1. Introduction
Most previous studies related to extratropical response to tropical heating were conducted for the boreal winter season over the Northern Hemisphere (NH). One reason is that the NH winter has the strongest extratropical westerly zonal flow, which is favorable for Rossby wave propagation. In contrast, the extratropical response in boreal summer and in the Southern Hemisphere (SH) has received little attention. Sardeshmukh and Hoskins (1988) investigated the Rossby wave source of tropical divergence induced by convective heating and found that an equatorial forcing in easterly winds can lead to a Rossby wave source in the subtropical westerlies, where it is extremely effective. This implies that in the boreal summer, even though the major tropical convection variability is in the NH off the equator, upper divergence could still lead to an effective Rossby wave source in the SH subtropics and induce considerable extratropical response.
The Asian summer monsoon is a dominant climate phenomenon of the NH summer circulation, which is usually accompanied with heavy precipitation from India to the Philippine Sea. Numerous studies have been done on the Asian summer monsoon (e.g., Lau et al. 2004; Ding 2007), and numerical models have been used to study the effects of mountains and diabatic heating on the mean state of the monsoon system (e.g., Hoskins and Rodwell 1995; Rodwell and Hoskins 1995). The monsoon possesses abundant variabilities with different time scales. The diabatic heating associated with the monsoon precipitation variability represents an important forcing to the global atmosphere. In the tropical region, the Asian summer monsoon is composed of two relatively independent subsystems: the Indian summer monsoon (ISM) and the western North Pacific summer monsoon (WNPSM), with strong variabilities in convection over the India–Bay of Bengal and South China Sea–Philippine Sea regions, respectively (e.g., Wang and Fan 1999; Wang et al. 2001).
The atmospheric teleconnections associated with the Asian summer monsoon are an interesting topic. Kripalani et al. (1997) found significant simultaneous positive correlations between summer India rainfall and 500-hPa geopotential height in three midlatitude regions: northeastern China, North Africa, and north of Iran. This positive correlation along the midlatitude Eurasian continent between the geopotential height and the ISM precipitation is likely a part of the circumglobal teleconnection pattern in the NH summer found in Ding and Wang (2005, 2007). On the other hand, the connection of the extratropical circulation to the WNPSM has a quite different structure. Nitta (1987) found a teleconnection pattern associated with the anomalous convective activities over the tropical western Pacific. This teleconnection pattern tends to have a wave train structure that extends from Philippine Sea to Japan, referred to as the Pacific–Japan (PJ) pattern. Little is known, however, about the connection of the SH circulation to the Asian summer monsoon. Nitta (1989) used 6 yr of data to study the global aspect of the PJ pattern and suggested that there could be a connection between the SH circulation anomalies and the convective activity in the tropical western Pacific. However, a clear picture of the teleconnection patterns associated with the ISM and WNPSM and their mechanisms has not yet been achieved.
In this study, connections between the monsoon convective anomalies and the global circulations are analyzed using monthly data of the 29 most recent years. Numerical experiments with idealized tropical forcing are performed to understand such global connections.
Section 2 describes the data used in the analysis. In section 3 precipitation indices are defined for the convective activity in different monsoon regions that are to be used to relate to the global circulations. Section 4 presents the observed global teleconnections associated with the monsoon convective activity using correlations and singular value decomposition (SVD) analysis. In section 5, model experiments are performed to assess whether the circulation anomalies are a response to monsoon heating. Linear experiments are conducted in section 6 to analyze the dynamical processes. Conclusions and discussions follow in section 7.
2. The data and analysis procedures
We make use of the National Centers for Environmental Prediction National Center for Atmospheric Research (NCEP–NCAR) global reanalysis (Kalnay et al. 1996). Variables used here include monthly geopotential height at 200 hPa and zonal and meridional winds at 200 and 850 hPa. Streamfunctions at 200 and 850 hPa are calculated from the wind data. To describe the tropical convection associated with the Asian summer monsoon, the monthly data of the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) are used. Other data applied to represent a proxy for tropical convection are the monthly satellite-observed outgoing longwave radiation (OLR) data from the National Oceanic and Atmospheric Administration (NOAA) polar-orbiting series of satellites (Liebmann and Smith 1996). These data are provided by the NOAA Earth System Research Laboratory (ESRL) Physical Sciences Division (PSD) in Boulder, Colorado, from their Web site (http://www.cdc.noaa.gov/).
All the datasets have a 2.5° longitude × 2.5° latitude horizontal resolution. The analysis is conducted for 29 NH summers from 1979 to 2007, where the summer is defined to be the months of June–September (JJAS), with a total of 116 months. The annual cycle of the 29-yr monthly climatology is removed for each grid point before the analysis.
3. Precipitation indices for the Asian summer monsoon
Figures 1a and b depict the time average and standard deviation, respectively, of the monthly precipitation calculated from the CMAP data for the 116 JJAS months from 1979 to 2007. As can be seen, the time average and the standard deviation have a similar distribution; that is, maxima exist over western India, the Bay of Bengal, and the tropical western North Pacific region from the South China Sea to the Philippine Sea. These features are consistent with previous studies (e.g., Ding 2007). Shown respectively in Figs. 1c and 1d are the time average and standard deviation of the OLR, a proxy for tropical convective activity. The strong summertime seasonal mean convective activity in the South Asian region is reflected by the weak OLR values (less than 220 W m−2 in Fig. 1c). The variability of the OLR agrees with that of the precipitation in the western India and western Pacific regions, where maximum standard deviations for both variables are found. In the Bay of Bengal, however, where the precipitation has a strong variability, the OLR only has weak standard deviation values, indicating that in this region the precipitation variability is weakly associated with the convective activity.
