter during ENSO cycles. As a result, the FWFinter induced feedback is also misrepresented to be negative. What are the effects on interannual variability simulations of SSS and SST in the HCM?
The HCM simulations are presented in Figs. 3 –5. Although anomalous FWF is miscalculated and the FWF forcing-induced feedback is misrepresented to be negative in the HCM, the basic features of simulated interannual variability in this run are still similar to those in the (P − E)inter run, including the oscillation periods. This indicates that interannual variability associated with ENSO in the tropical Pacific is predominantly driven by the SST–winds–thermocline feedback.
However, significant quantitative differences exist. The effects on the mean ocean state and its interannual variability can be seen in the SST and SSS fields (Figs. 3 –5 and Figs. 9 –10). Although the OGCM and the SVD-based wind stress model have not been modified at all, the simulated interannual variability from the Einter run is diminished significantly, as compared to that in the (P − E)inter run. (Note that the stronger negative SST anomalies appear during the years 24–26 in the Einter run that can be attributed to initial adjustment of the coupled system to the imposed FWF forcing). For example, temporal variability of SSS is weak in the central basin (Figs. 3b and 4b); the east–west migration of the SSS front is not as strong as in the (P − E)inter run (Fig. 3b).
The total SSS and SST fields are further compared in Fig. 9 for the two runs. SST in the central and eastern basin is high during El Niño but low during La Niña, while SSS in the western and central basin is low during El Niño but high during La Niña. The ranges of interannual SST and SSS variabilities in the Einter run are reduced compared with those in the (P − E)inter run. In particular, the Einter run shows no freshening in the central basin during El Niño since the El Niño–induced positive anomalous FWF is misrepresented to be negative in the central basin. The commonly used indices for Niño-3 SST, Niño-4 zonal wind stress, and Niño-4 SSS are further compared in Fig. 10 for the Einter run and the (P − E)inter run. Again, a systematic difference is evident. The ranges of these interannual anomalies in the Einter run are reduced during ENSO cycles. In particular, the amplitude of interannual SST anomalies is significantly underestimated in the eastern equatorial Pacific (Fig. 5b). This indicates that the anomalous FWF forcing can have pronounced effects not only on SSS (directly) but also SST and winds (indirectly).
Some biases in the Einter run can be explained in terms of the FWFinter bias induced. As shown above, ENSO-induced FWF anomalies should be positive in the central and western regions during El Niño (Fig. 7b), but negative during La Niña (Fig. 8b). When these anomalous FWF patterns are reversed in the Einter run during ENSO cycles, large errors are induced in the HCM simulations.
For example, during El Niño, FWFinter is misrepresented to be negative in the Einter run; the biased FWFinter forcing exerts direct effects on SSS and QB in the central basin. The former results in less freshening in the surface mixed layer (e.g., the years 2031–32 in Fig. 9a), which tends to enhance the vertical mixing of heat and entrainment at the base of mixed layer, leading to a decrease in SST in the central basin (e.g., the years 2031–32 in Fig. 9b). At the same time, the artificially induced negative FWFinter acts to increase the amplitude of the negative QB in the central region (i.e., becoming more negative), which leads to an increase in MLDs, with stronger entrainment of subsurface water into the mixed layer. These oceanic processes act to enhance their cooling effects on the surface layer in the central basin, reducing the warming produced by El Niño. The situations for La Niña operate in an opposite way. Therefore, the FWF-induced oceanic processes in the Einter run act in such a way that the warming during El Niño and cooling during La Niña are weakened in the central basin, leading to a reduced interannual variability. Clearly, the Einter run has misrepresented the FWF-induced positive feedback to be negative in the HCM, acting to damp interannual SST variability. As a result, ENSO cycles are damped significantly in the Einter run.
The Einter can also be calculated using the SVD-based empirical model (Fig. 2c). The effects are similar to those using the AML model shown above. That is, the FWF-induced positive feedback is artificially reversed to be negative and interannual variability is damped (figures not shown).
b. The interannual P forcing
Next, we consider another approximation in which only Pinter is taken into account in the FWFinter calculation [i.e., FWFTotal = (P − E)clim + Pinter], referred as the Pinter run. This approximation has been adopted previously in several forced ocean model simulations to examine the ocean responses to a prescribed precipitation anomaly (e.g., Reason 1992; Yang et al. 1999). Here we will examine the effects and the nature of the biases induced in a coupled ocean–atmosphere modeling context.
