• Akiyama, T., 1973: Frequent occurrence of the heavy rainfalls along the north side of the low-level jet stream. Pap. Meteor. Geophys.,24, 379–388.

  • Arakawa, A., and W. H. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large scale environment. Part I. J. Atmos. Sci.,31, 674–701.

  • Blackadar, A. K., 1957: Boundary layer wind maxima and their significance for the growth of nocturnal inversions. Bull. Amer. Meteor. Soc.,38, 283–290.

  • Bonner, W. D., 1968: Climatology of the low level jet. Mon. Wea. Rev.,96, 833–850.

  • Chang, C.-P., and G. T. J. Chen, 1995: Tropical circulations associated with southwest monsoon onset and westerly surges over the South China Sea. Mon. Wea. Rev.,123, 3254–3267.

  • Chen, C., C. H. Bishop, G. S. Lai, and W.-K. Tao, 1997: Numerical simulations of an observed narrow cold frontal rainband. Mon. Wea. Rev.,125, 1027–1045.

  • Chen, G. T. J., 1983: Observational aspects of the mei-yu phenomenon in subtropical China. J. Meteor. Soc. Japan,61, 306–312.

  • ——, and C. C. Yu, 1988: Study of low-level jet and extremely heavy rainfall over northern Taiwan in the mei-yu season. Mon. Wea. Rev.,116, 884–891.

  • Chen, Q., 1982: The instability of the gravity–inertial wave and its relation to low-level jet and heavy rainfall. J. Meteor. Soc. Japan,60, 1041–1057.

  • Chen, X. A., and Y.-L. Chen, 1995: Development of low-level jets during TAMEX. Mon. Wea. Rev.,123, 1695–1719.

  • Chen, Y.-L., X. A. Chen, and Y.-X. Zhang, 1994: A diagnostic study of the low-level jet during TAMEX IOP 5. Mon. Wea. Rev.,122, 2257–2284.

  • Chou, L. C., C. P. Chang, and R. T. Williams, 1990: A numerical simulation of the mei-yu front and the associated low level jet. Mon. Wea. Rev.,118, 1408–1428.

  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci.,46, 3077–3107.

  • ——, 1993: A nonhydrostatic version of the Penn State–NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev.,121, 1493–1513.

  • Emanuel, K. A., 1983: On assessing local conditional symmetric instability from atmospheric soundings. Mon. Wea. Rev.,111, 2016–2033.

  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev.,121, 764–787.

  • Hoskins, B. J., 1974: The role of potential vorticity in symmetric stability and instability. Quart. J. Roy. Meteor. Soc.,100, 480–482.

  • ——, I. Draghici, and H. C. Davies, 1978: A new look at the ω-equation. Quart. J. Roy. Meteor. Soc.,104, 31–38.

  • Hsu, W.-R., and W.-Y. Sun, 1994: A numerical study of a low-level jet and its accompanying secondary circulation in a mei-yu system. Mon. Wea. Rev.,122, 324–340.

  • Keyser, D., and M. J. Pecnick, 1985: A two-dimensional primitive equation model of frontogenesis forced by confluence and horizontal shear. J. Atmos. Sci.,42, 1259–1282.

  • ——, and M. A. Shapiro, 1986: A review of the structure and dynamics of upper-level frontal zones. Mon. Wea. Rev.,114, 452–499.

  • Matsumoto, S., 1972: Unbalanced low-level jet and solenoidal circulation associated with heavy rainfalls. J. Meteor. Soc. Japan,50, 194–203.

  • ——, 1973: Lower tropospheric wind speed and precipitation activity. J. Meteor. Soc. Japan,51, 101–107.

  • ——, K. Ninomiya, and S. Yoshizumi, 1971: Characteristic features of “baiu” front associated with heavy rainfall. J. Meteor. Soc. Japan,49, 267–281.

  • Ninomiya, K., 1984: Characteristics of baiu front as a predominant subtropical front in the summer Northern Hemisphere. J. Meteor. Soc. Japan,62, 880–894.

  • ——, and T. Akiyama, 1974: Band structure of mesoscale clusters associated with low-level jet stream. J. Meteor. Soc. Japan,52, 300–313.

  • Ray, P. S., Ed., 1986: Mesoscale Meteorology and Forecasting. Amer. Meteor. Soc., 793 pp.

  • Schlesinger, R. E., 1973: A numerical model of deep moist convection: Part I. Comparative experiments for variable ambient moisture and wind shear. J. Atmos. Sci.,30, 835–856.

  • Stauffer, D. R., and N. L. Seaman, 1990: Use of four-dimensional data assimilation in a limited-area mesoscale model. Part I: Experiments with synoptic-scale data. Mon. Wea. Rev.,118, 1250–1277.

  • Tao, S., and L. Chen, 1987: A review of recent research on the East Asian summer monsoon in China. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 60–92.

  • Tseng, S.-F., 1993: A case study of heavy rainfall in the mei-yu season (in Chinese). M. S. thesis, Department of Atmospheric Physics, National Central University, Chung-Li, Taiwan, 121 pp.

