Intraseasonal Variability in Coupled GCMs: The Roles of Ocean Feedbacks and Model Physics

Charlotte A. DeMott Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

Search for other papers by Charlotte A. DeMott in
Current site
Google Scholar
PubMed
Close
,
Cristiana Stan Department of Atmospheric, Oceanic and Earth Sciences, George Mason University, and Center for Ocean-Land-Atmosphere Studies, Fairfax, Virginia

Search for other papers by Cristiana Stan in
Current site
Google Scholar
PubMed
Close
,
David A. Randall Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

Search for other papers by David A. Randall in
Current site
Google Scholar
PubMed
Close
, and
Mark D. Branson Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

Search for other papers by Mark D. Branson in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The interaction of ocean coupling and model physics in the simulation of the intraseasonal oscillation (ISO) is explored with three general circulation models: the Community Atmospheric Model, versions 3 and 4 (CAM3 and CAM4), and the superparameterized CAM3 (SPCAM3). Each is integrated coupled to an ocean model, and as an atmosphere-only model using sea surface temperatures (SSTs) from the coupled SPCAM3, which simulates a realistic ISO. For each model, the ISO is best simulated with coupling. For each SST boundary condition, the ISO is best simulated in SPCAM3.

Near-surface vertical gradients of specific humidity, (temperature, ), explain ~20% (50%) of tropical Indian Ocean latent (sensible) heat flux variance, and somewhat less of west Pacific variance. In turn, local SST anomalies explain ~5% (25%) of variance in coupled simulations, and less in uncoupled simulations. Ergo, latent and sensible heat fluxes are strongly controlled by wind speed fluctuations, which are largest in the coupled simulations, and represent a remote response to coupling. The moisture budget reveals that wind variability in coupled simulations increases east-of-convection midtropospheric moistening via horizontal moisture advection, which influences the direction and duration of ISO propagation.

These results motivate a new conceptual model for the role of ocean feedbacks on the ISO. Indian Ocean surface fluxes help developing convection attain a magnitude capable of inducing the circulation anomalies necessary for downstream moistening and propagation. The “processing” of surface fluxes by model physics strongly influences the moistening details, leading to model-dependent responses to coupling.

Corresponding author address: Charlotte A. DeMott, Department of Atmospheric Science, 1371 Campus Delivery, Colorado State University, Fort Collins, CO 80523. E-mail: demott@atmos.colostate.edu

Abstract

The interaction of ocean coupling and model physics in the simulation of the intraseasonal oscillation (ISO) is explored with three general circulation models: the Community Atmospheric Model, versions 3 and 4 (CAM3 and CAM4), and the superparameterized CAM3 (SPCAM3). Each is integrated coupled to an ocean model, and as an atmosphere-only model using sea surface temperatures (SSTs) from the coupled SPCAM3, which simulates a realistic ISO. For each model, the ISO is best simulated with coupling. For each SST boundary condition, the ISO is best simulated in SPCAM3.

Near-surface vertical gradients of specific humidity, (temperature, ), explain ~20% (50%) of tropical Indian Ocean latent (sensible) heat flux variance, and somewhat less of west Pacific variance. In turn, local SST anomalies explain ~5% (25%) of variance in coupled simulations, and less in uncoupled simulations. Ergo, latent and sensible heat fluxes are strongly controlled by wind speed fluctuations, which are largest in the coupled simulations, and represent a remote response to coupling. The moisture budget reveals that wind variability in coupled simulations increases east-of-convection midtropospheric moistening via horizontal moisture advection, which influences the direction and duration of ISO propagation.

These results motivate a new conceptual model for the role of ocean feedbacks on the ISO. Indian Ocean surface fluxes help developing convection attain a magnitude capable of inducing the circulation anomalies necessary for downstream moistening and propagation. The “processing” of surface fluxes by model physics strongly influences the moistening details, leading to model-dependent responses to coupling.

Corresponding author address: Charlotte A. DeMott, Department of Atmospheric Science, 1371 Campus Delivery, Colorado State University, Fort Collins, CO 80523. E-mail: demott@atmos.colostate.edu

I. Introduction

Why does coupling an atmospheric general circulation model (AGCM) to an interactive ocean model improve the simulated intraseasonal oscillation (ISO)? The ISO is a dominant mode of tropical weather variability characterized by a large-scale, convectively coupled equatorial disturbance that propagates eastward with a phase speed of ~5 m s−1 and a period of 30–70 days (Madden and Julian 1971, 1972, 1994). The convective signal maximizes over warm sea surface temperatures (SSTs) in the tropical Indian and west Pacific Oceans. During boreal winter, the eastward-propagating ISO is often referred to as the Madden–Julian oscillation, or MJO. The boreal summer ISO (BSISO) is more complex than its wintertime counterpart, when the eastward propagation is modulated by westward- and northward-propagating modes, and is closely related to active and break periods of the Indian and East Asian summer monsoons (DeMott et al. 2013, and references therein).

Krishnamurti et al. (1988) documented intraseasonal SST variability in the Indian and west Pacific Oceans, and suggested that air–sea interaction must be important to maintaining the long-lived convection associated with the MJO. Weller and Anderson (1996) and Lau and Sui (1997) analyzed data collected during the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE; Webster and Lukas 1992) and concluded that SSTs were modulated by surfaces fluxes associated with the passage of MJO convection. Hendon and Glick (1997), Shinoda et al. (1998), and Woolnough et al. (2000) analyzed multiyear time series of SST and reanalysis surface fluxes to document the ocean’s response to intraseasonal atmospheric variability. Their results are consistent with the in situ TOGA COARE observations that point to surface insolation east of MJO convection and latent heat fluxes coincident with and west of MJO convection for modulating warm pool SSTs on intraseasonal time scales.

Woolnough et al. (2000) argued that for the MJO to be considered a coupled phenomenon, two criteria must be satisfied: 1) the atmospheric ISO must affect the ocean and 2) the MJO must be sensitive to the ocean response. Focusing on the first criterion, they found a coherent quadrature relationship between intraseasonal convection and SSTs throughout the tropical Indian and west Pacific Oceans. They concluded that intraseasonal SST anomalies are primarily the result of atmospheric processes: they maximize ~10 days prior to (east of) MJO convection in response to increased surface shortwave heating, and decrease during and after (west of) maximum convection in response to decreased surface shortwave heating and increased latent heat fluxes associated with the low-level westerly winds following MJO convection. They conclude that the atmospheric response to the ocean (the second criterion) can only be determined through modeling studies.

A variety of model experiments have examined the atmospheric response to tropical intraseasonal SST variations. These experiments provide insights into the role of air–sea coupling on tropical intraseasonal variability (ISV), although each study has its limitations. In particular, studies that compare coupled GCM (CGCM) simulations to either observations or AMIP simulations with the same AGCM (e.g., Kemball-Cook et al. 2002; Fu et al. 2002; Sperber et al. 2005; Zhang et al. 2006; Wu et al. 2008; Bollasina and Nigam 2009; Wang and Seo 2009; Roxy et al. 2013; DeMott et al. 2011) are complicated by differences in the basic state climate associated with coupled model SST biases.

One approach to reducing climate differences between CGCMs and AGCMs is to prescribe time-varying CGCM SSTs as the boundary condition for the AGCM (e.g., Fu et al. 2003; Fu and Wang 2004a,b; Zheng et al. 2004; Seo et al. 2007; Pegion and Kirtman 2008; Lestari et al. 2011; Levine and Turner 2012). Limitations of such “replay” simulations are the reduction or complete disappearance of the SST–precipitation phase lag in simulations that retain intraseasonal SST variability, a reduced dynamic range of SST anomalies in simulations based on CGCM SSTs averaged to monthly mean values, and potential differences in the magnitude and phase of surface fluxes with respect to SST.

Slab ocean or mixed-layer models coupled to AGCMs are an attractive option for studying intraseasonal air–sea interaction. The SST climatology may be specified via relaxation to a given mean state, while still allowing realistic phasing of precipitation and SST anomalies (Waliser et al. 1999; Watterson 2002; Maloney and Sobel 2004; Watterson and Syktus 2007; Marshall et al. 2008; Benedict and Randall 2011). Single-layer ocean models require a choice of mixed-layer depth, and MJO behavior can be affected by the specified depth and the relaxation time scale. Furthermore, these models cannot represent the effects of oceanic advection, or interactions with deeper water associated with entrainment, Ekman pumping (such as that which occurs in the Somali upwelling region), and/or vertical mixing (Duncan and Han 2009; Lloyd and Vecchi 2010; Vialard et al. 2012), which may help initiate some MJO events (Webber et al. 2010, 2012). Recently, Klingaman et al. (2011) coupled an AGCM to the multilevel K-profile parameterization (KPP) mixed-layer model (Large et al. 1994) and found that MJO simulation is sensitive to the ocean model’s vertical resolution and the coupling frequency.

