A global survey is conducted for atmospheric anomaly patterns of meridional teleconnection over the summer hemisphere associated with anomalous tropical convection. The patterns may be akin to the Pacific–Japan (PJ) teleconnection pattern analyzed in detail in the companion paper. From the survey, meridional teleconnections are identified over five regions, namely, the western North Pacific and Central/North America in boreal summer, as well as the western South Indian Ocean, central South Pacific, and western South Atlantic in austral summer. All of the patterns are observed in the western peripheries of the summertime surface subtropical anticyclones over the individual ocean basins. Although all of the patterns can convert available potential energy (APE) efficiently from the vertically sheared subtropical westerly jets, the efficiencies of barotropic energy conversion from the mean flow and diabatic APE generation differ from one pattern to another. Still, all of the patterns gain energy as the net, to maintain themselves against dissipative processes. Both the anomalous moisture convergence near the surface and the midtropospheric anomalous ascent required for the vorticity and thermal balance act to sustain the anomalous tropical convection, while the wind-evaporation feedback contributes positively only to the PJ pattern over the western North Pacific. Examination of common features and discrepancies among the five teleconnection patterns with respect to their structures and energetics reveals that climatological background features, including the largest horizontal extent of the Asian monsoon system and the North Pacific subtropical anticyclone, in addition to particularly high SST over the Pacific warm pool, render the PJ pattern an outstanding mode of variability.
Tropical–extratropical teleconnections have gained increasing attention since the early 1980s, when Hoskins and Karoly (1981) presented a theoretical framework for interpreting the Pacific–North American pattern (Horel and Wallace 1981) as a manifestation of a quasi-stationary equivalent-barotropic Rossby wave train excited by anomalous convective heating in the tropics. Many studies on the tropical–extratropical teleconnections focused on the winter season. This is because the upper-tropospheric time-mean westerlies are stronger and extend into lower latitudes than in summer, which is favorable for the poleward propagation of stationary Rossby waves forced in the tropics. Nevertheless, tropical–extratropical teleconnections are observed also in summer, including the Pacific–Japan (PJ) pattern associated with anomalous convection around the northern Philippines (Nitta 1987).
In our companion paper (Kosaka and Nakamura 2010, hereafter Part I), characteristics of the PJ pattern as a moist dynamical mode have been postulated. Kosaka and Nakamura (2006, 2008) and Part I have indicated that Rossby wave–like poleward and equatorward energy dispersion occurs in the lower and upper troposphere, respectively, in association with the PJ pattern, which is embedded in a zonally varying climatological-mean flow with a vertically sheared meridional flow. The anomalies associated with the PJ pattern can effectively gain kinetic energy (KE) and available potential energy (APE) through barotropic and baroclinic energy conversions, respectively, from the climatological-mean state, in addition to effective APE generation caused by anomalous diabatic heating in the tropical western North Pacific (WNP). The anomalous circulation, in turn, tends to reinforce the anomalous convective activity over climatologically high SST. Furthermore, Part I has identified a “brother” of the PJ pattern, which is associated with subtropical anomalous convection around the Bonin Islands south of Japan and benefits from efficient energy conversions from the climatological-mean flow.
Results from Part I indicate that the PJ pattern owes its existence to the configuration of the climatological-mean field, which is characterized by the warm Asian monsoon system to its west, the cool North Pacific subtropical anticyclone to its east, the vertically sheared westerly jet (the Asian jet) in midlatitudes, and climatologically active cumulus convection over high SST. Offering a plausible interpretation why the PJ pattern is observed as the dominant mode of variability in the particular maritime region in summer, Part I suggests the possibility that PJ-like anomaly patterns may be observed in other regions where the climatological-mean fields are similar to those over the summertime WNP. In the present study, an attempt is made to seek “relatives” of the PJ pattern over the entire Northern Hemisphere (NH) in boreal summer as well as over the entire Southern Hemisphere (SH) in austral summer.
Section 2 gives a description of the data used in the present study. Global distribution of meridional teleconnections is investigated in section 3 based on a composite analysis. Structure and energetics of each of the teleconnection patterns are examined in section 4. Discussion and concluding remarks are given in section 5.
2. Data and diagnosis methods
The data used in this study are the same as in Part I, including monthly-mean data of the Japanese 25-yr reanalysis (JRA-25; Onogi et al. 2007), the National Oceanic and Atmospheric Administration (NOAA) interpolated outgoing longwave radiation (OLR), and the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) for a 29-yr period from 1979 to 2007. The NOAA optimum interpolation SST data, available only from December 1981, are also used. As in Part I, horizontal smoothing has been applied only to the vorticity fields.
