Simulation of the South Asian Monsoon in a Coupled Model with an Embedded Cloud-Resolving Model

V. Krishnamurthy Center for Ocean-Land-Atmosphere Studies, Institute of Global Environment and Society, and Department of Atmospheric, Oceanic and Earth Sciences, George Mason University, Fairfax, Virginia

Search for other papers by V. Krishnamurthy in
Current site
Google Scholar
PubMed
Close
,
Cristiana Stan Center for Ocean-Land-Atmosphere Studies, Institute of Global Environment and Society, and Department of Atmospheric, Oceanic and Earth Sciences, George Mason University, 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
,
Ravi P. Shukla Center for Ocean-Land-Atmosphere Studies, Institute of Global Environment and Society, Fairfax, Virginia

Search for other papers by Ravi P. Shukla in
Current site
Google Scholar
PubMed
Close
, and
James L. Kinter III Center for Ocean-Land-Atmosphere Studies, Institute of Global Environment and Society, and Department of Atmospheric, Oceanic and Earth Sciences, George Mason University, Fairfax, Virginia

Search for other papers by James L. Kinter III in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The simulation of the South Asian monsoon by a coupled ocean–atmosphere model with an embedded cloud-resolving model is analyzed on intraseasonal and interannual time scales. The daily modes of variability in the superparameterized Community Climate System Model, version 3 (SP-CCSM), are compared with those in observation, the superparameterized Community Atmospheric Model, version 3 (SP-CAM3), and the control simulation of CCSM (CT-CCSM) with conventional parameterization of convection. The CT-CCSM fails to simulate the observed intraseasonal oscillations but is able to generate the atmospheric El Niño–Southern Oscillation (ENSO) mode, although with regular biennial variability. The dominant modes of variability extracted from daily anomalies of outgoing longwave radiation, precipitation, and low-level horizontal wind in SP-CCSM consist of two intraseasonal oscillations and two seasonally persisting modes, in good agreement with observation. The most significant observed features of the intraseasonal oscillations correctly simulated by the SP-CCSM are the northward propagation of convection, precipitation, and circulation as well as the eastward and westward propagations. The observed spatial structure and the periods of the oscillations are also well captured by the SP-CCSM, although with lesser magnitude. The SP-CCSM is able to simulate the chaotic variability and spatial structure of the seasonally persisting atmospheric ENSO mode, while the evidence for the Indian Ocean dipole mode is inconclusive. The SP-CAM3 simulates two intraseasonal oscillations and the atmospheric ENSO mode. However, the intraseasonal oscillations in SP-CAM3 do not show northward propagation while their periods and spatial structures are not comparable to observation. The results of this study indicate the necessity of coupled models with sufficiently realistic cloud parameterizations.

Corresponding author address: V. Krishnamurthy, Center for Ocean-Land-Atmosphere Studies, Research Hall, MS 6C5, George Mason University, 4400 University Drive, Fairfax, VA 22030. E-mail: krishna@cola.iges.org

Abstract

The simulation of the South Asian monsoon by a coupled ocean–atmosphere model with an embedded cloud-resolving model is analyzed on intraseasonal and interannual time scales. The daily modes of variability in the superparameterized Community Climate System Model, version 3 (SP-CCSM), are compared with those in observation, the superparameterized Community Atmospheric Model, version 3 (SP-CAM3), and the control simulation of CCSM (CT-CCSM) with conventional parameterization of convection. The CT-CCSM fails to simulate the observed intraseasonal oscillations but is able to generate the atmospheric El Niño–Southern Oscillation (ENSO) mode, although with regular biennial variability. The dominant modes of variability extracted from daily anomalies of outgoing longwave radiation, precipitation, and low-level horizontal wind in SP-CCSM consist of two intraseasonal oscillations and two seasonally persisting modes, in good agreement with observation. The most significant observed features of the intraseasonal oscillations correctly simulated by the SP-CCSM are the northward propagation of convection, precipitation, and circulation as well as the eastward and westward propagations. The observed spatial structure and the periods of the oscillations are also well captured by the SP-CCSM, although with lesser magnitude. The SP-CCSM is able to simulate the chaotic variability and spatial structure of the seasonally persisting atmospheric ENSO mode, while the evidence for the Indian Ocean dipole mode is inconclusive. The SP-CAM3 simulates two intraseasonal oscillations and the atmospheric ENSO mode. However, the intraseasonal oscillations in SP-CAM3 do not show northward propagation while their periods and spatial structures are not comparable to observation. The results of this study indicate the necessity of coupled models with sufficiently realistic cloud parameterizations.

Corresponding author address: V. Krishnamurthy, Center for Ocean-Land-Atmosphere Studies, Research Hall, MS 6C5, George Mason University, 4400 University Drive, Fairfax, VA 22030. E-mail: krishna@cola.iges.org

1. Introduction

The South Asian monsoon exhibits strong variability on subseasonal and interannual time scales. These variations include intraseasonal oscillations with different periods and seasonal-mean patterns that may be related to the sea surface temperature (SST) of the Indian and Pacific Oceans (see, e.g., Krishnamurthy and Kinter 2003). The simulation and prediction of the monsoon have been a challenging problem even for advanced general circulation models (GCMs) used in operational forecasts (Achuthavarier and Krishnamurthy 2011a; Drbohlav and Krishnamurthy 2010; Rai and Krishnamurthy 2011). Some of the most important physical processes in the tropics, such as convection and cloud formation, are represented through parameterizations in the models. The lack of explicit treatment of physical processes leads to large systematic errors in climate simulation on different space and time scales (Guilyardi et al. 2009), with the largest errors arising from the parameterization of clouds (Randall et al. 2007). The reason for using parameterization is related to the fact that the horizontal resolution of the coupled models is still not high enough to permit the explicit representation of physical processes. An approach to solve this problem is based on a multiscale modeling framework (MMF) in which large-scale atmospheric circulation is represented on a coarse grid, while an embedded high-resolution cloud-resolving model (CRM) explicitly treats convection and other physical processes (Grabowski 2001; Khairoutdinov and Randall 2001). A recent study by Stan et al. (2010) has demonstrated that a coupled GCM (CGCM) with a two-dimensional CRM improves the simulation of such phenomena as the Madden–Julian oscillation (MJO) and El Niño–Southern Oscillation (ENSO).

Based on the hypothesis of Charney and Shukla (1981) that the interannual variability of the tropical climate is mainly determined by slowly varying components such as SST, a conceptual model describes the monsoon variability as a combination of seasonally persisting large-scale patterns and subseasonal oscillations (Krishnamurthy and Shukla 2000). Observational studies have provided evidence for the existence of two dominant nonlinear intraseasonal oscillations and two seasonally persisting modes in precipitation, outgoing longwave radiation (OLR), and three-dimensional circulation (Krishnamurthy and Shukla 2007, 2008; Krishnamurthy and Achuthavarier 2012). The leading intraseasonal oscillation was also noted by other studies (e.g., Annamalai and Sperber 2005). The first oscillation has a period centered at 45 days and propagates northeastward from the Indian Ocean to the Indian subcontinent and western Pacific while the second oscillation propagates northwestward with a period of 30 days. These two oscillations describe the active and break phases of the monsoon but do not contribute much to the seasonal mean. There are other higher frequency oscillations and synoptic disturbances but with less variance. The two seasonally persisting modes consist of large-scale patterns varying with the same sign throughout the monsoon season but changing on the interannual scale. These modes are related to ENSO and the Indian Ocean dipole (IOD) with strong lead–lag relation with the SST of the Pacific and Indian Oceans (Krishnamurthy and Kirtman 2009), respectively, and together account for the interannual variability of the seasonal-mean monsoon rainfall. Any good model should properly simulate the period, propagation, and spatial structure of the intraseasonal oscillations and the seasonal persistence and interannual variation of large-scale patterns.

The MMF approach is expected to improve the spatiotemporal organization of convection because of the interactions among processes of wide ranging spatial and temporal scales present in the CRM and GCM. The embedding of CRMs in a coarser GCM is also called a superparameterization (SP) of small-scale processes. A study by Stan et al. (2010) embedded CRMs in the atmospheric component of the Community Climate System Model (CCSM) of the National Center for Atmospheric Research (NCAR) and analyzed its climate simulation. This model, referred to as SP-CCSM, was shown to reproduce the observed period, spatial structure, and eastward propagation of MJO and the chaotic variability of ENSO well, while the CCSM with conventional parameterization (without SP) was not so successful.

