The role of atmospheric internal variability on the tropical instability wave dynamics

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117,, doi: /2012jc007906, 2012 The role of atmospheric internal variability on the tropical instability wave dynamics Balachandrudu Narapusetty, 1 Cristiana Stan, 1,2 Ben P. Kirtman, 1,3 Paul S. Schopf, 1,2 Lawrence Marx, 1 and James L. Kinter III 1,2 Received 19 January 2012; revised 12 June 2012; accepted 17 June 2012; published 25 August [1] The impact of atmospheric internal variability on tropical instability wave (TIW) activity in the eastern equatorial Pacific is examined. To diagnose the atmospheric internal variability, two simulations were performed with a state-of-the-art coupled general circulation model that uses an eddy permitting ocean component model. Standard coupling procedures are implemented in the control simulation. In the experimental simulation, the so-called interactive ensemble coupling is used, which systematically reduces the contribution of internal atmospheric dynamics to the air-sea fluxes of heat, momentum and fresh water. In the eastern equatorial Pacific, the reduction of the atmospheric internal variability leads to an enhancement of the available potential energy and higher exchanges from mean to eddy potential energy. The perturbations in the available potential energy and the eddy potential energy contribute to the enhancement in the TIW activity through the increased eddy kinetic energy. Due to the negative correlation between the atmospheric internal variability and TIW activity, the covariance between the momentum flux at the air-sea interface and the ocean surface currents as well as heat flux at the air-sea interface and the sea surface temperatures were nearly conserved west of 120 W between the control and the experimental simulations. Citation: Narapusetty, B., C. Stan, B. P. Kirtman, P. S. Schopf, L. Marx, and J. L. Kinter III (2012), The role of atmospheric internal variability on the tropical instability wave dynamics, J. Geophys. Res., 117,, doi: /2012jc Introduction [2] In the eastern equatorial Pacific, the sea surface temperature anomalies exhibit a persistent variability commonly known as tropical instability wave activity on days in time and km in spatial scales [Düing et al., 1975; Legeckis, 1977]. Tropical instability waves (TIWs) are shown to significantly impact the ocean-atmosphere interactions [Hayes et al., 1989; Jochum and Murtugudde, 2004; An, 2008]. Satellite observations show a significant zonal as well as meridional wind covariability with the TIWs [Chelton et al., 2001]. TIWs are generated and maintained by the barotropic and the baroclinic instability mechanisms in the tropical oceans [e.g., Düing et al., 1975; Qiao and Weisberg, 1995]. These instability mechanisms influence the TIW strengths by modulating the interactions between mean and eddy kinetic energies of the ocean currents. 1 Center for Ocean-Land-Atmosphere Studies, Calverton, Maryland, USA. 2 Department of Atmospheric, Oceanic and Earth Sciences, George Mason University, Fairfax, Virginia, USA. 3 Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida, USA. Corresponding author: B. Narapusetty, Center for Ocean-Land- Atmosphere Studies, 4041 Powder Mill Rd., Ste, 302, Calverton, MD 20705, USA. (bala@cola.iges.org) American Geophysical Union. All Rights Reserved /12/2012JC [3] In the eastern equatorial Pacific, the barotropic instability arises due to the meridional shear between the weak and the strong branches of south equatorial current (SEC) located at the equator and 2 N, respectively. Similar meridional shear exists between the opposing SEC and the north equatorial counter currents (NECC) located around 4 N [Düing et al., 1975; Philander, 1976; Hansen and Paul, 1984]. Barotropic instability also arises due to the similar meridional shear between SEC and equatorial under current (EUC) around and north of the equator, respectively [Qiao and Weisberg, 1995]. [4] The baroclinic instability is due to the vertical shear in the mean currents and is generally associated with the sharp meridional sea surface temperature (SST) gradients [Yu et al., 1995]. The TIW induced SST anomalies are stronger in typical La Niña years and tend to be weaker in strong El Niño years. [5] TIWs significantly impact the heat and the momentum budgets in the oceanic mixed layer [Hansen and Paul, 1984]. Since the strength of TIWs is highly nonlinear in space and time, the resulting rectification could be observed on the interannual timescales. For example, TIWs influence the meridional SST gradients in the eastern equatorial Pacific and have been suggested to play an important role in the onset of El Niño events [Jochum and Murtugudde, 2004; An, 2008]. Ham and Kang [2010] study suggests that better representation of TIWs in the tropical Pacific leads to improvements in the skill of ENSO forecasts. Several atmospheric general circulation model (AGCM) studies 1of23