  • Uccellini, L. W., and D. R. Johnson, 1979: The coupling of upper and lower tropospheric jet streaks and implications for the development of severe convective storms. Mon. Wea. Rev.,107, 682–703.

  • Zhang, D.-L., and R. A. Anthes, 1982: A high-resolution model of the planetary boundary layer—Sensitivity tests and comparisons with SESAME-79 data. J. Appl. Meteor.,21, 1594–1609.

  • View in gallery

    Contour analyses of 300-hPa geopotential (m, solid), wind vectors, and isotherms (K, thin dashed) for (a) 23 June 1991 and (b) 24 June 1991 (after Tseng 1993). The panels (c) and (d) show the corresponding analyses for 850 hPa. The isotherms at 300 hPa are for temperature, while those at 850 hPa are for equivalent potential temperature. The heavy dashed line represents the location of the trough.

  • View in gallery

    Surface map analyses of surface pressure (hPa, solid) and temperature (K, dashed) for (a) 0000 UTC 23 June 1991 and (b) 0000 UTC 24 June 1991 (after Tseng 1993).

  • View in gallery

    Geostationary Meteorological Satellite (GMS) IR imagery at 2332 UTC 23 June 1991 (after Tseng 1993).

  • View in gallery

    Schematic diagram showing the second nested grid domain. The grid resolution in this domain is 45 km. The solid lines AB and CD represent the orientation of the xz cross sections for the following analyses.

  • View in gallery

    Cross sections (xy) of the accumulated precipitation (cm, shaded) at the surface and contours of potential temperature θ (solid, every 1 K) at the height of 500 m (MSL). The time sequence of the plots is (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The time interval used to calculate the accumulation is shown in each panel. The dotted line represents the 301-K isentrope.

  • View in gallery

    Vertical cross sections (xz) of θ (solid lines, every 4 K) and hydrometeor content (g kg−1, shaded) for 2400 UTC 23 June 1991 along the lines (a) AB and (b) CD. The horizontal distance between the tick marks represent a spatial scale of 45 km. The dotted line represents the 308-K isentrope.

  • View in gallery

    Atmospheric sounding sampled at the intersection of lines AB and CD at 1200 UTC.

  • View in gallery

    Vertical cross sections (xz) of ageostrophic wind vectors and hydrometeor content (g kg−1, shaded) along the line AB for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The horizontal spatial scale is the same as in Fig. 6a.

  • View in gallery

    Vertical cross sections (xz) of ageostrophic wind vectors and normal wind speed (m s−1, shaded) along the line AB for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The character J denotes the LLJ position. The horizontal spatial scale is the same as in Fig. 6a.

  • View in gallery

    Cross sections (xy) of the wind speed (m s−1, shaded) and streamlines for 1200 UTC 23 June 1991 at the height of z = (a) 13 km and (b) 10 km.

  • View in gallery

    Same as in Fig. 8 except for the vertical velocity (every 2 m s−1, shaded) and overlayed cloud outline.

  • View in gallery

    Same as in Fig. 10 except for the ageostrophic wind vectors and the divergence of the ageostrophic wind (s−1, shaded).

  • View in gallery

    Schematic diagram showing the flow structure of an observed mei-yu front (after Chen et al. 1994). The thin solid line depicts the direct (D) circulation while the thin dashed line depicts the indirect (I) circulation. The heavy solid line shows the frontal position. The character J denotes the jet positions. The thick heavy line represents the tropopause boundary. Regions with relative humidity greater than 70% are shaded.

  • View in gallery

    Cross sections (xy) of the wind speed (m s−1, shaded) and streamlines at the height of z = 1.5 km for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The horizontal extent of the LLJ is represented by the dark shaded area with wind speeds greater than 20 m s−1.

  • View in gallery

    Cross sections (xy) of the accumulated precipitation (cm, shaded) at the surface from 1200 to 2400 UTC 23 June 1991 for cases (a) NOPBL, (b) NOCU, (b) NOEXP, and (d) CTRL.

  • View in gallery

    Same as in Fig. 14 except at 2400 UTC for cases (a) NOPBL, (b) NOCU, (b) NOEXP, and (d) NOCLD.

  • View in gallery

    Same as in Fig. 9 except at 2400 UTC for cases (a) NOPBL, (b) NOCU, (b) NOEXP, and (d) NOCLD.

  • View in gallery

    Cross section (xy) of wind speed (m s−1, solid line) and pressure (hPa, shaded) at z = 1.5 km showing the difference between two cases (CTRL minus NOCLD).

  • View in gallery

    Same as in Fig. 8 except for showing the difference between two cases (CTRL minus NOCLD).

  • View in gallery

    Conceptual schematic diagram illustrating the role of (a) the upper-level jet, and (b) the vertical and slantwise convection in the development of the LLJ. The heavy solid line with triangles shows the frontal position. The character J denotes the jet position, and the boldness of the J represents the jet strength. The thick heavy line represents the tropopause boundary. The dashed line depicts the weaker circulation, and the thin sold line represents the stronger circulation. Regions with cloud are shaded.