The SP-CCSM simulation in the Asian monsoon region was analyzed by DeMott et al. (2011, 2013) to show that the propagation of equatorial Rossby and mixed Rossby–gravity waves were better simulated compared to CCSM and atmospheric GCM with SP and that the mechanisms for the northward propagation of the boreal summer intraseasonal oscillation are well captured by the SP-CCSM. However, these studies did not examine the lower-frequency oscillations that are more dominant on the intraseasonal time scale nor the seasonally persisting modes. Goswami et al. (2012) analyzed the same SP-CCSM simulation in the monsoon region and concluded that the model did not show northward propagation of intraseasonal oscillation at a 30–60-day time scale. However, their analysis was performed on bandpass-filtered data under a priori assumptions on the periods of the oscillations. Their analysis did not isolate the oscillations with 45 and 30 periods but filtered the data in a much broader frequency range and also used different frequency bands for observations and model data. Their study also did not address the simulation of SST-related seasonally persisting modes that determine the interannual variability of the seasonal-mean monsoon. The simulation of the South Asian monsoon in coupled models has also been studied in model intercomparison projects. For example, the monsoon simulations in phases 3 and 5 of the Coupled Model Intercomparison Project (CMIP3 and CMIP5) were analyzed by Sperber et al. (2013). However, their study did not assess the performance of the models based on convection parameterization.

The objective of this study is to investigate how realistically the monsoon variability is simulated by SP-CCSM on intraseasonal and interannual time scales. Specifically, the model’s ability to simulate the periods, spatial structures, and propagation properties of the observed intraseasonal oscillations of the 45- and 30-day periods will be examined. Further, whether the model can simulate the seasonally persisting modes related to ENSO and IOD will be investigated. For this purpose, this study will adopt the same method used by earlier studies of observations (Krishnamurthy and Shukla 2007, 2008; Krishnamurthy and Achuthavarier 2012). To also show the importance of ocean–atmosphere coupling, comparison will be made with simulation by the superparameterized atmospheric GCM forced by observed SST.

Section 2 describes the model simulations, observed data, and the method used in this study. The mean monsoon and the basic modes of monsoon variability in the models will be compared with those of observations in section 3. The intraseasonal oscillations will be discussed in section 4, while section 5 will present the analysis of seasonally persisting modes. Section 6 will provide summary and discussion.

2. Models, data, and method of analysis

a. Models and data

The NCAR CCSM, version 3 (Collins et al. 2006), used in this study consists of the Community Atmospheric Model, version 3 (CAM3), with T42 horizontal resolution as its atmospheric component and the low-resolution (3.6° in longitude and varying in latitude) Parallel Ocean Program (POP) model as the ocean component. The conventional parameterization of convection by Zhang and McFarlane (1995) is used in CCSM. The two-dimensional CRM used in SP and its incorporation in CAM are described by Khairoutdinov and Randall (2001, 2003), and a similar implementation of SP in CCSM was performed by Stan et al. (2010). In this study, a 20-yr-long simulation by SP-CCSM is analyzed and compared with a similar control simulation by CCSM with conventional parameterization (denoted by CT-CCSM). Comparison is also made with the simulation by the SP version of CAM3 (denoted by SP-CAM3) with observed SST prescribed as the boundary condition for the period 1986–2003 (Khairoutdinov et al. 2008). Daily values of OLR, precipitation, 850-hPa horizontal wind, relative humidity, total cloud, and SST from these simulations have been analyzed.

Daily means of observed OLR on a 2.5° × 2.5° grid used in this study were obtained from the National Oceanic and Atmospheric Administration (NOAA; Liebmann and Smith 1996) for the period 1984–2003. From the European Centre for Medium-Range Forecasts Interim Re-Analysis (ERA-Interim; Dee et al. 2011), daily averages of the zonal wind u, meridional wind υ, and relative humidity on the T255 horizontal grid have been used for the period 1984–2003. Daily values of SST from the optimally interpolated SST (OISST, version 2) dataset developed by NOAA on a 0.25° × 0.25° grid were obtained for the period 1984–2003 (Reynolds et al. 2007). For all the observed and model data, daily climatology has been removed from the daily means to obtain daily anomalies.

b. Method of analysis

The method used in this study to extract the space–time modes of the monsoon is the multichannel singular spectrum analysis (MSSA). This is a powerful data-adaptive method to extract nonlinear oscillations, persistent modes, and trends from time series of spatial data (Ghil et al. 2002) and is the multivariate version of the singular spectrum analysis introduced by Broomhead and King (1986). While this method is equivalent to the extended empirical orthogonal function analysis, the MSSA has gone through more advancement in nonlinear dynamical systems theory, differing in the choice of key parameters such as the width of the lag window and in the interpretation of the results (Ghil et al. 2002). The given time series X(t) at L grid points (channels), where time t = 1, 2, 3, …, N, is added with M lagged copies, and a grand covariance matrix is constructed. The diagonalization of yields LM eigenvectors, which are the space–time empirical orthogonal functions (ST-EOFs), with LM eigenvalues. By projecting the original time series on the ST-EOFs, the corresponding space–time principal components (ST-PCs) are obtained. The MSSA extracts both spatial and temporal patterns of variability.

The ST-EOF and ST-PC of a particular eigenmode can be combined to form the reconstructed component (RC) [see Ghil et al. (2002) for the formula]. The RCs are simply the decomposition of the original time series into the corresponding eigenmodes and have the same space and time dimensions and the phase of the original time series. A pair of eigenmodes having nearly equal eigenvalues and periods with their ST-EOFs and ST-PCs in quadrature is identified as an oscillation (Plaut and Vautard 1994).

3. Mean monsoon and monsoon modes

a. Mean and standard deviation

A brief description of the mean and standard deviation is given before discussing the monsoon modes. The climatological mean for the June–September (JJAS) season is shown in Fig. 1 for OLR and 850-hPa horizontal wind in model simulations and observation. In the observation, the maximum convective activity is located in the Bay of Bengal and surrounding region (Fig. 1a). The CT-CCSM simulates two centers of maximum convection, one in the equatorial Indian Ocean and the other in the Bay of Bengal (Fig. 1b). The general spatial structure of the OLR over India and adjoining ocean region in observation (Fig. 1a) is reasonably well simulated by SP-CCSM but with slightly higher values (Fig. 1c). There is less convection over central India compared to observation. The SP-CAM3 shows stronger convection at about 10°N (Fig. 1d). Although the CT-CCSM captures the general direction of the 850-hPa horizontal wind, the magnitude is smaller (Fig. 1f) compared to the climatology in the reanalysis (Fig. 1e). The cross-equatorial flow, the southwesterly flow in the Arabian Sea, and the westerly flow over the Indian peninsula in the 850-hPa horizontal wind of the reanalysis (Fig. 1e) are well simulated by SP-CCSM (Fig. 1g) and also by SP-CAM3 but with slightly higher magnitude (Fig. 1h).

Fig. 1.
Fig. 1.

JJAS seasonal climatological-mean OLR (W m−2) in (a) observation, (b) CT-CCSM, (c) SP-CCSM, and (d) CT-CAM3. JJAS seasonal climatological-mean 850-hPa horizontal wind (streamlines, m s−1) in (e) observation, (f) CT-CCSM, (g) SP-CCSM, and (h) SP-CAM3.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The standard deviations of daily OLR and zonal wind u during the JJAS season are presented in Fig. 2. The standard deviation of OLR in CT-CCSM is generally lower (Fig. 2b) compared to observation (Fig. 2a). The spatial structure and the magnitude of the standard deviation are well simulated by SP-CCSM except for slightly higher values in the Bay of Bengal and west Pacific (Fig. 2c). The standard deviation of OLR is higher in SP-CAM3 with the maximum range covering a very large region north of the equator (Fig. 2d). The standard deviation of u in CT-CCSM is slightly lower in several areas (Fig. 2f) compared to reanalysis (Fig. 2e). The standard deviation of u is similar in SP-CCSM and SP-CAM3 (Figs. 2g,h) but slightly higher over most of the region compared to reanalysis.

Fig. 2.
Fig. 2.

Standard deviation of daily-mean OLR (W m−2) for JJAS season in (a) observation, (b) CT-CCSM, (c) SP-CCSM, and (d) SP-CAM3. Standard deviation of daily-mean 850-hPa zonal wind (m s−1) for JJAS season in (e) observation, (f) CT-CCSM, (g) SP-CCSM, and (h) SP-CAM3.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

b. MSSA modes

To extract the space–time modes of the monsoon, the MSSA was applied on daily anomalies of OLR during JJAS season in the domain (35°S–35°N, 40°–160°E). This procedure was carried out separately for observation and simulations by SP-CCSM, SP-CAM3, and CT-CCSM for the entire period of the respective data (as specified in section 2) using a lag window of 61 days. No prefiltering of the OLR anomalies was done before applying the MSSA. The eigenmodes are arranged in descending order of the variance explained, which is related to the eigenvalue. The ST-EOFs and ST-PCs of the MSSA eigenmodes were examined in each case, and oscillatory modes and persistent modes were identified. The oscillatory modes appear in pairs of eigenmodes and satisfy several criteria specified in section 2 (also see Plaut and Vautard 1994). The RCs of the relevant eigenmodes were constructed in each case. Similar MSSA was also performed on daily anomalies of precipitation and 850-hPa horizontal wind (u, υ) in the case of SP-CCSM to gain further insight into the performance of that model.

