The Origin of the Subtropical Anticyclones

1JULY 001 CHEN ET AL. 187 The Origin of the Subtropical Anticyclones PING CHEN, MARTIN P. HOERLING, AND RANDALL M. DOLE NOAA CIRES Climate Diagnostics Center, Boulder, Colorado (Manuscript received 6 April 1999, in final form 18 November 000) ABSTRACT The origin of the Northern Hemisphere summer subtropical anticyclones is investigated using a linear quasigeostrophic model. It is found that the broad features in the model solutions forced by realistic heating fields acting on observed zonal flows agree well with those in the observations. The realistic features of the model solutions include the subtropical continental lows and oceanic highs in the lower troposphere, and continental ridges and midoceanic troughs in the upper troposphere. The forced responses are largest near the surface and the tropopause with a vertical node around 500 hpa as observed. The results indicate that the model stationary waves owe their existence largely to the Asian heat source. The authors thus propose that the observed low-level subtropical anticyclones over the North Pacific and North Atlantic Oceans be interpreted as a remote response of Rossby waves forced by the large-scale heat sources over Asia. The results support the existing theory that the observed low-level cyclones and upper-level anticyclones over Asia and North America are a local response to monsoonal latent heat release in the midtroposphere. The sensitivity of model solutions to the basic state is discussed, emphasizing the effects of the meridional and vertical shear of the zonal flow on the structure of the stationary waves. 1. Introduction The summer stationary waves in the Northern Hemisphere are dominated by the continental cyclones and oceanic anticyclones in the lower troposphere, and continental ridges and oceanic troughs in the upper troposphere (Fig. 1). The low-level oceanic anticyclones are commonly referred to as the subtropical anticyclones. As documented in previous studies (e.g., White 198), the stationary waves possess maximum amplitude both near the surface and the tropopause, with a vertical node around 500 hpa, and they exhibit a westward phase tilt in the vertical and a southwest northeast phase tilt in the horizontal. The origin of the summer circulations over Southeast Asia is well understood. Webster (197) demonstrated in a linear two-layer primitive equation model that the model circulations in response to a localized subtropical heating resemble those observed over Southeast Asia. He showed that strong cyclonic (anticyclonic) circulations exist in the lower (upper) layer in the vicinity of the heat source, and the phase of the circulation centers exhibits a southwest northeast tilt. Gill (1980) showed that the gross features of an analytic solution forced by a localized subtropical heating are also similar to those observed over Southeast Asia. Subsequent investiga- Corresponding author address: Dr. Ping Chen, NOAA CIRES Climate Diagnostics Center, Mail Code: R/CDC1, 35 Broadway, Boulder, CO 80303-338. E-mail: pc@cdc.noaa.gov tions using more complex models (e.g., Lin 1983) confirmed the important role of latent heating in forcing and maintaining the summer circulations over Southeast Asia. In contrast, the origin of the summer circulations over the oceans, and the low-level subtropical anticyclones in particular, continues to be an open question. A recent hypothesis holds that the subtropical anticyclones in the eastern ocean basins are directly related to the adacent monsoon heating to the east (Hoskins 1996; Rodwell and Hoskins 1996). This theory is based on idealized model results, which predict that a localized subtropical heating produces a region of descent northwest of the heating. This hypothesis implies that the North Atlantic anticyclone would be forced by heating over North Africa, whereas the North Pacific anticyclone would be a forced response to heating over North America. The purpose of the current paper is to describe the observed summer stationary waves and to provide a dynamical explanation for their origin. We will suggest that the zonally asymmetric circulations are manifestations of stationary Rossby waves forced by the zonally asymmetric diabatic heating. In particular, we will propose that the downstream (eastward) energy propagation in the midlatitude westerlies from the heat source regions over East Asia and the western Pacific is the ultimate mechanism for the existence and maintenance of subtropical anticyclones and their associated upper-level troughs over both the eastern North Pacific and eastern North Atlantic. We will show that the background flow 001 American Meteorological Society