2 show that in the atmospheric boundary layer over the tropical Pacific (180 W 90 W; 8 S 8 N) TIWs influence the low-level stratiform cloudiness, meridional wind strength, eddy momentum and moisture transports, and diabatic heating rates on intraseasonal to interannual timescales [Xie et al., 1998; Hashizume et al., 2001; Jochum et al., 2007]. [6] In the eastern equatorial Pacific, the ocean-atmosphere coupled feedbacks play a major role in the oceanic and atmospheric general circulation on all timescales. Imada et al. [2012] study suggests that resolution of the atmospheric and oceanic components of a general circulation model is important in representing the realistic TIW activity and their study also shows that the atmospheric mean state impacts the TIW energetics. Since the TIWs appear on the ocean surface and systematically impact the atmospheric circulation, there ought to be an atmospheric feedback that either positively or negatively impacts the TIW strength. Pezzi et al. [2004] ocean-atmosphere coupling sensitivity experiments and Seo et al. [2007] analysis show that TIWs influence atmospheric zonal and merdional eddies, which in turn negatively feedback onto the TIW activity. However, it is not known how the weather noise (atmospheric variability due to internal dynamics) in general (which may also include the TIW-prompted atmospheric eddy component) impacts the energetics associated with TIWs. Moreover, there is no study explaining mechanisms through which the atmospheric internal variability influences TIW activity. One approach for understanding the mechanisms through which the weather noise influences the TIWs is to compare coupled general circulation model experiments that realistically simulate TIWs in the presence as well as in the absence of weather noise. [7] Our intent in this study is to deconstruct how the weather noise locally influences the individual TIWs. In order to isolate the atmospheric feedback we use the socalled interactive ensemble coupling strategy [e.g., Kirtman and Shukla, 2002; Kirtman et al., 2005, 2009, 2011; Stan and Kirtman, 2008], which uses ensemble averages to reduce the imprint of internal atmospheric dynamics on the air-sea fluxes of heat, momentum and fresh water. This ensemble averaging procedure has the effect of significantly reducing the noisiness of the air-sea fluxes. Perhaps, the most obvious result of this reduced noisiness would be a shallower ocean mixed layer with reduced eddy activity; this was our first assumption. The results presented here, conversely, show that the ensemble averaging actually leads to significantly enhanced ocean eddies on TIW timescales, which leads to a more turbulent and deeper ocean mixed layer. The enhanced ocean eddies are due to the reduced atmospheric forcing associated with the interactive ensemble averaging. [8] The main objective of this study is to understand how the atmospheric internal variability influences the TIW variability. The model description and experiments are explained in section 2, results are presented in section 3, discussion of the results is given in section 4, and conclusions are summarized in section Description of Numerical Experiments [9] Two different coupled general circulation model simulations are performed, namely, Climate Forecasting System version 2 (CFS) [Saha et al., 2010, and references therein], and CFS in Interactive Ensemble configuration [Stan and Kirtman, 2008]. [10] The atmospheric component of CFS is the latest version of the global forecasting system (GFS). The dynamical core is set up in the Eularian spectral framework with triangular truncation at wave number 126. The atmospheric component has 64 sigma pressure hybrid levels in the vertical from surface up to the model top located at 0.2 hpa. The ocean component of CFS is the Geophysical Fluid Dynamics Laboratory Modular Ocean Model version 4 (MOM4). The spatial resolution of the ocean component in the zonal direction is 0.5, while the grid resolution in the meridional direction is 0.25 from 10 Sto10 N, progressively decreases to 0.5 from 10 to 30, and is fixed at 0.5 beyond 30 in either hemisphere. This is an eddy-permitting model and has 40 levels in the vertical with 27 layers in the upper 400 m. The vertical resolution is 10 m up to 240 m depth, increasing to a maximum of 511 m in deeper levels. MOM4 includes the Arctic Ocean dynamics and an interactive threelayer sea ice model. In the coupled framework of CFS, the atmosphere and the ocean components exchange freshwater, heat, and momentum fluxes every 30 minutes during the simulation. [11] The CFS Interactive Ensemble (referred to as CFSIE hereinafter) is a modified version of CFS to allow the coupling of the ocean component of CFS with the averaged fluxes of six atmospheric components of CFS. Therefore, during the CFSIE model integration, at the air-sea interface, the ocean component receives freshwater, heat, and momentum fluxes derived as the ensemble mean of six atmospheric realizations at every thirty minutes. The interactive ensemble coupling is known to lead to large changes in the sea ice climatology [e.g., Kirtman et al., 2011]. To prevent this, over the sea ice, fluxes simulated by the first atmospheric component only are applied to the oceanic component. Ideally, to isolate the atmospheric signal and to completely remove the impact of atmospheric internal dynamics in the air-sea fluxes, an infinitely many number of atmospheric realizations are required. This is not computationally feasible, however, our experience with the interactive ensemble indicates that the mean of six ensemble realizations significantly reduces the atmospheric internal variability acting at the air-sea interface. For a more detailed discussion on the reduction of atmospheric internal variability using Interactive Ensemble approach refer to Kirtman et al. [2005]. Both CFS and CFSIE are integrated from 1 November 1980 to 31 October In both models, the atmosphere and the ocean initial conditions are derived from CFS reanalysis products. The six atmospheres of CFSIE are initialized with CFS reanalysis data separated by 6 h starting from 0000 Z of 1 November Results [12] In this paper, all the analyses are performed over eastern equatorial Pacific between 180 to 90 W and from 5 S to 5 N. Anomalies of the variables are obtained as deviations from optimally estimated seasonal cycles [Narapusetty et al., 2009]. Eddies are estimated by applying the zonal high-pass filtering that retains the only variability associated with the zonal wavelength lower than 13 in the 2of23