In the observed OLR, the first six MSSA eigenmodes were found to consist of two oscillatory pairs and two persistent modes, exactly like the eigenmodes obtained by Krishnamurthy and Shukla (2008). Eigenmode pairs (1, 2) and (5, 6) are the two oscillatory pairs and explain 3.7% and 2.1% of the total variance of the raw daily anomalies, respectively. The eigenmodes 3 and 4 are seasonally persistent modes explaining 1.6% and 1.2% of the total variance. It must be noted that the low values of the variance explained here are typical of MSSA eigenmodes of daily anomalies since both spatial and time patterns are resolved.

The monsoon intraseasonal oscillations (MISOs) represented by eigenmode pairs (1, 2) and (5, 6) will be referred to as MISO-1 and MISO-2, respectively. Since the RCs are additive, the RCs of eigenmodes 1 and 2 are added together to obtain the RC of MISO-1 and similarly for MISO-2. Since RCs are simply the decomposition of the total anomaly, they can be analyzed by the same methods used for any time series of spatial maps. For this purpose, a spatial EOF analysis was performed separately for the RCs of MISO-1 and MISO-2. In each case, the first two spatial EOFs (S-EOFs) explain almost all the variance of the RC while the corresponding spatial PCs (S-PCs) are in quadrature. The power spectra of the first S-PC (S-PC1) for MISO-1 and MISO-2 plotted in Fig. 3a show peaks centered at 48 and 28 days, respectively, consistent with earlier results (Krishnamurthy and Shukla 2008). An examination of the S-EOFs and S-PCs of MSSA eigenmodes 3 and 4 revealed that they are seasonally persistent and correspond to the seasonally persistent OLR modes related to ENSO and IOD as obtained by Krishnamurthy and Shukla (2008). Therefore, MSSA eigenmodes 3 and 4 will be referred to as ENSO and IOD modes, respectively, and further description will be provided in section 5. The power spectra of these ENSO and IOD modes are red in nature, as shown in Fig. 3a.

Fig. 3.
Fig. 3.

Power spectra of the S-PC1 of the RCs of MISO-1 (green) and MISO-2 (blue), ENSO (red) in (a) observation, (c) SP-CCSM, and (d) SP-CAM3, and IOD (purple) in (a) and (c). (b) CT-CCSM, the power spectra of the S-PC1 of first eight RCs are shown.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

An examination of the first 20 MSSA eigenmodes of daily OLR anomalies in CT-CCSM did not reveal the presence of any oscillatory pairs. To demonstrate this, the power spectra of S-PC1 of the RCs of first eight eigenmodes are plotted in Fig. 3b showing all of them to have red spectra. The eigenmode 1 that explains 4.9% of the total variance is identified as the seasonally persisting ENSO mode and will be further described in section 5. The IOD mode was not resolved in CT-CCSM.

In the case of the SP-CCSM simulation, the first six eigenmodes from MSSA of daily OLR anomalies revealed two oscillatory pairs and two seasonally persistent modes. Similar to the analysis of observation, the RCs of the oscillations and persistent modes were constructed and spatial EOF analyses of the RCs were performed. An examination of the S-EOF1 and S-PC1 of the RCs revealed that the MSSA modes are similar to those of observation except that there is a slight difference in the order of the modes. Eigenmode pairs (2, 3) and (5, 6) form MISO-1 and MISO-2, explaining 3.2% and 2.5% of the total variance, respectively, in SP-CCSM. The power spectra of S-PC1 of MISO-1 and MISO-2 (Fig. 3c) are similar to those in observation but with peaks centered at 59 and 29 days, respectively. The eigenmodes 1 and 4 correspond to the seasonally persisting ENSO mode and IOD mode with red power spectra (Fig. 3c) and explain 3.6% and 1.4% of the total variance, respectively.

Two oscillatory pairs of eigenmodes were also found in the MSSA of OLR in SP-CAM3. The eigenmode pairs (2, 3) and (4, 5) can be identified as MISO-1 and MISO-2 that explain 3.7% and 2.8% of the total variance. However, the power spectra of PC1 of the two oscillations show much broader and overlapping spectra centered at 69 and 48 days (Fig. 3d). After examining the remaining dominant eigenmodes, it was found that eigenmode 1 in SP-CAM3 is the seasonally persistent ENSO mode, explaining 4.2% of the total variance and displaying a red power spectrum (Fig. 3d). The seasonally persisting IOD mode was not resolved in SP-CAM3.

Thus, SP-CCSM simulates all the dominant modes of observations fairly well. While SP-CAM3 seems to resolve some of the observed modes (although not so properly), CT-CCSM fails to simulate the dominant intraseasonal oscillation and the second persisting mode.

4. Intraseasonal oscillations

The intraseasonal oscillations MISO-1 and MISO-2 obtained from the MSSA of daily OLR anomalies are further described in this section. The space–time structures of the oscillations in SP-CCSM are compared with those of observation and SP-CAM3. As discussed in section 3, the MSSA of the OLR anomalies in CT-CCSM did not yield any oscillations. The evolution of the spatial patterns during an oscillatory cycle can be best studied by constructing phase composites. The amplitude A(t) and the phase angle θ(t) of the oscillation are determined from the S-PC1 of the corresponding RC using the method suggested by Moron et al. (1998). Since θ varies from 0 to 2π, each oscillatory cycle is divided into eight phase intervals each of length π/4. To construct the phase composite for a particular interval, the RC (or any other field) is averaged for all θ in the concerned phase interval over the entire period of the data.

a. MISO-1

The phase composites of the OLR RC of MISO-1 for the half cycle of the oscillation (four phase intervals) are shown in Fig. 4 for observation, SP-CCSM, and SP-CAM3. In the observation (Fig. 4a), phase 1 consists of negative anomalies (convection) in the equatorial Indian Ocean and positive anomalies in the western Pacific and the Bay of Bengal. The convective anomalies move northward to the Indian peninsula and eastward and form a tilted band extending from the Arabian Sea to the western Pacific in phase 2. The tilted band of convection propagates northeastward in phases 3 and 4 during which time the Indian subcontinent is in the active phase of the monsoon. The positive anomalies in the northwestern Pacific move northward and diminish from phase 1 to 4. In phase 2, a small region of positive anomalies appears in the equatorial Indian Ocean and expands and intensifies during phase 3 and 4. These positive anomalies organize into a band and move northeastward in phases 5–8 and bring the break phase of the monsoon over India. The structure and propagation of the anomalies during phases 5–8 (second half of the oscillatory cycle) are exactly the same as those during phases 1–4 but with anomalies of the opposite sign (figure not shown; see Krishnamurthy and Shukla 2008).

Fig. 4.
Fig. 4.

Phase composites of the OLR RCs (W m−2) for four phase intervals of an average cycle of MISO-1 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. Each phase interval is of length π/4, and the phase number is given at the top-right corner of each panel.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The phase composites of the OLR RC in MISO-1 of SP-CCSM (Fig. 4b) also show the observed sequence fairly well. In SP-CCSM also, MISO-1 starts with convection in the Indian Ocean and forms a tilted band along with positive anomalies in the western Pacific during phases 1 and 2. The northeastward propagation of the tilted convection band is clearly evident during phases 2–4. As in the observation, positive anomalies develop in the equatorial Indian Ocean and expand during phases 3–4 in SP-CCSM also. The anomalies are slightly weaker in SP-CCSM, and the maximum convection appears slightly to the west in phases 1–2. The other half of the MISO-1 cycle in SP-CCSM is exactly similar to Fig. 4b but with anomalies of opposite sign.

In the SP-CAM3, however, MISO-1 does not capture the observed structure and the northeastward propagation (Fig. 4c). Weak convection appears in the equatorial Indian Ocean in phase 1 and moves over India without intensifying and forming a tilted band during phases 2–4. Strong positive anomalies are present in the northwestern Pacific in all the four phases while strong convection develops to the south, intensifies, and moves westward and merges with the positive anomaly region over India in phase 4.

The propagation characteristics of MISO-1 are more clearly demonstrated through Hovmöller diagrams. For this purpose, the phase composites of OLR RC of MISO-1 are constructed in 24 phase intervals of equal length for θ between 0 and 2π. The eastward or westward propagation in the equatorial Indian Ocean is shown by longitude–phase diagram of RC averaged over 5°S–10°N, as plotted in Fig. 5. Both the observation and SP-CCSM show eastward propagation in the eastern Indian Ocean and western Pacific and a standing pattern in the western Indian Ocean (Figs. 5a,b). The standing pattern of SP-CCSM extends a bit more to the east, and the maxima are slightly to the west compared to the observed pattern. The SP-CAM3, however, shows westward propagation between 100° and 160°E (Fig. 5c). The northward or southward propagation is inferred from the latitude–phase diagram of OLR RC averaged over the monsoon region 60°–90°E. In observation and SP-CCSM, northward propagation is clearly evident between 5°S and 25°N, although with different spatial structures at 5°N (Figs. 5d,e). However, SP-CAM3 shows a clear southward propagation in the same region (Fig. 5f).