Based on the understanding of the FWFinter effects during ENSO cycle, some arguments can be inferred as follows. In the Pinter run in which E is kept to its climatology and only Pinter is included in the FWFinter calculation, the offset by Einter to Pinter is not taken into account. This will lead to a bias in the FWFinter calculation. For example, during El Niño, Pinter and Einter both exhibit a positive anomaly; the resultant (P − E)inter pattern is similar to Pinter but with smaller amplitude because of the Einter offset. When the Einter contribution to FWFinter is neglected, the calculated FWFinter is positively overestimated during El Niño (i.e., an exaggerated positive FWF into the ocean). Similarly, neglecting the offset of a negative Einter anomaly during La Niña will lead to the calculated FWFinter that is negatively overestimated (i.e., more negative). Thus, without the offset by Einter contribution, the amplitude of the calculated FWFinter in the Pinter run is overestimated during ENSO cycles (i.e., more positive in the central basin during El Niño and more negative during La Niña). This will lead to a larger positive feedback in the HCM. It can be inferred that interannual variability in the Pinter run will be larger than that in the (P − E)inter run.
This is indeed confirmed in the HCM experiments, as shown in Figs. 3 –5. In the Pinter run, the longitudinal displacements of the fresh–warm pool in the west are more pronounced in association with ENSO events (Fig. 3). The exaggerated anomalous FWF forcing acts to enhance interannual SSS variability in the central region (Fig. 4c), as compared to that in the (P − E)inter run (Fig. 4a) and in the Einter run (Fig. 4b). Furthermore, the strengthening effect on interannual SST variability is also evident in Fig. 5. In fact, the Pinter run has largest interannual variability in all HCM runs performed in this article (Figs. 4 –5).
Oceanic processes and related effects are similar to those in the (P − E)inter run as analyzed above, but are opposite to those in the Einter run. For example, as compared with the (P − E)inter run, a more positive FWF anomaly during El Niño in the Pinter run (without the offset by Einter) leads to more freshening and more reduction in the negative QB anomaly in the central equatorial Pacific. The freshening in the surface mixed layer acts to increase the stability of the upper ocean and restrict vertical mixing of the heat in the central basin. At the same time, a less negative QB acts to lead to a mixed layer that is shallower, with less entrainment of subsurface waters into the mixed layer. These ocean processes tend to increase SSTs in the central region, which reinforces the warm SST anomalies produced by El Niño. As a result, both SSS and SST anomalies are enhanced while QB anomalies are reduced. Situations during La Niña operate in a similar way but with opposite sense. As a result, the Pinter run has a stronger feedback between SST and FWF, acting to enhance SSS and SST anomalies during ENSO cycles.
c. The climatological (P − E) forcing
Lastly, an experiment is performed in which both P and E are kept to their climatology without taking into account FWFinter in the HCM [FWFTotal = (P − E)clim]. This case has been examined in Zhang and Busalacchi (2009). When FWFinter is not taken into account in the HCM, there is no interannually varying FWF forcing effects on SSS and QB during ENSO evolution. The interannual FWF forcing-induced positive feedback is absent. As such, interannual variability in this case is expected to be weak compared with the (P − E)inter run. Furthermore, it can be inferred from the above analyses that in terms of the FWF forcing effects, the level of interannual variability in this run should be between the Einter run and (P − E)inter run. This is confirmed in Fig. 11 showing the equatorial distributions of the standard deviation of SSS and SST variability for the four experiments. Indeed, the simulated interannual variability is weakened in the (P − E)clim run since the positive feedback is excluded; it is even more damped in the Einter runs because the positive feedback is misrepresented to be negative; but it is exaggerated in the Pinter run because the positive feedback is overestimated.