3 Figure 1. Time longitude cross sections at 1.5 N of zonally high-pass filtered SST anomalies from July to January of a typical La Niña year in (a) CFS and (b) CFSIE. SST anomalies. The estimated eddies represent the variability on time and spatial scales similar to that of the observed TIWs and they are denoted with superscript * throughout this paper. Since the TIW activity is predominantly present from July to December months (referred to as J2D hereinafter), only these months are considered in the analyses throughout this paper. The significance of mean differences in variables as well as in flux quantities of 19 yearly averaged values between CFS and CFSIE are determined by performing two-tail Student s t test at 5% significant levels. [13] Figure 1 shows that both models realistically simulate the westward propagation of TIWs from July to December at 1.5 N in the eastern equatorial Pacific for a randomly chosen La Niña year. An inspection of Figures 1a and 1b suggest that the TIWs in CFSIE are stronger at 1.5 N than in CFS, but this is a particular example and any generalization is not justified. The wave number frequency spectra in Figures 2a and 2b, however, do indicate that the variability of TIWs in CFSIE is indeed stronger for the period The 19 year ( ) J2D averaged wave number frequency spectra [Straus and Shukla, 1981] at 1.5 N over the longitudes encompassing eastern equatorial Pacific shows that the TIW variance is higher ( 40%) in CFSIE than in CFS (compare Figure 2b with Figure 2a). The eastward propagation is very small in both CFS and CFSIE; the variance associated with the eastward propagation is around 100 times smaller compared to the westward propagation (Figures 2c and 2d). [14] The increased TIW activity in CFSIE is also pronounced in the ocean currents. The ocean eddy currents simulated in CFSIE are stronger than those simulated in CFS. This results in increased eddy heat fluxes (t*u*+t*v*) throughout the ocean depths up to 100 m. Figure 3 compares the standard deviation of the eddy heat flux simulated by CFS and CFSIE. The fluxes are averaged over W and from the surface to 55 m. The eddy heat flux in CFSIE is larger than in CFS at all the latitudes from 5 Sto5 N. As noted earlier, this enhanced variability was somewhat unexpected, and suggests that atmosphere forcing acts as a damping factor for TIW growth. Nevertheless, it is possible that the interactive ensemble coupling could have changed the mean state of the ocean and thereby changed the associated energy mechanisms that support the development of the TIWs. This possibility is discussed below. [15] Figure 4 compares the vertical profiles of the W averaged zonal currents simulated by CFS and CFSIE. Both models realistically represent the asymmetry in the southern and northern branches of the SEC, located on south and north side of the equator, respectively [Philander et al., 3of23