Fig. 5.
Fig. 5.

Longitude–phase cross section of the phase composites of OLR RC averaged over 5°S–10°N for one complete cycle of MISO-1 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. Latitude–phase cross section of OLR RC over 60°–90°E for one complete cycle of MISO-1 in (d) observation, (e) SP-CCSM, and (f) SP-CAM3.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

Further evidence for the ability of SP-CCSM to correctly simulate the observed features of MISO-1 is provided by a similar analysis of precipitation and 850-hPa horizontal wind. The MSSA was applied on daily anomalies of precipitation in SP-CCSM in exactly the same manner as it was done for the OLR anomalies. Separately, a similar analysis was performed on the combined fields of u and υ at 850 hPa. These analyses yielded MISO-1 and MISO-2 for both cases, with the same periods as in the case of OLR. The RCs of the two oscillations were computed for precipitation and the horizontal wind from their respective ST-EOFs and ST-PCs. The phase composites of their RCs for the half cycle of MISO-1 are shown in Fig. 6. The phase composites of precipitation (Fig. 6a) are consistent with those of OLR (Fig. 5b). The positive precipitation anomalies form a large longitudinal band and propagate northward from the equatorial Indian Ocean to the Indian subcontinent and western Pacific. At the same time, negative anomalies occur to the north of the regions with positive anomalies and subsequently move northeastward. The phase composites of the 850-hPa horizontal wind (Fig. 6b) reveal cross-equatorial flow, southwesterly flow in the Arabian Sea, and a cyclonic flow in the western Pacific developing in phase 1 and intensifying in phase 2 and consistent with the positive precipitation anomalies in Fig. 6a. During phases 3 and 4, the westerly flow over India and the cyclonic flow over the Pacific together move northward. This space–time evolution of the horizontal wind in SP-CCSM is similar to the phase composites of MISO-1 in reanalysis (Krishnamurthy and Achuthavarier 2012).

Fig. 6.
Fig. 6.

Phase composites of (a) precipitation RC (mm day−1) and (b) 850-hPa horizontal wind RC (streamlines, m s−1) for four phase intervals of an average cycle of MISO-1 in SP-CCSM. The RCs were obtained by performing MSSA on precipitation and wind anomalies separately, in exactly the same manner as the MSSA of the OLR anomalies. The phase number is given at the top-right corner of each panel.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

To provide more direct evidence of the role played by the CRMs in the SP-CCSM, the phase composites of relative humidity and total cloud for MISO-1 are examined. The daily anomalies of relative humidity was averaged over the monsoon region (70°–90°E) between 1000 and 200 hPa in both observation and SP-CCSM simulation. The phase composites of the relative humidity were constructed for MISO-1 for both cases and shown for the first half cycle of MISO-1 in Fig. 7. The positive anomalies in the observation have a maximum value at about 500 hPa, and the vertical structure seems to expand and propagate northward from phase 1 to 4 (Fig. 7a). The negative anomalies develop to the south, intensify, and move northward following the positive anomalies. These composites are consistent with the active phase followed by the break phase in the monsoon as shown also by OLR, precipitation, and 850-hPa horizontal wind (Figs. 4 and 6). The phase composite of the relative humidity in SP-CCSM in MISO-1 (Fig. 7b) captures the observed vertical structure of the positive anomalies with the maximum at about 500 hPa and the northward propagation. The negative anomalies develop to the south, intensify, and move northward in the SP-CCSM also.

Fig. 7.
Fig. 7.

Phase composites of relative humidity anomalies (%) for a half cycle of MISO-1 in (a) observation and (b) SP-CCSM. The vertical structures are shown as a function of latitude by averaging the specific humidity over 70°–90°E. The phase number is given at the top-right corner of each panel.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The phase composites of the vertically averaged relative humidity reveal the horizontal structure of the relative humidity in MISO-1, as shown in Fig. 8a for half the cycle. Similar phase composites of the vertically integrated total cloud are shown in Fig. 8b. In phase 1, positive anomalies are present in the equatorial Indian Ocean while negative anomalies cover a part of India and the northwestern Pacific in both relative humidity (Fig. 8a) and cloud (Fig. 8b), consistent with the corresponding composite of OLR (Fig. 4b) and precipitation (Fig. 6a). The positive anomalies of relative humidity and cloud move northward and align along a tilted band in phase 2. Farther northward movement brings positive anomalies over central India in phase 3 and marks the peak active phase of the monsoon. In phase 4, the positive anomalies over India diminish while they intensify and move farther north over the Pacific. The location and propagation of the negative anomalies in both the relative humidity and total cloud in phases 1–4 are also consistent with the positive anomalies in the OLR composites (Fig. 4b). Thus, the space–time structures of the relative humidity and total cloud in MISO-1 (Figs. 7 and 8) are in tune with the OLR (Fig. 4b), precipitation, and low-level horizontal wind (Fig. 6).

Fig. 8.
Fig. 8.

Phase composites of (a) vertically averaged relative humidity anomalies (%) and (b) vertically integrated total cloud for a half cycle of MISO-1 in SP-CCSM. The phase number is given at the top-right corner of each panel.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The results presented so far, showing the better performance of SP-CCSM compared to SP-CAM3 and CT-CCSM, indicate that the ocean–atmosphere interaction is also important in addition to cloud representation. Simple evidence to support this assertion is provided through a correlation analysis with the SST. For this purpose, an index of MISO-1 over India is defined as the OLR RC of MISO-1 averaged over 10°–30°N, 70°–110°E and referred to as the extended Indian monsoon (EIM) index. This index was introduced for rainfall by Goswami et al. (1999) and has been commonly used in monsoon studies (e.g., Drbohlav and Krishnamurthy 2010; Rai and Krishnamurthy 2011). The point correlation between the daily time series of the MISO-1 EIM index and the daily SST anomalies was computed for observation, SP-CCSM, and SP-CAM3, as plotted in Fig. 9. In the observation, there is a band of negative correlation spanning from the Arabian Sea to the western Pacific with a positive correlation band to the south (Fig. 9a). This structure is collocated with the tilted band of OLR RC during the peak phase of MISO-1 (Fig. 4a). The correlation pattern in the observation has good correspondence with the pattern obtained by Krishnamurthy and Kirtman (2009), who also showed that the negative band is statistically significant even though the magnitude is small. The correlation pattern of MISO-1 in SP-CCSM (Fig. 9b) has good correspondence with the observed pattern. However, the correlation in SP-CAM3 (Fig. 9c) does not have any structure that resembles the observed pattern. Although the correlation analysis suggests the need for SP in a coupled model, the exact role of the ocean–atmosphere interaction in MISO-1 cannot be deduced, especially with low values of correlation.

Fig. 9.
Fig. 9.

Simultaneous point correlation of daily EIM index of OLR RC of MISO-1 with daily SST anomalies in (a) observation, (b) SP-CCSM, and (c) SP-CAM3.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

b. MISO-2

In Fig. 10, the phase composites of OLR RC of MISO-2 for its half cycle are shown for observation, SP-CCSM, and SP-CAM3. In the observation (Fig. 10a), phase 1 has a quadrupole structure that intensifies in phase 2 and becomes an east–west structure in phase 4. The dipole pattern with positive anomalies over India and negative anomalies in the Indian Ocean moves northward, bringing the active phase of the monsoon over the Indian subcontinent. The positive anomalies in the western Pacific move westward and merge with those coming from the Indian Ocean. The space–time structure of MISO-2 is consistent with the earlier study by Krishnamurthy and Shukla (2008).

Fig. 10.
Fig. 10.

Phase composites of the OLR RCs (W m−2) for four phase intervals of an average cycle of MISO-2 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. The phase number is given at the top-right corner of each panel.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The MISO-2 in SP-CCSM (Fig. 10b) also starts with a quadrupole structure in phase 1, although with weaker anomalies in the southern part, and becomes better defined in phase 2. As in the observation, the dipole part over India and the Indian Ocean moves northward, although the positive anomalies are weaker. The positive anomalies in the western Pacific are slightly stronger than observed but move westward and merge with the positive anomalies moving from the Indian Ocean as in the observation. In SP-CAM3 (Fig. 10c), the spatial structure in phases 1 and 2 has some semblance to a quadrupole: the positive anomaly regions are shifted eastward while the negative anomalies over India are weaker.