Several studies have used hindcast simulations to assess the impact of air–sea coupling on ISO forecasts (Fu et al. 2007, 2008; Woolnough et al. 2007; Kim et al. 2008, 2010; Seo et al. 2009; Kim and Kang 2008; Wang et al. 2009). These studies employ a variety of ocean boundary conditions (BCs), including fully dynamic ocean models, intermediate complexity ocean models, single- or multilayer slab ocean models, and specified daily observed SSTs or SSTs generated from coupled simulations. Inclusion of high-frequency SST variability or interactive coupling generally extends forecast skill by ~1 week and, in some cases, improves the northward propagation of the BSISO.

The studies summarized above differ from one another in a variety of ways, including season studied (boreal summer vs boreal winter), region of interest (Indian Ocean vs west Pacific Ocean), ocean BC, and model physics (i.e., parameterizations of convection, boundary layer, microphysics, etc.). In all but a few cases (Ajayamohan et al. 2011; Levine and Turner 2012), coupling and/or specified SSTs that retain intraseasonal variability improved the simulation of the ISO over that of the AGCM with monthly mean climatological SSTs, even in cases where the AGCM produced very little ISO activity or propagation (Zheng et al. 2004; Watterson and Syktus 2007; Klingaman et al. 2008). Among the improvements cited were stronger ISO variability in AGCMs with a weak or nonexistent MJO, reduced ISO variability in AGCMs with excessive variability, more realistic phase speed (either faster or slower than in the AGCM) and wavenumber–frequency spectra, more realistic horizontal and vertical structure of the MJO, better eastward propagation (not always accompanied by better northward propagation during boreal summer), and better northward propagation during boreal summer season (not always accompanied by better eastward propagation). Clearly, there are nuanced model-to-model differences in the nature of improvement due to coupling, and a few studies suggest either that the coupling strength is too strong in some models (Wu et al. 2008; Bollasina and Nigam 2009) or that the coupled SST variability can be too high (Roxy et al. 2013).

Factors attributed to the improved ISO in coupled simulations include enhanced low-level convergence and convective instability east (north) of convection, which improves eastward (northward) propagation, and enhanced latent heat fluxes and SST cooling west (south) of convection, which could help regulate ISO (BSISO) periodicity. In a series of boreal summer modeling studies that limited ocean coupling to either the Indian Ocean or the west Pacific Ocean, Weng and Yu (2010), Lin et al. (2011), and Achuthavarier and Krishnamurthy (2011) concluded that coupling in the Indian Ocean is more important than west Pacific Ocean coupling to the BSISO northward propagation. In a similar study, Wu and Kirtman (2005) found that replacing coupling with monthly mean SSTs in either ocean basin led to excessive intraseasonal variability due to a lack of negative feedback on SSTs.

Despite the wealth of experiments focused on understanding the impacts of air–sea interaction on the ISO, the physical mechanisms by which coupling affects convection on intraseasonal time scales remain obscure. For example, several studies (e.g., Fu et al. 2003; Roxy and Tanimoto 2007; Roxy et al. 2013) attribute the increased low-level convergence to increased shallow convection, which results from enhanced SST-induced convective instability. Some studies suggest a wind–evaporation feedback (e.g., Marshall et al. 2008) whereby increased shortwave fluxes lead to enhanced surface evaporation and low-level convergence, while others (Hsu and Li 2012) suggest that intraseasonal SST gradients themselves can induce modest low-level convergence anomalies according to the ideas of Lindzen and Nigam (1987) and Back and Bretherton (2009). Benedict and Randall (2011) advocate a dynamic generation of the low-level convergence via Kelvin waves that are induced by ISO convection.

Regardless of which mechanisms influence convection, they all involve temporal and spatial variations in turbulent surface fluxes. Therefore, understanding the sensitivity of the ISO to ocean coupling hinges on understanding its sensitivity to these fluxes, since the atmosphere responds to the fluxes, and not the SST anomalies themselves (Zhang 2005). Evaluation of ISO flux sensitivity is complicated by the range of model physics employed in AGCMs. With few exceptions (e.g., Zhang et al. 2006; Seo and Wang 2010), the studies summarized above focused solely on the role of ocean BC, rather than model physics, on the ISO. However, given the sensitivity of the ISO to cumulus parameterization (Lin et al. 2008; Jia et al. 2010) it is possible that different physics may yield different sensitivities to coupling.

Here we examine the interaction of coupling and model physics on the ISO in a suite of CGCM and AGCM simulations. Several questions motivate this study: Why does coupling improve intraseasonal variability? How does ocean BC affect air–sea interaction? How is coupling sensitivity influenced by the choice of model physics? The models used to address these questions are described in section 2, while their seasonal mean states, variability, and rainfall–SST relationships are described in section 3. The sensitivity of surface fluxes to ocean BC and model physics is analyzed in section 4, while their relation to the intraseasonal moisture budget is presented in section 5. Results are discussed and summarized in section 6.

2. Models and data

This study is motivated by the improved ISO simulation in the superparameterized (SP) Community Climate System Model (CCSM), version 3 (SPCCSM3; Stan et al. 2010; DeMott et al. 2011, 2013) compared to the atmosphere-only and traditionally parameterized versions of that model. The atmospheric component of SPCCSM3 is the Community Atmospheric Model, version 3 (CAM3; Collins et al. 2006a,b). Rather than using a cumulus parameterization, the cloud and radiation physics tendencies in SPCCSM3 are calculated with a two-dimensional cloud-resolving model in each GCM grid column (Khairoutdinov and Randall 2003).

We compare SPCCSM3 to three different versions of the National Center for Atmospheric Research (NCAR) Community Atmospheric Model (CAM): CAM3, version 4 (CAM4), and superparameterized version 3 (SPCAM3). In this study, CAM3 is run with a semi-Lagrangian dynamical core (as is SPCAM3) and the cumulus parameterizations of Zhang and McFarlane (1995) and Hack (1994). CAM4 adopts a new land surface model, a finite-volume dynamical core and, importantly, modifies the Zhang–McFarlane parameterization by adding an entraining plume assumption and convective momentum transports (Neale et al. 2008). Additional model details are given in Table 1. While CAM3 provides the best comparison to SPCAM3, CAM4 is a more recent version of the NCAR AGCM and produces better intraseasonal variability than CAM3 (Subramanian et al. 2011).

Table 1.

Model physics and ocean boundary condition for each of the nine simulations.

Table 1.

Using the “replay” approach described in section 1, SSTs from SPCCSM3 are applied to all AGCM simulations to eliminate differences in mean state SST. Two sets of AGCM experiments are performed: 1) a 5-day running mean filter is applied to SPCCSM3 SSTs, and 2) SPCCSM3 SSTs are averaged to monthly mean values and interpolated to daily resolution according to Taylor et al. (2000). Two coupled simulations are also performed with CAM3 and CAM4 (CCSM3 and CCSM4, respectively). The coupled simulations have different mean SST states than SPCCSM3, but allow comparison of the fully coupled rainfall–SST relationship in those models.

The following observational and reanalysis datasets are used for comparison: daily rainfall and SST time series (1998–2010) from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI; Kummerow et al. 1998), National Oceanic and Atmospheric Administration (NOAA) outgoing longwave radiation (OLR; Liebmann and Smith 1996), and Interim European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-Interim, hereafter ERAI; Simmons et al. 2006) surface fluxes and atmospheric data (1990–2010). TRMM Microwave-Infrared (Gentemann et al. 2004) global 0.1-degree SST (2006–12) is averaged to model grid resolution for sensitivity tests requiring daily observed SSTs (see Table 1).

3. Rainfall and SST: Variability and covariability

Annual mean tropical rainfall biases in coupled and “_5d” simulations are shown in Fig. 1. Coupled simulations are compared to TMI rainfall (Figs. 1a–c), while the _5d AGCM simulations—all of which have the SPCCSM3 SST climatology—are compared to TMI (Figs. 1d–f) and SPCCSM3 rainfall (Figs. 1g–i). Precipitation biases in the coupled models (top row) reveal common biases in all three models, such as the southern Pacific and western Indian Oceans, but large differences in the northern Pacific Ocean. When the AGCMs are driven with SPCCSM3 SSTs, the mean biases are more similar (Figs. 1d–f). AGCM-SPCCSM3 mean rainfall differences (Figs. 1g–i) are smaller and more localized than the biases for coupled simulations. Therefore, utilizing SPCCSM3 SSTs does not completely eliminate mean-state differences among the AGCMs, but it reduces the differences relative to those in free-running coupled simulations. Results for the “_mon” simulations are similar and not shown.

Fig. 1.
Fig. 1.

Annual mean precipitation bias of coupled models compared to (a)–(c) TRMM 3B42 1998–2010 (mm day−1) and _5d models, (d)–(f) TRMM 3B42 and _5d models, and (h)–(j) SPCCSM3. Results for _mon models (not shown) are similar to those in (d)–(i).