To represent the convective activity and precipitation in different monsoon regions, three areas are selected that have maximum precipitation variabilities. Region 1 represents western India (10°–30°N, 65°–80°E), region 2 the Bay of Bengal (10°–30°N, 85°–100°E), and region 3 the vicinity of the Philippines (10°–25°N, 120°–150°E). Regions 1 and 3 are outlined with solid rectangles in Figs. 1b and 1d, and region 2 is outlined with a dashed rectangle. Three precipitation indices (PI1, PI2, and PI3) are defined as the precipitation rate anomaly averaged over these three regions.
Three OLR indices can also be defined using the OLR anomaly averaged over the same three regions (CI1, CI2, and CI3). Because the seasonal cycle has been removed, the monthly values of the precipitation and OLR indices include interannual and intermonthly variabilities. Shown in Table 1 are the correlations between pairs of the three precipitation indices and the three OLR indices. Strong correlations are found between PI1 and CI1 (−0.79) and between PI3 and CI3 (−0.93), indicating that the precipitation variability in the western Indian and western Pacific regions is closely associated with the convective activity. For this reason, we will focus our following analysis on the atmospheric circulation associated with convective activity in these two regions. In comparison, the correlation between PI2 and CI2 is significantly weaker, implying that in the Bay of Bengal region the precipitation variability is only weakly associated with the convective activity. It can also be found that the three precipitation indices are nearly independent, with almost no correlation to each other.
Wang and Fan (1999) and Wang et al. (2001) divided the Asian summer monsoon into ISM and WNPSM subsystems. They defined the intensity of convective heating of the ISM and WNPSM by averaging the rainfall over the India–Bengal region and the vicinity of the Philippines. In the present study, the region used to define PI3 matches well that of the WNPSM rainfall. Therefore, we will refer to PI3 as a convection index for WNPSM. The ISM has two maximum precipitation variability regions (i.e., western India and the Bay of Bengal; Fig. 1b). Because the rainfall in the Bay of Bengal is weakly associated with deep convections, for simplicity we will refer to PI1 as a convection index for ISM.
4. Observed global teleconnections associated with monsoon convective activity
a. Global correlations with precipitation indices
Temporal correlations are calculated to examine the association between the global circulations and the monsoon convective activity using monthly anomaly data. As the zonally symmetric structure is not the focus of the present study, the zonal means of the atmospheric variables are removed before the calculations. Shown in Fig. 2 are correlation coefficients between the global circulation and the ISM convective activity (PI1). Large areas of significant correlations are found not only in the NH but also in the SH. In 200-hPa geopotential height field (Fig. 2a), significant correlations appear along the subtropical latitudes in both hemispheres. A general feature is that the circulation anomaly is quite symmetric about the equator. A positive ISM convection anomaly is associated with positive subtropical geopotential height anomalies (Fig. 2a) and large anticyclonic gyres (Fig. 2b) in the upper troposphere in both hemispheres, which extend westward from Indian Ocean to the American continents. Between the anticyclonic gyres lies an easterly zonal wind anomaly over the equator. This structure resembles an upper equatorial Rossby wave response to a tropical heating located near India (Gill 1980). In the lower troposphere, two cyclonic flow anomalies are found to the west of India, straddling the equator, but they are quite local and have a smaller extent compared to the upper tropospheric anticyclones. At 850 hPa, significant correlations in streamfunction are mainly located over Pacific, corresponding to large anticyclonic flow anomalies in the NH and SH with easterly equatorial wind across the Pacific (Fig. 2c).
In the middle latitudes of the NH, significant positive 200-hPa height anomalies are found to exist along the upper westerlies (Fig. 2a). Five positive correlation centers are found over North Africa, west-central Asia, northeast China, the North Pacific, and North America. This wave pattern looks very similar to the circumglobal pattern as observed in Ding and Wang (2005) and to the teleconnection pattern associated with the ISM in Wang et al. (2001). It is also similar to that in Branstator (2002), even though the latter is for the winter season. When using geopotential height or streamfunction fields, the structure is dominated by positive centers. On the other hand, if one uses meridional wind, wave patterns of both positive and negative centers would be evident, as pointed out in Branstator (2002). The positive correlations of 200-hPa geopotential height with ISM convection along the midlatitude Eurasian continent also agrees with the observations of Kripalani et al. (1997). In the SH, a wave train is seen to originate from the subtropical western Indian Ocean, propagating first southeastward and then northeastward to influence the South Pacific.
Figure 3 depicts the correlation between the global circulation and the WNPSM convective activity (PI3). As can be seen, the teleconnection associated with PI3 is very different from that associated with PI1. A striking feature is a significant Rossby wave train in the SH that propagates from the subtropical eastern Indian Ocean southeastward to the high-latitude South Pacific and then moves northeastward to influence South America. In the SH extratropical regions, the circulation anomalies seems to have an equivalent barotropic structure with the same sign in the upper and lower troposphere (Figs. 3b and 3c), consistent with previous studies (e.g., Jin and Hoskins 1995). In the tropical region, little evidence of equatorial wave generation is found except for the baroclinic gyres in streamfunction straddling the equator near the Philippine longitudes (Figs. 3b and 3c). The geopotential height anomalies are mainly in the extratropical regions. In the NH, the only area with significant correlations is over Japan and eastern China. Above-normal WNPSM convective activity is associated with a positive 200-hPa geopotential height anomaly over Japan and eastern China, a feature similar to the PJ teleconnection pattern (Nitta 1987; Kosaka and Nakamura 2006).
