1. Introduction
Convectively coupled equatorial waves (CCEWs) are significant parts of the synoptic-scale convective variability in the tropics. There has been systematic research on CCEWs from the synoptic-scale perspective in observation and theory for the past several decades. This paper focuses on the climatology and interannual variability of CCEW activity including four modes of the mixed Rossby–gravity (MRG), tropical-depression-type (TD-type), equatorial Rossby (ER), and Kelvin waves.
Classical “dry” linear equatorial beta-plane shallow-water theory (Matsuno 1966; Lindzen 1967) described and predicted successfully the basic structure of these tropical trapped waves. However, the theory alone is insufficient to explain the observed equatorial waves in the tropical atmosphere. The nonlinear wave–wave interaction and wave–flow interaction may modify the wave structure and dispersion characteristics (Zhang and Webster 1989; Straub and Kiladis 2003a). Many tropospheric equatorial waves are associated with moist deep convection, which is why they are referred to as convectively coupled waves. Thus, the wave–convection interaction may also modify the characteristics of CCEWs (Wheeler and Kiladis 1999, hereafter referred to as WK99; Lindzen 2003). The prominent spectral peak of convection, represented by satellite-derived precipitable water, outgoing longwave radiation (OLR), and infrared brightness temperature (Tb) datasets, correspond with the dispersion curves and structure predicted by the linear equatorial shallow-water theory, although the corresponding equivalent depth of CCEWs is shallower related to the theoretical value for dry air (e.g., WK99; Roundy and Frank 2004, hereafter referred to as RF04; Kiladis et al. 2009).
Previous studies on CCEWs have mainly focused on their synoptic-scale properties. The dynamics structure and propagation characteristics of CCEWs were thoroughly analyzed by Wheeler et al. (2000) and Yang et al. (2007a,b), based on various methodologies for identifying CCEWs from observational datasets (Takayabu 1994; WK99; Yang et al. 2003). In WK99, the variance of several tropical disturbance bands in boreal summer and winter was presented. As in the extended work of WK99, RF04 analyzed the daily climatology of tropical CCEW activity using OLR and precipitable water datasets. The activity of various CCEWs has distinctively different regional and seasonal distributions (RF04; Kiladis et al. 2009). More precise seasonal cycles of the various CCEWs over different regions are required to extend previous results.
Mekonnen et al. (2008) analyzed the interannual variability of convectively coupled Kelvin waves and revealed its significant relationship with the tropical Africa rainfall anomalies and sea surface temperature (SST) anomalies. The relationship between the ER, MRG, and TD-type wave activity and tropical cyclogenesis on interannual time scales was disclosed by Frank and Roundy (2006). Chen and Huang (2009) also investigated the impacts of interannual variation of the MRG wave on tropical cyclogenesis over the northwestern Pacific. Because of the significant impacts of CCEWs on tropical weather on an interannual time scale, it is meaningful to investigate the interannual variability of CCEW activity.
The primary problem of studying the interannual variability of CCEW activity is the description of monthly or yearly CCEW activity. Some discrete methods, such as counting the number of wave events propagating through a selected region (Mekonnen et al. 2008) and counting the days when wave variance is above a threshold value (Frank and Roundy 2006), have been utilized to describe the activity of CCEWs. Apart from these discrete methods, the seasonal variance of CCEWs and the seasonal wavenumber–frequency spectrum of raw data were also used (Cho et al. 2004; Masunaga 2007). Here a new method, the root-mean-square of daily Tb filtered for various CCEWs in each month, is used to represent the monthly CCEW activity. By utilizing this method, the climatology and interannual variance of CCEW activity are described, and their interannual relationship related to background conditions and tropical SST anomaly events are investigated.
The data and methodology are described in section 2. Section 3 presents the climatology of CCEW activity. The interannual variance of CCEW activity is displayed in section 4. The interannual relationship of CCEW activity with background conditions and tropical SST anomalies pattern is investigated in section 5. Finally in section 6, major findings of the study are summarized and some discussions are given.
2. Data and methodology
a. Data
The studies of deep convection here are based on the satellite-observed brightness temperature Tb datasets obtained from the Cloud Archive User Service (CLAUS) of the European Union, which utilizes geostationary and polar-orbiting images and the infrared channel (Hodges et al. 2000). These data are available from the British Atmospheric Data Center (BADC; http://badc.nerc.ac.uk/data/claus/) eight times daily on a global 0.5° latitude–longitude grid from January 1984 to December 2004; in advance, spatial and temporal averaging is performed to convert the dataset to a 1° grid and four times daily. This resolution is sufficient for the present study. The few missing values over the Indian Ocean are filled using linear temporal interpolation. Some studies have shown that the CLAUS Tb data are an excellent proxy for deep convection coupled with equatorial waves (Yang et al. 2003; Mekonnen et al. 2008; Kiladis et al. 2009).
The monthly Hadley Center Global Sea Ice and Sea Surface Temperature (HadISST version 2; Rayner et al. 2003) analyses datasets from January 1980 to December 2008 with 1° horizontal resolution are also used.
b. Methodology
The zonal space–time spectral analysis and filtering described in WK99 is employed on the Tb data to determine the distribution of power in the wavenumber–frequency domain associated with zonally propagating waves. This method has also been successfully applied on other datasets such as the OLR (WK99; Straub and Kiladis 2002), precipitable water (RF04), Tropical Rainfall Measuring Mission rainfall (Cho et al. 2004), and Tb data (Kiladis et al. 2009) in previous studies. The seasonal and semiannual cycles, which are calculated by averaging over 21 years for each day and then filtering by a fast Fourier transform, are removed from each grid point. As in WK99, a 96-day segment, such that each segment overlaps with the previous segment for 65 days, is used in the space–time spectral analysis. The results are independent of the length of the segment (WK99). The raw spectra are calculated for the equatorial band from 20°S to 20°N. A spectrum of Tb data, with wavenumber as the abscissa and frequency as the ordinate in Fig. 1 is consistent with that in WK99. The predominant bands are the Madden–Julian oscillation (MJO), Kelvin, n = 1 ER, MRG, and TD-type waves.