As in Part I, composite analyses are conducted for extreme monthly events of enhanced convection observed at every grid point (with 2.5° lat–lon intervals) in the tropics. The composited anomaly fields are used for evaluating barotropic and baroclinic energy conversions (CK and CP, respectively) and diabatic APE generation (CQ) as well, in the same manner as in Part I. Efficiencies of the energy conversions and generation are evaluated as time scales on which the area-integrated energy could be replenished by a particular conversion/generalization integrated over a given domain: τCK = [KE]/[CK], τCP = [APE]/[CP], τdry = [KE + APE]/[CK + CP], τmoist = [KE + APE]/[CQ], and τtotal = [KE + APE]/[CK + CP + CQ], with square brackets indicating spatial integrals. In addition, dynamically induced anomalous ascent caused by anomalous vorticity and heat transport is diagnosed at the 400-hPa level. As in Part I, the diagnosis based on a linearized omega equation around the zonally asymmetric climatological-mean flow is applied to the composited anomalies. Likewise, a wave-activity flux, which is formulated by Takaya and Nakamura (2001) for stationary Rossby waves embedded in the zonally asymmetric climatological-mean flow, is also evaluated based on the composited anomalies.
3. Global distribution of the PJ-like anomaly patterns
In the climatological lower-tropospheric circulation (Fig. 1b) for NH summer (June–July–August, hereafter JJA), a pair of basin-scale semipermanent anticyclones is observed, namely, the North Pacific subtropical anticyclone and the North Atlantic subtropical anticyclone (i.e., the Azores high). In the upper troposphere (Fig. 1a), the Asian monsoon accompanies the predominant Tibetan high. Though much weaker, the Mexican monsoon accompanies a well-defined anticyclone aloft (Fig. 1a). The associated upper-level northerlies are collocated with the low-level southerlies associated with the Azores high, suggesting that a PJ-like anomaly pattern may develop associated with anomalous convection over high SST in the Caribbean Sea or tropical eastern North Pacific (Fig. 1d). The trade winds associated with the Azores high transport moist air into the Caribbean Sea (Fig. 1c), where the climatological condition thus appears to be favorable for strong convective activity. However, the climatological moisture convergence is most conspicuous around the Yucatan Peninsula and its south (Fig. 1c), suggesting that the maritime regions off the Pacific and Caribbean coasts of Central America may be more favorable for anomalous convection to force a meridional teleconnection.
In the SH summer (December–January–February, hereafter DJF), three basin-scale anticyclones are evident in the lower troposphere (Fig. 2b), namely the anticyclone in the Indian Ocean (the Mascarene high), the South Pacific subtropical anticyclone, and the South Atlantic subtropical anticyclone. In the upper troposphere (Fig. 2a), three planetary-scale anticyclones are found in correspondence with the South African, Australian, and South American (Amazon) monsoons. Upper-level southerlies associated with the upper-level anticyclones over the Amazon and South Africa overlap the low-level northerlies associated with the surface subtropical anticyclones over the South Atlantic and Indian Ocean, respectively, although the upper-tropospheric southerlies and the associated vertical shear of the mean meridional wind are weaker in the latter region. The upper-tropospheric anticyclone associated with the Australian monsoon is not well separated from the South African counterpart, and the former is so distant from the surface South Pacific subtropical anticyclone that the overlapping of the low-level northerlies and upper-level southerlies is not apparent. Rather, convective activity along the South Pacific Convergence Zone (SPCZ) may play a role equivalent to a monsoon system, accompanying poleward and equatorward mean flow in the lower and upper troposphere, respectively, to the east of the date line. The tropical regions over the central South Pacific (CSP) and western South Indian Ocean (WIO) are characterized by high SST, abundant precipitable water, and pronounced moisture flux convergence (Figs. 2c–d). In contrast, SST is somewhat lower over the tropical western South Atlantic (WSA) and the climatological moisture convergence is found primarily over the Amazon and to its southeast. Nevertheless, the characteristics of the climatological-mean summer fields over the three tropical regions (i.e., the WIO, CSP, and WSA/Amazon sectors) share some of the features observed in the summertime WNP, setting conditions favorable for PJ-like teleconnection patterns.