To see if the strong rainfall variability in the Bay of Bengal (Fig. 1b) is connected to the global circulation, correlations with PI2 are presented in Fig. 4. This time few significant correlations are found. As discussed in the last section, although there exists a maximum precipitation variability over the Bay of Bengal, the OLR variance is small. The rainfall variability in this region is not associated with significant changes in deep convections and diabatic heating; thus, its global influence is limited.
In summary, we found two distinct global teleconnection patterns associated with the ISM and WNPSM convection anomalies. The ISM variability is correlated with circulations not only in the tropics but also in the extratropical regions in both hemispheres, suggesting that the ISM convection variability has a significant far-reaching global connection. On the other hand, the circulation anomaly associated with the WNPSM convection anomaly is mainly in the extratropical SH, which takes the form of a wave train. The WNPSM variability is significantly correlated with a Rossby wave train that influences the high-latitude South Pacific and South America.
b. Singular value decomposition analysis
We have seen that two distinct teleconnection patterns are respectively associated with the ISM and WNPSM convective activities. In this subsection, we use the singular value decomposition technique (Bretherton et al. 1992) to confirm that these two teleconnection patterns are indeed the dominant global patterns coupled with the major convective activities of the Asian summer monsoon. The SVD analysis is conducted from the covariance between the global monthly 200-hPa geopotential height anomaly and the OLR in the Asian summer monsoon region (0°–30°N, 60°–150°E) in the 29 JJAS seasons. The data of 200-hPa geopotential height for SVD analysis are on a 5° × 5° grid between 85°S and 85°N. The effect of unequal areas represented by different grid points is taken into account by multiplying the value at a grid by the square root of the cosine of latitude at that grid point. OLR data are on a 2.5° × 2.5° grid, so there is no reduction of resolution. Again, the zonal mean of the 200-hPa geopotential height is first removed because we are interested in the departure from the zonally symmetric flow. It is noted that the streamfunction may be a better variable to represent tropical circulation (Hsu and Lin 1992), but the inverse Laplacian applied when calculating the streamfunction acts as a strong spatial filter and thus the streamfunction may not well represent local circulations. Because our interest is mainly in looking at extratropical circulation anomalies, upper-level geopotential height is therefore used in the SVD analysis here and in the model output analysis in sections 5 and 6.
The percentages of squared covariance explained by the first four pairs of SVDs, along with the correlations between the time series of expansion coefficients of 200-hPa geopotential height and that of OLR, are listed in Table 2. SVD1 and SVD2 explain more than 60% of the total squared covariance and clearly are the dominant coupled modes of the global circulation and the Asian monsoon convective activity. Shown in Fig. 5 are (top) global 200-hPa geopotential height and (bottom) monsoon region OLR distributions of SVD1 and SVD2, presented as heterogeneous regressions. The magnitude corresponds to one standard deviation of each time coefficient. The main feature of the OLR distribution of SVD1 (Fig. 5c) shows an increased convective activity (negative OLR anomaly) over western India and the adjacent Arabian Sea region. The temporal correlation between the expansion coefficient of the OLR component of SVD1 and PI1 is 0.55, well passing the 0.01 statistical significance level. Associated with this Indian monsoon OLR anomaly, the 200-hPa geopotential height component of SVD1 (Fig. 5a) shows a structure similar to the circulation anomaly associated with PI1 (Fig. 2a). The circumglobal teleconnection pattern in the NH is evident with five positive centers, consistent with the correlation map of Fig. 2a. The pattern correlation coefficient, measuring the spatial correlation between Figs. 2a and 5a over the whole global region, is 0.67.
The OLR distribution of SVD2 (Fig. 5d) corresponds to an enhanced convection and precipitation over the WNPSM region. In fact, the expansion coefficient of the OLR component of SVD2 is highly correlated with PI3 (0.84). The 200-hPa geopotential height component of SVD2 (Fig. 5b) is almost identical to the circulation anomaly associated with PI3 (Fig. 3a), and the pattern correlation coefficient between these two fields reaches 0.85.
5. Model experiments of the global response to monsoon heating
The observed circulation anomalies associated with the ISM and WNPSM precipitation have two distinct global-scale patterns. To test the hypothesis that the global circulation anomalies result from a direct response to the monsoon convective heating, a series of numerical model experiments are performed. We use the simple general circulation model (SGCM) as described in Hall (2000). In brief, it is a global spectral model with no moisture representation. The resolution used in this study is T31 with 10 vertical levels. A detailed description of the model parameters can be found in Hall and Derome (2000).
An important feature of this model is that a time-independent forcing is used to maintain a climatology similar to the observations that is calculated empirically from observed daily data. This forcing is obtained as a residual for each time tendency equation by computing the dynamical terms of the model, together with the dissipation, with daily global analyses and averaging in time. Such a procedure is very similar to that used in Roads (1987) and also in a quasigeostrophic setting by Marshall and Molteni (1993) and Lin and Derome (1996). All the processes that are not resolved by the model’s dynamics are thus included in the forcing. No topography is prescribed in the model, but its time-mean effect is accounted for by the forcing. To calculate the model forcing of JJAS for the purpose of the present study, the daily data of the NCEP–NCAR reanalyses of 30 JJAS seasons from 1978 to 2007 are used. Forcing fields are calculated separately for each JJAS, and the time average of the 30 forcing fields is obtained as the constant climatological forcing.