Wavenumber–frequency spectrums of Tb from 1984 to 2004 between 20°S and 20°N. The heavy solid polygons indicate the domains for filtering the Kelvin, n = 1 ER, MRG, and TD-type waves. The thin solid lines indicate the dispersion curves for the equivalent heights of 8, 25, and 90 m.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Wavenumber–frequency spectrums of Tb from 1984 to 2004 between 20°S and 20°N. The heavy solid polygons indicate the domains for filtering the Kelvin, n = 1 ER, MRG, and TD-type waves. The thin solid lines indicate the dispersion curves for the equivalent heights of 8, 25, and 90 m.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Wavenumber–frequency spectrums of Tb from 1984 to 2004 between 20°S and 20°N. The heavy solid polygons indicate the domains for filtering the Kelvin, n = 1 ER, MRG, and TD-type waves. The thin solid lines indicate the dispersion curves for the equivalent heights of 8, 25, and 90 m.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
In the present study, four prominent subseasonal CCEWs including the convectively coupled ER, MRG, Kelvin, and TD-type waves are isolated. The wave-filtering bands shown by the heavy solid polygons in Fig. 1 are almost identical to those in RF04 for comparing with their results. However, a low-frequency cutoff of 17 days for the Kelvin wave is used as in Straub and Kiladis (2002) instead of 30 days for separating it from the MJO band, and a high-frequency cutoff of 2.5 days for the TD-type wave is used instead of 2 days for separating it from the westward inertia–gravity waves. Following Straub and Kiladis (2002), the filtering is applied to raw data without decomposing into symmetric and antisymmetric components about the equator in advance since the convective disturbances tend to propagate along the latitude of the intertropical convergence zone (Kiladis et al. 2009). The details of the wavenumber–frequency filtering technique were described in WK99 and Straub and Kiladis (2002).
The root mean square (rms) of the daily-filtered Tb for each CCEW is calculated in every calendar month to represent corresponding monthly wave activity. The seasonal cycles of wave activity are the long-term mean monthly wave activity for 1984 to 2004. The meridional ranges of the tropical MRG, TD-type, and ER waves are chosen from 20°S to 20°N, while a smaller range from 12.5°S to 12.5°N is selected for the Kelvin wave to avoid the contamination of extratropical Rossby waves (Straub and Kiladis 2003b; RF04) as in Masunaga (2007). Since the extraction of CCEWs involves temporal filtering, the monthly CCEW activity represented by rms of daily filtered convection also includes some signal of other months. This leakage of the monthly wave activity is like performing a very weak running averaging on the true monthly wave activity and does not influence the results on the seasonal cycles of CCEW activity.
The spatial distributions of mean CCEW activity (not shown) based on the monthly rms of filtered Tb are well consistent with those of the seasonal variance of filtered CCEWs in previous works (WK99; RF04; Masunaga 2007; Kiladis et al. 2009), although the dimension of CCEW activity in the present study is different from previous works in which the total variance of filtered fields was often used. This method is also performed on the MJO band (not shown) and the result of the MJO is basically consistent with that in Zhang and Dong (2004). These coincident results demonstrate that the rms of filtered Tb is credible for describing the monthly CCEWs activity.
3. Climatology of CCEW activity
a. Seasonal cycles of the meridional mean CCEW activity
The seasonal cycles of CCEW activity are derived from the long-term mean monthly activity of various CCEWs from 1984 to 2004. Overall, the seasonal cycles of the meridional mean activity of the MRG, TD-type, and ER waves (Figs. 2a–c) in the Northern Hemisphere are approximately seasonally antisymmetric corresponding to that in the Southern Hemisphere. However, Kelvin wave activity (Fig. 2d) has basically synchronized seasonal cycles in the Northern and Southern Hemispheres.
Seasonal cycles of meridional mean CCEW activity in the Northern (shading) and Southern (contours) Hemisphere.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Seasonal cycles of meridional mean CCEW activity in the Northern (shading) and Southern (contours) Hemisphere.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Seasonal cycles of meridional mean CCEW activity in the Northern (shading) and Southern (contours) Hemisphere.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
The maximum MRG wave activity (Fig. 2a) over the northern Indian Ocean (60°–120°E) and the northeastern Pacific (180°–60°W) occurs during northen summer, while it occurs during the northern fall over the northwestern Pacific (120°E–180°) and the northern Atlantic (60°W–0°). This result is different from a rough description in RF04 that the maximum OLR variance for the MRG wave occurs during the northern summer. The MRG wave in the Southern Hemisphere is much weaker than in the Northern Hemisphere, and its maximum occurs during the southern late summer and fall over the central and western Pacific. The TD-type wave activity has seasonal cycles similar to the MRG wave activity except that the activity in the Atlantic and Africa is relatively greater and peaks during the northern summer (Fig. 2b).
The ER wave has the strongest activity of all CCEWs. Unlike the short active season of the MRG and TD-type waves, the active ER wave persists nine months from April to December over the warm-pool region, the Indian Ocean, and the western Pacific in the Northern Hemisphere (Fig. 2c). In the Southern Hemisphere, the active ER wave is located over the Indian Ocean, central Pacific, and Atlantic with a maximum during the southern summer. Another noteworthy characteristic of ER wave is that there are similar active ER waves in the South and North Atlantic during the Southern Hemisphere summer.