4 Figure 2. J2D spectral wave number frequency representation of SSTA variance in the domain of W at 1.5 N. Westward moving variances in (a) CFS and (b) CFSIE and eastward moving variances in (c) CFS and (d) CFSIE. The eastward moving variances are multiplied by 10 2 for better comparison with the westward moving variances. 1987] along with a strong undercurrent centered at the equator and located below 40 m (Figures 4a and 4b). In CFS, the mean zonal current in SEC as well as in the lower half of the EUC is stronger than in CFSIE with maximum located in SEC north of the equator (2 4 N), while in CFSIE, the mean zonal current is stronger than in CFS in the upper edge of the EUC branch (Figure 4c). Student s t test performed at 5% significant levels on 19 years of zonal mean currents reveals that CFS is significantly stronger in SEC around 2.5 N and in the lower branch of EUC below 130 m, while CFSIE produced stronger currents in the upper branch of EUC (Figure 4c). [16] In contrast to the differences in zonal mean currents, the meridional mean currents are similar in both CFS and CFSIE. In both models, above the mixed layer, the meridional mean currents are directed poleward in either hemisphere. Below the mixed layer, the meridional mean currents are directed equatorward. In the southern hemisphere, the poleward directed current peaks around 3 S and in the northern hemisphere at 3 N, with a maximum value of 8 cm s 1 near the surface (Figures 5a and 5b). The differences between meridional mean currents reveal that CFSIE is stronger in the northward branch of SEC and upper edge of the EUC (Figure 5c). These differences are statistically significant at 5% levels. However, the differences are not as extensive as in the case of mean zonal currents (compare Figures 4c and 5c). 4of23

5 Figure 3. J2D standard deviation of the eddy heat flux (t*u* +t*v*) in K cm s 1 in the upper 50 m over W. The solid line depicts CFS and the dashed line depicts CFSIE. [17] The eddy kinetic energy (EKE; 1 2 ½ u*2 þ v* 2 Š) due to the TIWs is examined and compared with the mean kinetic energy (MKE; 1 2 ½ u2 þ v 2 Š) in CFS and CFSIE to understand the eddy growth in the ocean in both models. In the Northern Hemisphere, CFS and CFSIE simulate two major MKE cells, one centered on the equator and the other at 2 N, owing to the existence of EUC and SEC, respectively. In both cells, the MKE in CFS is around 15% higher than the MKE in CFSIE (Figures 6a and 6b). The higher MKE in CFS is understood as a straightforward consequence of higher zonal mean currents. The differences in MKE between CFS and CFSIE closely follows the differences in zonal mean currents. CFS is significantly stronger in the northward branch of SEC, while CFSIE is significantly stronger in the upper edge of the EUC branch (Figure 6c). [18] In both CFS and CFSIE, the EKE peaks at the surface around the equator and has an asymmetric structure throughout the ocean depths with higher values in the Northern Hemisphere (Figures 7a and 7b). It is important to note that in CFS, the ratio of EKE to MKE is around 15%, on the other hand, in CFSIE, this ratio is about 35%. This suggests that in CFSIE, due to the reduced weather noise at the air-sea interface, the accelerated mean eddy interactions fuel the growth of oceanic eddies. Compared to the differences in MKE between CFS and CFSIE, the differences in EKE are rather intriguing. CFSIE is significantly stronger from 2 S 2 N within the mixed layer. Note that the difference in EKE between CFS and CFSIE relative to the total EKE in CFS is much higher (30%; compare Figure 7c with Figures 7a and 7b) compared to the differences in MKE relative to the total MKE (15%; compare Figure 6c with Figures 6a and 6b). [19] The mechanisms for the increased EKE activity in CFSIE are explored in detail. The available potential energy (APE) and exchanges between mean and eddy potential energies [Ivchenko et al., 1997] are examined as the volume integrals over the domain delimited by W in longitude, 5 S 5 N in latitude, and from the surface to 50 m in depth (this volume is referred as REG hereafter). In REG the APE in CFSIE is 10% larger compared with the APE in CFS. In CFSIE, the large value of the APE is explained by the increased mean density of ocean water in the REG. The higher APE in CFSIE has the potential to yield to an increase in the eddy potential energy (EPE) if the conversion rate also increases. In the REG, the EPE in CFSIE is also 10% larger compared to CFS. [20] The conversion rate of APE is controlled by the baroclinic processes. The baroclinic conversion rate, g(r*w*), is a measure of the energy transformation from APE to EKE [Masina et al., 1999]. Figures 8a and 8b show that in the mixed layer, the baroclinic conversion rate from the equator to 3 N is around 10% higher in CFSIE. The baroclinic conversion rate is significantly stronger in CFSIE above the mixed layer from the equator to 1 N (Figure 8c). [21] Reynolds stresses also provide a source of energy for the instabilities associated with barotropic processes and the strength of barotropic conversion rate of zonal currents is also an indicator for the strength of the eddy kinetic energy associated with TIWs. The barotropic conversion rate, r 0 U y ðu*v* Þ, measures the energy transformation between 5of23