  • View in gallery

    Vertical cross sections (xz) of M surfaces (m s−1, solid line), equivalent potential temperature (348 K, shaded), and cloud profile (thick solid line) along the line AB for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 20 20 2
PDF Downloads 13 13 1

The Intensification of the Low-Level Jet during the Development of Mesoscale Convective Systems on a Mei-Yu Front

View More View Less
  • 1 Mesoscale Atmospheric Processes Branch, Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, Maryland, and Science Systems and Applications, Inc., Lanham, Maryland
  • | 2 Mesoscale Atmospheric Processes Branch, Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, Maryland
  • | 3 Department of Atmospheric Physics, National Central University, Chung-Li, Taiwan
  • | 4 Mesoscale Atmospheric Processes Branch, Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, Maryland, and Science Systems and Applications, Inc., Lanham, Maryland
  • | 5 Department of Atmospheric Physics, National Central University, Chung-Li, Taiwan
© Get Permissions
Full access

Abstract

During the period of 21–25 June 1991, a mei-yu front, observed by the post–Taiwan Area Mesoscale Experiment, produced heavy precipitation along the western side of the Central Mountain Range of Taiwan. Several oceanic mesoscale convective systems were also generated in an area extending from Taiwan to Hong Kong. Numerical experiments using the Penn State–NCAR MM5 mesoscale model were used to understand the intensification of the low-level jet (LLJ). These processes include thermal wind adjustment and convective, inertial, and conditional symmetric instabilities.

Three particular circulations are important in the development of the mei-yu front. First, there is a northward branch of the circulation that develops across the upper-level jet and is mainly caused by the thermal wind adjustment as air parcels enter an upper-level jet streak. The upper-level divergence associated with this branch of the circulation triggers convection.

Second, the southward branch of the circulation, with its rising motion in the frontal region and equatorward sinking motion, is driven by frontal vertical deep convection. The return flow of this circulation at low levels can produce an LLJ through geostrophic adjustment. The intensification of the LLJ is sensitive to the presence of convection.

Third, there is a circulation that develops from low to middle levels that has a slantwise rising and sinking motion in the pre- and postfrontal regions, respectively. From an absolute momentum surface analysis, this slantwise circulation is maintained by conditionally symmetric instability located at low levels ahead of the front. The presence of both the LLJ and moisture is an essential ingredient in fostering this conditionally symmetric unstable environment.

Corresponding author address: Dr. Chaing Chen, Mesoscale Atmospheric Processes Branch, NASA/Goddard Space Flight Center, Code 912, Greenbelt, MD 20771.

Email: chen@betsy.gsfc.nasa.gov

Abstract

During the period of 21–25 June 1991, a mei-yu front, observed by the post–Taiwan Area Mesoscale Experiment, produced heavy precipitation along the western side of the Central Mountain Range of Taiwan. Several oceanic mesoscale convective systems were also generated in an area extending from Taiwan to Hong Kong. Numerical experiments using the Penn State–NCAR MM5 mesoscale model were used to understand the intensification of the low-level jet (LLJ). These processes include thermal wind adjustment and convective, inertial, and conditional symmetric instabilities.

Three particular circulations are important in the development of the mei-yu front. First, there is a northward branch of the circulation that develops across the upper-level jet and is mainly caused by the thermal wind adjustment as air parcels enter an upper-level jet streak. The upper-level divergence associated with this branch of the circulation triggers convection.

Second, the southward branch of the circulation, with its rising motion in the frontal region and equatorward sinking motion, is driven by frontal vertical deep convection. The return flow of this circulation at low levels can produce an LLJ through geostrophic adjustment. The intensification of the LLJ is sensitive to the presence of convection.

Third, there is a circulation that develops from low to middle levels that has a slantwise rising and sinking motion in the pre- and postfrontal regions, respectively. From an absolute momentum surface analysis, this slantwise circulation is maintained by conditionally symmetric instability located at low levels ahead of the front. The presence of both the LLJ and moisture is an essential ingredient in fostering this conditionally symmetric unstable environment.

Corresponding author address: Dr. Chaing Chen, Mesoscale Atmospheric Processes Branch, NASA/Goddard Space Flight Center, Code 912, Greenbelt, MD 20771.

Email: chen@betsy.gsfc.nasa.gov

1. Introduction

The mei-yu, which literally translated means plum rain, is a major annual rainfall event that begins in mid-May over southern China and moves northward to the Yangtze River and southern Japan between mid-June and mid-July. During the mei-yu season, the quasi-stationary mei-yu front can extend from southern Japan to southern China and is characterized by a weak temperature gradient, strong vertical wind shear, and large moisture gradient across the front at low levels. Precipitation systems imbedded in the mei-yu front are essentially convective in nature and long lived (Chen 1983; Matsumoto et al. 1971). Statistically, the heavy precipitation produced by the mei-yu front is closely related to the existence of a low-level jet (LLJ) (Chen and Yu 1988), a southwesterly flow located between 850 and 700 hPa with a speed of around 20 m s−1 and a horizontal length scale of 400–500 km. This LLJ1 is not the classical boundary layer jet of Blackadar (1957) and Bonner (1968) but rather a strengthening of the wind between the 600- and 900-hPa levels (Chen et al. 1994).