Three experiments are conducted, namely 1) the control run, with climatological JJAS forcing; 2) the ISM run, with climatological JJAS forcing plus an anomalous ISM heating anomaly; and 3) WNPSM run, with climatological JJAS forcing plus an anomalous WNPSM heating anomaly. For each experiment, an integration of 3700 days is performed from an initial condition that is taken from 1 July 1978 observations. Note that the result does not depend on the selection of initial condition because the analysis is conducted for the period after the climate equilibrium is reached. For both the ISM and WNPSM runs, a tropical heating anomaly that is added to the temperature equation is switched on at t = 0 and persists during the integration. No forcing anomaly is applied to the vorticity, divergence, or mass equations. The heating perturbation represents deep convection in the tropics and has an elliptical form in the horizontal. Both the ISM and WNPSM heatings have a semimajor axis of 28° of longitude. The semiminor axis for the ISM heating is 17° of latitude, whereas that for the WNPSM forcing is 12°, reflecting the fact that the latter has a narrower meridional extent according to Figs. 1b and 1d. The magnitude of the heating is proportional to the squared cosine of the distance from the center. The heating anomaly has a vertical profile of (1 − σ) sin[π(1 − σ)], which peaks at σ = 0.35 with a vertically averaged heating rate at the center of 2.5°C day−1. This rate is equivalent to a latent heating associated with a precipitation of 1 cm day−1. The center locations of the heating anomalies are 20°N, 75°E for the ISM and 15°N, 130°E for the WNPSM, respectively. The vertically averaged heating anomalies for the ISM and WNPSM runs are illustrated separately (in Figs. 7a and 8a, respectively).
Integrations starting from observed states and the adjustment to model forcings can be expected to cause an initial model spinup; therefore, the output of the first 100 days is not used. The average of the last 3600 days represents the climate of each run, and the difference between a perturbation run and the control run represents a forced signal by the anomalous forcing field.
Hall (2000) demonstrated that with an NH winter climatological forcing the SGCM simulates reasonably well the wintertime global time mean circulation and variability. To illustrate that the SGCM is also able to reproduce the NH summertime climate, Fig. 6 compares the time mean JJAS streamlines of the NCEP reanalysis at 200 and 850 hPa with those of the 3600-day averages of the control perpetual summer integration. The areas in orange are those with a time mean zonal wind U greater than 15 m s−1 at 200 hPa and 5 m s−1 at 850 hPa, whereas the areas in blue indicate where U < −10 m s−1 at 200 hPa and U < −5 m s−1 at 850 hPa. As can be seen, the SGCM is doing a reasonably good job in simulating the large-scale features in both the upper and lower levels of the troposphere. In both the NCEP reanalysis and the SGCM simulation, strong upper westerlies are observed in the SH middle latitudes. Plots of U200 (not shown) indicate that the SH westerly jet occurs around 30°S with maximum winds near the east coast of Australia. In the NH midlatitudes, the upper westerly jet is weaker and narrower than in the SH, with maximum winds extending from North Africa to the middle Pacific and near the east coast of North America. The monsoon circulation is also reproduced well, which includes the South Asian high over the Tibetan Plateau, the tropical easterly jet, and the southward cross-equatorial flow over the equatorial Indian Ocean and the Maritime Continent at 200 hPa (Figs. 6a and 6c). At 850 hPa, the monsoon westerly flow over the Arabian Sea, India, and the Bay of Bengal is simulated, as well as the northward cross-equatorial flows near Somalia and the 120°E longitudes (Figs. 6b and 6d). The large-scale structure provides a background for the atmospheric response to tropical heating anomalies. We will discuss the role of the basic flow in section 6 when linear experiments are analyzed.
Shown in Figs. 7b and 7c are the model’s response to the ISM forcing at 200-hPa geopotential height and 850-hPa streamfunction, after the zonal means are removed. They are calculated as the difference between the 3600-day time means of the ISM perturbation run and the control run. Compared to the correlation map between the observed 200-hPa geopotential height and PI1 (Fig. 2a) and to the 200-hPa geopotential height component of SVD1 (Fig. 5a), the main features are reproduced in the model. The pattern correlation coefficient, measuring the spatial correlation between Figs. 2a and 7b over the whole global region, is 0.70. The subtropical response is quite symmetric about the equator, with the anticyclonic gyres in the upper troposphere in both hemispheres extending westward from the Indian Ocean to the tropical Atlantic. The positive anomalies with five centers along the NH midlatitude westerlies are in general consistent with the observations. This indicates that the ISM convective anomaly is likely responsible for generating the global teleconnection pattern. The main discrepancy occurs in the midlatitude South Pacific, where the simulated ridge is too weak. At 850 hPa (Fig. 7c), the streamfunction response is also comparable to the observations (Fig. 2c).
The model’s 200-hPa geopotential height and 850-hPa streamfunction responses to the WNPSM forcing are illustrated in Figs. 8b and 8c, respectively. A comparison of Fig. 8b with Figs. 3a and 5b reveals that the SH wave train is reproduced in the model. The 850-hPa streamfunction response (Fig. 8c) also in general agrees with the observations (Fig. 3c). It can be concluded that the convective anomaly in the western North Pacific monsoon region is responsible for the teleconnection pattern, which has an important impact on the SH weather.
6. Linear experiments
To better understand the dynamical processes that determine the extratropical response to the Asian summer monsoon forcing, a linear perturbation model is used that is based on the SGCM, with the approach as described in Hall and Derome (2000). The basic state is chosen to be the model climate of the control perpetual summer integration as described in the last section, whose 200-hPa flow was illustrated in Fig. 6c. A forcing is applied to maintain this basic state, which is calculated as a residual in the model equations with only time mean quantities. The same perturbation forcings as for the nonlinear integrations as described in the last section are used. The linear model is integrated for 30 days starting from the SGCM’s control run climate. To ensure a small-amplitude linear response, the forcing amplitude is scaled by 10−2 times the anomalous forcing, and the solutions are scaled back for display purposes.