The most active Kelvin wave occurs over the central Pacific during the northern early summer and over Africa during the northern spring (Fig. 2d). Although the Kelvin wave activity over the central Pacific is relatively large, there are almost zonally uniform Kelvin waves during April to June. This nature is consistent with the basic zonally uniform ITCZ. Apart from the most active season, the Kelvin wave in the Northern Hemisphere has a secondary active season from October to December over the central Pacific and Atlantic. That is, the Kelvin wave activity has a distinctive semiannual variation in contrast to the MRG, TD-type, and ER waves dominated by the annual variation. The Kelvin wave activity in the Southern Hemisphere is approximately similar to that in the Northern Hemisphere with a maximum during the northern late spring and early summer.
b. Seasonal cycles of the zonal mean CCEW activity
The seasonal cycles of zonal mean CCEW activity are shown in Fig. 3. The active MRG wave and the TD-type wave are nearly identical and mainly located between 5° and 15° latitude in both hemispheres, while the ER wave activity occurs at higher latitudes, 10°–15°, in both hemispheres; the Kelvin wave peaks in April–June near 5°N.
Seasonal cycles of zonal mean CCEW activity.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Seasonal cycles of zonal mean CCEW activity.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Seasonal cycles of zonal mean CCEW activity.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
The MRG and TD-type wave activity has almost identical seasonal cycles so that the maximum zonal mean occurs in the Northern Hemisphere during the northern late summer and early fall (Figs. 3a,b) due to the dominant wave over the northwestern and central Pacific with a maximum during the northern fall (Figs. 2a,b).
There are asymmetric seasonal variations of the ER wave activity in the two hemispheres such that the maximum Northern Hemisphere part of the ER wave occurs during the northern fall and that of the Southern Hemisphere during the northern late winter (Fig. 3c). Although the Northern Hemisphere part of ER wave has an active period persisting for nine months from April to December over the warm-pool region, the maximum zonal mean ER wave activity in the Northern Hemisphere occurs during the northern fall owing to the Atlantic ER wave peaks during this period (Fig. 2c). The Southern Hemisphere part of ER wave activity has comparable magnitude with the part in the Northern Hemisphere unlike the MRG and TD-type waves.
The active convection associated with Kelvin waves is located near the equator in the northern late spring, and then it is strengthened and moves to the Northern Hemisphere during the early summer. There also is a secondary active Kelvin wave season during the northern early winter with active convection between 0° and 10°N due to the semiannual variation of Kelvin wave activity over the northern central Pacific and Atlantic (Fig. 2d). The seasonal cycles of the Kelvin wave appear to be similar to that of the ITCZ. This could be attributed to the most active convection associated with the Kelvin wave being located near the equator (Fig. 3d) coincident with the ITCZ.
In the present study, the CCEWs are only filtered from the convection represented by the Tb data but not from the dynamical fields. Although the structure of CCEWs is not derived from the linear wave theory, the spectrum of convection and the cross-spectrum of convection–wind in WK99 showed that CCEWs also approximately follow the linear wave theory. Therefore, the CCEW activity analyzed for the convection should not be distinctively different from the result of the dynamical fields. For confirming this supposition, a similar analysis is performed on the 850-hPa zonal and meridional winds of European Centre for Medium-Range Weather Forecasts (ECMWF) 6-hourly reanalysis data. Similar results in the present study are obtained (not shown), although there are some differences such as the most active Kelvin wave filtered from the zonal wind is during boreal winter in contrast to the boreal summer—the most active Kelvin season determined from Tb. Thus, the CCEW activity based on the convection reflects the main characteristics.
4. Interannual variance of CCEW activity
Interannual variance of the CCEW activity is calculated to represent their interannual variation amplitudes after removing the long-term mean. The spatial distributions of the interannual variance of the MRG, TD-type, and ER wave activity (Fig. 4) are almost identical to the climatological mean of CCEW activity (not shown, but in Kiladis et al. 2009). However, the interannual variance of Kelvin wave activity has a significantly different distribution from its climatological mean such that the maximum variance of Kelvin wave activity is located on the equator over the central and eastern Pacific (CEP) (Fig. 4d). That is, the strongest interannual variability of Kelvin wave activity is over the CEP although its climatology is relatively weak over this region. Figure 4 also shows the ratio of the interannual standard deviations of CCEW activity to the long-term mean CCEW activity that expresses the relative amplitude of CCEW activity variation corresponding to its climatology. The ratio for each CCEW, including the MRG, TD-type, and ER wave, has almost identical distribution in which the greatest values are located on the equatorial CEP.
Interannual variance (K2) of CCEW activity (shading) and the ratios of the interannual standard deviations to the long-term mean CCEW activity (contours).
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Interannual variance (K2) of CCEW activity (shading) and the ratios of the interannual standard deviations to the long-term mean CCEW activity (contours).
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Interannual variance (K2) of CCEW activity (shading) and the ratios of the interannual standard deviations to the long-term mean CCEW activity (contours).
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
The seasonal variations or zonal distributions of the MRG, TD-type, and ER wave activity variance are not distinctively different from the corresponding distributions of climatologies. However, the maximum interannual variance of Kelvin wave activity occurs over the eastern Pacific during the early summer in the Northern Hemisphere (Fig. 5d) compared with the corresponding maximum climatology over the central Pacific (Fig. 2d). In addition, the interannual variance of the eastern Pacific Kelvin wave activity during the northern spring is stronger relative to the climatology. The interannual variance of Kelvin wave activity appears to have a zonally asymmetric distribution and differs from its climatology characterized by almost zonally uniform distribution.