For the four regions (one over the summertime North Atlantic and three in the summertime SH) other than the WNP, lower-tropospheric cyclones to their west are not as conspicuous as the Asian monsoon low (Figs. 1b, 2b). Around those regions, lower-tropospheric westerlies are absent, which can be regarded as a counterpart of the monsoon westerlies associated with the Asian summer monsoon. Nevertheless, the presence of “monsoon westerlies” may not be a necessary condition energetically for the presence of a PJ-like pattern with characteristics of a dry dynamical mode, because it is not the sign of u but its zonal gradient ∂u/∂x that is crucial for CK and the zonal gradient is negative in exits of either the trade winds or monsoon westerlies. From a viewpoint of a moist dynamical mode, however, the absence of monsoon westerlies means that anomalous low-level westerlies, if induced by anomalous tropical convection, could not enhance surface evaporation locally, which otherwise could reinforce the anomalous convection. This deficiency in moist feedback may act against the dominance of a PJ-like pattern in those regions. For the Central/North American (CNA), South African, and South American monsoons, the continental monsoonal lows and associated upper-tropospheric anticyclones are narrower in longitude than for the Asian monsoon, which may also be unfavorable for teleconnection with Rossby waves.
It should be noted that, compared to the NH summer, the climatological-mean westerlies extend farther equatorward in the upper troposphere than in the lower troposphere at almost every longitude over the summertime SH (Miyasaka and Nakamura 2010). Both the climatological planetary waves and quasi-stationary Rossby waves associated with anomalous convective activity in the subtropics can propagate farther equatorward over the summertime SH than over the summertime NH. The associated vertical shear along the western flanks of the three surface subtropical anticyclones over the SH is favorable for poleward-tilted vorticity perturbations.
b. Meridional teleconnections over the globe
As in Part I, analysis in the present paper is based on composite maps constructed locally by using the OLR index (Iϕ,λ) defined almost in the same manner as in Part I. Specifically, the index Iϕ,λ is defined for each month as the strongest monthly OLR anomaly, either positive or negative, identified within a given 10° × 10° domain centered at (latitude ϕ, longitude λ). For the present analysis, the index has been obtained for each of the 1008 domains that altogether cover the entire tropics and subtropics, namely, ϕ = (5°, 10°, 15°, 20°, 25°, 30°, and 35°N) for boreal summer and (5°, 10°, 15°, 20°, 25°, 30°, and 35°S) for austral summer, with λ = (0°, 5°E, 10°E, … , 10°W, and 5°W) for each of the hemispheres. As in Part I, composite maps have been constructed locally based on the index. In this study, however, we utilize composite difference maps between “positive” and “negative” events based on Iϕ,λ (i.e., positive minus negative). The positive (negative) events are those months when the negative (positive) OLR index values at a given location exceed half the standard deviation in magnitude for a particular calendar month. The composite difference maps are adopted because the convection-related teleconnection is not necessarily a dominant signal in the interannual summertime variability except over WNP.
The results of our composite analysis are summarized in Figs. 3, 4 for boreal and austral summers, respectively. Figures 3a, 4a show geographical distributions of local meridional difference in the composited lower-tropospheric vorticity anomalies associated with locally enhanced convection, as a measure of a PJ-like vorticity dipole. In Fig. 3a, meridional vorticity dipoles are strongest in association with locally enhanced convection over the tropical WNP east of the Philippines. This is the region of highly efficient energy conversions plus diabatic generation (Table 3c of Part I), including the region for the PJ pattern (Kosaka and Nakamura 2006, 2008). In addition to WNP, PJ-like vorticity dipoles are observed in association with anomalous convection over a tropical region from the eastern North Pacific into Mexico (Fig. 3a). Likewise, vorticity dipoles are found in austral summer over WIO, CSP, and WSA/Amazon (Fig. 4a). In these regions, the associated wave-activity flux based on the composited anomalies is predominantly poleward and equatorward in the lower and upper troposphere (Figs. 3c, 4c, 3b, 4b), respectively, as in the composite for the PJ pattern shown in Part I. As discussed in the preceding subsection, these regions are characterized by the equatorward shear in the climatological-mean flow, which favors the opposing direction of the wave-activity flux between the upper and lower troposphere. The vorticity dipoles are also characterized by climatological-mean moisture convergence and high underlying SST, except over the WSA/Amazon region.