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Intraseasonal (20–100 day) precipitation variance biases are shown in Fig. 2. Tropical ISV biases in SPCCSM3 (Fig. 2c) are small in the Indian and west Pacific Oceans, with large positive biases in the subtropics. In contrast, CCSM3 and CCSM4 both produce large areas of negative ISV bias in the Indian and west Pacific Oceans (Figs. 2b,c). For the _5d simulations (Figs. 2d–f), the absolute value of the biases increases slightly in all simulations, except for the South Pacific convergence zone in SCPAM3_5d. Compared to SPCCSM3, ISV differences in SPCAM3_5d remain small, suggesting that ISV in that model is not particularly sensitive to ocean BC (Fig. 2i). In contrast, CAM3_5d and CAM4_mon ISV differences are large (especially in the northern Indian Ocean) and remain widespread (Figs. 2g,h). Despite the same SST climatology, ISV in CAM3 and CAM4 is much weaker than in SPCAM3, highlighting the importance of model physics for ISV. Spatially, results for the _mon simulations (not shown) are similar to the _5d simulations, but with further reductions in variance.

Fig. 2.
Fig. 2.

As in Fig. 1, but for 20–100-day precipitation variance biases [(mm day−1)2].

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

In Fig. 2, it is impossible to tell if the larger negative ISV differences in CAM3 and CAM4 (compared to CCSM3 or CCSM4) result from the absence of coupling or from differences in mean SST or SST variability. Coupled SST biases and SST variance biases are shown in Fig. 3. Compared to CCSM3 and CCSM4, SPCCSM3 SSTs are colder and less variable, especially in the Indian Ocean. It is possible that at least some of the ISV decrease in CAM3 and CAM4 simulations (compared to CCSM3 or CCSM4) arises from the use of prescribed SSTs. The combined effects of mean ocean temperature and variability among all models can be seen in Fig. 4, which shows the region where any given ocean grid point warms to 29°C at least 5% of the time. SPCCSM3 produces the coldest SSTs of the coupled models. The effects of temporally averaging the SPCCSM3 SSTs for the replay experiments are seen in the family of red curves. The SST climatology with the red curves is identical, but the region of intermittently high SSTs is reduced, particularly in the equatorial western Pacific Ocean.

Fig. 3.
Fig. 3.

(a)–(c) Annual mean SST bias (K) and (d)–(f) 20–100-day SST variance bias (K2) of coupled models compared to 1998–2010 TRMM TMI SST.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Fig. 4.
Fig. 4.

Regions where the probability of SST ≥ 29°C exceeds 5% for CCSM3 (blue), CCSM4 (green), and SPCCSM3 (solid red). Results for the _5d (red dashed) and _mon (red dotted) simulations are also shown.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Eastward-propagating anomalies of 20–100-day filtered rainfall and SST from all seasons for observations and each model are shown in Fig. 5. All of the coupled models (Figs. 5b–d) simulate at least some degree of eastward-propagating rainfall, which lags warm SST anomalies by ~10 days. Evidence of standing wave behavior is seen in CCSM3 and, to a lesser extent, CCSM4, while SPCCSM3 simulates true eastward propagation. When the AGCMs are run using SPCCSM3 SSTs, a variety of responses are seen. Both CAM3 simulations (Fig. 5e,h) generate a westward-propagating disturbance while CAM4 has both westward and eastward propagation (Figs. 5f,i). SPCAM3 (Figs. 5g,j) retains robust eastward propagation with 5-day running mean SSTs, and weaker eastward propagation with monthly mean SSTs. When coupling is removed, precipitation maxima tend to be collocated with SST maxima, but the correlations are lower compared to the coupled simulations. This is because the atmosphere no longer modulates the SSTs, and precipitation is only weakly coincident with warm SSTs.

Fig. 5.
Fig. 5.

All-season tropical (10°S–10°N) 20–100-day rainfall (shaded) and SST (red contoured) anomalies correlated with 20–100-day tropical Indian Ocean (10°S–5°N, 75°–100°E) rainfall as a function of lag and longitude. Rainfall contour interval is 0.2, beginning at ±0.1 and positive (negative) correlations are dark (light) shaded. Positive SST anomalies are contoured (0.1 contour interval) with local 95% confidence of r ~ 0.2 emphasized.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Within each model (i.e., panels in the same column), eastward propagation weakens with the removal of coupling and intraseasonal SST variability. For each ocean BC (i.e., panels in the same row), eastward propagation improves from left to right as a function of model physics. Although it is difficult to identify the relative importance of model physics, ocean BC, and mean state from this analysis, it appears that the SP ISO simulations are the least sensitive to ocean BC of the three models. Composites of the poleward-propagating ISO are shown in Fig. 6. Similar to the eastward propagation results, coupling produces the most realistic behavior within each model class (column), and SP produces the most realistic behavior for each ocean BC (row).

Fig. 6.
Fig. 6.

All-season Bay of Bengal (80°–100°E) 20–100-day rainfall (shaded) and SST (red contoured) anomalies correlated with 20–100-day tropical Indian Ocean (10°S–5°N, 75°–100°E) rainfall as a function of lag and latitude. Shading and contours as in Fig. 5.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

The intraseasonal precipitation–SST phase angle relationship on GCM grid scales is shown in Fig. 7. The phase angle was computed as the arctangent of the 20–100-day averaged real and imaginary components of the precipitation–SST cross-spectrum. Six (three) 3-yr, nonoverlapping time series were used in the model (observation) calculations. In observations (Fig. 7a) the quarter-cycle offset of precipitation and SST is seen near the equator with slightly larger offsets in the subtropics. The coupled models (Figs. 7b–d) all exhibit smaller offsets than the observations, with CCSM3 having the smallest offsets and SPCCSM3 having the largest. CCSM4 and SPCCSM3 both simulate entraining convection, which may help increase the phase angle over that of the nonentraining CCSM3 convection. For the _5d and _mon simulations (Figs. 7e–g and 7h–j, respectively), the phase angles are nearly zero and less significant (as indicated by the 0.1 Coh2 contour) than the coupled simulations. The widespread low coherence-squared (Coh2) values in the _5d simulations imply that precipitation is not strongly phase-locked to SST anomalies. At GCM grid scales, intraseasonal SST anomalies in the _5d simulations appear as nearly random noise to intraseasonal precipitation anomalies.

Fig. 7.
Fig. 7.

All-season anomalous rainfall-SST 20–100-day mean phase angle (shaded) and Coh2 (contour interval 0.1) for (a) TMI, (b)–(d) coupled models, (e)–(g) atmosphere-only models with SPCCSM3 5-day running mean SSTs, and (h)–(j) atmosphere-only models with SPCCSM3 monthly mean SSTs interpolated to daily mean values. Values are plotted only when the result exceeds p = 0.95 confidence interval.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

4. Sensitivity of surface turbulent fluxes to model physics and ocean boundary condition

Our analysis of the sensitivity of surface fluxes to model physics and ocean BC begins with a survey of terms that contribute to surface fluxes. For latent heat flux (LH),
e1
where is the density of near-surface air, L is the latent heat of vaporization, CH is the exchange coefficient, is the lowest model level wind speed, and is the near-surface vertical moisture gradient, defined as the difference between surface saturation specific humidity (estimated using the surface temperature) and the lowest model layer specific humidity. Overbars represent the seasonal average of unfiltered variables, and primed quantities are time-filtered departures from the mean annual cycle. Sensible heat flux (SH) is decomposed in a similar manner.

Assessing surface flux sensitivities to model physics and ocean BC requires an understanding of how both the mean (i.e., , , ) and time-varying quantities (i.e., , , ; here and hereafter the primes are omitted) in Eq. (1) change as a function of each. Figure 8 presents December –February (DJF) LH and related terms averaged over the Indian Ocean (10°S–10°N, 50°–100°E). For a given variable, the columns in Fig. 8 (from left to right) present the area averages of the mean, 20–100-day variance, and regression coefficient of the 20–100-day filtered, normalized (i.e., mean of zero, standard deviation of one) variable onto the filtered and normalized SST anomaly (expressed as a percentage of the standard deviation, ). Across the suite of simulations and observations, , , and (the near-surface mean specific humidity) are approximately equal, as seen by the small vertical displacements and slopes in the left-hand column. Variances of these quantities, however (column 2), are sensitive to both model physics (vertical displacements) and ocean BC (sloped solid lines). For each variable, the variance and SST regression coefficient generally decrease from left to right (i.e., from coupled to monthly mean SST ocean BCs). The variance fraction that is directly tied to SST fluctuations is the square of the fractional σ in column 3. SST variance accounts for ≤5% of the total LH, and variance, and up to ~5% of variance (Fig. 11, right column). This implies that the local influence of SST on LH is small, even in coupled simulations. Instead, SST fluctuations—especially in a coupled ocean–atmosphere system—enhance intraseasonal LH variability remotely via increased wind speed anomalies. The sensitivity of wind speed anomalies to ocean BC is probably complex and tied to larger-scale circulation changes.