The day-15 200-hPa geopotential height perturbations in the linear model forced by the ISM and WNPSM forcings are shown in Figs. 9a and 9b, respectively. By comparing Fig. 9a with Fig. 7b and comparing Fig. 9b with Fig. 8b, it is clear that almost all of the features in the full model nonlinear responses are reproduced in the linear integrations for both the ISM and WNPSM cases, except that the response in the SH extratropics is weaker in the linear runs. This indicates that the linear dynamics is largely responsible for the global response to the ISM and WNPSM forcings. In the SH, the nonlinear processes are likely to enhance the height anomaly in the midlatitudes.
Because the global response to the summer monsoon heating anomaly in both the ISM and WNPSM regions can be explained reasonably well by the linear model, a detailed analysis of the linear integrations could help explain how the responses develop and why the extratropical responses to the ISM and WNPSM forcings are different. Figure 10 gives the evolutions of the 200-hPa geopotential height anomalies of the two linear integrations from days 1 to 5. As the equator is located near the edge of the heating source in both cases, a weak Kelvin wave symmetric about the equator is generated to the east of the heating. This Kelvin wave signal propagates fast eastward, traveling about half of the equator in 5 days, consistent with previous studies (e.g., Gill 1980; Jin and Hoskins 1995; Rodwell and Hoskins 1996). The Kelvin wave, however, is a weak transient feature. By day 15 (Figs. 9a and 9b), its signature can hardly be seen. Differences of response between the ISM and WNPSM runs begin to build up from day 1 to the west of the heating. For the ISM case (Figs. 10a–e), to the west of the heating there is a pair of anticyclones straddling the equator starting to develop right after the forcing is switched on, corresponding to the equatorially trapped Rossby wave response. The NH branch of the equatorial Rossby wave is much stronger than its SH counterpart. The Rossby wave anticyclonic gyres develop in strength and spread in extent westward, and by day 15 (Fig. 9a) their west front reaches the tropical American continent. Therefore, it is this Rossby wave response that constitutes the component that is symmetric about the equator. The westward expansion of the anticyclonic gyres influences the circulation in a wide region including India, the Middle East, Africa, and subtropical North and South America. To the north of the NH Rossby wave gyre, a wave train starts to develop on day 2 along the midlatitude westerly jet, and by day 15 (Fig. 9a) it becomes a chain of waves around the global NH midlatitudes. For the WNPSM heating run (Figs. 10f–j), however, the equatorially trapped Rossby wave response to the west of the heating is not symmetric about the equator. This time, the NH branch is very weak and does not extend westward. In contrast, the SH branch of the equatorial Rossby wave is much stronger and expand in extent westward. By day 5 it covers a wide zonally elongated area from Australia to southern Africa. Wave energy seems to emanate southeastward, and by day 15 significant anomalies can be found near the South Pole region (Fig. 9b) and the wavetrain pattern is formed.
We calculate each term on the right-hand side of (3) at 200 hPa on day 5 for the two linear integrations. Shown in Fig. 12 are the first two terms: dominant and total RWS (S′). A spectral filter has been applied to smooth the Rossby wave source fields. This filter was discussed in Sardeshmukh and Hoskins (1984) and takes the form of e−K[n(n+1)]2, with K chosen so that the highest wavenumber spectral coefficients are multiplied by 0.1. Term 1 represents the generation of wave vorticity by the anomalous divergence, and term 2 the advection of mean absolute vorticity by the perturbation divergent winds. In the case of the ISM heating, there is a series of RWS with alternating positive and negative centers along the NH midlatitude westerlies from North Africa to East Asia, which is believed to be responsible for the generation of the wavelike anomalies over the midlatitude Eurasian continent and the circumglobal patterns in Fig. 9a. For the case of WNPSM heating, a small area of RWS in the NH is generated only over north China and Japan, which may explain why there is only a limited response in the NH extratropics. In the SH, both the ISM and WNPSM heatings generate a positive RWS in the middle latitudes. Terms 1 and 2 both contribute to the RWS in the SH midlatitudes, although term 1 is stronger. The RWS in the ISM and WNPSM heating cases, however, occur at different longitudinal locations. The RWS of the ISM heating is located in the South Indian Ocean, whereas that for the WNPSM heating appears over Australia. The different Rossby wave propagations in these two cases may be related to the relative location of the RWS to the SH zonal westerly jet. For the WNPSM heating case (Fig. 12f), the RWS is located right at the westerly jet core region, where the generation of Rossby wave is more effective and the Rossby wave can propagate farther than in the ISM case. According to the analysis of Hoskins and Karoly (1981), the long wave structure of the RWS in the SH midlatitudes also implies that the generated Rossby wave can propagate poleward, whereas the short wavelength of RWS in the midlatitude NH in the ISM case would result in meridionally trapped waves along the westerly jet.