In Figs. 3a, 4a, signatures of meridional vorticity dipoles are also indicated in subtropical regions (30°–35° latitude) over the North Pacific and North Atlantic in boreal summer and over the Indian Ocean and eastern South Pacific in austral summer. Over these subtropical domains, anomalous convective activity is generally modest relative to the corresponding anomalous activity in the tropics and the associated wave-activity flux is eastward in the upper troposphere. Over the subtropical western North Atlantic at ∼20°N, the wave-activity flux is poleward and equatorward in the lower and upper troposphere, respectively (Figs. 3b,c), although the signature of lower-tropospheric vorticity dipoles is modest (Fig. 3a). The dipole signature is also detected over the tropical Indian Ocean (Fig. 3a) and to the north of Australia (Fig. 4a), but the associated wave-activity flux tends to be weak or even equatorward in the lower troposphere (Figs. 3c, 4c).
4. “Relatives” of the PJ pattern
a. Central and North America
Composited anomalies based on anomalous convective activity over (10°–20°N, 100°–90°W) (I15°N,95°W) for JJA are shown in Fig. 5 as an example of the meridional teleconnection around CNA, including the region over the eastern North Pacific/Mexican sector through North America. In association with enhanced precipitation off the west coast of Mexico and over the Caribbean Sea (Fig. 5a), a zonally elongated cyclonic anomaly extending from the Yucatan Peninsula to ∼140°W and an anticyclonic anomaly to its northeast are evident in the lower troposphere, accompanying poleward wave-activity flux (Fig. 5b). Another significant anticyclonic anomaly is found to the south of the primary cyclonic anomaly in a zonally elongated structure (Fig. 5b). In the upper troposphere (Fig. 5c), cyclonic and anticyclonic anomalies that correspond to the lower-tropospheric dipole are observed over the southern portion of the United States and over the Great Lakes, respectively. The strong anticyclonic anomalies are characterized by the prominent southeastward wave-activity flux. These upper-level vorticity anomalies are apparently shifted poleward relative to their low-level counterpart, as observed with the PJ pattern. A similar pattern can be found in the leading empirical orthogonal function defined for the spatially smoothed monthly anomalies of 850-hPa vorticity over (0°–60°N, 120°–60°W) (figure not shown). Interestingly, weak but significant reduction in precipitation is observed off the Carolina coast (Fig. 5a), which appears to be a counterpart of the anomalous precipitation observed around the baiu (mei-yu) front in association with the PJ pattern (cf., Figs. 2a and 14a of Part I).
In the upper troposphere (Fig. 5c), another anticyclonic anomaly is inconspicuous slightly to the south of the enhanced convection, whose counterpart for the PJ pattern may be found closer to the equator over the Maritime Continent (Figs. 2c and 14c of Part I). The anticyclonic anomaly is accompanied by strong northeastward wave-activity flux (Fig. 5c) along the climatological-mean southwesterlies over the western United States (Fig. 1a). The upper-tropospheric poleward wave-activity flux that can strengthen midlatitude anomalies is not observed with the PJ pattern, which is embedded in the climatological northerlies on the eastern flank of the Tibetan high. These are main structural differences from the PJ pattern.
In Fig. 5e, KE gain by the low-level zonally elongated vorticity anomalies is evident off the Pacific coast of Central America, which corresponds to the equatorward flank of the exit region of the trade winds associated with the Azores high. In the upper troposphere (Fig. 5f), KE gain and loss are found over and to the south of the midlatitude jet, respectively. Unlike for the PJ pattern (Figs. 3b and 14e of Part I), however, the KE gain and loss are associated primarily with the meridional shear of the westerly jet. In Fig. 5g, positive and negative CP are consistent with midtropospheric anomalous southerlies and northerlies, respectively, associated with the warm midlatitude anticyclonic anomaly situated in a baroclinic zone over North America (Fig. 5d). The positive contribution apparently dominates to yield the net APE gain (Table 1). These characteristics of energy conversion, including the dominant contribution from baroclinic conversion, are similar to those for the PJ pattern. It takes, however, more than two months to replenish the total energy of this anomaly pattern through the energy conversion only over CNA (Table 1), indicating less efficient energy conversion than for the PJ pattern. In our evaluation, the total KE + APE includes contributions from statistically insignificant anomalies away from the CNA/western Atlantic sector. If the domain for integrating KE and APE is limited to (0°–60°N, 150°W–0°), time scales for the energy conversion terms are generally reduced to about one-third or a quarter of those listed in Table 1. Still, lower-tropospheric CK is much less efficient than for the PJ pattern, whereas CK is slightly more efficient in the upper troposphere (Table 1). Strong APE generation associated with the enhanced convection is observed over the subtropical eastern North Pacific and the Caribbean Sea around 20°N (Fig. 5h). This strong generation is compensated by broadly distributed APE destruction associated with warm anomalies (Fig. 5d) collocated with anomalous diabatic cooling (perhaps radiative in addition to condensation) over North America, yielding no substantial net generation of APE (Table 1).