188 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 58 FIG. 1. Observed climatological stationary waves for Jul at (a) 1000 hpa, (b) 150 hpa, and (c) 35 N, based on 40 yr (1960 99) of NCEP NCAR reanalysis data. The contour interval is 0 m. has a strong impact on the structure of these thermally forced circulations. The paper is organized as follows: section describes the data and model, section 3 examines the relationship between the observed circulation and heating, section 4 presents solutions of the linear quasigeostrophic (QG) system in response to a realistic heating, and section 5 contains the conclusions and a discussion of the results.. Data and model We analyze the geopotential height from the National Centers for Environmental Prediction National Center for Atmospheric Research (NCEP NCAR) reanalysis (Kalnay et al. 1996) and the outgoing longwave radiation (OLR) measured by the National Oceanic and Atmospheric Administration Advanced Very High Resolution Radiometer (NOAA AVHRR) instruments (Gruber and Krueger 1984; Liebmann and Smith 1996). The data of geopotential height are from 1960 to 1999 and those of OLR are from 1974 to 1999 (excluding 1978). The OLR is a proxy for atmospheric latent heating, and, in general, regions of intense deep convection associated with the summer monsoons exhibit low OLR values, while cloud-free regions of the subtropics exhibit high values. The model is the QG beta-plane model described in the appendix. Specifically, the linear system is solved numerically using harmonic expansion in longitude and centered finite-difference schemes in latitude and height. The beta plane has a width of 90 latitude and is centered at 35 N. The upper boundary is at 50 km. The model has a resolution of 71 zonal harmonics and 0.5 latitude and 0.5-km height. A Rayleigh friction timescale of 5 days and a Newtonian cooling timescale of 15 days are used. The maor conclusions of this paper are not affected by the form and rate of the damping. A more detailed discussion about the effects of dampings can be found in Chen (001). 3. Observational relationships between the circulation and heating The observed summer circulation in the Northern Hemisphere is strongly asymmetric in longitude. Figure 1 shows the horizontal and vertical distributions for the zonally asymmetric component of the time-mean geopotential height field during July. It is seen that the stationary waves have a fundamental baroclinic structure, suggesting the effect of thermal forcing. An additional distinctive feature is the steep southwest northeast phase tilt in the horizontal, particularly in the lower troposphere. Figure shows the observed time-mean OLR for July. The lowest values occur over Southeast Asia and the intertropical convergence zone around 10 N. The minimum values over Southeast Asia and the western Pacific

1JULY 001 CHEN ET AL. 189 FIG.. Observed OLR climatology for Jul. Areas with OLR values 10Wm, 10 5Wm, and 5 40 W m are shaded with dark, intermediate, and light tones, respectively. reflect the strong convection associated with the Asian summer monsoon. Weak monsoonal heat sources also occur over tropical Africa and central America. In contrast, large values of OLR over North Africa, the eastern North Pacific, and North Atlantic between 15 and 45 N imply an absence of latent heating over these regions. Comparing Figs. 1 and, it is seen that the strong lowlevel cyclone and upper-level anticyclone over Asia are associated with the heating over Southeast Asia, and the weak low-level cyclone and upper-level anticyclone over North America are associated with the Mexican monsoon. The oceanic subtropical anticyclones appear to coincide with regions of radiative cooling. The stationary waves exhibit a marked seasonal cycle as shown in Fig. 3. The phase of the warm season (April September) stationary waves is roughly opposite to their cold season (November February) counterparts, with rapid transitions occurring in the equinox seasons. The main features develop and mature in tandem at all longitudes in both the upper and lower levels. Furthermore, the evolution of the waves appears phase-locked with that of the OLR. Figure 4 shows a longitude time cross section of the time-mean OLR averaged between 10 and 35 N, and the most prominent feature is the Asian summer monsoonal heat source. By comparison, latent heating over North America is weak in summer, and there is no indication for latent heating over the African continent between 10 and 35 N at any time. A comparison of Figs. 3 and 4 reveals that the maor features in the warm season circulation in the Northern Hemisphere develop in tandem with the monsoon heat source over East Asia and the western Pacific. It is, therefore, reasonable to hypothesize that the maor features in warm season stationary waves in the Northern Hemisphere are forced by a common source of heating over East Asia and the western Pacific. This hypothesis will now be tested by analyzing results from the linear QG model forced by idealized heating fields. 4. Model results The basic states of the linear model are derived from the observed time-mean and zonal-mean zonal wind for July. It is seen from Fig. 5 that the tropospheric zonal flow is westerly in middle and high latitudes and easterly in lower latitudes. For the domain between 15 and 55 N and 1000 100 mb, the averaged zonal-mean zonal wind is about 5 m s 1. For most of the experiments, the heating fields are given by the analytic formula Q (,, z) [Q (, ) Q (, )]Q(z), ˆ 1 (1) where and [ ] 90 cos, if 50 130 and 10 45 Q 1 (, ) 80 () 0, otherwise [ ] 55 cos, if 44 66 and 10 45 Q (, ) (3) 0, otherwise and Qˆ (z) is specified as in Eq. (4.1) in Chen (001). Here is east longitude (degrees) and is north latitude (degrees). Figure 6 shows a horizontal and vertical cross section of the heating field. The heating is zero at the surface and above 14 km. It is maximum at 7 km, with a maximum heating rate of about 3.5 dk day 1. This heating field