The importance of the LLJ is not simply its ability to transport moisture to fuel convection but also its signaling of the onset of the East Asia summer monsoon (EASM). This monsoon begins over southern China in mid-May (Chang and Chen 1995) and is subsequently followed by the mei-yu season. Thus, the LLJ has been an interesting and important topic because of its relationship with heavy rainfall (Akiyama 1973; Matsumoto 1972; Ninomiya and Akiyama 1974; Tao and Chen 1987) and the development of the EASM and associated mei-yu front.

The origin of the LLJ has been explored extensively. Matsumoto (1973) and Ninomiya and Akiyama (1974) suggested that it results from the downward mixing of southwesterly momentum from upper levels by convection. Because the LLJ usually does not lie directly underneath the upper-level jet, Chen (1982) suggested that the LLJ is produced by thermal wind adjustment that occurs in the entrance region of the upper-level jet, rather than by momentum mixing. Chen (1982) demonstrated that, by including convective heating, thermal wind adjustment can generate two cells; the convective heating occurs between these two circulation cells. To the left (north) of the convective heating, a cell develops across an upper-level jet at the jet streak entrance where frontal forcing is strong. To the right (south), the other cell, with its returning flow in low levels, can lead to the LLJ generation.

In their two-dimensional numerical experiments, Chou et al. (1990) found that convective heating can generate a direct circulation with rising motion in the convective region and sinking motion equatorward. The LLJ is produced by the Coriolis deflection of the return flow. The development of the convection is associated with confluent frontal forcing. Therefore, the theory proposed by Chou et al. (1990) can be regarded as an example of Chen’s (1982) theory.

Other theories have been proposed. For example, from their numerical experiments, Hsu and Sun (1994) suggested that heating by stratiform cloud deepens the low-level pressure trough and thereby leads to the development of the LLJ. Chen et al. (1994) and Chen and Chen (1995) argued that the intensification of the LLJ is linked to the development of a lee trough to the east of the Tibetan Plateau. Their argument rests on the finding that the LLJ intensification is collocated with those regions where the cross-isobar wind increases. The Coriolis deflection of this wind results in the development of the LLJ.

In summary, most theories agreed that the LLJ is produced by the Coriolis deflection of the low-level cross-isobar wind. However, they differ on the origin of this cross-isobar wind. One camp suggests that the mesoscale precipitation systems are the cause, whereas others attribute it to the deepening of the lee trough system.

The objective of this paper is to conduct numerical simulations to examine the mechanisms responsible for the development of the LLJ in the lower troposphere in a mei-yu front. We have chosen to examine the actual mei-yu front event of 23–24 June 1991. In section 2, a brief overview of weather during this period is given. In section 3, MM5 is described along with some of its key features. The methodology of the experiment design is also discussed. In section 4, the results from the control case are presented. To show the effects of clouds, results from the sensitivity experiments are also presented. In section 5, the physical processes that can lead to the development of the LLJ are discussed. These include a cloud triggering mechanism associated with the upper-level jet and the development of the vertical and the slantwise convection. Finally, a summary of the findings are presented in section 6.

2. Overview of synoptic situation: 23–24 June 1991

During 20–25 June 1991, heavy rainfall, with rates of 100–300 mm day−1, occurred over the western plain and mountains of Taiwan. This weather system, analyzed and reported by Tseng (1993), was characterized by the passage of a slow moving mei-yu front. During this period, a deep, northeast–south-southwest trough developed at upper levels over China. The associated LLJ extended from Japan to Taiwan into the South China Sea. This LLJ appeared to result from the strengthening and the westward movement of the Pacific high.

Although the weather event lasted from 20 to 25 June 1991, this paper concentrates on the most intense mesoscale convection that occurred during the period of 23–24 June 1991. Figure 1, showing the 300-hPa and 850-hPa analyses from 0000 UTC 23 June through 0000 UTC 24 June, reveals a trough extending from the northeastern to southern China. Figure 1a shows that this trough consists of three shortwave troughs: one over southeastern China, another one over Yellow/Yangtze River, and a more northern one over Manchuria. The development of the southernmost trough was closely related to the orographic effects due to the Tibetan Plateau, while the other two troughs resulted from atmospheric baroclinicity. Over a 24-h period (Fig. 1b), the system evolved into two troughs. The Yellow/Yangtze River trough merged with the southernmost trough, while the northernmost trough deepened and extended southward to the Yellow River.

Chen and Chen (1995) noted that a LLJ can develop simultaneously with the deepening of an upper-level lee trough to the east of the Tibetan Plateau. As the lee trough deepens, a low-level frontal cyclone can intensify the LLJ due to the Coriolis deflection of the cross-isobar wind. On the other hand, our present case does not have a prominent deepening of the upper-level lee trough over a 24-h period (Figs. 1a and 1b). However, the intensification of the LLJ (Figs. 1c and 1d) is still very significant over a large area extending from Taiwan to Japan. Although the increase of the wind speed and the formation of the LLJ can be explained by the increase of 850-hPa height gradients (Fig. 1d), there is no evidence that the increase of the low-level pressure gradient force is directly associated with the development of the Tibetan lee cyclone. Therefore, it would be interesting to investigate the differences between this case and that analyzed by Chen and Chen (1995).