6. Discussion and conclusions
Many factors have been identified that can contribute to large model biases in simulation and prediction of interannual variability in the tropical Pacific ocean–atmosphere system. Most previous modeling studies have focused on the roles of atmospheric forcing components of wind and heat flux to the ocean. FWF forcing and its related salinity variations have not been getting much attention; their effects and related feedbacks have been examined mostly in a forced ocean-only modeling context. When the ocean and atmosphere is coupled, changes in SSTs induced by FWF can affect the atmosphere, indicating the potential for its modulating effects on interannual variability. While FWF exerts no direct and immediate influence on SST, its variations can perturb SSS and QB, which can further modify the oceanic density, MLDs, the mixing and entrainment, all of which affects SST, the principal oceanic quantity felt by the atmosphere. More recently, Zhang and Busalacchi (2009) have demonstrated that FWF forcing can have significant effects on interannual variability in the coupled ocean–atmosphere system of the tropical Pacific, involving a positive feedback between SST and FWF.
Large interannual FWF variability is observed in the tropical Pacific during ENSO evolution, with its coherent positive relationship with SST. The ENSO-induced FWF anomalies are positive in the central and western regions during El Niño, but negative during La Niña. If these anomaly patterns are not represented adequately, a bias is introduced in the FWF forcing, which can be transferred to model biases in the coupled ocean–atmosphere system. Previously, various approximations have been made to represent FWF forcing in coupled models. But the extent to which this atmospheric forcing plays a role in interannual biases has not been examined in a coupled ocean–atmosphere modeling context. Detailed processes are not clearly identified.
In this work, we address this issue from a coupled modeling perspective. We have designed various experiments to illustrate possible relationships between interannual biases and the way anomalous FWF forcing (FWFinter) is represented in an HCM (Table 1). The climatological FWF fields, as represented by (P − E)clim, are combined with various specifications of its anomalous part to estimate the total FWF. Four cases are considered (Table 1). The results indicate that the time evolution of SST in the four different runs is qualitatively similar. This is because wind forcing is of primary importance for dynamics of interannual variability in the tropical Pacific climate system, and that interannually anomalous FWF forcing only plays a modulating role.
However, the results also indicate that interannual biases in the HCM simulations can be clearly attributed to different FWFinter specifications. When the anomalous FWF forcing is explicitly included in the HCM, a positive feedback is represented between SST and FWF, giving rise to a reasonably realistic simulation of interannual variability. When the anomalous FWF part is approximated somehow, interannual variability is impacted since the FWF-induced feedback is misrepresented. When FWFinter forcing is not included as in the (P − E)clim run, a weakening effect is seen since there is no positive feedback represented. When the relationship between variations in SST and FWFinter is artificially reversed as in the Eclim run, the induced feedback is misrepresented to be negative and correspondingly, interannual variability is damped significantly. On the other hand, if the feedback is overestimated as in the Pinter run, a strengthening effect is found. These results indicate that anomalous FWF forcing can modulate interannual variability in a substantial way.
Figures 12 –13 further show the standard deviation (std) of SSS and SST from these experiments. The corresponding observations for SSS and SST are presented in Fig. 14, which is estimated during the periods 1969–95. Large variability of SSS is located in the central basin near the date line while that of SST is in the central and eastern equatorial Pacific. Approximations in representing anomalous FWF forcing can lead to large changes in the strength of interannual variability (Fig. 11). For example, the levels of simulated SSS and SST variabilities are directly related with the representations of FWF forcing-induced feedback. The stronger the feedback between SST and FWF, the larger the interannual variability simulated in the HCM. Since the Pinter run is representing the strongest feedback between SST and FWF, its simulated interannual variability is strongest among all these experiments, followed by the (P − E)inter, (P − E)clim, and Einter runs, respectively. Note that the HCM produces unrealistic spatial distribution of interannual standard deviation of SST along the equator, with its maximum around the date line and second maximum in the eastern Pacific (between 120° and 100°W). This is different from observations (Fig. 14b), which show almost monotonically increased interannual standard deviation toward the east along the equator.