Figure 6b shows that the circulation anomalies weaken the surface wind to the southeast and far to the west of the enhanced convection in the absence of the climatological-mean low-level westerlies, while strengthening it to the west and east of Mexico in the presence of the mean trade winds. As shown in Fig. 6c, anomalous surface evaporation is consistent overall with the surface wind speed anomalies. Off the west coast of Mexico, however, where SST is significantly lower than the climatology (Fig. 6a), virtually no significant enhancement occurs in the surface evaporation (Fig. 6c) in spite of the significantly intensified surface wind (Fig. 6b). In the vicinity of the enhanced convection center, negative evaporation anomalies to its south thus dominate and render the wind-induced surface heat exchange (WISHE) less effective than for the PJ pattern. Nevertheless, anomalous intensification in surface wind and associated influx of moisture reinforces anomalous convection around the primary cyclonic anomaly (Fig. 6d), as observed in the PJ pattern (Kosaka and Nakamura 2006; Part I). In addition, the omega equation linearized about the climatological-mean flow diagnoses an anomalous ascent around the enhanced convection center, induced primarily by upper-tropospheric vorticity advection (Fig. 6e). These features, except the ineffective WISHE, suggest that the anomalous circulation acts to reinforce the associated convective activity.
The overall features of the aforementioned anomaly pattern and the climatological-mean field over the CNA region are summarized in Table 2. The pattern shares qualitatively the same features as observed for the PJ pattern, including its structure and energetics, while differences are evident especially over the subtropical–midlatitude eastern Pacific. In that sense, the pattern over CNA is a “relative” of the PJ pattern, but not its “brother.”
b. Southern Hemisphere
Composited anomalies based on anomalous convective activity in DJF over (25°–15°S, 50°–60°E) (I20°S,55°E; WIO), (20°–10°S, 165°–155°W) (I15°S,160°W; CSP), and (20°–10°S, 35°–25°W) (I15°S,30°W; WSA) are shown in Figs. 7, 8, 9, respectively. In each of the three composites for the lower troposphere (Figs. 7b, 8b, 9b), an anomalous vorticity dipole is observed that consists of a cyclonic anomaly around the center of enhanced precipitation and a zonally elongated anticyclonic anomaly located poleward, as consistent with Fig. 4a. Though less organized than associated with the PJ pattern, poleward wave-activity flux is also observed in the lower troposphere around the node of each of the vorticity dipoles over the SH, which is consistent with Fig. 4c. In the upper troposphere, subtropical cyclonic anomalies between the midlatitude and tropical anticyclonic anomalies are common among the three composites (Figs. 7c, 8c, 9c). The associated upper-level wave-activity flux is predominantly equatorward in a manner consistent with Fig. 4, though less conspicuous for the WSA pattern (Fig. 9c). In each of these anomaly patterns, the primary cyclonic anomaly in the vicinity of the anomalous convection in the tropics exhibits a poleward phase tilt with height, but the tilt is less apparent for the primary anticyclonic anomaly in midlatitudes, especially for the pattern over WIO (Fig. 7). This particular vertical structure of the anomalies may be attributable to the configuration of the climatological-mean flow over the summertime SH. As pointed out by Miyasaka and Nakamura (2010), the climatological midlatitude westerlies extend farther equatorward in the upper troposphere (Fig. 2a) than in the lower troposphere (Fig. 2b). Furthermore, the climatological southerly shear is confined to the domain equatorward of 25°S (Figs. 2a,b), and the upper-tropospheric southerlies are weak over WIO (Fig. 2a). The equatorward extension of the midlatitude westerlies is consistent with apparent downstream wave propagations observed in the subtropics for the anomaly patterns over WIO and WSA (Figs. 7c, 9c).
In the lower troposphere, positive CK associated with zonally elongated anomalies embedded in the exits of the trade winds is evident for the three SH anomaly patterns, as indicated by the westward-pointing Eliassen–Palm (EP) flux (Figs. 7e, 8e, 9e). The anomaly patterns over WIO (Fig. 7e) and CSP (Fig. 8e) lose their KE through anomalous southeasterlies (u′υ′ < 0) associated with the midlatitude anticyclonic anomalies under the anticyclonic shear of the climatological westerlies (∂u/∂y < 0), resulting in the inefficient net KE gain or even the slight net loss of KE (Table 1). This structural relationship between the circulation anomalies and the mean midlatitude westerlies extends into the upper troposphere, to render the net negative CK in the upper troposphere (Figs. 7f, 8f; Table 1). In the tropics, strong upper-tropospheric KE gain occurs (Figs. 7f, 9f) through anomalies associated with the Rossby wave that appears to be forced by anomalous convection over WIO (Fig. 7c) or WSA (Fig. 9c) and propagated through the upper-tropospheric westerlies (Fig. 2a).