The propagation features of MISO-2 are also studied by examining Hovmöller diagrams. The phase composites of the MISO-2 RC are constructed for phase intervals of length π/12. The eastward/westward propagation is inferred from the longitude–phase diagram of RC averaged over 10°–25°N. Both the observation (Fig. 11a) and the SP-CCSM (Fig. 11b) display same propagation characteristics with an almost standing pattern in the 40°–90°E region and a clear westward propagation in the 90°–160°E region, confirming the earlier discussion of the phase composites in Fig. 10. However, the SP-CAM3 (Fig. 11c) shows standing patterns in both the western and eastern part of the domain and with the maxima shifted compared to observation and SP-CCSM. The northward or southward propagation in the monsoon region is ascertained from the latitude–phase diagram of the RC averaged over 60°–90°E, as shown in Fig. 11. There is a clear northward propagation in the observation from 5°S (Fig. 11d) and in the SP-CCSM from the equator. However, the anomalies of SP-CCSM are weaker compared to observation. In the SP-CAM3, there is a hint of northward propagation of weaker anomalies from 5°N.

Fig. 11.
Fig. 11.

Longitude–phase cross section of the phase composites of OLR RC averaged over 10°–25°N for one complete cycle of MISO-2 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. Latitude–phase cross section of OLR RC over 60°–90°E for one complete cycle of MISO-2 in (d) observation, (e) SP-CCSM, and (f) SP-CAM3.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

As mentioned in section 4a, separate MSSA of precipitation and 850-hPa horizontal wind (u, υ) in the SP-CCSM simulation also yielded MISO-2. The phase composites of the precipitation RC of MISO-2 for its half cycle are shown in Fig. 12a. There is a quadrupole structure in phase 1 and somewhat in phase 2, with stronger anomalies in the northern part, consistent with the OLR structure (Fig. 10b). Here also, the dipole part of the positive anomalies over the Indian subcontinent and negative anomalies over the Indian Ocean move northward. This propagation contributes to the active/break cycle of monsoon over India. The negative anomalies in the western Pacific present in phase 1 propagate westward in subsequent phases and merge with the negative anomalies coming from the Indian Ocean, again similar to the OLR composites.

Fig. 12.
Fig. 12.

Phase composites of (a) precipitation RC (mm day−1) and (b) 850-hPa horizontal wind RC (streamlines, m s−1) for four phase intervals of an average cycle of MISO-2 in SP-CCSM. The RCs were obtained by performing MSSA on precipitation and wind anomalies separately, in exactly the same manner as the MSSA of the OLR anomalies. The phase number is given at the top-right corner of each panel.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The phase composites of (u, υ) RC of MISO-2 are shown in Fig. 12b for the half cycle. Phase 1 consists of southwesterly flow over the Indian peninsula and an anticyclonic circulation in the western Pacific (Fig. 12b), consistent with the positive and negative rainfall anomalies in the respective locations (Fig. 12a). The flow over India moves northward in phases 2 and 3, while the anticyclonic circulation moves westward, similar to the rainfall anomalies. The anticyclonic circulation moves farther west in phase 4, expands, and brings northeasterlies over the Indian peninsula, which opposes the mean monsoon flow, as also evidenced in the negative precipitation anomalies in that region (Fig. 12a). In phase 3, a cyclonic circulation appears in the east (around 12°N, 150°E), intensifies, and moves westward in phase 4 (Fig. 12b), coinciding with the positive precipitation anomalies (Fig. 12a). This cyclonic circulation will grow and propagate westward, while the northeasterlies move northward over India, bringing the break phase of the monsoon over central India during the second half of the MISO-2 cycle (figures not shown). The phase composites of the 850-hPa winds (u, υ) are similar to those of the observations as described by Krishnamurthy and Achuthavarier (2012).

5. Seasonally persistent modes

a. ENSO and IOD modes

Certain eigenmodes obtained from the MSSA of daily OLR anomalies were identified as seasonally persisting modes related to ENSO and IOD in section 3. The ENSO mode was obtained as eigenmode 3 in the observation and eigenmode 1 in the simulations of SP-CCSM, SP-CAM3, and CT-CCSM. The percentage variance explained by the models is slightly higher than that in the observation (see section 3). The daily RC of the ENSO mode was constructed in each case and a new spatial EOF analysis of the RC was performed. The spatial EOF1 of the RC explains more than 95% of the variance of the RC since it is very coherent. The EOF1 of the ENSO RC is presented in Fig. 13 for observation and model simulations. The ENSO mode in the observation (Fig. 13a) consists of negative anomalies covering the entire Indian Ocean and the Indian subcontinent and positive anomalies in the warm pool region of the western Pacific. This pattern varies with the same sign throughout most of or the entire JJAS season but changes on an interannual time scale. The daily variation and the interannual variability of this mode are demonstrated in more detail by Krishnamurthy and Shukla (2008).

Fig. 13.
Fig. 13.

Spatial EOF1 of the RC of ENSO mode (W m−2) in (a) observation, (b) SP-CCSM, (c) SP-CAM3, and (d) CT-CCSM. Spatial EOF1 of the RC of IOD mode (W m−2) in (e) observation and (f) SP-CCSM. The spatial EOF analysis was performed on the daily values of the RCs. Note that the scales are different for observation and model simulations.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The ENSO mode in the SP-CCSM also shows the same coverage by the negative anomalies over the Indian Ocean and India and positive anomalies in the western Pacific (Fig. 13b) as in the observed mode. However, the magnitude of the anomalies is about 1.5 times higher than that of the observed mode with some differences in the structure of the regions of maximum values. The ENSO mode in SP-CAM3 (Fig. 13c) has quite a bit of resemblance to the SP-CCSM pattern, but the positive anomalies extend much farther up to the central Indian Ocean. The magnitude of the negative anomalies in the western part of the domain is higher. The CT-CCSM also has negative anomalies of higher magnitude covering the Indian Ocean and Indian subcontinent (Fig. 13d), but with differences in the spatial structure compared to observation (Fig. 13a) and SP-CCSM (Fig. 13b). Although all the three models are able to simulate the general pattern of the ENSO mode to varying degrees of success, the SP-CCSM seems to have better coverage over India.

In both observation and SP-CCSM, the seasonally persisting IOD mode was obtained as eigenmode 4 from the MSSA of daily OLR anomalies with about the same percentage of variance explained (see section 3). The MSSA of OLR in SP-CAM3 and CT-CCSM simulations did not yield any IOD mode. The IOD mode in the observation consists of negative anomalies covering the western part of the Indian Ocean, the Indian subcontinent, and the western Pacific (Fig. 13e). With positive anomalies in the eastern Indian Ocean, the OLR pattern also reveals the east–west dipole pattern similar to its oceanic counterpart in the SST. The IOD mode in the SP-CCSM (Fig. 13f) also has negative anomalies over the western Indian Ocean and India, similar to the observed pattern but with half the magnitude. The positive anomalies in the eastern Indian Ocean are weaker and cover a smaller region but extend into the western Pacific, compared to observation. However, negative anomalies are present to the north and south of the equator in the western Pacific. In the IOD mode also, the pattern over India in SP-CCSM is quite close to the observed pattern although the magnitude is weaker.

b. Relation with SST

To establish that the seasonally persisting modes in OLR are indeed associated with ENSO and IOD, a correlation analysis with the SST is performed. The point correlations of the daily time series of PC1 of the ENSO mode in OLR (corresponding to EOF1 shown in Figs. 13a–d) with the daily anomalies of the corresponding SST were computed. The correlation pattern in the observation (Fig. 14a) reveals the well-known ENSO SST pattern in the Pacific with strong negative values (up to −0.8) in central and eastern Pacific surrounded by a horseshoe pattern of positive values, consistent with the study by Krishnamurthy and Kirtman (2009). This pattern also shows weaker positive values in the Indian Ocean.

Fig. 14.
Fig. 14.

Simultaneous point correlation of daily PC1 of spatial EOF analysis of ENSO mode in (a) observation, (b) SP-CCSM, (c) SP-CAM3, and (d) CT-CCSM. Simultaneous point correlation of daily PC1 of spatial EOF analysis of IOD mode in (e) observation and (f) SP-CCSM. The PCs correspond to the EOFs shown in Fig. 13.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

A similar pattern is also present in the correlation of the ENSO mode in SP-CCSM (Fig. 14b) with some differences. The negative correlation region extends farther to the west and to higher latitudes in the Pacific, and the positive correlation extends farther west in the Indian Ocean. The correlation pattern of the ENSO mode in SP-CAM3 (Fig. 14c) is similar to that in the observation (Fig. 14a) except that the correlation is slightly weaker in the equatorial Pacific and positive correlation (although weak) in the equatorial and western Indian Ocean. The ENSO mode in CT-CCSM has a correlation pattern (Fig. 14d) with strong negative correlation extending over the entire equatorial Pacific and positive correlation covering a significant part of the Indian Ocean. The better correlation pattern in SP-CAM3 may not be surprising as the model is forced by observed SST. Although the three models are able to capture the observed ENSO–monsoon relation in the equatorial Pacific reasonably well, all of them have a larger region of positive correlation in the Indian Ocean compared to observation. The role of the Indian Ocean may be the reason for the differences in the details of the spatial structure of the ENSO mode in the models shown in Fig. 13.