Fig. 8.
Fig. 8.

(a)–(c) Latent heat flux mean, variance, and percent of standard deviation for a 1-σ SST anomaly, respectively, averaged over the Indian Ocean (10°S–10°N, 50°–100°E; see Fig. 10) in DJF for ERAI (black), SPCAM3 (red), CAM3 (blue), and CAM4 (green) model results, and coupled (C), 5-day running mean (5d), and monthly mean (mon) SPCCSM3 SSTs ocean BCs. Mean, variance, and percent-of-standard deviation values are also shown for (d)–(f) wind speed, (g)–(i) Δq, and (j)–(l) qair. Area averages are indicated with a dot. Solid (dashed) lines connect simulations with the same (different) mean SST climatology. Vertical displacements among dots that share the same SST climatology arise from differences in model physics. Sloped solid lines arise from differences in ocean BC for a given model. Percent of variance explained by SST anomalies for any given term is the square of the fractional percent of standard deviation shown in column 3. A 20–100-day bandpass filter is applied to all time series before the regressions are performed.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Results for SH are shown in Fig. 9. There is more intermodel spread in mean quantities (vertical displacements in , , and ) but the larger differences are again seen in the variances of these terms. Similar to the LH results, variance and SST dependence decrease from left to right. A notable difference between the LH and SH results is the larger dependence of SH and variability on SST.

Fig. 9.
Fig. 9.

As in Fig. 8, but for sensible heat flux, ΔT, and Tair.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

The sensitivity of surface fluxes to SST is examined by applying the chain rule to Eq. (1)
e2
and evaluating each quotient. All terms in Eq. (2) are evaluated locally (at each grid point), but may be influenced by remote large-scale mean and/or transient circulations. This is especially true of the terms in Eqs. (1) and (2), which exert a “remote” control on local LH fluxes, whereas terms involving or tend to exert more “local” control.

Note that and are estimated by regressing LH anomalies onto time series of and , where all time series are filtered and normalized as described above. Results for ERAI DJF are shown in Fig. 10. Near 10°S, LH and are out of phase (negative regression coefficients) because large positive vertical moisture gradients are observed during the suppressed phase of the ISO when surface winds are weak (Fig. 10a). In contrast, tropical LH- regressions are large (Fig. 10b). Applying the regression coefficient to the or time series and multiplying by the local or standard deviation and the factor that approximates allows computation of the LH variance predicted by each term in Eq. (1) (Figs. 10c,d). Tropical LH variability is small compared to that observed in the Northern Hemisphere subtropics, where cold air outbreaks increase its variability. The fraction of total LH accounted for by each term is shown in Figs. 10e and 10f and illustrates the regional contrast of LH control. The difference between Figs. 10e and 10f (Fig. 10g) depicts the boundary where local LH control changes from to .

Fig. 10.
Fig. 10.

Regression coefficients of normalized LH regressed onto (a) normalized and (b) near-surface . Also shown are (c) -predicted and (d) -predicted latent heat flux variance ([W m−2]2) and the ratio of (e) -predicted and (f) -predicted latent heat flux variance to total latent heat flux variance. (g) Difference [(f) − (e)] of fractional -predicted variance and fractional -predicted variance. All time series were filtered to retain 20–100-day variability. Indian Ocean and west Pacific averaging regions used in Figs. 8, 12, and 13 are indicated with black outlines.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

The same analysis is applied to SH anomalies (Fig. 11). The tropics–subtropics differences between the contributions of vertical temperature gradient and wind speed are not as distinct as for LH anomalies. Although SH variance is smaller than LH variance (Figs. 11c,d vs Fig. 10c,d), plays a larger role in tropical SH than does in tropical LH (Fig. 11a vs Fig. 10a). The ratio difference between - and -predicted SH (Fig. 11g) points to the importance of the local for tropical sensible heat fluxes, particularly in the Indian Ocean.

Fig. 11.
Fig. 11.

As in Fig. 10, but for SH, , and .

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

The analysis shown in Figs. 10 and 11 was applied to all nine simulations for DJF and June–August (JJA) and the results are broadly similar to those for ERAI. Namely, LH anomalies are dominated by near the equator, and away from the equator, while SH anomalies exhibit less distinct equator–subtropics boundaries between and wind speed control. Figure 12 summarizes the DJF and JJA area-averaged regression coefficients for all simulations averaged over the Indian Ocean and west Pacific Ocean regions shown in Figs. 10 and 11. For both seasons, Indian Ocean is generally larger than west Pacific (Figs. 12a–d), but always less than (Figs. 12e–h), which explains ~80%–90% of LH fluctuations (~70% of the total variance). CAM4 (SPCAM3) generally has the largest (smallest) values. Results for SH (Figs. 12i–p) illustrate the larger role of than , and larger intermodel differences in both the and relationships. With few exceptions (CAM3 in Fig. 12a), LH (SH) in CAM3 and CAM4 is more (less) sensitive to than in SPCAM3, suggesting greater (less) local control of LH (SH) for those two models.

Fig. 12.
Fig. 12.

Indian Ocean and west Pacific area averages of (a)–(d) and (e)–(h) during (left) DJF and (right) JJA. Also shown are the (i)–(l) and (m)–(p) averages.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Area-averaged and [the remaining terms in Eq. (2)] for each ocean basin and season are shown in Fig. 13. SST influence on and is largest in the coupled simulations, yet still accounts for less than ~50% of those fluctuations. This suggests that and are most sensitive to the overlying specific humidity or air temperature, respectively. Except for ERAI and SP during DJF, local SST anomalies play a larger role in and variability (and therefore LH and SH) in the Indian Ocean than in the west Pacific Ocean, which is consistent with previous studies (Wu and Kirtman 2005; Weng and Yu 2010; Lin et al. 2011; Achuthavarier and Krishnamurthy 2011).

Fig. 13.
Fig. 13.

DJF and JJA regression coefficients (expressed as a percentage of one standard deviation) of (a)–(d) and (e)–(h) for a 1-σ SST anomaly for the Indian Ocean and west Pacific Ocean averaging boxes shown in Fig. 10. A 20–100-day bandpass filter is applied to all time series before the regressions are performed.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

The conclusions of this analysis can be summarized as follows: Tropical LH anomalies are small compared to those in the subtropics. Tropical LH anomalies are primarily driven by wind speed anomalies, whereas subtropical LH anomalies are determined by both and . Mean latent heat fluxes and their associated variables (, , and ) are only slightly sensitive to model physics and ocean BC. Variability about the mean, however, is sensitive to both model physics and ocean BC. Local SST anomalies account for a small but nonnegligible fraction (~5%) of LH variance; the rest is largely determined by wind speed fluctuations. For a given model, the sensitivity of wind speed variability to ocean BC probably involves changes in convective intensity and large-scale circulation anomalies, which suggests a spatially distributed (rather than local) adjustment to SST fluctuations. SH variance in both the tropics and subtropics is more directly influenced by SST anomalies (15%–25% total variance).

5. Atmospheric moistening on intraseasonal time scales

In the previous section, we demonstrated that ocean coupling induces important changes to wind speed variability. In the tropics, large-scale circulation anomalies are induced by convective heating anomalies (i.e., Webster 1972; Gill 1980), which exhibit a close relationship to the vertical moisture profile (Thayer-Calder and Randall 2009). In this section, we analyze the moisture budget to understand the interaction of surface turbulent fluxes, wind speed variability, atmospheric moistening, and ISO convection.

Maloney (2009) analyzed the moist static energy (MSE) budget of a modified version of CAM3 and found that lower-tropospheric intraseasonal MSE is dominated by specific humidity fluctuations, consistent with the weak temperature–gradient approximation. Furthermore, during the onset of an ISO convective event, the increase in vertically integrated MSE is dominated by horizontal moisture advection. Kiranmayi and Maloney (2011) analyzed ERAI and National Centers for Environmental Prediction (NCEP) reanalysis datasets and observed horizontal advective moistening during ISO preconditioning, but also a larger contribution from vertical moisture advection compared to the Maloney (2009) results.

We focus our moisture budget analysis on a composite ISO event when convection is located in the eastern Indian Ocean. This location was chosen to address the role of coupling on the eastward propagation of ISO convection across the Maritime Continent and into the west Pacific Ocean, which is an especially challenging process for many models. For the ERAI and each model, all-season 20–100-day filtered rainfall, surface variables, and vertical profiles of atmospheric moisture budget terms—tendency, plus horizontal and vertical advection—are regressed onto the all-season 20–100-day filtered rainfall anomaly averaged over 10°S–5°N and 70°–100°E, which is the “base point” time series defined by Waliser et al. (2009). As discussed by Maloney (2009), the vertically integrated vertical moisture advection may be calculated as either or , where the brackets indicate vertical integration over the depth of the troposphere. The second expression (vertical moisture advection) results in a smaller residual in the vertically integrated moisture budget (not shown) and is used for our vertical profile analyses. When vertically resolved, the lowest model level of the first expression (moisture convergence) is useful for identifying regions where low-level convergence may induce convection.