The effect can be further quantified using the Niño-3 and Niño-4 standard deviation for some selected parameters (Table 2). For example, the std of Niño-3 SST (Niño-4 SSS) anomalies is 0.76°C (0.16 psu) in the standard (P − E)inter run, 0.84°C (0.18 psu) in the Pinter run, 0.67°C (0.11 psu) in the (P − E)clim run, and 0.56°C (0.10 psu) in the Einter run. Relative to the reference (P − E)inter run, these values are representing an increase of 11% (13%) in the Pinter run, but a reduction of 12% (31%) in the (P − E)clim run and of 26% (38%) in the Einter run. Thus, a significant fraction of the SSS and SST variability can be attributed to anomalous FWF forcing.
Physical processes involved in the FWF bias-induced effects can be illustrated as follows. As one component of the atmospheric forcings to the ocean, FWF directly forces changes in SSS, which modifies the oceanic density, the stability of the stratification and the vertical mixing. At the same time, as a part of QB, FWFinter presents a compensating effect on heat flux. This acts to reduce interannual QB variability, which in turn modulates MLDs and the entrainment of subsurface water into the mixed layer. All these oceanic processes influence SSTs, which in turn feed back to the atmosphere. Anomalies of SSS, MLDs, and SST in the ocean induced by FWF forcing and related feedback act to reinforce those originally produced by ENSO cycles.
Since significant model tropical biases are still prominent in prediction of ENSO in the tropical Pacific, there is a clear need to improve coupled models for better simulation of interannual climate variability. In addition to physical understanding, another motivation of this work is to see if interannual variability simulations can be improved in terms of the way anomalous FWF forcing is represented in the HCM. As mentioned above, FWF forcing has not been adequately taken into account in many coupled ocean–atmosphere models. For example, most intermediate coupled models (ICMs) and HCMs currently used for ENSO prediction have not explicitly included the FWF forcing (e.g., Zebiak and Cane 1987; Barnett et al. 1993; Zhang et al. 2003). Since large FWF anomalies are observed during ENSO cycles and have direct effects on SSS and QB, they can be a clear source for tropical biases. Here, we clearly demonstrate that the strength of interannual variability can be modulated in a substantial way by different representations of anomalous FWF forcing. Also, it is evident that the amplitude of the SST variability simulated by the HCMs can be easily tuned in terms of the strength of FWF forcing-induced feedback. For example, the weakness of SST variability in the eastern tropical Pacific is a common bias in many coupled models. Sensitivity experiments with different strength of FWF feedback indicate a clear effect on the SST simulation there. Thus, FWF forcing can be a clear source for model biases in the SST simulation and thus needs to be taken into account adequately. As demonstrated here, when the FWF forcing is taken into account in the HCM, SST variability can be increased significantly in the eastern equatorial Pacific cold tongue region. Clearly, an improved simulation in the eastern region can be achieved by taking into account this atmospheric FWF forcing adequately. Note, however, that including FWF feedback has not improved the unrealistic spatial distribution of interannual SST in the central basin (Fig. 12).
These results are also useful for other model improvements. Because of its important effect, FWF forcing needs to be taken into account explicitly in simplified coupled models, such as ICMs and HCMs. In this article, a simple empirical model is constructed to capture interannual FWF variability in the tropical Pacific, which provides an effective parameterization that can be used to take into account FWF forcing-induced positive feedback for simplified coupled models used for ENSO simulation and prediction (e.g., Syu et al. 1995; Barnett et al. 1993; Latif et al. 2001; Zhang et al. 2003). It is expected that improved simulations of the ocean mean state and its variability (such as total SST and upwelling fields) can also have significant impacts upon simulations of other physical and biogeochemical parameters.
Acknowledgments
We would like to thank Phil Arkin, Boyin Huang, and J. Carton for their comments. The authors wish to thank three anonymous reviewers for their numerous comments that helped to improve the original manuscript. This research is supported in part by NSF Grant ATM-0727668 and NOAA Grant NA08OAR4310885, and NASA Grants NNX08AI74G, NNX08AI76G, NNX08AT50G, and NNX09AU74G. D. Chen and G. Wang are supported by the National Natural Science Foundation of China (40730843) and the National Basic Research Program of China (2007CB816005) and International Corporation Program of China (2008DFA22230).
REFERENCES
Ballabrera-Poy, J., R. Murtugudde, and A. J. Busalacchi, 2002: On the potential impact of sea surface salinity observations on ENSO predictions. J. Geophys. Res., 107 , 8007. doi:10.1029/2001JC000834.