The point correlations of the PC1 of the OLR IOD mode (corresponding to EOF1 in Figs. 13e,f) with the corresponding daily SST anomalies are shown in Figs. 14e and 14f for observation and SP-CCSM, respectively. In the observation, negative correlation is present in the Arabian Sea, Bay of Bengal, and the region around the Maritime Continent (Fig. 14e). There is a region of weak positive correlation in the equatorial Indian Ocean, hinting a weak signature of the IOD. In the Pacific, a horseshoe pattern of positive correlation surrounds the negative correlation in the eastern Pacific. The observed pattern is consistent with the study by Krishnamurthy and Kirtman (2009). In the SP-CCSM, the observed negative correlation is reproduced in the Arabian Sea and part of the Bay of Bengal but not in the region around the Maritime Continent. The negative correlation pattern and the positive horseshoe pattern in the equatorial Pacific extend too far to the west (Fig. 14f), which results in the positive correlation region in the eastern Indian Ocean. The intrusion of the correlation pattern from the Pacific Ocean into the Indian Ocean may explain the lack of a well-defined positive branch of the IOD mode in OLR (Fig. 13f). While the OLR mode shows the signature of the IOD, the support from the correlation with the SST is not strong.

c. Interannual variability

The hypothesis by Charney and Shukla (1981) suggested that the interannual variability of the seasonal-mean monsoon is mostly determined by boundary forcings such as the SST. Confirming this hypothesis, Krishnamurthy and Shukla (2008) showed that the relative strengths of the ENSO and IOD modes in OLR accounted for almost all of the interannual variability of the seasonal-mean OLR over India. To verify the role of the seasonally persisting modes in the models, the EIM index of the JJAS seasonal mean of OLR RCs of the ENSO mode is computed for the model simulations and observation. The EIM index of the ENSO mode is plotted in Figs. 15a–d.The EIM index in observation shows strong ENSO events during 1987, 1988, 1997, and 1998 (Fig. 15a).

Fig. 15.
Fig. 15.

Time series of EIM index of JJAS seasonal-mean RC of ENSO mode in (a) observation, (b) SP-CCSM, (c) SP-CAM3, and (d) CT-CCSM. Time series of EIM index of JJAS seasonal-mean RC of IOD mode in (e) observation and (f) SP-CCSM.

Citation: Journal of Climate 27, 3; 10.1175/JCLI-D-13-00257.1

The EIM index of the ENSO mode in SP-CCSM (Fig. 15b) shows strong chaotic interannual variability with magnitude comparable to that of observation. The EIM index in SP-CAM3 (Fig. 15c), which is forced by observed SST, captures the observed interannual variability quite well. In CT-CCSM, however, the EIM index of the ENSO mode (Fig. 15d) shows more regular variation on the biennial time scale, although not perfectly periodic. The variability of the OLR ENSO mode in the models is consistent with the results of Stan et al. (2010), who showed that the ENSO index in SST also varies regularly with a 2-yr spectral peak in CT-CCSM, while it varies chaotically in SP-CCSM on a longer time scale.

The EIM index of the JJAS seasonal mean of OLR RC of the IOD mode is plotted in Fig. 15e for observation. The IOD mode also exhibits strong interannual variability and captures the strong IOD events in 1987 and 1994. The EIM index of the IOD mode in SP-CCSM also shows interannual variability with magnitude comparable to that of the observation.

6. Summary and discussion

The simulation of the South Asian monsoon by a coupled ocean–atmosphere model with embedded cloud-resolving models was analyzed in this study. Specifically, the ability of the SP-CCSM to simulate the leading intraseasonal oscillations and the seasonally persisting large-scale patterns was examined. To test whether the explicit treatment of convection and cloud processes in the superparameterization leads to improvement in the simulation of convection and rainfall in the monsoon region, the SP-CCSM was compared with CT-CCSM that uses a conventional parameterization of convection. Since the MMF involves interaction between large-scale and small-scale processes, comparison with SP-CAM3 was also made to determine the importance of ocean–atmosphere interaction. The SP-CCSM was found to simulate the dominant intraseasonal oscillations and the seasonally persisting modes close to the observation. While the SP-CAM3 was able to simulate some of the observed modes, though not satisfactorily, the CT-CCSM simulated only one persisting mode but none of the intraseasonal oscillations.

The nonlinear intraseasonal oscillations and seasonally persisting modes were obtained by a data-adaptive method using raw anomalies without applying any prefilter. By applying MSSA, no preassumption was made on the time scales and periods of the eigenmodes extracted. The space–time structures of the two leading intraseasonal oscillations, MISO-1 and MISO-2, in SP-CCSM are similar to those in the observation. The most significant observed feature of MISO-1 and MISO-2 that SP-CCSM has captured is the northward propagation of convection, precipitation, and low-level circulation from the Indian Ocean to the Indian subcontinent and western Pacific. Additionally, the SP-CCSM has correctly simulated the eastward propagation of MISO-1 in the equatorial Indian Ocean and the westward propagation of MISO-2 from the western Pacific to the Indian subcontinent. The spatial structures of MISO-1 and MISO-2 in SP-CCSM have fairly good correspondence with those in the observation, although there are slight shifts in the locations of the maxima. The magnitude of the oscillations is slightly lower in the model. While the period of MISO-1 is longer in SP-CCSM, the period of MISO-2 is the same as that in the observation. Although SP-CAM3 simulates MISO-like oscillations, the periods are longer and the spatial structure and the propagation characteristics are incorrect. The CT-CCSM fails to generate any intraseasonal oscillation. The results of this study at leading intraseasonal time scales complement the findings of DeMott et al. (2011) showing the ability of SP-CCSM to better simulate oscillations at Rossby and Rossby–gravity frequencies in the monsoon region. It would be of further interest to investigate whether the mechanism suggested by DeMott et al. (2013) can explain the northward propagation in the two distinct intraseasonal oscillations discussed in this study.

The comparison of three versions of the model has clearly demonstrated that both a realistic representation of convection and cloud processes and a coupled ocean–atmosphere model are essential for correctly simulating the intraseasonal oscillations in the monsoon. In this regard, the role of CRM was confirmed with SP-CCSM providing evidence for better representation of the vertical and horizontal structure of relative humidity and total cloud in MISO-1. The SP-CCSM also captured the observed relation between MISO-1 OLR and daily SST in the Indian Ocean and western Pacific Ocean. Although the SP-CAM3 was forced by observed SST, it failed to show any kind of noticeable relation with the SST. It may be tempting to attribute the successful results in SP-CCSM entirely to coupling; however, this is not the case because of the different mean states of the coupled (Figs. 1c,g) and uncoupled (Figs. 1d,h) versions of the model.

Although all the three versions of the model simulated the seasonally persisting ENSO mode in OLR, the spatial structure of the ENSO mode in SP-CCSM showed the best correspondence with the observed mode, especially over India. The magnitude of the ENSO mode in all three models is higher in certain regions compared to observation. The interannual variability of the seasonal mean of the ENSO mode over India in SP-CCSM has variance and time scale comparable to those in observation and is more chaotic than the regular biennial variation in CT-CCSM. The SP-CCSM is the only model that was able to simulate the seasonally persisting IOD mode. While the spatial structure of the IOD mode in SP-CCSM is similar to the observed pattern over India, the pattern over the eastern Indian Ocean is not well simulated. Because of the lack of strong support from the SST correlation pattern, the evidence for the IOD mode in SP-CCSM is inconclusive.

While the performance of SP-CCSM is remarkable, certain aspects of the intraseasonal oscillations and the seasonally persisting modes need further investigation. The higher period of MISO-1, the higher amplitude of the ENSO mode and the weaker anomalies of the IOD mode in the eastern Indian Ocean may be related to the role of the Indian Ocean. Similar problems in another coupled model were found to indicate the importance of the ocean–atmosphere interaction in the Indian Ocean (Achuthavarier and Krishnamurthy 2011b; Achuthavarier et al. 2012). With the better performance of the SP-CCSM, it would be of interest to determine the predictability of SP-CCSM by performing retrospective seasonal-scale forecasts.

Acknowledgments

This research was supported by grants from the National Science Foundation (0334910 and 1211848), the National Oceanic and Atmospheric Administration (NA040AR4310034), and the National Aeronautics and Space Administration (NNG04GG46G). A portion of this work has been supported by the National Science Foundation Science and Technology Center for Multiscale Modeling of Atmospheric Processes, managed by Colorado State University under Cooperative Agreement ATM-0425247. The authors acknowledge the support of the Computational and Information Systems Laboratory at NCAR for providing computer time for this work.

REFERENCES

  • Achuthavarier, D., and V. Krishnamurthy, 2011a: Daily modes of South Asian summer monsoon variability in the NCEP Climate Forecast System. Climate Dyn., 36, 19411958.

    • Search Google Scholar
    • Export Citation
  • Achuthavarier, D., and V. Krishnamurthy, 2011b: Role of Indian and Pacific SST in Indian summer monsoon intraseasonal variability. J. Climate, 24, 29152930.