Composite surface fields for ERAI and each model are shown in Fig. 14. For a given variable, x, the units plotted are (x)/(mm day−1). In all cases, eastern Indian Ocean ISO rainfall is associated with east-of-convection surface moisture convergence that extends over part or most of the Maritime Continent (MC), and is collocated with negative LH anomalies. The western and central Pacific Ocean are characterized by positive LH anomalies (see also lower panels of Figs. 1518) in the ERAI, CAM4, and SP simulations.

Fig. 14.
Fig. 14.

Regression coefficients of latent heat flux anomalies (shaded), lowest-level winds (vectors), horizontal moisture convergence (red contours), and precipitation (green contours) regressed onto tropical Indian Ocean precipitation anomalies averaged over 10°S–5°N, 75°–100°E. Contour interval is [(1 × 10-6 g kg−1 s−1)/(mm day−1)] for moisture convergence and [(0.5 mm day−1)/(mm day−1)] for rainfall. Minimum contour corresponds to significant values at the p = 0.95 confidence level. Wind vectors (drawn every other grid point) and latent heat flux are only plotted when their regression meets the same significance criteria. A 20–100-day bandpass filter as applied to all time series before the regressions are performed.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Fig. 15.
Fig. 15.

(a)–(c) 10°N–10°S averaged regression coefficients of longitude–pressure cross sections of 20–100-day filtered moisture budget equation terms, plus (d) the residual term when regressed onto 20–100-day filtered eastern Indian Ocean precipitation. Contour interval = (1 × 10−7 g kg−1 s−1)/(mm day−1). In (a)–(d), mean specific humidity profile (minus zonal mean) is shaded, and surface rainfall is overlaid (green; peak value is 1 mm day−1). In (a), standard deviation of base point rainfall. In (e), scaled regression coefficients of surface latent and sensible heat flux [units = (1 W m−2)/(mm day−1)], SST anomaly [units = (0.01 K)/(mm day−1)], and surface pressure [units = (0.1 mb)/(mm day−1)]. Maritime Continent landmasses are indicated with heavy black lines.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Fig. 16.
Fig. 16.

As in Fig. 15, but for (a)–(e) SPCCSM3, (f)–(j) SPCAM3_5d, and (k)–(o) SPCAM3_mon.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Fig. 17.
Fig. 17.

As in Fig. 15, but for (a)–(e) CCSM4, (f)–(j) CAM4_5d, and (k)–(o) CAM4_mon.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Fig. 18.
Fig. 18.

As in Fig. 15, but for (a)–(e) CCSM3, (f)–(j) CAM3_5d, and (k)–(o) CAM3_mon.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

Evidence of frictional wave–conditional instability of the second kind (wave-CISK) (Wang and Rui 1990) is seen in all panels of Fig. 14, where low-level winds over the MC and west Pacific converge into the convectively induced low pressure anomaly in that region. Presumably, this low-level moisture convergence is enhanced by central Pacific LH in ERAI, CAM4 (coupled and uncoupled), and SP simulations, and helps maintain a strong convective anomaly as the ISO propagates across the MC. Closer inspection of Fig. 14, however, suggests that surface moisture convergence cannot account for the differences in propagation behavior among models. For instance, CAM4 and the SP models simulate nearly identical surface convergence, but SPCAM3 simulates better eastward propagation. Similarly, although CAM3 has weaker surface convergence overall, its presence over the MC in all three simulations is inconsistent with the westward propagation in the uncoupled runs (Figs. 14e,h). While frictional wave-CISK and surface turbulent fluxes are important processes in the ISO, they cannot explain ISO propagation characteristics in all models analyzed herein.

The vertical profile of the composite moisture budget illustrates the convective linkage between surface fluxes and ISO moistening processes. Following Yanai et al. (1973), we compute the apparent moisture sink (divided by the latent heat of vaporization, L), , by evaluating the 20–100-day filtered terms in the square bracket of Eq. (3) and computing as a residual, which represents the net moistening and subgrid-scale vertical eddy transport of water vapor by convection:
e3
where and are the horizontal and vertical wind velocity, respectively, c and e are the condensation and evaporation of water vapor, and is the unresolved vertical eddy transport of water vapor. For simulations that saved the computed subgrid-scale moistening diagnostic, there is essentially no difference between the residual and the explicitly calculated subgrid-scale moistening. When working with model output, the budget terms in Eq. (3) were computed on model hybrid pressure coordinate levels and then interpolated to fixed pressure levels. ERAI results were computed directly from fixed pressure levels. As in Fig. 14, filtered time series of each term in Eq. (3) at each vertical level are regressed onto the area-averaged filtered eastern Indian Ocean precipitation time series. Results are averaged from 10°N to 10°S and presented as longitude–pressure cross sections.

The composite moisture budget terms for ERAI are shown in Fig. 15. The mean background specific humidity (minus the zonal mean) is shaded, and the contour interval for all budget terms is (1 × 10−7 g−1 kg−1 s−1)/(mm day−1). Composite rainfall is shown with the thin green line (positive anomalies only; peak value is 1 mm day−1), and the standard deviation of rainfall in the base point region is indicated in the top panel. Regression coefficients for LH and SH, SST and surface pressure anomalies are shown in the bottom panel. When ISO precipitation is in the eastern Indian Ocean, peak moistening occurs immediately to the east at ~500 mb, and gradually descends and weakens over the MC and west Pacific. The maximum moistening is collocated with the minimum LH over the MC, highlighting the need for processes other than local surface evaporation to provide the moistening necessary for eastward convective development. Vertical moisture advection is the largest budget term, but is largely offset by condensation in rainy areas. The residual term (plotted as − for consistency) shows that the subgrid-scale convection, and the associated positive LH anomalies, moistens the lower troposphere in the central Pacific. Kiranmayi and Maloney (2011) caution that the residual term in the vertically integrated ERAI moisture budget is as large as the moistening term. This residual may be linked to missing or miscalculated moistening processes in the reanalysis data, and may lead to erroneously large contributions from the vertical advection term.

Moisture budgets for the three SP simulations are shown in Fig. 16. For all simulations, the results are qualitatively similar to those for ERAI. The residual profiles in the central and eastern Pacific Ocean indicate the crucial role of shallow cumulus and cumulus congestus in moistening the preconvective environment. Comparing all three simulations yields insight into the role of coupling and, for the uncoupled simulations, SST variability on ISO propagation. Recall that wind speed variability is largest (smallest) in SPCCSM3 (SPCAM3_mon). The reduction of LH in the central and eastern Pacific Ocean (bottom panels) results in less moisture being detrained into the lower troposphere by shallow convection (residual panels). Since SST anomalies in this region are close to zero in all three simulations, the reduced surface fluxes in the uncoupled simulations are the result of weaker wind—and not SST-related —fluctuations. The combination of weaker wind speed variability and reduced convectively driven eddy vertical moisture transports reduces the westward transport of moisture by horizontal advection, which leads to reduced total moistening (top panels). Compared to the fully coupled simulation, MC congestus developing in the drier midtroposphere of the uncoupled simulations will struggle to mature and produce the deep convective heating profile and associated circulation anomalies that drive the downstream (i.e., central Pacific) LH and convective moistening anomalies, thereby reducing subsequent advective moistening and limiting eastward propagation of the ISO.

Moisture budgets for the CAM4 and CAM3 simulations are shown in Figs. 17 and 18, respectively. Both CCSM4 and CCSM3 produce a top-heavy mean specific humidity profile over the west Pacific (shaded field in Figs. 17 and 18). The top-heaviness is reduced in the simulations driven with SPCCSM3 SSTs, but its persistence points to the dominance of convection over sea surface climatology in determining the mean tropical moisture profile. Resolved east-of-convection moistening in CCSM4 (Fig. 17a) and horizontal moisture advection over the MC (Fig. 17c) are weaker than those simulated by SPCCSM3. Moistening by convective processes in the west Pacific (Fig. 17d) is large in order to counteract the upper-level subsidence in this region (Fig. 17b), but there is a distinct gap near the date line, suggesting that the convective linkage between surface evaporation in the east Pacific and ISO convection in the west Pacific breaks down in this region. In CCSM3 (Fig. 18, left column) east Pacific LH anomalies are nonexistent, and there is simply no drying by vertical advection or compensating subgrid-scale moistening of the lower troposphere, so the MC moistening sequence is prevented.