Ballabrera-Poy, J., R. G. Murtugudde, R-H. Zhang, and A. J. Busalacchi, 2007: Coupled ocean–atmosphere response to seasonal modulation of ocean color: Impact on interannual climate simulations in the tropical Pacific. J. Climate, 20 , 353–374.
Barnett, T. P., M. Latif, N. Graham, M. Flugel, S. Pazan, and W. White, 1993: ENSO and ENSO-related predictability. Part I: Prediction of equatorial Pacific sea surface temperature with a hybrid coupled ocean–atmosphere model. J. Climate, 6 , 1545–1566.
Barnston, A. G., M. H. Glantz, and Y. He, 1999: Predictive skill of statistical and dynamical climate models in SST forecasts during the 1997–98 El Niño episode and the 1998 La Niña onset. Bull. Amer. Meteor. Soc., 80 , 217–243.
Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pacific. Mon. Wea. Rev., 97 , 163–172.
Boyer, T. P., J. I. Antonov, S. Levitus, and R. Locarnini, 2005: Linear trends of salinity for the World Ocean, 1955-1998. Geophys. Res. Lett., 32 , L01604. doi:10.1029/2004GL021791.
Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5 , 541–560.
Cane, M. A., and S. E. Zebiak, 1985: A theory for El Niño and Southern Oscillation. Science, 228 , 1084–1087.
Carton, J. A., 1991: Effect of seasonal surface freshwater flux on sea surface temperature in the tropical Atlantic Ocean. J. Geophys. Res., 96C , 12593–12598.
Chang, P., L. Ji, and R. Saravanan, 2001: A hybrid coupled model study of tropical Atlantic variability. J. Climate, 14 , 361–390.
Chen, D., L. M. Rothstein, and A. J. Busalacchi, 1994: A hybrid vertical mixing scheme and its application to tropical ocean models. J. Phys. Oceanogr., 24 , 2156–2179.
Chen, D., M. A. Cane, S. E. Zebiak, R. Canizares, and A. Kaplan, 2000: Bias correction of an ocean-atmosphere coupled model. Geophys. Res. Lett., 27 , 2585–2588.
Cooper, N. S., 1988: The effect of salinity on tropical ocean models. J. Phys. Oceanogr., 18 , 697–707.
Cravatte, S., T. Delcroix, D. Zhang, M. McPhaden, and J. Leloup, 2009: Observed freshening of the warming western tropical Pacific and extension of the Warm/Fresh pool in recent decades. Climate Dyn., 33 , 565–589. doi:10.1007/s00382-009-0526-7.
Delcroix, T., and C. Hénin, 1991: Seasonal and interannual variations of sea surface salinity in the tropical Pacific Ocean. J. Geophys. Res., 96C , 22135–22150.
DeWitt, D. G., 2005: Retrospective forecasts of interannual sea surface temperature anomalies from 1982 to present using a directly coupled atmosphere ocean general circulation model. Mon. Wea. Rev., 133 , 2972–2995.
Fedorov, A. V., R. C. Pacanowski, S. G. H. Philander, and G. Boccaletti, 2004: The effect of salinity on the wind-driven circulation and the thermal structure of the upper ocean. J. Phys. Oceanogr., 34 , 1949–1966.
Gent, P., and M. A. Cane, 1989: A reduced gravity, primitive equation model of the upper equatorial ocean. J. Comput. Phys., 81 , 444–480.
Hackert, E., A. Busalacchi, and R. Murtugudde, 2001: A wind comparison study using an ocean general circulation model for the 1997-1998 El Niño. J. Geophys. Res., 106C , 2345–2362.
Huang, B., and V. M. Mehta, 2004: Response of the Indo-Pacific warm pool to interannual variations in net atmospheric freshwater. J. Geophys. Res., 109 , C06022. doi:10.1029/2003JC002114.
Huang, B., and V. M. Mehta, 2005: Response of the Pacific and Atlantic oceans to interannual variations in net atmospheric freshwater. J. Geophys. Res., 110 , C08008. doi:10.1029/2004JC002830.