    • Search Google Scholar
    • Export Citation
  • Achuthavarier, D., V. Krishnamurthy, B. P. Kirtman, and B. Huang, 2012: Role of the Indian Ocean in the ENSO–Indian summer monsoon teleconnection in the NCEP Climate Forecast System. J. Climate, 25, 24902508.

    • Search Google Scholar
    • Export Citation
  • Annamalai, H., and K. R. Sperber, 2005: Regional heat sources and the active and break phases of boreal summer intraseasonal (30-50 day) variability. J. Atmos. Sci., 62, 27262748.

    • Search Google Scholar
    • Export Citation
  • Broomhead, D. S., and G. P. King, 1986: Extracting qualitative dynamics from experimental data. Physica D, 20, 217236.

  • Charney, J. G., and J. Shukla, 1981: Predictability of monsoons. Monsoon Dynamics, J. Lighthill and R. P. Pearce, Eds., Cambridge University Press, 99–109.

  • Collins, W. D., and Coauthors, 2006: The Community Climate System Model version 3 (CCSM3). J. Climate, 19, 21222143.

  • Dee, D. P., and Coauthors, 2011: The ERA-Interim Reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597.

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

    • 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.

    • Search Google Scholar
    • Export Citation
  • Drbohlav, H.-K. L., and V. Krishnamurthy, 2010: Spatial structure, forecast errors and predictability of South Asian monsoon in CFS monthly retrospective forecasts. J. Climate, 23, 47504769.

    • Search Google Scholar
    • Export Citation
  • Ghil, M., and Coauthors, 2002: Advanced spectral methods for climatic time series. Rev. Geophys., 40, 1003, doi:10.1029/2000RG000092.

  • Goswami, B. B., P. Mukhopadhyay, M. Khairoutdinov, and B. N. Goswami, 2012: Simulation of Indian summer monsoon intraseasonal oscillations in a superparameterized coupled climate model: Need to improve the embedded cloud resolving model. Climate Dyn., 41, 1497–1507, doi:10.1007/s00382-012-1563-1.

    • Search Google Scholar
    • Export Citation
  • Goswami, B. N., V. Krishnamurthy, and H. Annamalai, 1999: A broad-scale circulation index for the interannual variability of the Indian summer monsoon. Quart. J. Roy. Meteor. Soc., 125, 611633.

    • Search Google Scholar
    • Export Citation
  • Grabowski, W. W., 2001: Coupling cloud processes with the large-scale dynamics using the Cloud-Resolving Convection Parameterization (CRCP). J. Atmos. Sci., 58, 978997.

    • Search Google Scholar
    • Export Citation
  • Guilyardi, E., A. Wittenberg, A. Fedorov, M. Collins, C. Wang, A. Capotondi, G. J. van Oldenborgh, and T. Stockdale, 2009: Understanding El Niño in ocean–atmosphere general circulation models: Progress and challenges. Bull. Amer. Meteor. Soc., 90, 325340.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., and D. A. Randall, 2001: A cloud resolving model as a cloud parameterization in the NCAR Community Climate System Model: Preliminary results. Geophys. Res. Lett., 28, 36173620.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., and D. A. Randall, 2003: Cloud-resolving modeling of ARM summer 1997 IOP: Model formulation, results, uncertainties and sensitivities. J. Atmos. Sci., 60, 607625.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., C. A. DeMott, and D. A. Randall, 2008: Evaluation of the simulated interannual and subseasonal variability in an AMIP-style simulation using the CSU multiscale modeling framework. J. Climate, 21, 413431.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., and J. Shukla, 2000: Intraseasonal and interannual variability of rainfall over India. J. Climate, 13, 43664377.

  • Krishnamurthy, V., and J. L. Kinter III, 2003: The Indian monsoon and its relation to global climate variability. Global Climate, X. Rodó and F. A. Comín, Eds., Springer-Verlag, 186–236.

  • Krishnamurthy, V., and J. Shukla, 2007: Intraseasonal and seasonally persisting patterns of Indian monsoon rainfall. J. Climate, 20, 320.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., and J. Shukla, 2008: Seasonal persistence and propagation of intraseasonal patterns over the Indian monsoon region. Climate Dyn., 30, 353369.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., and B. P. Kirtman, 2009: Relation between Indian monsoon variability and SST. J. Climate, 22, 44374458.

  • Krishnamurthy, V., and D. Achuthavarier, 2012: Intraseasonal oscillations of the monsoon circulation over South Asia. Climate Dyn., 38, 23352353.

    • 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
  • Moron, V., R. Vautard, and M. Ghil, 1998: Trends, interdecadal and interannual oscillations in global sea-surface temperatures. Climate Dyn., 14, 545569.

    • Search Google Scholar
    • Export Citation
  • Plaut, G., and R. Vautard, 1994: Spells of low-frequency oscillations and weather regimes in the Northern Hemisphere. J. Atmos. Sci., 51, 210236.

    • Search Google Scholar
    • Export Citation
  • Rai, S., and V. Krishnamurthy, 2011: Error growth in CFS daily retrospective forecasts of South Asian monsoon. J. Geophys. Res., 116, D03108, doi:10.1029/2010JD014840.

    • Search Google Scholar
    • Export Citation
  • Randall, D. A., and Coauthors, 2007: Climate models and their evaluation. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 589–662.

  • Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey, and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 54735496.

    • Search Google Scholar
    • Export Citation
  • Sperber, K. R., H. Annamalai, I.-S. Kang, A. Kitoh, A. Moise, A. Turner, B. Wang, and T. Zhou, 2013: The Asian summer monsoon: An intercomparison of CMIP5 vs. CMIP3 simulations of the late 20th century. Climate Dyn., 41, 27112744, doi:10.1007/s00382-012-1607-6.

    • Search Google Scholar
    • Export Citation
  • Stan, C., M. Khairoutdinov, C. A. DeMott, V. Krishnamurthy, D. M. Straus, D. A. Randall, J. L. Kinter III, and J. Shukla, 2010: An ocean-atmosphere climate simulation with an embedded cloud resolving model. Geophys. Res. Lett., 37, L01702, doi:10.1029/2009GL040822.

    • Search Google Scholar
    • Export Citation
  • Zhang, G. J., and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.–Ocean, 33, 407446.

    • Search Google Scholar
    • Export Citation
Save
  • Achuthavarier, D., and V. Krishnamurthy, 2011a: Daily modes of South Asian summer monsoon variability in the NCEP Climate Forecast System. Climate Dyn., 36, 19411958.

    • Search Google Scholar
    • Export Citation
  • Achuthavarier, D., and V. Krishnamurthy, 2011b: Role of Indian and Pacific SST in Indian summer monsoon intraseasonal variability. J. Climate, 24, 29152930.

    • Search Google Scholar
    • Export Citation
  • Achuthavarier, D., V. Krishnamurthy, B. P. Kirtman, and B. Huang, 2012: Role of the Indian Ocean in the ENSO–Indian summer monsoon teleconnection in the NCEP Climate Forecast System. J. Climate, 25, 24902508.

    • Search Google Scholar
    • Export Citation
  • Annamalai, H., and K. R. Sperber, 2005: Regional heat sources and the active and break phases of boreal summer intraseasonal (30-50 day) variability. J. Atmos. Sci., 62, 27262748.

    • Search Google Scholar
    • Export Citation
  • Broomhead, D. S., and G. P. King, 1986: Extracting qualitative dynamics from experimental data. Physica D, 20, 217236.

  • Charney, J. G., and J. Shukla, 1981: Predictability of monsoons. Monsoon Dynamics, J. Lighthill and R. P. Pearce, Eds., Cambridge University Press, 99–109.

  • Collins, W. D., and Coauthors, 2006: The Community Climate System Model version 3 (CCSM3). J. Climate, 19, 21222143.

  • Dee, D. P., and Coauthors, 2011: The ERA-Interim Reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597.

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

    • 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.

    • Search Google Scholar
    • Export Citation
  • Drbohlav, H.-K. L., and V. Krishnamurthy, 2010: Spatial structure, forecast errors and predictability of South Asian monsoon in CFS monthly retrospective forecasts. J. Climate, 23, 47504769.

    • Search Google Scholar
    • Export Citation
  • Ghil, M., and Coauthors, 2002: Advanced spectral methods for climatic time series. Rev. Geophys., 40, 1003, doi:10.1029/2000RG000092.

  • Goswami, B. B., P. Mukhopadhyay, M. Khairoutdinov, and B. N. Goswami, 2012: Simulation of Indian summer monsoon intraseasonal oscillations in a superparameterized coupled climate model: Need to improve the embedded cloud resolving model. Climate Dyn., 41, 1497–1507, doi:10.1007/s00382-012-1563-1.

    • Search Google Scholar
    • Export Citation
  • Goswami, B. N., V. Krishnamurthy, and H. Annamalai, 1999: A broad-scale circulation index for the interannual variability of the Indian summer monsoon. Quart. J. Roy. Meteor. Soc., 125, 611633.