The corresponding mesoscale surface analyses near Taiwan during the simulation period are shown in Fig. 2. From 0000 UTC 23 June through 0000 UTC 24 June, the surface front moved slowly toward the southeast. Ahead of the front, the wind was mainly southwesterly. Behind the front, the wind varied from northeasterly to southwesterly during this period. The temperature gradient in the vicinity of the surface front to the southwest of Taiwan was weak at 0000 UTC 23 June (Fig. 2a). However, this temperature gradient at 0000 UTC 24 June (Fig. 2b) seemed to strengthen as a southwesterly flow ahead of the front advected warm air. This speculation is confirmed by model results presented in a later section.

Looking at IR imagery for 2332 UTC 23 June (Fig. 3), a cloud band ahead of the mei-yu front extended from Japan to the Luzon Strait (north of the Philippines). This cloud band was not responsible for the heavy precipitation that occurred on 23 June over southwestern Taiwan. Rather, there was another band of mesoscale convective systems (MCSs), labeled A and B, extending southwestward from Taiwan to Hong Kong. These MCSs were located southeast of the upper-level jet and northwest of the LLJ. In this study, the initiation and the development of these MCSs and their interactions with upper- and low-level jets are investigated.

In Fig. 3, there is also a MCS that developed over southwestern Taiwan whose generation was closely associated with the terrain. We have chosen not to analyze this particular MCS. Because we are primarily interested in the oceanic MCSs that occurred near Taiwan, the LLJ that we will examine is the low-level southwesterly flow that intensified to the south and southwest of Taiwan.

3. Mesoscale model and experiment design

a. Model description

MM5 was used to conduct numerical simulations of the mei-yu frontal system. Elements of the Penn State–NCAR MM5 have been described by Dudhia (1993). In brief, MM5 is a three-dimensional, nonhydrostatic, and elastic mesoscale model. It uses finite differences and a time-splitting scheme to solve prognostic equations on an Arakawa type-B staggered grid. Its vertical coordinate, though defined as a function of the reference-state pressure, is similar to a terrain-following coordinate. For case studies, MM5 employs observed wind, temperature, and humidity as the initial and boundary conditions and incorporates realistic topography and sophisticated physical processes to represent the appropriate forcing for the development of the observed weather system. These physical processes include clouds, long- and shortwave radiative processes, and surface fluxes of heat, moisture, and momentum. Because of the wide variety of physical process schemes that can be used in MM5, only an overview of those that are most important are listed as follows:

  1. The model is initialized from National Centers for Environmental Prediction (NCEP) archived global analyses with 2.5° latitude–longitude resolution that are interpolated to the model gridpoint locations. The first-guess fields are then enhanced by blending in the observational data using an objective analysis technique to introduce mesoscale features. The time-varying lateral boundary condition is provided by repeating the above procedure at 12-h intervals. The time interval of interest is from 0000 UTC 23 June to 0000 UTC 24 June of 1991.
  2. Three nested domains were constructed with a grid resolution of 135, 45, and 15 km, and had numbers of grid points in (x, y, σ) of 43 × 28 × 23, 55 × 37 × 23, and 73 × 49 × 23, respectively. The number of σ levels is 24 (1, 0.99, 0.98, 0.96, 0.93, 0.89, 0.85, 0.8, 0.75, 0.7, 0.65, 0.6, 0.55, 0.5, 0.45, 0.4, 0.35, 0.3, 0.25, 0.2, 0.15, 0.1, 0.05, 0.0), which gives 23 layers at which the temperature, moisture, and wind variables are defined. The model top is located at 50-hPa level. A time step of 300 s is used in the coarse-grid domain. To gain a comprehensive understanding of the weather system of interest, only results from the second domain are shown, of which the schematic representation is illustrated by Fig. 4. There are feedbacks between the various domains.
  3. Grell’s cumulus parameterization scheme (Grell 1993)2 was used in this study. It is a simple one-cloud scheme based on the quasi-equilibrium assumption (Arakawa and Schubert 1974). It tests for instability in the sounding and the height of lifting for free convection. This scheme is a mass flux transport scheme with detrainment at cloud top and compensating subsidence outside the cloud. This scheme also includes downdrafts.The cloud is triggered by detecting a deep enough layer of available buoyant energy in the sounding, which is caused by the destabilization of the large-scale environment. The mass flux through the cloud base is then calculated by balancing the cloud stabilization rate with the large-scale change of this energy. Subsequently, the parameterized rate of the heating, moistening, and precipitation of the convection is determined by the cloud base mass flux. In this study, the vertical motion associated with the ageostrophic flow across the upper-level jet can destabilize the large-scale environment by the vertical lifting of moisture.
  4. A three-class cloud microphysics scheme (Dudhia 1989) is used to account for the resolvable scale convection. This scheme allows for ice-phase processes in which cloud water is treated as cloud ice and rainwater as snow when the temperature is below the freezing point.
  5. Blackadar’s high-resolution PBL scheme (Zhang and Anthes 1982) was used to calculate vertical fluxes of heat, moisture, and momentum at each vertical layer. The fluxes from the surface are based on similarity theory. In the nocturnal regime, the vertical fluxes are computed from K theory above the surface layer. In the free-convection regime, these vertical fluxes within the mixed layer are not determined by local gradients but rather the thermal structure of the whole mixed layer (transilient approach). Above the mixed layer, the calculation of fluxes are again based on K theory.Based on the results shown in this paper, Blackadar’s PBL scheme seems to perform well in simulating the boundary layer structure over the ocean. Furthermore, results reported by Chen et al. (1997) indicate that the same PBL scheme also works very well over the land.
  6. To produce more realistic initial conditions, a four-dimensional data assimilation (FDDA) scheme (Stauffer and Seaman 1990) was employed from 0000 to 1200 UTC 23 June 1991. It was probably not necessary to use the FDDA scheme in the simulation. Our previous experience (Chen et al. 1997) showed that the model still produces realistic features without FDDA. During the FDDA period, the model fields were nudged toward analyses of the observed data. Once the assimilation period was completed at 1200 UTC, the model simulation was continued for another 12 h.