Huang, B., Y. Xue, and D. W. Behringer, 2008: Impacts of Argo salinity in NCEP Global Ocean Data Assimilation System: The tropical Indian Ocean. J. Geophys. Res., 113 , C08002. doi:10.1029/2007JC004388.
Huang, R. X., and R. W. Schmitt, 1993: The Goldsbrough–Stommel circulation of the World Ocean. J. Phys. Oceanogr., 23 , 1277–1284.
Ji, M., D. W. Behringer, and A. Leetmaa, 1998: An improved coupled model for ENSO prediction and implications for ocean initialization. Part II: The coupled model. Mon. Wea. Rev., 126 , 1022–1034.
Jin, F-F., and S-I. An, 1999: Thermocline and zonal adjective feedbacks within the equatorial ocean recharge oscillator model for ENSO. Geophys. Res. Lett., 26 , 2989–2992.
Kessler, W. S., 1999: Interannual variability of the subsurface high salinity tongue south of the equator at 165°E. J. Phys. Oceanogr., 29 , 2038–2049.
Kirtman, B., Y. Fan, and E. K. Schneider, 2002: The COLA global coupled and anomaly coupled ocean–atmosphere GCM. J. Climate, 15 , 2301–2320.
Lagerloef, G. S. E., 2002: Introduction to the special section: The role of surface salinity on upper ocean dynamics, air-sea interaction and climate. J. Geophys. Res., 107 , 8000. doi:10.1029/2002JC001669.
Large, W. G., and G. Danabasoglu, 2006: Attribution and impacts of upper-ocean biases in CCSM3. J. Climate, 19 , 2325–2346.
Latif, M., and Coauthors, 1998: A review of the predictability and prediction of ENSO. J. Geophys. Res., 103 , 14375–14393.
Latif, M., and Coauthors, 2001: ENSIP: The El Niño simulation intercomparison project. Climate Dyn., 18 , 255–276.
Levitus, S., J. I. Antonov, T. P. Boyer, H. E. Garcia, and R. A. Locarnini, 2005: Linear trends of zonally averaged thermosteric, halosteric, and total steric sea level for individual ocean basins and the world ocean, (1955–1959)–(1994–1998). Geophys. Res. Lett., 32 , L16601. doi:10.1029/2005GL023761.
Lukas, R., and E. Lindstrom, 1991: The mixed layer of the western equatorial Pacific Ocean. J. Geophys. Res., 96 , (Suppl.). 3343–3357.
Maes, C., 2000: Salinity variability in the equatorial Pacific Ocean during the 1993-98 period. Geophys. Res. Lett., 27 , 1659–1662.
Meehl, G. A., P. R. Gent, J. M. Arblaster, B. Otto-Bliesner, E. Brady, and A. P. Craig, 2001: Factors that affect the amplitude of El Niño in global coupled climate models. Climate Dyn., 17 , 515–526.
Miller, J. R., 1976: The salinity effect in a mixed layer ocean model. J. Phys. Oceanogr., 6 , 29–35.
Misra, V., L. Marx, M. Brunke, and X. Zeng, 2008: The equatorial Pacific cold tongue bias in a coupled climate model. J. Climate, 21 , 5852–5869.
Murtugudde, R., and A. J. Busalacchi, 1998: Salinity effects in a tropical ocean model. J. Geophys. Res., 103 , 3283–3300.
Murtugudde, R., R. Seager, and A. J. Busalacchi, 1996: Simulation of tropical oceans with an ocean GCM coupled to an atmospheric mixed-layer model. J. Climate, 9 , 1795–1815.
Murtugudde, R., J. Beauchamp, C. R. McClain, M. Lewis, and A. J. Busalacchi, 2002: Effects of penetrative radiation on the upper tropical ocean circulation. J. Climate, 15 , 470–486.
Reason, C. J. C., 1992: On the effect of ENSO precipitation anomalies in a global ocean GCM. Climate Dyn., 8 , 39–47.
Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, and W. Wang, 2002: An improved in-situ and satellite SST analysis for climate. J. Climate, 15 , 1609–1625.
Roeckner, E., and Coauthors, 1996: The atmospheric general circulation model ECHAM4: Model description and simulation of present day climate. Rep. 218, Max-Planck-Institut fur Meteorologie, 90 pp. [Available from MPI fur Meteorologie, Bundesstr. 55, 20146 Hamburg, Germany].