6 Figure 4. J2D latitude and depth cross sections averaged over the domain W of the zonal mean currents in (a) CFS, (b) CFSIE, and (c) the differences between CFS and CFSIE. Stippling depicts 5% significant values and the thick black line marks the position of the mean mixed layer depth. MKE and EKE [Masina et al., 1999]. The barotropic conversion rate in CFSIE is higher than in CFS (Figures 9a and 9b). In both CFS and CFSIE, two barotropic conversion cells occur centered at 1 N and 3 N, the former having larger magnitudes. In the vertical direction, the barotropic conversion cell centered at 1 N extends from the surface to 100 m, but the peak value occurs within the mixed layer. The barotropic conversion cell centered at 3 N is stronger in the CFS compared to the CFSIE, whereas the cell centered at 1 N is stronger in the CFSIE compared to the CFS. However, the differences at 1 N are twice as strong and extensive vertically compared to the cell located at 3 N. The differences in barotropic conversion between CFS and CFSIE are statistically significant for both cells centered at 1 N and 3 N (Figure 9c). 4. Discussion [22] CFS simulations are compared with observations to understand the fidelity of the model and further to assess the reliability of the results presented in this study. The zonal mean currents simulated by CFS at 140 W and 6of23

7 Figure 5. Same as Figure 4 but for meridional mean currents. 110 W are very similar to the observations provided by Johnson et al. [2002] (compare Figure 10 with Figure 2 of Johnson et al. [2002]). The north equatorial countercurrent is slightly weaker in CFS at 140 W. Also, beyond 100 meter depths, the northward branch of south equatorial current as simulated by CFS at 110 W is weaker compared to the observations. Note that all the years of continuous data are used in producing Figure 10. [23] The time mean and standard deviation of eddy kinetic energy simulated by CFS are 20% weaker than the observed counterparts. Observations are obtained from the Tropical Atmosphere Ocean project and data are available only at 140 W on the equator (Figure 11). [24] To understand whether TIW variability is dependent on the mean density, a high- and a low-tiw regimes (8 years each) are examined in a 30 yearlong control simulation of CFS (referred to as CFSL hereinafter). Figure 12 shows the spectral variance in wave number frequency domain is much stronger in the high-tiw regime. The high- TIW years are marked with high time mean density that is significant in the top 150 meters (Figure 13). Similar analysis between CFS and CFSIE shows that the mean density is 7of23

8 Figure 6. J2D latitude depth cross sections of mean kinetic energy (MKE) (a) CFS, (b) CFSIE, and (c) differences between CFS and CFSIE. Stippling depicts 5% significant values and in all the subplots the thick black line marks the position of the mean mixed layer depth. slightly higher in the CFSIE in the mixed layer (Figure 14). The higher time mean density in the mixed layer is conductive to the growth in TIW strengths through higher values in the APE. The higher mean density in the CFSIE is very well correlated with the variance in the salinity anomaly. In CFSIE, the variance in salinity anomalies is significantly higher than in CFS on TIW timescales (Figure 15). [25] As noted in the introduction, our expectation was that the reduced noisiness in the CFSIE air-sea fluxes would reduce the ocean mixed layer depth. The increased TIW activity in CFSIE leads to precisely the opposite result with a deepening of the mixed layer at significant levels (Figure 16) as a direct result of the enhanced ocean turbulence. The higher variability of wind stress at the air-sea interface in CFS (not shown) dissipates the eddy activity closer to spatial and temporal scales of TIWs by mixing. This in turn results in eddy spectra cascading toward low wave numbers and ultimately results in reduction of TIWs (Figures 2a and 2b). 8of23