When CAM4 is driven with SPCCSM3 SSTs (center and right columns of Fig. 17), the results resemble those seen in the SP models (Fig. 16). Compared to CCSM4, CAM4_5d resolved midlevel moistening is enhanced east of and near existing convection, horizontal advective moistening occurs over a greater portion of the MC, central Pacific LH is enhanced, and convective moistening (the residual) exhibits a continuous low- to upper-tropospheric moistening. In CAM4_mon, the east-of-convection moistening disappears, as horizontal advective moistening is replaced with weak advective drying.

When CAM3 is driven with SPCCSM3 SSTs (center and right columns of Fig. 18), midlevel moistening (top panels) is seen west of existing convection, particularly in the CAM3_mon simulation. Since these are the only two simulations analyzed with westward-propagating ISO convection, the evidence suggests that ISO propagation direction is toward regions that are moistened by midlevel horizontal moisture advection.

While the sensitivity of ISO propagation to midlevel moisture advection seems clear, its connection to ocean surface fluxes is less so. Latent heat fluxes in the Pacific Ocean are largely determined by convectively generated wind speed anomalies (Figs. 10 and 12). In the central (far west) Pacific, the wind anomalies enhance (diminish) the climatological winds, leading to positive (negative) surface fluxes. Indian Ocean heating anomalies are larger in SPCCSM3 than in the uncoupled SP simulations, as shown by the standard deviations of the rainfall anomaly in Fig. 16 (top panels). This is consistent with the greater surface fluxes at peak rainfall location in SPCCSM3 (Fig. 16 bottom panels), which are somewhat enhanced by local SST variations in the coupled configuration (Figs. 8b,c). The enhanced Indian Ocean convection forces stronger central Pacific compensating subsidence and subgrid-scale moistening, and induces a radiatively driven west Pacific circulation response that enhances midlevel moisture advection (Kim et al. 2014). In summary, the coupling processes that lead to more robust Indian Ocean convection induce stronger low-level moistening by subgrid-scale processes in the central and west Pacific, and stronger midlevel MC moisture advection, enabling the ISO to propagate into the west Pacific.

As discussed in section 4, Indian Ocean and sensitivity to SST (and therefore LH and SH sensitivity) is larger in the Indian Ocean than in the west Pacific. We hypothesize that, during the developing stages of the ISO, Indian Ocean surface fluxes modulate the (de)stabilization of the lower troposphere and, through their interaction with model physics, help organize ISO convection to a sufficiently large spatial scale that generates circulation anomalies necessary for eastward propagation. Our findings suggest that future studies of the effects of coupling on the ISO should focus on the interactions of surface fluxes and model physics during the period of developing eastern Indian Ocean convection.

6. Discussion and conclusions

In this study, we have examined the sensitivity of the ISO to three different ocean BCs with three different GCMs (CAM3, CAM4, and SPCAM3). Each model was coupled to a fully dynamic ocean model, and then run as an atmosphere-only model using 1) 5-day filtered and 2) monthly mean SSTs from the best-performing coupled model, SPCCSM3. For each atmospheric model, the coupled simulation always produced the best ISO. For a given ocean BC (fully coupled, prescribed 5-day or monthly SSTs), the superparameterized model always produced the best ISO. Both CAM3 and CAM4 ISO simulations degraded in the uncoupled simulations, while the SPCAM3 ISO degraded only slightly.

The means of surface turbulent fluxes, near-surface vertical moisture and temperature gradients, and wind speeds are not very sensitive to ocean BC, but their variability is. Furthermore, the variability of these terms differs from one model to the next (i.e., they are sensitive to model physics). In all models, local SST variations account for about <5% of tropical LH variability; the rest is controlled by wind speed variability, which is not closely related to local SST fluctuations. This underscores the fact that understanding ISO sensitivity to ocean BC requires study of both local surface flux changes and remote circulation changes. Sensible heat fluxes are much smaller than latent heat fluxes (~one order of magnitude), but are more sensitive to local SST fluctuations, especially in the Indian Ocean, where 15%–25% of sensible heat flux variance is explained by SST variability.

To understand the role of SST-influenced surface fluxes on eastward propagation of the ISO, we analyzed the spatial structure of moisture budget terms when ISO convection is located in the eastern Indian Ocean. All simulations—regardless of ISO propagation direction—simulated at least some low-level moisture convergence east of ISO convection over the Maritime Continent. In the ERAI and all simulations that produce at least some eastward propagation, net midlevel moistening is observed east of ISO convection over the Maritime Continent despite the collocated negative surface flux anomalies. Consistent with the findings of Benedict and Randall (2009), Maloney (2009), Kiranmayi and Maloney (2011), and Wu and Deng (2013), this midlevel moistening has significant contributions from horizontal moisture advection, which is centered at ~700 hPa. The residual term, which represents moistening by unresolved convective condensation/evaporation and vertical eddy moisture fluxes, is characterized by shallow moistening in the east Pacific that gradually deepens toward the west Pacific. This vertical transport of moisture from the surface to the lower troposphere in the east Pacific is a source of midlevel moisture that is advected over the Maritime Continent, enabling the slow eastward propagation of the ISO into the west Pacific. For the two simulations that produce a westward-propagating ISO (CAM3_5d and CAM3_mon), midlevel moistening and horizontal moisture advection maximize west of convection. The link between propagation direction and midlevel horizontal advection for both eastward and westward propagating ISO convection supports the paradigm that the ISO is a moisture mode disturbance, in which convective development is governed by the environmental moisture supply (e.g., Raymond and Fuchs 2009; Sobel and Maloney 2013).

Our moisture budget analysis shows more robust moistening in coupled than uncoupled simulations, but the moistening processes—enhanced central Pacific latent heat fluxes and upward moisture transport, and midlevel moisture advection over the Maritime Continent—are primarily associated with enhanced wind speed variability, and not SST fluctuations. We argue that the enhanced wind speed variability in the coupled models is the result of more robust Indian Ocean convection that develops in coupled simulations. This leads us to conclude that coupling, and its influence on surface turbulent fluxes, helps to organize ISO convection as it develops in the Indian Ocean. Coupling may more directly impact developing convection in the Indian Ocean, where and contribute relatively more to latent and sensible heat fluxes (Figs. 10 and 11) and are more sensitive to SST variations (Fig. 13) than in the west Pacific.

Enhanced surface fluxes by themselves, however, are not sufficient to produce an ISO when coupling is introduced. The final result ultimately depends upon how the model physics incorporates these fluxes into the growth and organization of convection to achieve a robust, realistic heating profile capable of generating the crucial wind anomalies. More realistic convective entrainment of environmental air that arises from either explicit simulation of convection (e.g., Benedict and Randall 2011) or modifications to cumulus parameterizations (Hannah and Maloney 2011; Zhou et al. 2012; Crueger et al. 2013; Klingaman and Woolnough 2014) is important for improving the links between surface fluxes and convective development. This is why coupling does not always produce the same degree of improvement when applied to different models.

These processes are illustrated schematically in Fig. 19 for models that do (bottom panels) and do not (top panels) produce an ISO in their uncoupled state. A model that simulates frequent, disorganized convection (Fig. 19a) cannot generate the circulation anomalies needed for east-of-convection moistening. However, if surface fluxes in that model are sufficiently or overly sensitive to or , the time-varying SSTs of a coupled simulation may provide a level of organization that allows those circulations to develop (Fig. 19b). For models that simulate the ISO in their uncoupled state (Fig. 19c), the ingredients for eastward propagation already exist, and coupling enhances them (Fig. 19d).

Fig. 19.
Fig. 19.

Plan view illustration of equatorial (thin horizontal line in each panel) circulation anomalies and moistening processes for an eastward propagating ISO for a model with a (top) weak and (bottom) realistic ISO in its uncoupled configuration. Convective organization (strength) is indicated by the degree of clustering (size) of gray cloud elements. Convectively driven circulation anomalies are shown with red ellipses. Warm (cold) SST anomalies in the coupled simulations are shown with light red (blue) filled ovals. Moderate (strong) latent heat flux anomalies are shown with light (dark) green shading. Black dots are low-level moistening by unresolved convective processes. Vertically oriented green arrows represent low-level moisture convergence into the equatorial trough, and the left-pointing green arrow is midlevel moisture advection.

Citation: Journal of Climate 27, 13; 10.1175/JCLI-D-13-00760.1

To fully understand the effects of ocean coupling on the ISO, it would be useful to study the moisture budget under realistic coupling conditions (where the ocean responds to the atmosphere) while also being able to constrain the SST climatology. A potential approach to this problem is to couple many different AGCMs to the same ocean mixed layer model, which is continuously nudged to climatological SSTs. Marshall et al. (2008) proposed such a study using a slab ocean model, but coupling to a multiple-layer mixed layer model, such as done in Klingaman et al. (2011), would allow a more realistic representation of the upper ocean. Finally, a uniform set of air–sea interaction diagnostics could help illuminate the broader sensitivity of ocean coupling to model physics.