Schneider, N., and T. P. Barnett, 1995: The competition of freshwater and radiation in forcing the ocean during El Niño. J. Climate, 8 , 980–992.
Seager, R., M. Blumenthal, and Y. Kushinir, 1995: An advective atmospheric mixed layer model for ocean modeling purposes: Global simulation of surface heat fluxes. J. Climate, 8 , 1951–1964.
Sprintall, J., and M. Tomczak, 1992: Evidence of the barrier layer in the surface layer of the Tropics. J. Geophys. Res., 97 , (C5). 7305–7316.
Syu, H-H., J. D. Neelin, and D. Gutzler, 1995: Seasonal and interannual variability in a hybrid coupled GCM. J. Climate, 8 , 2121–2143.
Vialard, J., and P. Delecluse, 1998: An OGCM study for the TOGA decade. Part I: Role of salinity in the physics of the western Pacific fresh pool. J. Phys. Oceanogr., 28 , 1071–1088.
Wittenberg, A. T., A. Rosati, N-C. Lau, and J. J. Ploshay, 2006: GFDL’s CM2 global coupled climate models. Part III: Tropical Pacific climate and ENSO. J. Climate, 19 , 698–722.
Xie, P., and P. Arkin, 1995: An intercomparison of gauge observations and satellite estimates of monthly precipitation. J. Appl. Meteor., 34 , 1143–1160.
Yang, S., K-M. Lau, and P. S. Schopf, 1999: Sensitivity of the tropical Pacific ocean to precipitation-induced freshwater flux. Climate Dyn., 15 , 737–750.
Yu, L., and R. A. Weller, 2007: Objectively analyzed air–sea heat fluxes for the global ice-free oceans (1981–2005). Bull. Amer. Meteor. Soc., 88 , 527–539.
Zebiak, S. E., and M. A. Cane, 1987: A model El Niño/Southern Oscillation. Mon. Wea. Rev., 115 , 2262–2278.
Zhang, R-H., and A. J. Busalacchi, 2005: Interdecadal changes in properties of El Niño in an intermediate coupled model. J. Climate, 18 , 1369–1380.
Zhang, R-H., and A. J. Busalacchi, 2009: Freshwater flux (FWF)-induced oceanic feedback in a hybrid coupled model of the tropical Pacific. J. Climate, 22 , 853–879.
Zhang, R-H., S. E. Zebiak, R. Kleeman, and N. Keenlyside, 2003: A new intermediate coupled model for El Niño simulation and prediction. Geophys. Res. Lett., 30 , 2012. doi:10.1029/2003GL018010.
Zhang, R-H., S. E. Zebiak, R. Kleeman, and N. Keenlyside, 2005: Retrospective El Niño forecast using an improved intermediate coupled model. Mon. Wea. Rev., 133 , 2777–2802.
Zhang, R-H., A. J. Busalacchi, and R. G. Murtugudde, 2006a: Improving SST anomaly simulations in a layer ocean model with an embedded entrainment temperature submodel. J. Climate, 19 , 4638–4663.
Zhang, R-H., A. J. Busalacchi, R. G. Murtugudde, P. A. Arkin, and J. Ballabrera-Poy, 2006b: An empirical parameterization for the salinity of subsurface water entrained into the ocean mixed layer (Se) in the Tropical Pacific. Geophys. Res. Lett., 33 , L02605. doi:10.1029/2005GL024218.