b. Experiment design

The experiment design is summarized in Table 1. A total of five experiments, including the control, were carried out. All experiments used FDDA during the first 12 h. For the control (CTRL) case, MM5 was run with cumulus parameterization, cloud microphysics, and PBL processes. The NOPBL case consists of running the model with the PBL scheme deactivated. The motivation for this test is to rule out the possibility that inertial oscillations can couple with PBL processes to generate a nocturnal boundary layer LLJ (Blackadar 1957). This test may not have been necessary, because the mei-yu LLJ occurs at levels much higher than that of the boundary layer jet.

The remaining three cases (NOCU, NOEXP, and NOCLD) tested the hypothesis proposed by Hsu and Sun (1994), in which they conjectured that the latent heat release by the stratiform clouds is the primary cause for the development of the LLJ. They conducted three numerical experiments, which are denoted by without-cumulus (no subgrid-scale cumulus heating), without-latent-heat (neither resolvable nor cumulus heating), and with-cumulus (with resolvable and cumulus heating). Our corresponding experiments are represented by NOCU, NOCLD, and CTRL, respectively. The additional NOEXP case examined the stratiform clouds argument from a different perspective, because the NOEXP case cannot generate stratiform clouds. Therefore, the cumulus parameterization and explicit cloud microphysics were disabled in the NOCU and NOEXP cases, respectively. A simulation that deactivated the latent heat processes completely is denoted by NOCLD. The NOCLD case can be regarded as a fake dry simulation in which the effect of the moisture is still included in the buoyancy term of the vertical wind acceleration equation.

4. Results

a. The control experiment

Figures 5b, 5c, and 5d show the accumulated rainfall for 4 h ending at the time specified on the figure; the rainfall shown in Fig. 5a is for 12 h (0000–1200 UTC). These figures also show the potential temperature at 500 m (MSL, notation omitted hereafter) at the end of the period. During the first 12 h, there was a northeast–southwest band of precipitation, with the heaviest precipitation at the southern edge of China. However, in the hours that follow, the system over the East China Sea became weaker. In contrast, the system over the South China Sea exhibited strong convection and propagated in a southwesterly direction from Hainan toward Taiwan. At the end of the simulation (Fig. 5d), the precipitation pattern west of Taiwan resembled that shown by satellite IR imagery (Fig. 3). However, the precipitation pattern was much more widespread in the model than that suggested by the satellite picture. Because the purpose of this study is to understand the LLJ rather than the exact precipitation pattern, an absolute agreement between the model simulation and observation is not necessary.

In general, the frontal temperature gradient at the earth’s surface was weak at 1200 UTC (Fig. 5a) over the region from southern Japan to Taiwan to the South China Sea. However, the gradient subsequently strengthened in two areas. The first region extended from Japan to the Ryukyu Islands. Midlatitude baroclinic frontogenesis was probably responsible for the tightening. The second region was located over the southern edge of the Taiwan Strait. In this area, the increase of gradient seemed to be associated with the development of convection.

Cross-sectional analysis shows the structure of the precipitation system. The location of these cross sections are shown in Fig. 4 by lines AB and CD. These cross sections slice through the precipitation system MCS A located to the west of Taiwan. The lines AB and CD are essentially perpendicular (from northwest to southeast) and parallel (from southwest to northeast) to the surface front (see Fig. 2), respectively. The cross-sectional analysis for MCS B is not presented because its evolution is very similar to that of MCS A.

Upon examining Fig. 6a (cross section AB), two prominent features of the cloud system can be identified:deep vertical convection and a relatively shallow convection slanted from the surface up to a height of 6 km. The temperature gradient at the leading edge of the surface front is weaker than that at the upper levels. In contrast to the AB cross section (Fig. 6a), the cross section CD (Fig. 6b) shows that a band of MCSs, containing primarily vertical convection, developed over a wide area along the front, in a direction from northeast to southwest.