Meridional mean interannual variance (K2) of CCEW activity in the Northern (shading) and Southern (contours) Hemisphere.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Meridional mean interannual variance (K2) of CCEW activity in the Northern (shading) and Southern (contours) Hemisphere.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Meridional mean interannual variance (K2) of CCEW activity in the Northern (shading) and Southern (contours) Hemisphere.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
The interannual variances of the MRG and TD-type waves appear predominantly in the Northern Hemisphere with a maximum during early fall (Figs. 6a,b), almost identical to their zonal mean climatology. Also, there is a secondary large interannual variance of the Northern Hemisphere ER wave activity during late spring and early summer, except for the most active season of the northern fall. It may be attributed to the large variance of ER wave activity over the central Pacific during this period (Fig. 5c). The zonal mean interannual variance of Kelvin wave activity exhibits a simple seasonal characteristic that the maximum variance occurs during the northern spring on the equator.
Zonal-mean interannual variance (K2) of CCEW activity.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Zonal-mean interannual variance (K2) of CCEW activity.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Zonal-mean interannual variance (K2) of CCEW activity.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
5. Interannual variability of CCEW activity
a. Relationship between CCEW activity and background conditions
As presented by the climatological distribution of CCEW activity (Figs. 2 and 3), the active CCEWs are generally located in active background convective regions. The relationship of the CCEW activity with local convection is shown in Fig. 7. Unlike the regional distribution of CCEW activity, the strong background convection–CCEW activity relationship is located over weak background convection. The large correlations of various CCEW activity–local convection are all located in the CEP, and the correlation over the South Atlantic is also significant. In contrast, the linear relationship over the warm pool, which is the most active convection region, is relatively weak.
Correlation coefficient between CCEW activity anomalies and local monthly Tb anomalies: shadings represent significance at the 95% confidence levels.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Correlation coefficient between CCEW activity anomalies and local monthly Tb anomalies: shadings represent significance at the 95% confidence levels.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Correlation coefficient between CCEW activity anomalies and local monthly Tb anomalies: shadings represent significance at the 95% confidence levels.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
However, Fink and Speth (1997) revealed that there is a somewhat reduced, but still good, correlation between convection and the intensity of high-frequency (of periods less than 25 days) OLR fluctuations over the warm pool, while it is significantly high over the tropical CEP. That is, the background convection over the warm pool is related to the total high-frequency convection activity but not to various CCEW bands activity, while the convection over the CEP is closely correlated to both. A possible explanation inspired by Fink and Speth is straightforward: the CCEW event is episodic and thus the CCEW activity is mainly determined by the amplitude of individual CCEW events. For the activity of total high-frequency convection, it is determined largely by the frequency and duration of weak unorganized signals over the warm pool, but it is basically consistent with the activity of CCEWs over the CEP in which convection is weak and the high-frequency fluctuations are infrequent. Therefore, the CCEW activity over the warm pool is not significantly correlated with local convection, though the correlation of total high-frequency activity with convection is high. Comparatively, the CCEW activity and total high-frequency activity over the CEP are both consistent with the background convection.
Apart from the background convection, background SST anomalies may also stimulate and impact CCEWs (Straub and Kiladis 2002; RF04; Masunaga 2007). Correlation coefficients between CCEW activity and the local monthly SST anomalies are shown in Fig. 8. Similar to the distribution of the CCEW activity–local convection relationship, a significant linear correlation between CCEW activity and local SST anomalies is also located over the CEP. In addition, there are weaker local wave activity–SST relationships over the southern Atlantic for all four CCEWs and over the western Pacific for the TD-type and Kelvin wave.
As in Fig. 7, but for the correlation coefficient between CCEW activity and local SST.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 7, but for the correlation coefficient between CCEW activity and local SST.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 7, but for the correlation coefficient between CCEW activity and local SST.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
The similar correlation of CCEW activity with local SST and convection shows that the background convection may be the bridge connecting the SST anomalies and the local CCEW activity. Waliser and Graham (1993) revealed that the relationship between tropical convection and the underlying SST is generally nonlinear; yet this relationship is quasi linear when the underlying SST is between 25° and 29.5°C . Thus the linear relationship of the convection with underlying SST should appear over the tropical CEP with moderate SST. Considering the close correlation of CCEW activity with the convection over the CEP, the strong CEP SST variability may dominate the variability of CCEW activity in this region through impacting the background convection.
In contrast to the absolute variance of CCEW activity (shadings in Fig. 4), the largest relative variance of various CCEW activity (contours in Fig. 4) as well as the high correlation of local SST–CCEW activity and the largest interannual SST variances (e.g., Rasmusson and Carpenter 1982) are all located over the CEP. It indicates that the variability of various CCEW activity is mainly modulated by the background SST, although the largest variances of various CCEW activity are differently distributed.
b. Relationship between CCEW activity and the traditional ENSO
The Hovmöller diagrams of CCEW activity anomalies from 1984 to 2004, after performing three-month running averaging and meridional averaging from 10°S to 10°N, are shown in Fig. 9. Approximately, the CCEW activity anomalies over the CEP are more continuous in contrast with that over the warm pool. Some active/inactive periods of ER and Kelvin waves can persist for a year over the CEP, although a three-month running average has been performed on the anomalies. The lead/lag autocorrelation of monthly CCEW activity anomalies also shows that the autocorrelation with 10-month lead/lag over the CEP is still significant while the autocorrelation over other regions is much poorer (not shown). Another noticeable finding is that the CCEW activity anomalies over the warm pool are generally reversed from that over the CEP.