We now have seen that the total RWS is dominated by that related to the perturbation divergence (term 1; Figs. 12a and 12d). An immediate question to ask is how the upper divergent flow is generated. Shown in Figs. 13a and 13d are the divergence responses at 200 hPa on day 5 to the ISM and WNPSM heatings, respectively. The forced divergent flow contributes to a major part of the RWS shown in Figs. 12a and 12d. The upper divergence is closely related to the tropospheric vertical motion, as can be seen from the ω fields at 350 hPa (Figs. 13b and 13e). The ISM-forced vertical motion near India and the midlatitude NH are in general consistent with those of Rodwell and Hoskins (1996), who simulated the response to the mean monsoon heating and attributed the Sahara desert to the monsoon-forced downward motion near North Africa. Their linear experiment with a resting basic state and a heating source at 25°N, 90°E indicates that the descending motion over the Middle East and North African region is part of the “pure” Rossby wave response. This descending motion and the associated upper convergence could thus explain the RWS generation in that region.
7. Conclusions and discussion
We have analyzed the global circulation anomalies associated with the Asian summer monsoon convective activity based on monthly data of 29 summers. Two distinct teleconnection patterns are identified and associated respectively with the convection anomalies of the two subsystems of the Asian summer monsoon (i.e., ISM and WNPSM). The ISM has a significant connection to atmospheric circulation in both hemispheres. A stronger than normal ISM convection is linked to positive subtropical geopotential height anomalies and large anticyclonic gyres on both sides of the equator in the upper troposphere extending westward from Indian Ocean to the American continents, suggesting that the ISM has a far-reaching zonal connection. In the NH midlatitudes, circumglobal patterns with wavenumbers 5–6 are observed along the upper westerlies. On the other hand, the circulation anomalies associated with the WNPSM are mainly in the SH. An anomalous WNPSM convection is connected to changes in circulation in the high-latitude South Pacific and South America through a Rossby wave train.
Numerical experiments using a primitive equation model with anomalous thermal forcings that resemble the ISM and WNPSM convective heatings are able to reproduce the observed teleconnections, suggesting that the observed circulation anomalies result from the forcings of the Asian summer monsoon. Linear experiments indicate that the global response to the summer monsoon heating anomaly in both the ISM and WNPSM regions is mainly controlled by linear processes. The equatorial symmetric subtropical response is closely related to the westward spread of the equatorially trapped Rossby wave gyres forced by the ISM heating. The most important difference between the responses to the ISM and to the WNPSM is the behavior of the NH branch of the equatorially trapped Rossby wave located to the west of the forcings, which is determined by the basic state zonal flow. In the ISM case, the NH Rossby wave response has a large amplitude in the Middle East and North African region, where the basic state zonal wind is favorable for its development. With the WNPSM forcing, however, the strong upper easterlies south of the South Asian high prevent the Rossby wave gyre from developing and propagating westward. Further analysis of the Rossby wave source reveals that the generation of the RWS in the NH subtropics is related to the descending motion of the equatorially trapped Rossby wave over the North African region. In the SH, upward motion and upper divergence that contribute to the RWS in the subtropics are associated with temperature advection by the SH westerly mean flow.
This study is based on monthly mean data. Similar analyses using NH summer seasonal mean data arrive at similar results. The teleconnection patterns largely represent the stationary responses to the ISM and WNPSM forcing anomalies, which are not sensitive to the time period used as the average so long as it is long enough for the extratropical response to be established. Because the direct global Rossby wave response to tropical heating is established in about 15 days (e.g., Jin and Hoskins 1995; Lin et al. 2007), similar features of circulation anomalies are expected for a time scale longer than 15 days.
Similar teleconnection patterns have been reported in previous studies. The NH midlatitude part of the ISM response is similar to the circumglobal pattern observed in previous studies (e.g., Branstator 2002; Ding and Wang 2005, 2007). The existence of the circumglobal pattern is closely related to the NH midlatitude jet stream waveguide as discussed in Branstator (2002). Ding and Wang (2005) found that the summertime circumglobal pattern has a significant correlation with the ISM, suggesting that the anomalous ISM may excite the circumglobal pattern. Our result indicates that in fact the ISM variability is a forcing mechanism for the circumglobal pattern. Several teleconnection patterns found in the SH [e.g., the South Pacific wave train (Kidson 1999) and the Pacific–South American pattern (Mo 2000; Cai and Watterson 2002; Renwick 2002)] bear similarities to the SH wave train related to the WNPSM variability. It can be concluded that the ISM and WNPSM variabilities excite teleconnection patterns that significantly influence the global extratropical circulation. It is, however, also possible that the extratropical teleconnection patterns observed in this study are intrinsic to the atmosphere rather than being unique forced patterns. In other words, the ISM and WNPSM forcings may not create new circulation patterns; instead, they act to modify the probability distributions of the atmospheric internal modes as described in Palmer (1999). This hypothesis may be tested by analyzing the atmospheric internal variability component of a large ensemble of GCM integrations by removing the effects of interannually varying boundary conditions. Note that for the NH winter case, by looking at a 22-member ensemble simulated by the Community Climate Model 3 (CCM3) of NCAR, Branstator (2002) found a similar circumglobal pattern for the internal variability. Another reason to speculate that the NH circumglobal pattern and the SH wave train are atmospheric internal modes is that the ISM and WNPSM variabilities themselves may be caused by atmospheric internal processes.
Both the ISM and the WNPSM variabilities are influenced by the El Niño evolution (e.g., Wang et al. 2001). Indeed, there are weak correlations with the Niño-3.4 index (5°N–5°S, 120°–170°W), which is the sea surface temperature anomaly over the east central tropical Pacific and is calculated by the U.S. Climate Prediction Center (http://www.cdc.noaa.gov/Correlation/nina34.data), and with PI1 (−0.20) and PI3 (−0.19). No correlation with PI2 (0.02) is found, however. It is interesting to know the circulation patterns associated with the Asian summer monsoon that are independent of El Niño. The correlation and SVD calculations as section 4 in are repeated, except that this time the El Niño signal is removed from all the variables involved by subtracting the part that is linearly associated with the Niño-3.4 index. It turns out that the correlation results remain almost unchanged (not shown). The two leading SVDs have distributions almost identical to that found without removing the El Niño, except that the order of these two SVDs is switched (not shown). This indicates that the observed connections between the global circulation and the summer monsoon convective activity are not influenced by the El Niño signal.