Positive and negative CP in the western and eastern flanks of the midlatitude anticyclonic anomalies, respectively, are common to all of the three anomaly patterns (Figs. 7g, 8g, 9g). The positive and negative conversions are associated with westward and eastward phase tilts, respectively, of the vorticity anomalies with height (Figs. 7b,c, 8b,c, 9b,c) along the vertically sheared mean westerlies. Midtropospheric westerly anomalies accompany warm anomalies around the enhanced convection center for CSP and WSA patterns (Figs. 8d, 9d), which, embedded in the climatological-mean westward temperature gradient, also contribute to the net positive CP (through u′T ′ > 0), as indicated by shading in Figs. 8g, 9g. For the NH patterns (i.e., the PJ and CNA patterns shown in Fig. 5), however, the eastward heat transport is manifested by anomalous easterlies with cool anomalies to the north of the convection centers. This discrepancy may be attributable to the equatorward expansion of the midlatitude jet and the associated equatorward retreatment of the mean southerly shear region (Figs. 2a,b).
The net baroclinic conversion CP for the three SH patterns can replenish APE within a month or so (Table 1), indicating its important contribution to their maintenance. Within the midlatitude baroclinic westerlies, wavy anomalies associated with anomalous convection over the WIO and CSP regions are found to induce effective CP. In fact, CP integrated solely to the south and north of 40°S (30°S) for the WIO (CSP) pattern yields comparable τCP values of 62.9 and 68.0 (41.4 and 49.1) days, respectively. Although positive and negative contributions considerably cancel out, the extratropical CP still contributes significantly to the energetics of these anomaly patterns.
Upper-tropospheric circulation anomalies associated with anomalous convective activity for the WSA pattern seem somewhat less coherent (Fig. 9c) than for the other SH patterns discussed earlier. In fact, the equatorward flux of Rossby wave activity is confined to the vicinity of the convection center, which may be attributable to the smaller zonal scale of the climatological-mean upper-tropospheric pressure trough and associated westerlies over the tropical South Atlantic than over the tropical South Pacific (Fig. 2a). Embedded in the anticyclonic shear of the mean westerlies over South America, upper-tropospheric vorticity anomalies with NE–SW tilt yield efficient CK for their maintenance (Table 1). In the lower troposphere (Figs. 9b,e), positive and negative values of CK are distributed incoherently over South America in association with relatively small-scale circulation anomalies, probably due to the orographic influence of the steep Andes (Fig. 9b). Unlike for its NH and SH counterparts, the net CK for the WSA pattern is much less efficient in the lower troposphere than in the upper troposphere (Table 1).
Diabatic energy generation (CQ) can also contribute substantially to the anomaly patterns over WIO and CSP (Figs. 7h, 8h; Table 1). For the CSP pattern, enhanced precipitation along the SPCZ accompanies warm anomalies (Figs. 8a,d), yielding the efficient APE generation (Fig. 8h). For the WIO pattern, the enhanced convective activity (Fig. 7a) is collocated with warm and cold anomalies over the western and central Indian Ocean, respectively (Fig. 7d). The latter yields diabatic energy loss, which lowers the efficiency of the net CQ (Table 1). For the WSA pattern (Fig. 9h), positive CQ is confined to the South American continent. Over the maritime region, a prominent upper-tropospheric cyclonic anomaly off Brazil collocated with the center of enhanced convection (Figs. 9a,c) accompanies a cool anomaly in the midtroposphere (Fig. 9d). The resultant negative CQ is strong enough to almost completely offset the positive CQ over the continent (Table 1). In this regard, the WSA pattern is essentially a dry dynamical mode, although the anomalous convection forces the anomalous circulation via anomalous vertical motion.