9 Figure 7. Same as Figure 6 but for EKE. On the other hand, the weaker wind stress at air-sea interface in CFSIE inhibits the dampening of the TIWs. The vigorous TIW eddies in CFSIE thus enhance the turbulence and deepen the mixed layer. [26] In CFSIE, the increase in the TIW associated eddy activity is conspicuous in the momentum fluxes throughout the mixed layer. The zonal mean depth cross sections show a significant increase of the eddy momentum ( 35%) in CFSIE compared to CFS (Figures 17a 17c). Similarly, the meridional heat flux in CFSIE is also significantly higher and marked with sharp gradients north of the equator compared to CFS (Figures 18a 18c). [27] Figure 19 shows the wave number frequency spectra of eddy kinetic energy for the westward and eastward flows. It is evident that in CFSIE the westward moving eddy kinetic energy is concentrated at smaller scales (compare Figure 19b with Figure 19a). This suggests that the TIWs contribute to heating of the ambient ocean as the eddy kinetic energy cascades from large to small scales. This is in direct agreement with increase in ambient heat flux in CFSIE (compare Figure 18b with Figure 18a). In both CFS and CFSIE, the variance associated with the eastward moving eddy kinetic energy is very small (Figures 19c and 19d). 9of23

10 Figure 8. J2D latitude and depth cross sections of the baroclinic conversion ( g(r*w*)) in g s 3 cm 1 averaged over the domain W as estimated in (a) CFS, (b) CFSIE, and(c) differences between CFS and CFSIE. Stippling depicts 5% significant values and in all the subplots the thick black line marks the position of the mean mixed layer depth. [28] The ocean energetic calculations discussed above indicate a strong negative correlation between the atmospheric internal variability and the ocean eddy activity in the eastern equatorial Pacific. Interestingly, the covariance between the wind stress and the ocean eddy activity (referred as CWE) in CFSIE is very similar in strength with the same in CFS (Figures 20a and 20b). Similarly, the covariance between heat flux and sea surface temperature anomalies (SSTA) (referred to as CHS hereinafter) is nearly conserved west of 120 W between CFS and CFSIE (Figures 21a and 21b). The differences between CFS and CFSIE in CWE as well as in CHS are statistically significant predominantly to the east of 120 W (120 W 90 W), while they are not significant to the west of 120 W ( W). Moreover, the calculations presented here indicate that the CFSIE and CFS are significantly different in terms of their TIW activity west 10 of 23

11 U Figure 9. Same as Figure 8 but for barotropic conversion r 0 y ðu*v* Þ in g s 3 rad 1. of 120 W (Figure 16) and this is the region where we find small difference in CWE and CHS. To the east of 120 W, there are small changes in the mixed layer depth that are consistent with the changes in CWE and CHS. 5. Summary [29] Both modeling and observational evidence indicate that TIWs exist primarily due to the instability mechanisms in the ocean, and yet our results show that the atmospheric internal variability influences the growth of TIWs. The atmospheric internal variability modulates the mean eddy interactions of the ocean currents and thus changes the eddy energetics. In this study, we used the interactive ensemble coupling strategy to diagnose how the atmospheric internal variability impacts the growth of the TIWs. In particular, the interactive ensemble reduces the noisiness in air-sea fluxes associated with internal atmospheric dynamics. Contrary to our assumptions, the direct result of this reduced noisiness is an enhancement of the TIWs leading to a deeper ocean mixed layer. [30] This study suggests that enhanced TIW activity is directly related to the reduced atmospheric forcing. The existence of stronger atmospheric internal variability at the air-sea interface directly results in damped TIW activity in the CFS. Examining the CFS and CFSIE simulations reveal two mechanisms responsible for the negative 11 of 23

12 Figure 10. Time mean zonal current as simulated by CFS at (a) 140 W and (b) 110 W. correlation between atmospheric internal variability and the TIW activity. In CFSIE, the reduction in the atmospheric forcing at the air-sea interface leads to enhanced mean density perturbations in the eastern equatorial Pacific. The enhanced mean density perturbations result in an increase of the available potential energy and cause higher exchanges from mean to eddy potential energies. The interactions from mean to eddy energies prompt the increase in the oceanic eddy activity. In this study, the estimated barotropic and baroclinic conversion rates, that indicate energy conversions from mean to eddy kinetic energy and potential to eddy kinetic energy, respectively, also show the reduction in atmospheric internal variability significantly enhances the eddy mean flow interactions. Because of the enhanced eddy activity in the CFSIE, the horizontal eddy momentum and heat fluxes in the ocean are significantly stronger. [31] Also somewhat surprising was the fact that ocean energy generation terms such as the covariance between momentum flux and surface currents or heat flux and SST were nearly conserved west of 120 W between CFS and CFSIE. This suggests that the coupled system is internally adjusting to the reduced air-sea flux variance in order to conserve larger-scale climatic energetic constraints. [32] One critique of these results is the fact that the ocean component model used in this study is eddy permitting not eddy resolving, and as a consequence, the TIWs differ from the observations in terms of their details. Despite this limitation, there is a robust eddy activity simulated in the model 12 of 23