Acknowledgments

This work was support by NSF Grants AGS-111999, AGS-1211848, and the National Science Foundation Science and Technology Center for Multi-Scale Modeling of Atmospheric Processes, managed by Colorado State University under cooperative agreement ATM-0425247. We acknowledge the support of the Computational and Information Systems Laboratory at NCAR for computer resources and the European Center for Medium-Range Forecasts for the ERA-Interim reanalysis data. TMI data are produced by Remote Sensing Systems and sponsored by the NASA Earth Science MEaSUREs DISCOVER Project. Data are available at www.remss.com. We thank N. Klingaman, K.-H. Seo, and E. Maloney for their insightful comments and suggestions on this paper.

REFERENCES

  • Achuthavarier, D., and V. Krishnamurthy, 2011: Role of Indian and Pacific SST in Indian summer monsoon intraseasonal variability. J. Climate, 24, 29152930, doi:10.1175/2010JCLI3639.1.

    • Search Google Scholar
    • Export Citation
  • Ajayamohan, R. S., H. Annamalai, J.-J. Luo, J. Hafner, and T. Yamagata, 2011: Poleward propagation of boreal summer intraseasonal oscillations in a coupled model: Role of internal processes. Climate Dyn., 37, 851867, doi:10.1007/s00382-010-0839-6.

    • Search Google Scholar
    • Export Citation
  • Back, L. E., and C. S. Bretherton, 2009: On the relationship between SST gradients, boundary layer winds, and convergence over the tropical oceans. J. Climate, 22, 41824196, doi:10.1175/2009JCLI2392.1.

    • Search Google Scholar
    • Export Citation
  • Benedict, J. J., and D. A. Randall, 2009: Structure of the Madden–Julian oscillation in the superparameterized CAM. J. Atmos. Sci., 66, 32773296, doi:10.1175/2009JAS3030.1.

    • Search Google Scholar
    • Export Citation
  • Benedict, J. J., and D. A. Randall, 2011: Impacts of idealized air–sea coupling on Madden–Julian oscillation structure in the superparameterized CAM. J. Atmos. Sci., 68, 19902008, doi:10.1175/JAS-D-11-04.1.

    • Search Google Scholar
    • Export Citation
  • Bollasina, M., and S. Nigam, 2009: Indian Ocean SST, evaporation, and precipitation during the South Asian summer monsoon in IPCC-AR4 coupled simulations. Climate Dyn., 33, 10171032, doi:10.1007/s00382-008-0477-4.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2006a: The formulation and atmospheric simulation of the Community Atmosphere Model version 3 (CAM3). J. Climate, 19, 21442161, doi:10.1175/JCLI3760.1.

    • Search Google Scholar
    • Export Citation
  • Collins, W. D., and Coauthors, 2006b: The Community Climate System Model version 3 (CCSM3). J. Climate, 19, 21222143, doi:10.1175/JCLI3761.1.

    • Search Google Scholar
    • Export Citation
  • Crueger, T., B. Stevens, and R. Brokopf, 2013: The Madden–Julian oscillation in ECHAM6 and the introduction of an objective MJO metric. J. Climate, 26, 32413257, doi:10.1175/JCLI-D-12-00413.1.

    • Search Google Scholar
    • Export Citation
  • Danabasoglu, G., W. G. Large, J. J. Tribbia, P. R. Gent, B. P. Briegleb, and J. C. McWilliams, 2006: Diurnal coupling in the tropical oceans of CCSM3. J. Climate, 19, 23472365, doi:10.1175/JCLI3739.1.

    • Search Google Scholar
    • Export Citation
  • DeMott, C. A., C. Stan, D. A. Randall, J. L. Kinter, and M. Khairoutdinov, 2011: The Asian monsoon in the superparameterized CCSM and its relationship to tropical wave activity. J. Climate, 24, 51345156, doi:10.1175/2011JCLI4202.1.

    • Search Google Scholar
    • Export Citation
  • DeMott, C. A., C. Stan, and D. A. Randall, 2013: Northward propagation mechanisms of the boreal summer intraseasonal oscillation in the ERA-Interim and SP-CCSM. J. Climate, 26, 19731992, doi:10.1175/JCLI-D-12-00191.1.

    • Search Google Scholar
    • Export Citation
  • Duncan, B., and W. Han, 2009: Indian Ocean intraseasonal sea surface temperature variability during boreal summer: Madden–Julian oscillation versus submonthly forcing and processes. J. Geophys. Res., 114, C05002,doi:10.1029/2008JC004958.

    • Search Google Scholar
    • Export Citation
  • Fu, X., and B. Wang, 2004a: Differences of boreal summer intraseasonal oscillations simulated in an atmosphere–ocean coupled model and an atmosphere-only model. J. Climate, 17, 12631271, doi:10.1175/1520-0442(2004)017<1263:DOBSIO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Fu, X., and B. Wang, 2004b: The boreal-summer intraseasonal oscillations simulated in a hybrid coupled atmosphere–ocean model. Mon. Wea. Rev., 132, 26282649, doi:10.1175/MWR2811.1.

    • Search Google Scholar
    • Export Citation
  • Fu, X., B. Wang, and T. Li, 2002: Impacts of air–sea coupling on the simulation of mean Asian summer monsoon in the ECHAM4 model. Mon. Wea. Rev., 130, 28892904, doi:10.1175/1520-0493(2002)130<2889:IOASCO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Fu, X., B. Wang, T. Li, and J. P. McCreary, 2003: Coupling between northward-propagating, intraseasonal oscillations and sea surface temperature in the Indian Ocean. J. Atmos. Sci., 60, 17331753, doi:10.1175/1520-0469(2003)060<1733:CBNIOA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Fu, X., B. Wang, D. E. Waliser, and L. Tao, 2007: Impact of atmosphere–ocean coupling on the predictability of monsoon intraseasonal oscillations. J. Atmos. Sci., 64, 157174, doi:10.1175/JAS3830.1.

    • Search Google Scholar
    • Export Citation
  • Fu, X., B. Yang, Q. Bao, and B. Wang, 2008: Sea surface temperature feedback extends the predictability of tropical intraseasonal oscillation. Mon. Wea. Rev., 136, 577597, doi:10.1175/2007MWR2172.1.

    • Search Google Scholar
    • Export Citation
  • Gent, P. R., and Coauthors, 2011: The Community Climate System Model version 4. J. Climate, 24, 49734991, doi:10.1175/2011JCLI4083.1.

    • Search Google Scholar
    • Export Citation
  • Gentemann, C. L., F. J. Wentz, C. M. Mears, and D. K. Smith, 2004: In situ validation of Tropical Rainfall Measuring Mission microwave sea surface temperatures. J. Geophys. Res., 109, C04021, doi:10.1029/2003JC002092.

    • Search Google Scholar
    • Export Citation
  • Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447462, doi:10.1002/qj.49710644905.

    • Search Google Scholar
    • Export Citation
  • Hack, J. J., 1994: Parameterization of moist convection in the National Center for Atmospheric Research community climate model (CCM2). J. Geophys. Res., 99, 55515568, doi:10.1029/93JD03478.

    • Search Google Scholar
    • Export Citation
  • Hannah, W. M., and E. D. Maloney, 2011: The role of moisture–convection feedbacks in simulating the Madden–Julian oscillation. J. Climate, 24, 27542770, doi:10.1175/2011JCLI3803.1.

    • Search Google Scholar
    • Export Citation
  • Hendon, H. H., and J. Glick, 1997: Intraseasonal air–sea interaction in the tropical Indian and Pacific Oceans. J. Climate, 10, 647661, doi:10.1175/1520-0442(1997)010<0647:IASIIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Hsu, P.-C., and T. Li, 2012: Role of the boundary layer moisture asymmetry in causing the eastward propagation of the Madden–Julian oscillation. J. Climate, 25, 49144931, doi:10.1175/JCLI-D-11-00310.1.

    • Search Google Scholar
    • Export Citation
  • Jia, X., C. Li, N. Zhou, and J. Ling, 2010: The MJO in an AGCM with three different cumulus parameterization schemes. Dyn. Atmos. Oceans, 49, 141163, doi:10.1016/j.dynatmoce.2009.02.003.

    • Search Google Scholar
    • Export Citation
  • Kemball-Cook, S., B. Wang, and X. Fu, 2002: Simulation of the intraseasonal oscillation in the ECHAM-4 model: The impact of coupling with an ocean model. J. Atmos. Sci., 59, 14331453, doi:10.1175/1520-0469(2002)059<1433:SOTIOI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M. F., and D. A. Randall, 2003: Cloud resolving modeling of the ARM summer 1997 IOP: Model formulation, results, uncertainties, and sensitivities. J. Atmos. Sci., 60, 607625, doi:10.1175/1520-0469(2003)060<0607:CRMOTA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kim, D., J.-S. Kug, and A. H. Sobel, 2014: Propagating versus nonpropagating Madden–Julian oscillation events. J. Climate, 27, 111125, doi:10.1175/JCLI-D-13-00084.1.