A schematic illustrating an HCM of the tropical Pacific climate system, consisting of an OGCM and a simplified atmospheric model, whose forcing fields to the ocean include three components: wind stress (τ), heat flux (HF), and FWF. The total wind stress is separated into climatological wind stress (τclim) and interannual anomaly (τinter); the former is prescribed from observations and the latter is associated with interannual SST anomalies (SSTinter). The heat flux is calculated using the Seager et al. (1995) advective AML model. Similarly, the total freshwater flux can be written as: FWFTotal = (P − E)clim + FWFinter, consisting of the prescribed climatological part, (P − E)clim, and interannual anomaly part FWFinter. Various approximations are made to FWFinter: anomalous P and E are both taken into account in the FWFinter calculation [FWFinter = (P − E)inter], anomalous E only (FWFinter = −Einter), anomalous P only (FWFinter = Pinter), and anomalous P and E are both not taken into account (FWFinter = 0.0). Empirical submodels for τinter, (P − E)inter, Einter and Pinter can be constructed from SVD analyses. In addition, Einter can also be calculated from the AML model as evaporative component of the heat flux. Buoyancy flux is calculated from the heat flux and freshwater flux to force a mixed layer model. The climatological SST fields (SSTclim) are specified from the spinup OGCM run forced by the τclim fields. The dashed lines indicate processes involved due to the FWF-induced effects.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
The spatial patterns of the first SVD mode derived for (a) SST, (b) (P − E), (c) E, and (d) P. The SVD analyses are performed during 1963–96. The contour interval is 0.2.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
Longitude–time sections along the equator of total SSS fields simulated from (a) the (P − E)inter run, (b) the Einter run, and (c) the Pinter run. The contour interval is 0.1 psu.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
As in Fig. 3, but for SSS anomalies. The contour interval is 0.1 psu.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
As in Fig. 3, but for SST anomalies. The contour interval is 0.5°C.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
A conceptual diagram illustrating the effects of a positive FWF anomaly in the central basin produced by El Niño, which act to induce a positive feedback on the coupled ocean–atmosphere system of the tropical Pacific.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
Horizontal patterns of interannual anomalies simulated from the (P − E)inter run for El Niño conditions in September 2031: (a) SST, (b) the QS part, (c) QB, (d) the QT part, (e) SSS, and (f) MLD. The contour interval is (a) 0.5°C, (b)–(d) 0.5 × 10−6 kg s−1 m−2, (e) psu, and (f) meters. The units in (b)–(d) are expressed in kg s−1 m−2 (the QB unit), which can be correspondingly converted to the commonly used units for heat flux and FWF: 1.0 × 10−6 kg s−1 m−2 is equivalent to 13.3 W m−2 for heat flux or to 3.4 mm day−1 for (P − E).
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
As in Fig. 7, but for La Niña conditions in July 2033.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
Time series of (a) SSS and (b) SST at 0°, 160°W simulated from the (P − E)inter run (line with open circles) and the Einter run (line with closed circles). The units are psu for SSS and °C for SST.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
Time series of (a) Niño-3 SST, (b) Niño-4 zonal wind stress, and (c) Niño-4 SSS simulated from the (P − E)inter run (line with open circles) and the Einter run (line with closed circles). The units are °C for SST, dyn cm−2 for zonal wind stress, and psu for SSS.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
Equatorial distributions of the standard deviation of (a) SSS and (b) SST variability for the (P − E)inter run, the Pinter run, the (P − E)clim run, and the Einter run. The units are psu for SSS and °C for SST.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
Horizontal distribution of the standard deviation for interannual SST variability: (a) the Einter run, (b) the (P − E)inter run, and (c) the Pinter run. The contour interval is 0.1. The std is calculated from the corresponding HCM simulations for 25 years.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
Horizontal distributions of the standard deviation for observed interannual (a) SSS and (b) SST variability estimated during 1969–95.
Citation: Monthly Weather Review 138, 5; 10.1175/2009MWR3054.1
The four model experiments designed in the article. The total FWF into the ocean can be written as: FWFTotal = (P − E)clim + FWFinter. The climatological part is specified as (P − E)clim; interannual anomaly part (FWFinter) can be represented in four cases: interannual (P − E) forcing [FWFinter = (P − E)inter, the (P − E)inter run], interannual P run (FWFinter = Pinter, the Pinter run), interannual E forcing (FWFinter = −Einter, the Einter run), and climatological (P − E) forcing [FWFinter = 0.0, the (P − E)clim run], respectively.
The standard deviation of some selected anomaly fields from the HCM simulations with different approximations made to FWF forcing. Shown in the Niño-4 region are for SSS, SST, MLD, and zonal wind stress (τx). Also shown in the last two rows are for SST in the Niño1 + 2 and Niño-3 regions. The unit is psu for SSS, °C for SST, m for MLD and dyn cm−2 for τx.