A 3-month running mean CCEW activity anomalies averaged between 10°S and 10°N from 1984 to 2004: negative anomalies dashed.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
A 3-month running mean CCEW activity anomalies averaged between 10°S and 10°N from 1984 to 2004: negative anomalies dashed.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
A 3-month running mean CCEW activity anomalies averaged between 10°S and 10°N from 1984 to 2004: negative anomalies dashed.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As shown in the above section, the CCEW activity has a significant connection with the local SST over the CEP, the key region of the El Niño–Southern Oscillation (ENSO). There are distinctive CCEW activity anomalies during strong El Niño events, such as 1987–88 and 1997–98 events (Fig. 9), while there are obvious convection anomalies pattern associated with the ENSO over the CEP (e.g., Yulaeva and Wallace 1994 and Fig. 10e). Thus, the relationship between CCEW activity and ENSO cycles are studied using the Niño-3 index (average of the SST anomalies over 5°S–5°N, 150°–90°W) to represent the ENSO cycles.
As in Fig. 8, but for the correlation coefficient of the Niño-3 index with CCEW activity and monthly Tb anomalies: negative values dashed.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 8, but for the correlation coefficient of the Niño-3 index with CCEW activity and monthly Tb anomalies: negative values dashed.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 8, but for the correlation coefficient of the Niño-3 index with CCEW activity and monthly Tb anomalies: negative values dashed.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
In Fig. 10, the correlation coefficient of the Niño-3 index with MRG and TD-type wave activity has a similar distribution such that a significant positive correlation appears over the equatorial CEP, while a negative correlation mainly appears over the off-equatorial western Pacific. The coefficient pattern is similar to the convection anomalies pattern (Fig. 10e) associated with the ENSO except that the negative correlation over the western Pacific is not centered on the equator. The negative correlation over the warm pool is much weaker compared with the positive correlation over the CEP, although it is at the 95% significance level. It may be due to the weaker background convection–CCEW activity relationship over the warm pool, especially on the equator. The significant negative correlation of the MRG and TD-type waves with the Niño-3 index over the warm pool may be a result induced by the local negative SST anomalies associated with the ENSO. However, the negative correlation of ER wave activity with the Niño-3 index over the warm pool is much weaker. This result could be explained by the propagating characteristic of the ER wave different from the MRG and TD-type wave. The dimension and propagation range of MRG and TD-type wave clusters are smaller than for ER and Kelvin waves (see Figs. 9, 15, 18 and 22 in RF04). Thus, during El Niño events, the positive ER anomalies originating from the CEP propagate westward into the warm pool and cancel each other out, with the negative ER anomalies modulated by local negative conditions.
On the other hand, the negative correlation of the MRG and TD-type activity with the Niño-3 index over the warm pool extends into the CEP beyond 10°N and 10°S while the significant positive correlation is over the equatorial CEP. It means that the MRG and TD-type waves over the CEP are more trapped near the equator during the El Niño events. This correlation distribution could be connected with the background convection changes associated with ENSO (Fig. 10e). However, Zhang and Webster (1989) argued that the MRG wave is less trapped in the westerlies than in easterlies, thus the westerly anomalies associated with El Niño should induce a less trapped MRG wave. Therefore, it can be concluded that the anomalous background convection contributes more to the MRG activity than the zonal wind anomalies associated with the ENSO.
For Kelvin waves, the positive correlation over the equatorial CEP is greater than those for other CCEWs since the maximum SST signal associated with ENSO, as well as the maximum Kelvin wave activity and highest SST–Kelvin wave activity correlation, is located over the equatorial CEP. Apart from the Pacific, significant positive Kelvin wave activity associated with the ENSO can be found over the northern Atlantic and Africa, and it may be the bridge of connecting the ENSO and African precipitation (Mekonnen et al. 2008; Nguyen and Duvel 2008).
c. Relationship between CCEW activity and the ENSO Modoki
The correlation coefficient between CCEW activity and the EMI is shown in Fig. 11. A significant positive correlation appears over the equatorial central Pacific while a significant negative correlation just emerges over the eastern Pacific for the TD-type and ER waves. The region with significant CCEW activity–EMI correlation is much smaller than that for the Niño-3 index. Figures 10e and 11e also show the different convection anomalies associated with two-flavor ENSO. The convection anomalies related to the ENSO Modoki are weaker than that for the traditional ENSO; the zonal range of each convection pole of the tripole convection anomaly pattern related to the ENSO Modoki is smaller than that of the dipole convection pattern related to the traditional ENSO. Thus, during the ENSO Modoki, the CCEW activity anomalies induced by the negative and positive convection anomalies can more easily cancel out each other due to the propagation of CCEWs. For example, the westward propagating MRG and TD-type waves are cancelled at the warm pool, and the far-eastern propagating Kelvin wave is mostly cancelled in the eastern Pacific. In addition, the smaller negative convection anomalies over the warm pool during the ENSO Modoki are also located westward and out of region with high CCEW activity–local convection correlation. Therefore, the negative correlation of CCEW activity associated with the ENSO Modoki is much weaker. However, the relationship for the MRG wave over the central Pacific is much weaker in contrast with TD-type wave, though these two kinds of waves have similar interannual variability in other aspects.