To describe the thermal structure of the convective system shown in Fig. 6a (or MCS A), a sounding, sampled at the intersection of lines AB and CD at 1200 UTC before the generation of a LLJ, is provided and shown in Fig. 7. From the sounding, a boundary layer capping inversion can be identified at 850 hPa. The air mass within the boundary layer is almost saturated. Therefore, the atmosphere is potentially unstable. The sounding is also characterized by the presence of a moist layer above the capping inversion from 750 to 550 hPa. This moist layer is probably the remnant of convection that occurred previously. As can be seen in the following plot (Fig. 8), this midlevel moist layer appears to play an important role in the onset of the MCS (Fig. 8a).

In Fig. 8, ageostrophic wind vectors (parallel to AB) are superimposed onto cloud profiles to show the development of the cloud system in the AB cross section. During the onset of convection at 1200 UTC (Fig. 8a), the upper-level wind is characterized by a transverse ageostrophic circulation across the upper-level baroclinic zone. The development of this circulation is similar to that discussed by Uccellini and Johnson (1979). The initial development of the midlevel cloud at z = 5 km results from the interaction of this upper-level ageostrophic circulation with the midlevel moist layer, as suggested by the sounding. Once the midlevel cloud formed, deep convection, as well as the intense vertical velocities, subsequently developed underneath the southern edge (right side) of the upper-level ageostrophic flow boundary. Meanwhile, a strong inflow at low levels on the right side of the convection developed and resulted in moisture transport into the cloud. From low to middle levels, some sporadic shallow convection (Fig. 8c) developed in association with this inflow.

Figure 8 also shows that slantwise convection occurs simultaneously with the development of the vertical convection. The associated flow indicates that a low- to midlevel flow with a sloped structure develops to the left of the vertical convection. This flow induces a weak circulation with a return branch from northern latitudes that provides convergence into the region near the surface underneath the vertical convection.

A circulation also develops to the right of the convection with rising motion in the frontal zone and sinking motion equatorward. Because of the horizontal extent of the second model domain, the sinking motion is not clearly shown by Fig. 8. However, in a similar cross section (not shown) from the coarse-grid domain, the existence of this sinking motion can be identified. Chou et al. (1990) called this direct circulation a “reversed Hadley” circulation. Despite the controversial usage of such terminology, we use it to signify this equatorward branch of the convectively driven circulation.

To see the mechanism responsible for generating this transverse circulation associated with the upper-level baroclinicity shown in Fig. 9, the normal wind V (perpendicular to AB) is plotted with the ageostrophic wind vectors to show the relationship between the upper-level jet and the ageostrophic circulation. Because upper-level jet dynamics may play an important role in driving this transverse circulation, it is not surprising to find that the strongest ageostrophic flow is located at the core of the jet. As shown in Fig. 10, an upper-level jet streak exists to the northeast of the cross section AB. The confluent flow associated with this upper-level jet streak may be the reason for the development of an ageostrophic circulation across the core of the upper-level jet.

Figure 11 shows a contour plot of the vertical velocity W. To correlate the relationship of W with respect to convection, we overlay solid lines that show the structure of the cloud. As can be seen in Fig. 11a, the transverse ageostrophic circulation across the upper-level jet produces vertical motions in the region where the initial midlevel cloud begins to form. Although the magnitude of this vertical motion is small, it appears to trigger the subsequent deep convection. Once the deep convection matures (Figs. 11b, 11c, and 11d), the maximum of W lies primarily in the middle to upper levels of the cloud.

The small vertical motion associated with the ageostrophic circulation can advect moisture in the vertical direction, resulting in the destabilization by the increase of the available buoyant energy. Grell’s cumulus parameterization scheme then activates and produces the subgrid-scale convection. The same process can also activate the cloud microphysics scheme. As shown by solid lines in Fig. 11a, the resolvable-scale cloud occurs in the low- to midlevels at 1200 UTC.

Because convection may be triggered by the divergence of ageostrophic flows at upper levels, it would be interesting to find the relationship between this divergence and the accumulated surface precipitation. Figure 12 is a plot of this divergence field (shaded area) and the associated ageostrophic wind vectors in an xy plane at heights of z = 13 and 10 km. Comparing with the surface accumulated rainfall pattern (Fig. 5), there is a very good correlation between precipitation and divergence of the upper-level ageostrophic wind, at least over the ocean. Furthermore, the outline of the entire mei-yu frontal rainband can be clearly identified by the dark area (divergence) in Fig. 12. Indeed, upper-level divergence was present before the development of MCSs A and B, as shown in the satellite IR imagery (Fig. 3).