13 Figure 11. Comparison of (a) mean and (b) standard deviation of eddy kinetic energy as simulated with CFS (grey) and observations (black) at 140 W on the equator. Figure 12. Westward moving wave number frequency spectral representation of SSTA variance for (a) CFSL low-tiw regime and (b) CFSL high-tiw regime. 13 of 23

14 Figure 13. J2D mean potential density (r 1000) profiles in (a) CFSL low-tiw regime, (b) CFSL high- TIW regime, and (c) differences between low- and high-tiw regimes. The zonal mean quantities are obtained by averaging over longitudes W. Stippling depicts 5% significant values. 14 of 23

15 Figure 14. Similar to Figure 13 but for CFS and CFSIE. 15 of 23

16 Figure 15. Variance in salinity anomalies on TIW time and spatial scales as shown in (a) CFS, (b) CFSIE, and (c) differences between CFS and CFSIE. stippling depicts 5% significant differences. 16 of 23

17 Figure 16. J2D mixed layer depth differences between CFSIE and CFS in centimeters. The stippling depicts 5% significant differences. 17 of 23

18 Figure 17. J2D mean meridional (over W) depth cross sections of horizontal momentum flux u*v* as estimated in (a) CFS, (b) CFSIE, and (c) differences between CFS and CFSIE. Stippling depicts 5% significant values and the thick black line marks the position of the mean mixed layer depth. 18 of 23

19 Figure 18. Same as Figure 17 but for meridional heat flux v*t*. 19 of 23

20 Figure 19. J2D spectral wave number frequency representation of eddy kinetic energy 1 2 ½ u*2 þ v* 2 Š variance in the domain of W and averaged over 2 S 2 N as shown for westward moving variances in (a) CFS and (b) CFSIE and eastward moving variances in (c) CFS and (d) CFSIE. 20 of 23

21 Figure 20. J2D covariance of horizontal surface current and wind stress eddies (~v*:~t* in g cm s 4 )in (a) CFS, (b) CFSIE, and (c) differences between CFS and CFSIE. Stippling depicts 5% significant values. 21 of 23

22 Figure 21. Same as Figure 20 but for the covariance between total downward heat flux and TIW induced SSTA eddies (Q*t*) inwm 2 C. 22 of 23