    • Search Google Scholar
    • Export Citation
  • Grabowski, W. W., 2001: Coupling cloud processes with the large-scale dynamics using the Cloud-Resolving Convection Parameterization (CRCP). J. Atmos. Sci., 58, 978997.

    • Search Google Scholar
    • Export Citation
  • Guilyardi, E., A. Wittenberg, A. Fedorov, M. Collins, C. Wang, A. Capotondi, G. J. van Oldenborgh, and T. Stockdale, 2009: Understanding El Niño in ocean–atmosphere general circulation models: Progress and challenges. Bull. Amer. Meteor. Soc., 90, 325340.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., and D. A. Randall, 2001: A cloud resolving model as a cloud parameterization in the NCAR Community Climate System Model: Preliminary results. Geophys. Res. Lett., 28, 36173620.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., and D. A. Randall, 2003: Cloud-resolving modeling of ARM summer 1997 IOP: Model formulation, results, uncertainties and sensitivities. J. Atmos. Sci., 60, 607625.

    • Search Google Scholar
    • Export Citation
  • Khairoutdinov, M., C. A. DeMott, and D. A. Randall, 2008: Evaluation of the simulated interannual and subseasonal variability in an AMIP-style simulation using the CSU multiscale modeling framework. J. Climate, 21, 413431.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., and J. Shukla, 2000: Intraseasonal and interannual variability of rainfall over India. J. Climate, 13, 43664377.

  • Krishnamurthy, V., and J. L. Kinter III, 2003: The Indian monsoon and its relation to global climate variability. Global Climate, X. Rodó and F. A. Comín, Eds., Springer-Verlag, 186–236.

  • Krishnamurthy, V., and J. Shukla, 2007: Intraseasonal and seasonally persisting patterns of Indian monsoon rainfall. J. Climate, 20, 320.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., and J. Shukla, 2008: Seasonal persistence and propagation of intraseasonal patterns over the Indian monsoon region. Climate Dyn., 30, 353369.

    • Search Google Scholar
    • Export Citation
  • Krishnamurthy, V., and B. P. Kirtman, 2009: Relation between Indian monsoon variability and SST. J. Climate, 22, 44374458.

  • Krishnamurthy, V., and D. Achuthavarier, 2012: Intraseasonal oscillations of the monsoon circulation over South Asia. Climate Dyn., 38, 23352353.

    • 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
  • Moron, V., R. Vautard, and M. Ghil, 1998: Trends, interdecadal and interannual oscillations in global sea-surface temperatures. Climate Dyn., 14, 545569.

    • Search Google Scholar
    • Export Citation
  • Plaut, G., and R. Vautard, 1994: Spells of low-frequency oscillations and weather regimes in the Northern Hemisphere. J. Atmos. Sci., 51, 210236.

    • Search Google Scholar
    • Export Citation
  • Rai, S., and V. Krishnamurthy, 2011: Error growth in CFS daily retrospective forecasts of South Asian monsoon. J. Geophys. Res., 116, D03108, doi:10.1029/2010JD014840.

    • Search Google Scholar
    • Export Citation
  • Randall, D. A., and Coauthors, 2007: Climate models and their evaluation. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 589–662.

  • Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey, and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 54735496.

    • Search Google Scholar
    • Export Citation
  • Sperber, K. R., H. Annamalai, I.-S. Kang, A. Kitoh, A. Moise, A. Turner, B. Wang, and T. Zhou, 2013: The Asian summer monsoon: An intercomparison of CMIP5 vs. CMIP3 simulations of the late 20th century. Climate Dyn., 41, 27112744, doi:10.1007/s00382-012-1607-6.

    • Search Google Scholar
    • Export Citation
  • Stan, C., M. Khairoutdinov, C. A. DeMott, V. Krishnamurthy, D. M. Straus, D. A. Randall, J. L. Kinter III, and J. Shukla, 2010: An ocean-atmosphere climate simulation with an embedded cloud resolving model. Geophys. Res. Lett., 37, L01702, doi:10.1029/2009GL040822.

    • Search Google Scholar
    • Export Citation
  • Zhang, G. J., and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.–Ocean, 33, 407446.

    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    JJAS seasonal climatological-mean OLR (W m−2) in (a) observation, (b) CT-CCSM, (c) SP-CCSM, and (d) CT-CAM3. JJAS seasonal climatological-mean 850-hPa horizontal wind (streamlines, m s−1) in (e) observation, (f) CT-CCSM, (g) SP-CCSM, and (h) SP-CAM3.

  • Fig. 2.

    Standard deviation of daily-mean OLR (W m−2) for JJAS season in (a) observation, (b) CT-CCSM, (c) SP-CCSM, and (d) SP-CAM3. Standard deviation of daily-mean 850-hPa zonal wind (m s−1) for JJAS season in (e) observation, (f) CT-CCSM, (g) SP-CCSM, and (h) SP-CAM3.

  • Fig. 3.

    Power spectra of the S-PC1 of the RCs of MISO-1 (green) and MISO-2 (blue), ENSO (red) in (a) observation, (c) SP-CCSM, and (d) SP-CAM3, and IOD (purple) in (a) and (c). (b) CT-CCSM, the power spectra of the S-PC1 of first eight RCs are shown.

  • Fig. 4.

    Phase composites of the OLR RCs (W m−2) for four phase intervals of an average cycle of MISO-1 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. Each phase interval is of length π/4, and the phase number is given at the top-right corner of each panel.

  • Fig. 5.

    Longitude–phase cross section of the phase composites of OLR RC averaged over 5°S–10°N for one complete cycle of MISO-1 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. Latitude–phase cross section of OLR RC over 60°–90°E for one complete cycle of MISO-1 in (d) observation, (e) SP-CCSM, and (f) SP-CAM3.

  • Fig. 6.

    Phase composites of (a) precipitation RC (mm day−1) and (b) 850-hPa horizontal wind RC (streamlines, m s−1) for four phase intervals of an average cycle of MISO-1 in SP-CCSM. The RCs were obtained by performing MSSA on precipitation and wind anomalies separately, in exactly the same manner as the MSSA of the OLR anomalies. The phase number is given at the top-right corner of each panel.

  • Fig. 7.

    Phase composites of relative humidity anomalies (%) for a half cycle of MISO-1 in (a) observation and (b) SP-CCSM. The vertical structures are shown as a function of latitude by averaging the specific humidity over 70°–90°E. The phase number is given at the top-right corner of each panel.

  • Fig. 8.

    Phase composites of (a) vertically averaged relative humidity anomalies (%) and (b) vertically integrated total cloud for a half cycle of MISO-1 in SP-CCSM. The phase number is given at the top-right corner of each panel.

  • Fig. 9.

    Simultaneous point correlation of daily EIM index of OLR RC of MISO-1 with daily SST anomalies in (a) observation, (b) SP-CCSM, and (c) SP-CAM3.

  • Fig. 10.

    Phase composites of the OLR RCs (W m−2) for four phase intervals of an average cycle of MISO-2 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. The phase number is given at the top-right corner of each panel.

  • Fig. 11.

    Longitude–phase cross section of the phase composites of OLR RC averaged over 10°–25°N for one complete cycle of MISO-2 in (a) observation, (b) SP-CCSM, and (c) SP-CAM3. Latitude–phase cross section of OLR RC over 60°–90°E for one complete cycle of MISO-2 in (d) observation, (e) SP-CCSM, and (f) SP-CAM3.

  • Fig. 12.

    Phase composites of (a) precipitation RC (mm day−1) and (b) 850-hPa horizontal wind RC (streamlines, m s−1) for four phase intervals of an average cycle of MISO-2 in SP-CCSM. The RCs were obtained by performing MSSA on precipitation and wind anomalies separately, in exactly the same manner as the MSSA of the OLR anomalies. The phase number is given at the top-right corner of each panel.

  • Fig. 13.

    Spatial EOF1 of the RC of ENSO mode (W m−2) in (a) observation, (b) SP-CCSM, (c) SP-CAM3, and (d) CT-CCSM. Spatial EOF1 of the RC of IOD mode (W m−2) in (e) observation and (f) SP-CCSM. The spatial EOF analysis was performed on the daily values of the RCs. Note that the scales are different for observation and model simulations.

  • Fig. 14.

    Simultaneous point correlation of daily PC1 of spatial EOF analysis of ENSO mode in (a) observation, (b) SP-CCSM, (c) SP-CAM3, and (d) CT-CCSM. Simultaneous point correlation of daily PC1 of spatial EOF analysis of IOD mode in (e) observation and (f) SP-CCSM. The PCs correspond to the EOFs shown in Fig. 13.

  • Fig. 15.

    Time series of EIM index of JJAS seasonal-mean RC of ENSO mode in (a) observation, (b) SP-CCSM, (c) SP-CAM3, and (d) CT-CCSM. Time series of EIM index of JJAS seasonal-mean RC of IOD mode in (e) observation and (f) SP-CCSM.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 141 34 5
PDF Downloads 69 20 1