    • Search Google Scholar
    • Export Citation
  • Kim, H.-M., and I.-S. Kang, 2008: The impact of ocean–atmosphere coupling on the predictability of boreal summer intraseasonal oscillation. Climate Dyn., 31, 859870, doi:10.1007/s00382-008-0409-3.

    • Search Google Scholar
    • Export Citation
  • Kim, H.-M., C. D. Hoyos, P. J. Webster, and I.-S. Kang, 2008: Sensitivity of MJO simulation and predictability to sea surface temperature variability. J. Climate, 21, 53045317, doi:10.1175/2008JCLI2078.1.

    • Search Google Scholar
    • Export Citation
  • Kim, H.-M., C. D. Hoyos, P. J. Webster, and I.-S. Kang, 2010: Ocean–atmosphere coupling and the boreal winter MJO. Climate Dyn., 35, 771784, doi:10.1007/s00382-009-0612-x.

    • Search Google Scholar
    • Export Citation
  • Kiranmayi, L., and E. D. Maloney, 2011: Intraseasonal moist static energy budget in reanalysis data. J. Geophys. Res., 116, D21117, doi:10.1029/2011JD016031.

    • Search Google Scholar
    • Export Citation
  • Klingaman, N. P., and S. J. Woolnough, 2014: The role of air–sea coupling in the simulation of the Madden–Julian oscillation in the Hadley Centre model. Quart. J. Roy. Meteor. Soc., doi:10.1002/qj.2295, in press.

  • Klingaman, N. P., P. M. Inness, H. Weller, and J. M. Slingo, 2008: The importance of high-frequency sea surface temperature variability to the intraseasonal oscillation of Indian monsoon rainfall. J. Climate, 21, 61196140, doi:10.1175/2008JCLI2329.1.

    • Search Google Scholar
    • Export Citation
  • Klingaman, N. P., S. J. Woolnough, H. Weller, and J. M. Slingo, 2011: The impact of finer-resolution air–sea coupling on the intraseasonal oscillation of the Indian monsoon. J. Climate, 24, 24512468, doi:10.1175/2010JCLI3868.1.

    • Search Google Scholar
    • Export Citation
  • Krishnamurti, T. N., D. K. Oosterhof, and A. V. Mehta, 1988: Air–sea interaction on the time scale of 30 to 50 days. J. Atmos. Sci., 45, 13041322, doi:10.1175/1520-0469(1988)045<1304:AIOTTS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kummerow, C., W. Barnes, T. Kozu, J. Shiue, and J. Simpson, 1998: The Tropical Rainfall Measuring Mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809817, doi:10.1175/1520-0426(1998)015<0809:TTRMMT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Large, W. G., J. C. McWilliams, and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization. Rev. Geophys., 32, 363403, doi:10.1029/94RG01872.

    • Search Google Scholar
    • Export Citation
  • Lau, K.-M., and C.-H. Sui, 1997: Mechanisms of short-term sea surface temperature regulation: Observations during TOGA COARE. J. Climate, 10, 465472, doi:10.1175/1520-0442(1997)010<0465:MOSTSS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lestari, R. K., M. Watanabe, and M. Kimoto, 2011: Role of air–sea coupling in the interannual variability of the South China Sea summer monsoon. J. Meteor. Soc. Japan, 89A, 283290, doi:10.2151/jmsj.2011-A18.

    • Search Google Scholar
    • Export Citation
  • Levine, R. C., and A. G. Turner, 2012: Dependence of Indian monsoon rainfall on moisture fluxes across the Arabian Sea and the impact of coupled model sea surface temperature biases. Climate Dyn., 38, 21672190, doi:10.1007/s00382-011-1096-z.

    • Search Google Scholar
    • Export Citation
  • Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 12751277.

    • Search Google Scholar
    • Export Citation
  • Lin, A., T. Li, X. Fu, J.-J. Luo, and Y. Masumoto, 2011: Effects of air–sea coupling on the boreal summer intraseasonal oscillations over the tropical Indian Ocean. Climate Dyn., 37, 23032322, doi:10.1007/s00382-010-0943-7.

    • Search Google Scholar
    • Export Citation
  • Lin, J.-L., K. M. Weickman, G. N. Kiladis, B. E. Mapes, S. D. Schubert, M. J. Suarez, J. T. Bacmeister, and M.-I. Lee, 2008: Subseasonal variability associated with Asian summer monsoon simulated by 14 IPCC AR4 coupled GCMs. J. Climate, 21, 45414567, doi:10.1175/2008JCLI1816.1.

    • Search Google Scholar
    • Export Citation
  • Lin, S.-J., 2004: A “vertically Lagrangian” finite-volume dynamical core for global models. Mon. Wea. Rev., 132, 22932307, doi:10.1175/1520-0493(2004)132<2293:AVLFDC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lin, S.-J., and R. B. Rood, 1996: Multidimensional flux-form semi-Lagrangian transport schemes. Mon. Wea. Rev., 124, 20462070, doi:10.1175/1520-0493(1996)124<2046:MFFSLT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lindzen, R. S., and S. Nigam, 1987: On the role of sea surface temperature gradients in forcing low level winds and convergence in the tropics. J. Atmos. Sci., 44, 24182436, doi:10.1175/1520-0469(1987)044<2418:OTROSS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Lloyd, I. D., and G. A. Vecchi, 2010: Submonthly Indian Ocean cooling events and their interaction with large-scale conditions. J. Climate, 23, 700716, doi:10.1175/2009JCLI3067.1.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in zonal wind in the tropical Pacific. J. Atmos. Sci., 28, 702708, doi:10.1175/1520-0469(1971)028<0702:DOADOI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci., 29, 11091123, doi:10.1175/1520-0469(1972)029<1109:DOGSCC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1994: Observations of the 40–50-day tropical oscillation—A review. Mon. Wea. Rev.,122, 814–837, doi:10.1175/1520-0493(1994)122<0814:OOTDTO>2.0.CO;2.

  • Maloney, E. D., 2009: The moist static energy budget of a composite tropical intraseasonal oscillation in a climate model. J. Climate, 22, 711729, doi:10.1175/2008JCLI2542.1.

    • Search Google Scholar
    • Export Citation
  • Maloney, E. D., and A. H. Sobel, 2004: Surface fluxes and ocean coupling in the tropical intraseasonal oscillation. J. Climate, 17, 43684386, doi:10.1175/JCLI-3212.1.

    • Search Google Scholar
    • Export Citation
  • Marshall, A. G., O. Alves, and H. H. Hendon, 2008: An enhanced moisture convergence–evaporation feedback mechanism for MJO air–sea interaction. J. Atmos. Sci., 65, 970986, doi:10.1175/2007JAS2313.1.

    • Search Google Scholar
    • Export Citation
  • Neale, R. B., J. H. Richter, and M. Jochum, 2008: The impact of convection on ENSO: From a delayed oscillator to a series of events. J. Climate, 21, 59045924, doi:10.1175/2008JCLI2244.1.

    • Search Google Scholar
    • Export Citation
  • Neale, R. B., J. H. Richter, S. Park, P. H. Lauritzen, S. J. Vavrus, P. J. Rasch, and M. Zhang, 2013: The mean climate of the Community Atmosphere Model (CAM4) in forced SST and fully coupled experiments. J. Climate, 26, 51505168, doi:10.1175/JCLI-D-12-00236.1.

    • Search Google Scholar
    • Export Citation
  • Pegion, K., and B. P. Kirtman, 2008: The impact of air–sea interactions on the simulation of tropical intraseasonal variability. J. Climate, 21, 66166635, doi:10.1175/2008JCLI2180.1.

    • Search Google Scholar
    • Export Citation
  • Raymond, D. J., and Z. Fuchs, 2009: Moisture modes and the Madden–Julian oscillation. J. Climate, 22, 30313046, doi:10.1175/2008JCLI2739.1.

    • Search Google Scholar
    • Export Citation
  • Richter, J. H., and P. J. Rasch, 2008: Effects of convective momentum transport on the atmospheric circulation in the Community Atmosphere Model, version 3. J. Climate, 21, 14871499, doi:10.1175/2007JCLI1789.1.

    • Search Google Scholar
    • Export Citation
  • Roxy, M., and Y. Tanimoto, 2007: Role of SST over the Indian Ocean in influencing the intraseasonal variability of the Indian summer monsoon. J. Meteor. Soc. Japan, 85, 349358, doi:10.2151/jmsj.85.349.

    • Search Google Scholar
    • Export Citation
  • Roxy, M., Y. Tanimoto, B. Preethi, P. Terray, and R. Krishnan, 2013: Intraseasonal SST–precipitation relationship and its spatial variability over the tropical summer monsoon region. Climate Dyn., 41, 4561, doi:10.1007/s00382-012-1547-1.

    • Search Google Scholar
    • Expo