The global connection of the circulation anomalies to the Asian summer monsoon provides a useful mechanism for the development of extended and long-range forecasting capability. The teleconnection pattern forced by the ISM convection indicates that the ISM variability has a significant impact on the subtropical and midlatitude regions in both hemispheres. The WNPSM variability sends a signal to regions in the SH as far as South America. It is therefore important to improve the representation of convection in general circulation models for the Asian summer monsoon region.
Acknowledgments
The author would like to thank Dr. Ayrton Zadra for helpful comments on an early version of the manuscript. The author would also like to thank three anonymous reviewers whose comments and suggestions helped to improve the paper. This research was supported in part by the Canadian Foundation for Climate and Atmospheric Sciences and by the Natural Science and Engineering Research Council of Canada.
REFERENCES
Branstator, G., 2002: Circumglobal teleconnections, the jet stream waveguide, and the North Atlantic Oscillation. J. Climate, 15 , 1893–1910.
Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5 , 541–560.
Cai, W., and I. G. Watterson, 2002: Modes of interannual variability of the Southern Hemisphere circulation simulated by the CSIRO climate model. J. Climate, 15 , 1159–1174.
Ding, Q., and B. Wang, 2005: Circumglobal teleconnection in the Northern Hemisphere summer. J. Climate, 18 , 3483–3505.
Ding, Q., and B. Wang, 2007: Intraseasonal teleconnection between the summer Eurasian wave train and the Indian monsoon. J. Climate, 20 , 3751–3767.
Ding, Y., 2007: The variability of the Asian summer monsoon. J. Meteor. Soc. Japan, 85B , 21–54.
Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106 , 447–462.
Hall, N. M. J., 2000: A simple GCM based on dry dynamics and constant forcing. J. Atmos. Sci., 57 , 1557–1572.
Hall, N. M. J., and J. Derome, 2000: Transients, nonlinearity, and eddy feedback in the remote response to El Niño. J. Atmos. Sci., 57 , 3992–4007.
Hoskins, B. J., and D. J. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38 , 1179–1196.
Hoskins, B. J., and M. J. Rodwell, 1995: A model of the Asian summer monsoon. Part I: The global scale. J. Atmos. Sci., 52 , 1329–1340.
Hoskins, B. J., and G. Yang, 2000: The equatorial response to higher-latitude forcing. J. Atmos. Sci., 57 , 1197–1213.
Hsu, H-H., and S-H. Lin, 1992: Global teleconnections in the 250-mb streamfunction field during the Northern Hemisphere winter. Mon. Wea. Rev., 120 , 1169–1190.
Jin, F., and B. J. Hoskins, 1995: The direct response to tropical heating in a baroclinic atmosphere. J. Atmos. Sci., 52 , 307–319.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Kidson, J. W., 1999: Principal modes of Southern Hemisphere low-frequency variability obtained from NCEP–NCAR reanalyses. J. Climate, 12 , 2808–2830.
Kosaka, Y., and H. Nakamura, 2006: Structure and dynamics of the summertime Pacific–Japan teleconnection pattern. Quart. J. Roy. Meteor. Soc., 132 , 2009–2030.
Kripalani, R. H., A. Kulkarni, and S. V. Singh, 1997: Association of the Indian summer monsoon with the Northern Hemisphere mid-latitude circulation. Int. J. Climatol., 17 , 1055–1067.
Lau, K-M., K-M. Kim, and J-Y. Lee, 2004: Interannual variability, global teleconnection, and potential predictability associated with the Asian summer monsoon. East Asian Monsoon, C. P. Chang, Ed., World Scientific, 153–176.
Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77 , 1275–1277.
Lim, H., and C-P. Chang, 1983: Dynamics of teleconnections and Walker circulations forced by equatorial heating. J. Atmos. Sci., 40 , 1897–1915.
Lin, H., and J. Derome, 1996: Changes in predictability associated with the PNA pattern. Tellus, 48A , 553–571.
Lin, H., J. Derome, and G. Brunet, 2007: The nonlinear transient atmospheric response to tropical forcing. J. Climate, 20 , 5642–5665.
Marshall, J., and F. Molteni, 1993: Toward a dynamical understanding of planetary-scale flow regimes. J. Atmos. Sci., 50 , 1792–1818.
Mo, K. C., 2000: Relationships between low-frequency variability in the Southern Hemisphere and sea surface temperature anomalies. J. Climate, 13 , 3599–3610.
Nitta, T., 1987: Convective activities in the tropical western Pacific and their impacts on the Northern Hemisphere summer circulation. J. Meteor. Soc. Japan, 65 , 373–390.
Nitta, T., 1989: Global features of the Pacific–Japan oscillation. Meteor. Atmos. Phys., 41 , 5–12.
Palmer, T. N., 1999: A nonlinear dynamical perspective on climate prediction. J. Climate, 12 , 575–591.
Renwick, J. A., 2002: Southern Hemisphere circulation and relations with sea ice and sea surface temperature. J. Climate, 15 , 3058–3068.
Roads, J. O., 1987: Predictability in the extended range. J. Atmos. Sci., 44 , 3495–3527.