23 along with definitive air-sea feedbacks associated with these eddies. Ideally, we would like to test these results using an eddy resolving ocean model, and this is a calculation that we intend to pursue in the future. [33] Acknowledgments. This research is funded with NOAA grants NA09OAR and NA09OAR431005, NSF grant , and NASA grant NNX09AN50G. Authors gratefully acknowledge the National Centers for Environmental Prediction (NCEP) for the CFS v2 model made available to COLA. We also acknowledge NCEP s assistance in porting the code to the computing platforms at the National Center for Atmospheric Research (NCAR) and the NASA Advanced Supercomputing (NAS) division. We particularly wish to thank Y. Hou, S. Moorthi and S. Saha for technical assistance and necessary data sets and W. Lapenta and L. Uccellini for enabling the collaborative activities. Authors also acknowledge Barry Klinger of George Mason University for discussion and comments. Computing resources provided by NCAR and NAS are also gratefully acknowledged. References An, S.-I. (2008), Interannual variations of the tropical ocean instability wave and ENSO, J. Clim., 21, Chelton, D. B., S. K. Esbensen, M. G. Schlax, N. Thum, M. H. Freilich, F. J. Wentz, C. L. Gentmann, M. J. McPhaden, and P. S. Schopf (2001), Observation of coupling between surface wind stress and sea surface temperature in the eastern tropical Pacific, J. Clim., 14, Düing, W., E. K. Hisard, J. Knauss, J. Meincke, L. Miller, K. Moroshkin, G. Philander, K. V. Rybnikov, and R. Weisberg (1975), Meanders and long waves in the equatorial Atlantic, Nature, 257, Ham, Y.-G., and I.-S. Kang (2010), Improvement of seasonal forecasts with inclusion of tropical instability waves on initial conditions, Clim. Dyn., 36, , /s Hansen, D., and C. Paul (1984), Genesis and effects of long waves in the equatorial Pacific, J. Geophys. Res., 89, 10,431 10,440. Hashizume, H., S.-P. Xie, W. T. Liu, and K. Takeuchi (2001), Local and remote response to tropical instability waves: A global review from space, J. Geophys. Res., 106, 10,173 10,185. Hayes, S. P., M. J. McPhaden, and J. M. Wallace (1989), The influence of sea surface temperature on surface wind in the eastern equatorial Pacific: Weekly to monthly variability, J. Clim., 2, Imada, Y., M. Kimoto, and X. Chen (2012), Impact of atmospheric mean state on tropical instability wave activity, J. Clim., 25, , doi: /2011jcli Ivchenko, V. O., A. M. Treguier, and S. E. Best (1997), A kinetic energy budget and internal instabilities in the fine resolution Antarctic model, J. Phys. Oceanogr., 27, Jochum, M., and R. Murtugudde (2004), Internal variability of the tropical Pacific Ocean, Geophys. Res. Lett., 31, L14309, doi: /2004gl Jochum, M., C. Desser, and A. Phillips (2007), Tropical atmospheric variability forced by oceanic internal variability, J. Clim., 20, Johnson, G. C., B. M. Sloyan, W. S. Kessler, and K. E. McTaggart (2002), Direct measurements of upper ocean currents and water properties across the tropical Pacific during the 1990s, Prog. Oceanogr., 52, Kirtman, B. P., and J. Shukla (2002), Interactive coupled ensemble: A new coupling strategy for CGCMs, Geophys. Res. Lett., 29(10), 1367, doi: /2002gl Kirtman, B. P., K. Pegion, and S. M. Kinter (2005), Internal atmospheric dynamics and tropical Indo-Pacific climate variability, J. Atmos. Sci., 62, Kirtman, B. P., D. M. Straus, D. Min, E. K. Schneider, and L. Siqueira, (2009), Toward linking weather and climate in the interactive ensemble NCAR climate model, Geophys. Res. Lett., 36, L13705, doi: / 2009GL Kirtman, B. P., E. K. Schneider, D. M. Straus, D. Min, and R. Burgman, (2011), How weather impacts the forced climate response, Clim. Dyn., 37, , doi: /s Legeckis, R. (1977), Long waves in the eastern equatorial Pacific Ocean: A view from a geostationary satellite, Science, 197, Masina, S., S. Philander, and A. Bush (1999), An analysis of tropical instability waves in a numerical model of the Pacific Ocean: 2. Generation and energetics of the waves, J. Geophys. Res, 104, 29,637 29,662. Narapusetty, B., T. DelSole, and M. K. Tippett (2009), Optimal estimation of the climatological mean, J. Clim., 22, Pezzi, L. P., J. Vialard, K. J. Richards, C. Menkes, and D. Anderson (2004), Influence of ocean-atmosphere coupling on the properties of tropical instability waves, Geophys. Res. Lett., 31, L16306, doi: /2004gl Philander, S. G. H. (1976), Instabilities of zonal equatorial currents, J. Geophys. Res., 83, Philander, S. G. H., W. J. Hurlin, and A. D. Seigel (1987), Simulation of the seasonal cycle of the tropical ocean, J. Phys. Oceanogr., 17, Qiao, L., and R. Weisberg (1995), Tropical instability wave kinematics: Observations from the tropical instability wave experiment, J. Geophys. Res., 100, Saha, S., et al. (2010), The NCEP Climate Forecast System Reanalysis, Bull. Am. Meteorol. Soc., 91(8), , doi: /2010bams Seo, H., M. Jochum, R. Murtugudde, A. J. Miller, and J. O. Roads (2007), Feedback of tropical instability wave-induced atmospheric variability onto the ocean, J. Clim., 20, Stan, C., and B. P. Kirtman, (2008), The influence of atmospheric noise and uncertainty in ocean initial conditions on the limit of the predictability in a coupled GCM, J. Clim., 21, , doi: /2007jcl Straus, D. M., and J. Shukla, (1981), Space-time spectral structure of a GLAS general circulation model and a comparison with observations, J. Atmos. Sci., 38, Xie, S.-P., M. Ishiwatari, H. Hashizume, and K. Takeuchi (1998), Coupled ocean-atmospheric waves on the equatorial front, Geophys. Res. Lett., 25, Yu, Z., J. P. R. McCreary, and J. A. Proehl (1995), Meridional asymetry and energetics of tropical instability waves, J. Phys. Oceanogr., 25, of 23