1830 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 58 FIG. 5. Observed 40-yr (1960 99) averaged zonal-mean zonal wind for Jul. The contour interval is 5 m s 1. FIG. 3. Observed seasonal cycle of 0 50 N averaged eddy geopotential height at (a) 1000 hpa and (b) 150 hpa. The contour interval is 0 m. mimics the maor summer heat sources in the Northern Hemisphere (e.g., Yanai and Tomita 1998). The results in the companion paper (Chen 001) demonstrated that, for a given heating field, the zonal-mean zonal wind has a profound impact on the structure of the forced responses. The solutions on an easterly or resting basic state are trapped to the heating region. The solutions on a westerly zonal flow can, on the other hand, propagate eastward and upward to the far field. In the following, we will show solutions on three different zonal-mean states to further understand the effects of the basic state on thermally forced stationary waves. Figure 7 shows the model solution in streamfunction forced by the heating on a uniform zonal flow of 5 m s 1 (the domain-averaged observed zonal-mean zonal wind in Fig. 5 is about 5 m s 1 ). Even with such a simplified basic state, several of the principal features in observations in Fig. 1 are realistically modeled, including the low-level oceanic anticyclones and continental cyclones (Fig. 7a), and the upper-level oceanic troughs and continental ridges (Fig. 7b). As in nature, the modeled low-level subtropical anticyclones over the Pacific and Atlantic are comparable in strength and size, while the Asian low is much stronger and broader than its American counterpart. 1 There are, however, several 1 The QG streamfunction relates to geopotential through / f 0 where f 0 8 10 5 s 1 at 35 N. Thus, a contour interval of.5 10 6 m s 1 for the streamfunction in Figs. 7, 8, 9, and 10 is equivalent to a contour interval of 0 m for the geopotential height in Figs. 1 and 3. FIG. 4. Longitude time cross section of the observed time-mean OLR averaged between 10 and 35 N. The shadings are the same as in Fig. 4.

1JULY 001 CHEN ET AL. 1831 FIG. 6. Distributions of the idealized heating: (a) horizontal structure at the level of maximum heating (7 km), and (b) longitude height cross section at 35 N. The contour interval is 0.5 K day 1. unrealistic features in the model solution, for example, the strong upward propagation and the large amplitude above 14 km. The solution also fails to produce the observed SW NE horizontal phase tilt of the tropospheric circulation systems. The failure to model the horizontal phase tilt stems from the omission of realistic meridional shear in the zonal-mean flow. Figure 8 shows a model solution, forced by the same heating field, but on a zonal-mean flow with meridional shear corresponding to the observed zonal-mean zonal wind at km (about 750 hpa). The modeled stationary waves now exhibit a pronounced SW NE phase tilt in the horizontal as observed, suggesting that the meridional shear of the zonal flow plays a critical role in generating the observed phase structure of the stationary waves. FIG. 7. Cross sections of model streamfunction at (a) the surface, (b) 14 km, and (c) 35 N on a uniform zonal flow of 5 m s 1. The contour interval is.5 10 6 m s 1.

183 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 58 FIG. 8. As in Fig. 7 but on the observed zonal flow at km. We also find that the sign of the zonal-mean flow in the stratosphere is very important for realistically modeling the vertical structure of the stationary waves. In the model responses in Figs. 7 and 8, the stationary waves are propagating upward above 14 km, while in the observations the waves are trapped to the troposphere and lower stratosphere. This discrepancy can be largely resolved by using easterlies in the stratosphere of the model, as occurs in observations (see Fig. 5). Figure 9 shows a model solution forced by the same heating field as used before, but with the full observed zonal-mean zonal wind as the basic state. The modeled stationary waves exhibit a pronounced SW NE phase tilt in the horizontal. At 0 N the low-level anticyclones are centered in the western Pacific and western Atlantic, respectively, while at 40 N they are centered in the eastern Pacific and eastern Atlantic, respectively. Vertically, the response (Fig. 9c) is now trapped to the troposphere and lower stratosphere, very similar to observations (Fig. 1c), which suggests that the forced stationary waves cannot propagate into the easterlies in the upper stratosphere. The model response in Fig. 9 strongly suggests that the observed summer stationary waves, including the near-surface subtropical anticyclones over the oceans, can be largely explained as the stationary Rossby waves forced by the zonally asymmetric diabatic heating. Figure 10 shows vertical cross sections at 35 N of the model solutions forced by the Asian and American part of the heating field, respectively, with the full observed zonal-mean zonal wind as the basic state. The response to the Asian heating has significant amplitude in all longitudes, while the response to the American heat source is largely confined to the longitudes of the heating. These striking differences between the two solutions arise because the zonal scale of the Asian heating is much greater, and as shown analytically in the companion paper (Chen 001), that the amplitude of the thermally forced stationary wave is proportional to the zonal scale of the heating. The results in Figs. 9 and 10 suggest that the observed subtropical anticyclones over the Pacific and Atlantic Oceans are largely a remote response to heat sources over Asia. It is the broad scale of the Asian heat sources (primarily the zonal wavenumber 1 component of that heating) that is responsible for forcing and maintaining the observed strong low-level cyclone over Asia and the low-level anticyclones over both the eastern Pacific and Atlantic Oceans. The narrow heating over North America is, on the other hand, responsible for forcing and maintaining the observed weak low-level cyclone over North America. It is interesting to note that, in the absence of this American heat source, a single low-level subtropical anticyclone would span the entire Western Hemisphere, rather than the two-cell pattern that is observed. 5. Conclusions and discussion The observational and model results of this study form the basis for a new hypothesis about the origin of