As in Fig. 10, but for the El Niño Modoki index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 10, but for the El Niño Modoki index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 10, but for the El Niño Modoki index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
d. Relationship between CCEW activity and other SST anomalies modes
For the Indian Ocean, one of the regions with the largest interannual CCEW activity variance, no significant CCEW activity–local SST relationship is found, as shown in Fig. 8; the remote impacts of ENSO also cannot extend there. However, Shinoda and Han (2005) pointed out that the intensity of submonthly variability over the Indian Ocean is largely reduced during positive Indian Ocean dipole (IOD) years, which is an interannual internal mode of the tropical Indian Ocean characterized by negative SST anomalies off Sumatra and positive SST anomalies in the western Indian Ocean (Saji et al. 1999). Thus, the possible relationship between CCEW activity and the IOD should be investigated.
Figure 12 shows the correlation coefficient of CCEW activity with the IOD index, which is often used to represent the IOD and is defined as the SST anomalies difference between the tropical western Indian Ocean (10°S–10°N, 50°–70°E) and the tropical southeastern Indian Ocean (10°S–equator, 90°–110°E). Significant negative correlation of CCEW activity with the IOD index is found over the southeastern Indian Ocean (Fig. 12). This negative correlation is much higher than the impact of local SST anomalies (Fig. 8) and the remote impact of the ENSO over this region (Fig. 10). This opposite correlation between the southeastern Indian Ocean CCEW activity and the IOD is similar to previous results on the relationship between the IOD and submonthly disturbances (Shinoda and Han 2005). The positive correlation of CCEW activity with the IOD index found in the central-eastern Pacific may be induced by the obvious relationship between the IOD and ENSO during the recent three decades (e.g., Baquero-Bernal et al. 2002).
As in Fig. 10, but for the Indian Ocean dipole index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 10, but for the Indian Ocean dipole index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 10, but for the Indian Ocean dipole index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
However, no significant correlation of CCEW activity–IOD or CCEW activity–local SST is found in the western and northern Indian Ocean, known as the Indian monsoon region. The interaction of CCEWs and atmospheric internal dynamic factors, such as the Indian monsoon, probably dominate the variability of CCEW activity over the western and northern Indian Ocean. For example, the propagation of Rossby waves accompanies both “break” and “active” phases of the Indian monsoon (Lau and Peng 1987).
For the Atlantic region, another active CCEW region, there is also significant correlation of CCEW activity with the underlying SST (Fig. 8). The Atlantic CCEW activity–SST correlation is asymmetric in the Southern and Northern Hemispheres. The correlation over the South Atlantic is much higher than that over the North Atlantic. Thus, the relationship of CCEW activity with a meridional Atlantic SST anomaly mode, the “Tropical Southern Atlantic mode” (TSA) (Enfield et al. 1999), is first considered. The correlation of CCEW activity with the TSA index (the average of the monthly SST anomalies within 0°–20°S, 10°E–30°W) is significant over the equatorial Atlantic (Fig. 13), especially for the TD-type wave. The significant correlation of the MRG wave and Kelvin wave activity with the TSA index is mainly located in the South Atlantic, while the correlation for the ER wave is relatively weak. These correlation patterns are much like the correlations of CCEW activity–convection (Fig. 7) and the CCEW activity–local SST (Fig. 8) over the Atlantic. Therefore, the background convection is also a bridge connecting the South Atlantic SST variability and the CCEW activity, similar to the situation over the Pacific.
As in Fig. 10, but for the tropical southern Atlantic index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 10, but for the tropical southern Atlantic index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
As in Fig. 10, but for the tropical southern Atlantic index.
Citation: Journal of Climate 24, 16; 10.1175/2011JCLI4021.1
Similar analysis is performed on other Atlantic SST anomalies modes, the “Atlantic Meridional Mode” (AMM) (Chiang and Vimont 2004) and the “Tropical Northern Atlantic mode” (TNA) (Enfield et al. 1999). The AMM is an opposite SST anomalies mode between the northern and southern Atlantic, while the TNA is just a SST anomaly mode over the northern Atlantic. However, there is no significant correlation of CCEW activity with the AMM index, as well as the TNA index (not shown), although the AMM index includes some signals of the southern Atlantic SST anomalies. Hence, the TSA mode is the most important mode for the variability of CCEW activity over the Atlantic. It is noteworthy for future studies on the Atlantic CCEWs.
6. Summary and discussions
This work studied the seasonal cycles of monthly activity for the four most important CCEWs including the mixed Rossby–gravity, tropical depression–type, equatorial Rossby, and Kelvin waves for extending the seasonal climatology in WK99 and RF04 based on satellite-observed Tb datasets, using the Wheeler–Kiladis filtering method. The root-mean-square of daily filtered convection was performed for every calendar month to represent the monthly CCEW activity. More precise seasonal cycles of CCEW activity were obtained from the meridional and zonal mean climatology. The seasonal cycles of the MRG, TD-type, and ER wave activity have similar characteristics so that the maximum activity occurs in the Northern Hemisphere during the northern late summer or fall over the western and central Pacific and there are approximately antisymmetric seasonal cycles in the Northern and Southern Hemisphere, while the Kelvin wave has almost synchronized seasonal cycles in both hemispheres and the maximum Kelvin wave activity occurs during the northern late spring and early summer over the CEP. The seasonal cycles of Kelvin waves appear similar to the ITCZ.
In the present study, the results of CCEW activity are mainly based on the convection dataset, although the dynamical fields of the 40-yr ECMWF Re-Analysis (ERA-40) are simply analyzed for validation and a similar climatology is obtained. However, using the convection to represent the activity variation of CCEW dynamics is based on a presupposition that the linear relationship between convection and dynamical fields is constant. Possible seasonal and interannual variations of the convection–dynamical field relationship are not considered. Thus, further study comparing the results of filtering the convection and dynamical fields and analyzing their relationship is needed to investigate the convectively coupling characteristics of tropical waves.