Anomalous latent heat flux composited for the WIO and CSP patterns is generally consistent with surface wind speed anomalies (Figs. 10a–c, 11a–c, respectively), indicating atmospheric thermodynamic forcing on SST. Anomalous evaporation over WSA (Fig. 12c), however, is weak overall and lacks statistical significance and thermodynamic consistency with the anomalies in wind speed (Fig. 12b) and SST (Fig. 12a). The anomaly patterns with enhanced convective activity over WIO (Fig. 10c) and CSP (Fig. 11c) accompany positive and negative anomalies in evaporation in comparable magnitudes, and anomalous evaporation is mostly negative around the enhanced convection center over WSA (Fig. 12c). For those patterns, the WISHE mechanism appears to be inefficient and thus inessential for their maintenance, which may be attributable, at least in part, to the lack of lower-tropospheric mean westerlies in these regions (Fig. 2b) as a counterpart of the Asian monsoon westerlies. For each of the three SH patterns, anomalous moisture convergence is nevertheless evident near the surface in association with the primary cyclonic anomaly around the center of enhanced convection (Figs. 10d, 11d, 12d).
Our diagnosis using the linearized omega equation reveals that anomalous ascent is required around the center of enhanced convection for each of the three SH patterns (Figs. 10e, 11e, 12e), for the vorticity and thermal balances. The results may be more or less affected by the equatorial boundary placed in solving the omega equation, which might cause poleward displacement of the diagnosed ascent from the enhanced convection center. This may be the case especially for the CSP pattern, whose primary convection center is near 10°S (Fig. 11e). Though possibly somewhat distorted, the anomalous ascent is nevertheless diagnosed in the vicinity of the convection center for each of the patterns even with the upper-tropospheric anomalous vorticity advection alone (Figs. 10e, 11e, 12e). The ascents can act to reinforce the anomalous tropical convection.
These results, except ineffective WISHE, indicate self-sustaining characteristics of the three SH anomaly patterns through interactions between their circulation and convection anomalies as dynamical modes. The WIO and CSP patterns (Figs. 7, 8) apparently show signatures of northeastward wave propagation through the upper-tropospheric midlatitude westerlies, which implies possible triggering of those patterns by Rossby wave trains from the upstream. This possibility is suggested also from our omega-equation diagnosis that anomalous ascent around the enhanced convection center can be induced by upper-tropospheric anomalous vorticity advection alone (Figs. 10e, 11e, 12e), which must be substantiated through further analysis on time evolution of daily anomaly events.
5. Discussion and concluding remarks
Motivated by implications from Part I, we have carried out a global survey of the presence of summertime meridional teleconnections that are structurally and dynamically similar to the PJ pattern observed over the WNP. Our survey has indeed identified PJ-like meridional teleconnections over western peripheries of the five lower-tropospheric subtropical anticyclones in summer, characterized by the lower-tropospheric vorticity dipoles and poleward and equatorward wave-activity flux in the lower and upper troposphere, respectively (Figs. 3, 4). This result confirms that each of the meridional teleconnections owes its existence to the configuration of the climatological-mean state characterized by a zonal temperature gradient between a region of climatologically active cumulus convection to the west and a cooler subtropical anticyclone to the east as well as a vertically sheared subtropical westerly jet. Furthermore, all of the five monthly teleconnection patterns over the globe convert APE efficiently (with time scales less than a month or so) from the climatological-mean westerlies in the subtropics (Table 1).
As summarized in Table 2, however, our investigation on the energetics of those teleconnection patterns reveals that their characteristics as dry dynamical modes vary somewhat depending on the efficiencies of their barotropic KE conversion. All of the patterns, other than the CSP pattern, gain KE as the net in the exits of the lower-tropospheric trade winds. For the WSA and CSP patterns, lower-tropospheric KE loss is enhanced through meridionally tilted anticyclonic anomalies embedded in the subtropical westerlies (Table 1). The efficiency of APE generation of a pattern through moist convection, and thereby its characteristics as a moist dynamical mode, also varies from one pattern to another. The efficiency for the CSP pattern is found as high as for the PJ pattern, while it is much lower for the other three patterns.
Our analysis has also revealed that these anomaly patterns tend to maintain anomalous convective activity in the tropics via anomalous moisture convergence and dynamically induced anomalous ascent, which can in turn maintain the anomalous circulation and convection. While WISHE also seems to be operative for the maintenance of anomalous convection associated with the PJ pattern, it is found ineffective for the other four patterns probably because of the lack of lower-tropospheric monsoon westerlies. From an energetic viewpoint, the meridional teleconnections over the CNA and WSA regions behave more like dry dynamical modes, while the CSP pattern exhibits a characteristic of a moist dynamical mode more apparently. The PJ pattern bears hybrid characteristics of dry and moist dynamical modes.