1JULY 001 CHEN ET AL. 1833 FIG. 9. As in Fig. 7 but on the observed zonal flow. the oceanic subtropical anticyclones. The good agreement between the model and observations suggest that the summertime stationary waves as a whole are primarily manifestations of stationary Rossby waves forced by the zonally asymmetric subtropical heating. It is proposed that the existence and maintenance of the subtropical anticyclones over the eastern ocean basins is ultimately related to the downstream (eastward) energy propagation of Rossby waves forced primarily by the large-scale heat sources over Asia. Here we wish to contrast our hypothesis with Hoskins (1996) hypothesis regarding the origin of the subtropical anticyclones. The emphasis there is on the local upstream (westward) effect of the continental monsoonal heating in forcing the oceanic anticyclones. Our hypothesis emphasizes the importance of downstream FIG. 10. Vertical cross sections of model streamfunction at 35 N forced only by (a) the Asian heating and (b) the American heating on the observed zonal wind. The contour interval is.5 10 6 m s 1.

1834 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 58 (eastward) energy propagation of thermally forced stationary Rossby waves in producing the zonally asymmetric circulation in the subtropics. Our theory predicts, for example, that it is the Asian heat source, rather than the American heat source as implied by the Hoskins hypothesis, that is primarily responsible for the formation and maintenance of the subtropical anticyclone over the North Pacific. Our conclusion, that the observed summertime circulation in the subtropics is forced primarily by diabatic heating over east Asia and the western Pacific, is in accord with results from earlier modeling and diagnostic studies. Using a linear, steady-state, two-layer primitive equation model, Webster (197) simulated the main features of the thermally forced stationary waves in the subtropics. He emphasized the role of latent heating over Southeast Asia in forcing the stationary waves. Using a linear, multilevel primitive equation model, Lin (1983) successfully reproduced many of the basic features of the summertime stationary waves, including the vertical structure and horizontal wavy pattern. He found that the latent heating over Asia is the most important forcing source for the development and maintenance of the summer circulation in the subtropics. More recently, Ting (1994) performed a budget study for the stationary waves in a general circulation model and confirmed the importance of the Asian monsoonal heating in maintaining these eddies. It is interesting that the main features of the thermally forced stationary waves in the linear QG system are so consistent with those of the observed summer stationary waves. The applicability of the QG system for understanding synoptic-scale low-frequency eddies in midlatitudes is well established (e.g., Holton 199), as is its utility in understanding planetary-scale stationary waves in the subtropics (e.g., Webster 1983). In addition, a scale analysis of the QG potential vorticity equation shows that, for planetary-scale disturbances and typical summer zonal-mean flows, the Rossby number (the ratio of the nonlinear advection of the relative vorticity to the linear advection of the planetary vorticity) is much less than 0.1 in the midlatitude westerlies where the subtropical anticyclones are located. This indicates that the potential vorticity generated by heating is mainly balanced by eddy advection of planetary vorticity. As such, the nonlinear advection of relative vorticity is not very important, and the linear QG theory applied here is suitable for describing the dynamics of planetaryscale waves. Lest the applicability of this simple model be overstated, it should be noted that some features of the observed summer stationary waves are not realistically modeled by the linear QG system. For example, the model response is weaker than the observations in the Western Hemisphere but stronger in the Eastern Hemisphere and the longitudinal extent of the subtropical anticyclones is narrower than that in observations. The precise cause for these discrepancies is not known. The physical processes neglected in the model, such as synoptic eddy feedback and more realistic treatment of the heating (including localized cooling over the oceans), are all likely to be important. It should be also noted that in nature the heating fields and circulations are strongly coupled. Nonetheless, these limitations notwithstanding, the essential features of the climatological summer stationary waves are reproducible by the thermally forced linear QG system. Acknowledgments. We thank Drs. Walter Robinson, Klaus Weickmann, and Jeffrey Whitaker for interesting discussions. We also thank Drs. Isaac M. Held and Theodore G. Shepherd for their constructive criticism and suggestions that led to much improvement on this paper. This research was supported by the NOAA Climate and Global Change Program. APPENDIX Model Equations and Solution Method The basic equations for the linear QG model used in this paper are 1 u [ 0 ( 0 ) z z] u 1 qy x x x 1 0 ( 0 Q ) z, (A.1a) u u z z x Q, x at z 0, (A.1b) x 0, at y (W/), (A.1c) where is the eddy streamfunction, u the zonal-mean 1 zonal wind, qy u yy 0 ( 0 u z ) z, and 1 and are the coefficients of the Rayleigh friction and Newtonian cooling. Other notation is defined in Chen (001). Expanding the primed variables into zonal harmonics, [ ] (y, z) ikx Re e, (A.) Q [ Q(y, z) ] where k is the zonal wavenumber, (A.1) can be written as 1 ( ) 0 0 z z yy, (A3.a), z 1 at z 0, (A.3b) 0, z at z, (A.3c) 0, at y (W/), (A.3d) where (y, z) (y, z)[iku(y, z) ], (A.4a) (y, z) [ikqy (iku 1)k ], 1 (A.4b) (y) (iku ), (y) r im (y), (A.4c) 1 z 1 (y, z) 0 ( 0 Q) z, (y) Q (A.4d)