Interannual variance of CCEW activity was further investigated based on monthly CCEW activity anomalies, which removed the long-term mean seasonal cycles. The Kelvin wave activity has maximum interannual variance over the eastern Pacific during the northern spring, while the activity of the other three types of CCEWs has similar distribution to their climatological activity. However, the relative variance of each CCEW activity, ratio of interannual standard deviations to the long-term mean CCEW activity, has the almost identical distribution that largest values are located over the equatorial CEP. Similarly, the MJO activity is also significantly correlated with local SST over the equatorial CEP (Fink and Speth 1997).
The relationship of CCEW activity with local convection and SST is dependent on the location, although previous studies show that the SST may dominate the ocean–CCEW coupled system in the tropics (Straub and Kiladis 2002; RF04; Masunaga 2007). The relationship is close over the relatively low SST and inactive convection regions, the equatorial CEP, and the southern Atlantic, but not significant over the tropical warm pool with the most active convection. Since the CCEW activity is mainly determined by the amplitude of individual CCEW events due to the episodic character of CCEW events, convection over the warm pool is determined largely by the frequency and duration of weak unorganized signals, while convection over the CEP is consistent with the CCEW activity due to weak convection and less high-frequency convection fluctuations over this region. On the other hand, the relationship of local SST and convection has a similar distribution to the convection–CCEW activity correlation since the SST–convection relationship is quasi linear for the SST range 25°–29.5°C (Waliser and Graham 1993). Therefore, convection is the bridge connecting CCEW activity and local SST with high correlation over the equatorial CEP and southern Atlantic.
The connections of CCEW activity with tropical Pacific SST anomaly events, known as the traditional ENSO and ENSO Modoki, were also investigated. Significant positive ENSO–CCEW activity correlation is located in the equatorial CEP, while negative correlation of the MRG and TD-type with the ENSO is high over the off-equatorial western and central Pacific. Therefore, the MRG and TD-type waves over the CEP are more trapped near the equator during El Niño events, although westerly anomalies associated with El Niño may induce less trapped MRG waves (Zhang and Webster 1989). The negative correlation of MRG and TD-type activity with Niño-3 index over the warm pool is much higher than that of the ER wave. This may be attributed to the dimension and propagating range of the MRG and TD-type wave clusters being smaller than the ER wave. Thus, the negative SST anomalies over the warm pool associated with the ENSO may dominate the local MRG and TD-type activity, and the positive ER anomalies from the CEP and local negative ER anomalies related to negative SST anomalies over the warm pool cancel each other out at the western Pacific.
The significant positive correlation of CCEW activity with the ENSO Modoki appears over the equatorial central Pacific, while the significant negative correlation only emerges over the eastern Pacific for the TD-type and ER waves. The ENSO Modoki has much weaker impact on the CCEWs activity than the traditional ENSO. The primary reason may be that the zonal range of each convection pole of the tripole convection anomaly pattern associated with the ENSO Modoki is smaller than that of the dipole convection pattern associated with the traditional ENSO. Thus, the positive and negative CCEW activity anomalies related to the tripole SST anomalies of ENSO Modoki can more easily cancel each other due to the propagation of CCEWs.
For the Indian Ocean dipole, there is a significant negative correlation with CCEW activity over the southeastern Indian Ocean, while no significant correlation of the external factors related to CCEW activity can be found over the western and northern Indian Ocean, the Indian monsoon region. This indicates that interannual variability of the CCEW activity over the Indian monsoon region may be determined by transient disturbances and atmospheric internal dynamic interaction. Over the Atlantic region, the tropical Southern Atlantic mode has the strongest impact on the Atlantic CCEWs activity, while the Atlantic meridional mode or the tropical Northern Atlantic mode has no significant relationship with CCEW activity. The correlation of CCEW activity with Atlantic SST anomaly modes may provide some teleconnections for Atlantic hurricanes and Atlantic SST anomalies.
Although the interannual relationship of CCEW activity with tropical SST anomalies events is investigated and some mechanisms are brought forward, the tropical SST anomaly events are treated as background and the dominant factor in the CCEW activity–SST system. In fact, CCEWs and SST may interact as in the MJO–ENSO relationship (e.g., Lau 1985). For example, the possible role of CCEWs on initiating or terminating ENSO events is a necessary investigation in further studies.
Acknowledgments
These results were obtained using the CLAUS Tb archive held at the British Atmospheric Data Centre, produced using ISCCP source data distributed by the NASA Langley Data Center, and we must thank Dr. Carl J. Schreck III for providing the preprocessed CLAUS Tb dataset in netCDF format. We also thank three anonymous reviewers for their constructive advice. This research was supported financially by the National Natural Science Foundation of China-Regional Cooperation Project (Grant 40921160379), the Special Scientific Research Project for Public Welfare of China Meteorological Administration (Grant 201006021), the National Basic Research Program of China (Grant 2010CB950403), and the National Natural Science Foundation of China (Grant 40975046).
REFERENCES
Ashok, K., S. K. Behera, S. A. Rao, H. Weng, and T. Yamagata, 2007: El Niño Modoki and its possible teleconnection. J. Geophys. Res., 112, C11007, doi:10.1029/2006JC003798.
Baquero-Bernal, A., M. Latif, and S. Legutke, 2002: On dipolelike variability of sea surface temperature in the tropical Indian Ocean. J. Climate, 15, 1358–1368.
Chen, G., and R. Huang, 2009: Interannual variations in mixed Rossby–gravity waves and their impacts on tropical cyclogenesis over the western North Pacific. J. Climate, 22, 535–549.