Rodwell, M. J., and B. J. Hoskins, 1995: A model of the Asian summer monsoon. Part II: Cross-equatorial flow and PV behavior. J. Atmos. Sci., 52 , 1341–1356.
Rodwell, M. J., and B. J. Hoskins, 1996: Monsoons and the dynamics of deserts. Quart. J. Roy. Meteor. Soc., 122 , 1385–1404.
Sardeshmukh, P. D., and B. J. Hoskins, 1984: Spatial smoothing on the sphere. Mon. Wea. Rev., 112 , 2524–2529.
Sardeshmukh, P. D., and B. J. Hoskins, 1988: The generation of global rotational flow by steady idealized tropical divergence. J. Atmos. Sci., 45 , 1228–1251.
Wang, B., and Z. Fan, 1999: Choice of South Asian summer monsoon indices. Bull. Amer. Meteor. Soc., 80 , 629–638.
Wang, B., R. Wu, and K-M. Lau, 2001: Interannual variability of the Asian summer monsoon: Contrasts between the Indian and the western North Pacific–East Asian monsoons. J. Climate, 14 , 4073–4090.
Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc., 78 , 2539–2558.
Zhang, C., 1993: Laterally forced equatorial perturbations in a linear model. Part II: Mobile forcing. J. Atmos. Sci., 50 , 807–821.
(a),(c) Time average of precipitation rate and OLR; (b),(d) standard deviation of month-to-month variability of precipitation rate and OLR for 116 JJAS months from 1979 to 2007. Contour intervals are 3 mm day−1 for (a), 1 mm day−1 for (b), 20 W m−2 for (c) and 3 W m−2 for (d). Areas shaded are those with (a) precipitation rates greater than 9 mm day−1, (b) precipitation rate standard deviation larger than 3 mm day−1, (c) average OLR < 220 W m−2, and (d) OLR standard deviation >12 W m−2. The areas outlined with solid rectangles in (b) and (d) represent the regions used to define the Indian and western Pacific precipitation indices (PI1 and PI3); the area outlined with a dashed rectangle is that used to define the Bay of Bengal precipitation index (PI2).
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
Correlations between PI1 and (a) 200-hPa geopotential height, (b) 200-hPa streamfunction, and (c) 850-hPa streamfunction. Contour interval is 0.1. Contours with negative values are dashed. Shaded areas represent those where the correlation passes a 0.05 significance level according to a Student’s t test.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
As in Fig. 2, but for PI3.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
As in Fig. 2, but for PI2.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
(a),(b) 200-hPa geopotential height and (c),(d) OLR distributions of SVD1 and SVD2, as calculated as heterogeneous regressions. The magnitude corresponds to one standard deviation of each time coefficient. The contour intervals are 8 m for (a) and (b), and 2 W m−2 for (c) and (d). Contours with negative values are dashed. Zero contours are omitted in (a) and (b).
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
Time-averaged streamlines of JJAS for (a) NCEP 200 hPa, (b) NCEP 850 hPa, (c) SGCM control run 200 hPa, and (d) SGCM control run 850 hPa. Areas in orange are those with a time mean zonal wind U > 15 m s−1 in (a) and (c), and U > 5 m s−1 in (b) and (d). Areas in blue indicate where U < −10 m s−1 in (a) and (c), and U < −5 m s−1 in (b) and (d).
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
(a) Vertically averaged anomalous heating rate for the ISM run, (b) 200-hPa geopotential height response, and (c) 850-hPa streamfunction response. The contour intervals are 0.5°C day−1 for (a), 10 m for (b), and 0.5 × 106 m2 s−1 for (c). Contours with negative values are dashed. The shaded areas represent those with a positive response over 10 m for (b) and 0.5 × 106 m2 s−1 for (c). The zero contour for (a) is omitted.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
As in Fig. 7, but for the WNPSM run.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
Linear 200-hPa geopotential height response at day 15 to (a) ISM and (b) WNPSM heatings. The contour interval is 10 m; contours with negative values are dashed. The shaded areas represent those with a positive response over 20 m.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
Linear 200-hPa geopotential height response to (a)–(e) ISM and (f)–(j) WNPSM heating from days 1 to 5. The contour interval is 5 m; contours with negative values are dashed. The zero line is not plotted.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
The amplitude factor A of response to a stationary forcing for equatorial Rossby waves of a zonal wavenumber 6 as a function of the basic state zonal wind U. Curves indicate Rossby wave mode n = 1 (solid) and n = 2 (dashed).
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
Day-5 Rossby wave sources at 200 hPa for the (left) ISM and (right) WNPSM heatings; (top) term 1 of (3), (middle) term 2, (bottom) total RWS. The exponential spectral filter as introduced in Sardeshmukh and Hoskins (1984) has been applied. The contour interval is 1.5 × 10−11 s−2. Zero contours are suppressed; negative contours are dashed.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
(top) Day-5 divergence at 200 hPa, (middle) ω at 350 hPa, and (bottom) temperature at 350 hPa for the (left) ISM and (right) WNPSM heatings. The same exponential spectral filter has been applied to the divergence fields. Contour intervals are 1.5 × 10−7 s−1 for divergence, 0.2 hPa s−1 for ω, and 0.2°C for temperature. Zero contours are omitted; negative contours are dashed.
Citation: Journal of the Atmospheric Sciences 66, 9; 10.1175/2009JAS3008.1
Temporal correlations between pairs of PI1, PI2, PI3, CI1, CI2, and CI3. Numbers in bold represent those that pass a 0.01 significance level according to a Student’s t test.
Squared covariance explained by the first four SVDs and correlation between the expansion coefficients of the OLR and Z200 components.