1JULY 001 CHEN ET AL. 1835 where and m are defined as (iku ) 1 and m { 1 [ / u k ]} z r, which is derived from the radiation condition. Using centered difference schemes, (A.3) can be written as 1 1 ak k 1 bk k ck k 1 dk k ek k f k, (A.5a) (4 ) g, 1 1 3 N (4 N 1 N ), (A.5b) 1 M k 0, k 0, (A.5c) where 1 1/[3 ( z) 1 (y)], 1/[3 ( z) (y)], and ] [ z ] y ] z ( z) H z k 1 k 1 k ak k, (A.6a) [ ( z) H bk ( z) k k k, (A.6b) k 1 k 1 k c, (A.6c) k k [ z dk k, (A.6d) y z ek k, (A.6e) y f k ( z) k, (A.6f) g ( z) 1. (A.6g) The system of linear equations in (A.5) is written as a block tridiagonal matrix equation, which is then solved using the FORTRAN programs from the NAG library. REFERENCES Chen, P., 001: Thermally forced stationary waves in a quasigeostrophic system. J. Atmos. Sci., 58, 1585 1594. Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447 46. Gruber, A., and A. F. Krueger, 1984: The status of the NOAA outgoing longwave radiation data set. Bull. Amer. Meteor. Soc., 65, 958 96. Holton, J. R., 199: An Introduction to Dynamic Meteorology. 3d ed. Academic Press, 551 pp. Hoskins, B. J., 1996: On the existence and strength of the summer subtropical anticyclones. Bull. Amer. Meteor. Soc., 77, 187 19. Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Proect. Bull. Amer. Meteor. Soc., 77, 437 471. Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 175 177. Lin, B.-D., 1983: Behavior of stationary waves and the summer monsoon. J. Atmos. Sci., 40, 1163 1177. Rodwell, M. J., and B. J. Hoskins, 1996: Monsoons and the dynamics of deserts. Quart. J. Roy. Meteor. Soc., 1, 1385 1404. Ting, M., 1994: Maintenance of northern summer stationary waves in a GCM. J. Atmos. Sci., 51, 386 3308. Webster, P. J., 197: Response of the tropical atmosphere to local, steady forcing. Mon. Wea. Rev., 100, 518 541., 1983: Large-scale structure of the tropical atmosphere. Large- Scale Dynamical Processes in the Atmosphere, B. J. Hoskins and R. P. Pearce, Eds., Academic Press, 35 75. White, G. H., 198: An observational study of the Northern Hemisphere extratropical summertime general circulation. J. Atmos. Sci., 39, 4 40. Yanai, M., and T. Tomita, 1998: Seasonal and interannual variability of atmospheric heat sources and moisture sinks as determined from NCEP NCAR reanalysis. J. Climate, 11, 463 48.