Chiang, J., and D. Vimont, 2004: Analogous Pacific and Atlantic meridional modes of tropical atmosphere–ocean variability. J. Climate, 17, 4143–4158.
Cho, H.-K., K. P. Bowman, and G. R. North, 2004: Equatorial waves including the Madden–Julian oscillation in TRMM rainfall and OLR data. J. Climate, 17, 4387–4407.
Enfield, D., A. Mestas-Nuñez, D. Mayer, and L. Cid-Serrano, 1999: How ubiquitous is the dipole relationship in tropical Atlantic sea surface temperatures? J. Geophys. Res., 104, 7841–7848.
Fink, A., and P. Speth, 1997: Some potential forcing mechanisms of the year-to-year variability of the tropical convection and its intraseasonal (25-70-day) variability. Int. J. Climatol., 17, 1513–1534.
Frank, W., and P. Roundy, 2006: The role of tropical waves in tropical cyclogenesis. Mon. Wea. Rev., 134, 2397–2417.
Hodges, K., D. Chappell, G. Robinson, and G. Yang, 2000: An improved algorithm for generating global window brightness temperatures from multiple satellite infrared imagery. J. Atmos. Oceanic Technol., 17, 1296–1312.
Kiladis, G., M. Wheeler, P. Haertel, K. Straub, and P. Roundy, 2009: Convectively coupled equatorial waves. Rev. Geophys., 47, RG2003, doi:10.1029/2008RG000266.
Lau, K.-M., 1985: Elements of a stochastic–dynamical theory of the long-term variability of the El Niño–Southern Oscillation. J. Atmos. Sci., 42, 1552–1558.
Lau, K.-M., and L. Peng, 1987: Origin of low-frequency (intraseasonal) oscillations in the tropical atmosphere. Part I: Basic theory. J. Atmos. Sci., 44, 950–972.
Lindzen, R. S., 1967: Planetary waves on beta planes. Mon. Wea. Rev., 95, 441–451.
Lindzen, R. S., 2003: The interaction of waves and convection in the Tropics. J. Atmos. Sci., 60, 3009–3020.
Masunaga, H., 2007: Seasonality and regionality of the Madden–Julian oscillation, Kelvin wave, and equatorial Rossby wave. J. Atmos. Sci., 64, 4400–4416.
Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44, 25–43.
Mekonnen, A., C. Thorncroft, A. Aiyyer, and G. Kiladis, 2008: Convectively coupled Kelvin waves over tropical Africa during the boreal summer: Structure and variability. J. Climate, 21, 6649–6667.
Nguyen, H., and J. P. Duvel, 2008: Synoptic wave perturbations and convective systems over equatorial Africa. J. Climate, 21, 6372–6388.
Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110, 354–384.
Rayner, N., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.
Roundy, P., and W. Frank, 2004: A climatology of waves in the equatorial region. J. Atmos. Sci., 61, 2105–2132.
Saji, N., B. Goswami, P. Vinayachandran, and T. Yamagata, 1999: A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363.
Shinoda, T., and W. Han, 2005: Influence of the Indian Ocean dipole on atmospheric subseasonal variability. J. Climate, 18, 3891–3909.
Straub, K. H., and G. N. Kiladis, 2002: Observations of a convectively coupled Kelvin wave in the Eastern Pacific ITCZ. J. Atmos. Sci., 59, 30–53.
Straub, K. H., and G. N. Kiladis, 2003a: Interactions between the boreal summer intraseasonal oscillation and higher-frequency tropical wave activity. Mon. Wea. Rev., 131, 945–960.
Straub, K. H., and G. N. Kiladis, 2003b: Extratropical forcing of convectively coupled Kelvin waves during austral winter. J. Atmos. Sci., 60, 526–543.
Takayabu, Y., 1994: Large-scale cloud disturbances associated with equatorial waves. I: Spectral features of the cloud disturbances. J. Meteor. Soc. Japan, 72, 433–449.
Waliser, D., and N. Graham, 1993: Convective cloud systems and warm-pool sea surface temperatures: Coupled interactions and self-regulation. J. Geophys. Res., 98, 12 881–12 893.
Wheeler, M., and G. N. Kiladis, 1999: Convectively coupled equatorial waves: Analysis of clouds and temperature in the wavenumber-frequency domain. J. Atmos. Sci., 56, 374–399.
Wheeler, M., G. N. Kiladis, and P. J. Webster, 2000: Large-scale dynamical fields associated with convectively coupled equatorial waves. J. Atmos. Sci., 57, 613–640.
Yang, G.-Y., B. Hoskins, and J. Slingo, 2003: Convectively coupled equatorial waves: A new methodology for identifying wave structures in observational data. J. Atmos. Sci., 60, 1637–1654.
Yang, G.-Y., B. Hoskins, and J. Slingo, 2007a: Convectively coupled equatorial waves. Part I: Horizontal and vertical structures. J. Atmos. Sci., 64, 3406–3423.
Yang, G.-Y., B. Hoskins, and J. Slingo, 2007b: Convectively coupled equatorial waves. Part II: Propagation characteristics. J. Atmos. Sci., 64, 3424–3437.
Yulaeva, E., and J. M. Wallace, 1994: The signature of ENSO in global temperature and precipitation fields derived from the microwave sounding unit. J. Climate, 7, 1719–1736.
Zhang, C., and P. J. Webster, 1989: Effects of zonal flows on equatorially trapped waves. J. Atmos. Sci., 46, 3632–3652.
Zhang, C., and M. Dong, 2004: Seasonality in the Madden–Julian oscillation. J. Climate, 17, 3169–3180.