It has also been revealed that midlatitude vorticity anomalies involved in the PJ-like meridional teleconnections associated with anomalous convection over the SH do not exhibit distinct poleward tilt with height. This interhemispheric difference may possibly be related to the climatological upper-tropospheric westerlies over the summertime SH that tend to extend farther equatorward than over the summertime NH. In addition, unlike in the NH, synoptic-scale eddies are active over the midlatitude SH even in summer (Nakamura and Shimpo 2004), suggesting a possibility of feedback from transient eddies to the SH anomaly patterns.
The meridional teleconnection pattern over CNA (Fig. 5) accompanies a midlatitude anticyclonic anomaly centered just to the northeast of the tropical cyclonic anomaly in the lower troposphere (Fig. 5b). The meridional scale of the dipole thus looks somewhat smaller than that of the PJ pattern, consistent with the tendency that the western portion of the North Atlantic subtropical anticyclone (Azores high) is meridionally narrower than that of the North Pacific anticyclone (Fig. 1b). This difference is consistent with our sensitivity experiment with our simple model in Part I where the meridional scale of the basic flow is changed (Fig. 9 of Part I).
Despite the common fundamental features in structure discussed above, our detailed analysis on energetics of the four summertime meridional teleconnection patterns has revealed nonnegligible differences (Tables 1, 2), which appear to be related to differences in the climatological-mean state in the atmospheric circulation and underlying SST/land–sea distributions. Specifically, a comparison of the four patterns with the PJ pattern gives us some insight into the factors for its predominance, including the meridionally broad Asian monsoon low and Tibetan high that accompany distinct vertically sheared mean meridional wind and the strong upper-level Asian jet. The strong near-surface monsoon westerlies over high SST can also contribute to the predominance of the PJ pattern through the effective WISHE mechanism. The longitudinally broad Asian monsoon and North Pacific subtropical anticyclone favor the coherent structure of circulation anomalies of the PJ pattern and coherent distribution of wave-activity flux.
Embedded in the downstream of the upper-tropospheric subtropical jet cores, the meridional teleconnections can be influenced by wavy anomalies propagated from their upstream. Anomalous ascent diagnosed with the linearized omega equation solely with anomalous upper-tropospheric vorticity advection suggests a possibility for wavy anomalies to trigger anomalous convection that accompanies a meridional teleconnection pattern. The circulation anomalies associated with the WIO and CSP patterns are indeed suggestive of such triggering as above by midlatitude wave trains (Figs. 7c, 8c). A similar triggering is also suggested for the PJ pattern (Kosaka et al. 2009; Part I).
Relationships with tropical interannual variability including ENSO are examined for the four teleconnection patterns (figure not shown). For its particular phase shown in Fig. 5, the CNA pattern is accompanied by significant La Niña–like SST anomalies. In contrast, no significant ENSO signals can be detected for any of the anomaly patterns over the SH, despite the tendency for El Niño/La Niña events to peak in austral summer.
Through Part I and the present study, mechanisms for the monthly PJ pattern have been generalized, leading to our identification of a similar anomaly pattern of meridional teleconnection inherent in each of the subtropical ocean basins characterized by notable zonal asymmetries in the climatological-mean state and climatologically active cumulus convection. Nevertheless, differences in detailed features of the mean state in the atmospheric circulation and SST among the ocean basins appear to modify some of the common characteristics, leading to their own characteristics. In that sense, each of the four anomaly patterns investigated in the present study should be regarded as a relative of the PJ pattern—an analog but not a replica of it.
The authors appreciate valuable comments and suggestions from Dr. R. Lu and other two anonymous reviewers. The authors also wish to acknowledge comments from Profs. M. Kimoto, H. Niino, K. Sato, T. Yamagata (University of Tokyo), and S.-P. Xie (University of Hawaii). This study is supported in part by the Grant-in-Aid 18204044 and 22340135 by the Japanese Ministry of Education, Culture, Sports, Science and Technology, also by the Global Environment Research Fund (S-5) of the Ministry of the Environment, Japan. The JRA-25 reanalysis dataset used for this study is provided from the cooperative research project of the JRA-25 long-term reanalysis by the Japan Meteorological Agency (JMA) and the Central Research Institute of Electric Power Industry (CRIEPI). NOAA/OLR and optimum interpolation SST and CMAP precipitation datasets are provided by the NOAA/Cooperative Institute for Research in Environmental Sciences (CIRES) Climate Diagnostics Center, Boulder, Colorado, from Web site (available online at http://www.cdc.noaa.gov/).
Corresponding author address: Yu Kosaka, International Pacific Research Center, School of Ocean and Earth Science and Technology, University of Hawaii, 1680 East-West Road, Honolulu, HI 96822. Email: firstname.lastname@example.org