Another important feature shown by Fig. 9 is the development of the LLJ located at the bottom-right corner of the cross section. The intensification of the LLJ is not caused by the downward mixing of momentum by clouds as proposed by Matsumoto (1973) and Ninomiya and Akiyama (1974), because the LLJ does not lie directly beneath the upper-level jet. Therefore, this simple cloud-mixing theory cannot account for the LLJ intensification. Rather, the development of the LLJ is closely related to the inflow structure that develops at lower levels to the right of the convection. Hereafter, we call this low-level, ageostrophic inflow structure: the mesoinflow. The Coriolis deflection of this mesoinflow results in the LLJ.

Circulations around MCSs have been thoroughly reviewed by Ray (1986). Although his Fig. 9.30b (figures by Shapiro) depicts the presence of a LLJ under an exit region of an upper-level jet, it serves well in the study of a LLJ under an entrance region. According to Shapiro, a northwesterly LLJ is expected under the upper-level jet entrance region, not a southwesterly jet as shown in this study. Therefore, the mechanism that produces the LLJ in a mei-yu front must be entirely different.

The flow structure described in Figs. 8 and 9 is in excellent agreement with that shown in the schematic diagram (Fig. 13) summarized by Chen et al. (1994) based on their diagnostic analysis of a mei-yu frontal system during TAMEX IOP 5 from 31 May to 2 June 1987. As in that case, there was an ageostrophic circulation across the upper-level jet and evidence of slantwise circulation from the surface to 600 hPa across the front. Furthermore, there was a reversed Hadley circulation that developed to the south of the vertical rising circulation. The location of the LLJ was along the returning branch of the reversed Hadley circulation at low levels.

Based upon our model results, there are some differences as well. For example, the location and the shape of their cloud, as suggested by the relative humidity (shaded), differs from our case. Furthermore, the mei-yu front in their case extended from upper levels down to the surface. In contrast, the present case does not show a distinct frontal structure extending from middle to low levels. In addition to these differences, their system developed a strong lee cyclone at low levels. Consequently, they attributed the development of the LLJ to the intensification of the lee cyclone. Despite the differences in the low-level cyclone’s intensity, the flow structures between these two cases are still amazingly similar. Therefore, the intensification of the lee cyclone must not be the only mechanism for the intensification of a LLJ.

Because the development of the LLJ is the focal point in this study of a mei-yu front, it is useful to examine the horizontal structure of the LLJ. Figure 14 shows a time sequence of streamlines and associated velocities at a height of 1.5 km. At an early stage of the simulation (1200 UTC), a patch of LLJ with wind speeds greater than 20 m s−1 prevails over a region east of the Ryukyu Islands. No significant jet appears in the vicinity of Taiwan. However, at later times, as the convective system develops, an intense LLJ becomes the primary feature in an area to the south and southwest of Taiwan.

Figure 14 also shows the development of mesocyclones along the coast of China. As shown in a later section, the mesocyclone that develops over southern China (northeast of Hainan) is not responsible for the intensification of the LLJ because the NOCLD case also shows such a mesocyclone over the same region without any LLJ intensification (see Fig. 16d). Rather, the LLJ intensification is caused by the development of the other mesocyclone which is located in an area over the southwestern edge of Taiwan Strait.

b. The sensitivity experiment

In this section, the model’s sensitivity to PBL processes (NOPBL), the cumulus parameterization (NOCU), and resolvable-scale explicit cloud processes (NOEXP) will be shown. Furthermore, the results from a fake dry simulation (NOCLD) will also be shown where the resolvable-scale and subgrid cloud processes are disabled. A summary of these sensitivity experiments is given in Table 1.

The 12-h accumulated rainfall from 1200 to 2400 UTC is shown in Fig. 15 for the four cases (NOPBL, NOCU, NOEXP, and CTRL). In general, all four cases show a similar precipitation pattern. However, the intensity and location of individual convective systems are different. Compared with the CTRL case (Fig. 15d), the NOPBL case (Fig. 15a) produced less rain. The model response for the NOPBL case is reasonable because the moisture flux from the ocean surface is severed. In contrast, the response for the NOCU (Fig. 15b) and the NOEXP (Fig. 15c) cases shows that with a partial activation of convective processes, the model tends to produce excessive amounts of precipitation in a narrow region with a bull’s-eye structure.

Figure 16 shows the horizontal flow pattern associated with the formation of the LLJ at the height of 1.5 km. For all cases involving clouds (NOPBL, NOCU, and NOEXP), the model produced the LLJ. These experiments bear some similarity when they are compared with the CTRL case (Fig. 14d). From Fig. 16a, PBL processes do not play any role in the formation of the LLJ. Rather, the development of the LLJ is only sensitive to cloud processes, as suggested by the NOCLD case (Fig. 16d).

The same sensitivity can also be seen in Fig. 17, an xz cross-sectional plot of ageostrophic wind vectors and normal wind V. The NOCLD case (Fig. 17d) did not produce a LLJ, despite the presence of the upper-level jet and the associated ageostrophic flow across the jet. Although this case did produce a weak secondary circulation underneath the upper-level jet, no significant vertical motion was generated on either side of the jet. Therefore, there was no strengthening of any circulation, such as the low-level mesoinflow, which can lead to the development of the LLJ. As for the other three cases (NOPBL, NOCU, and NOEXP), they showed a similar flow structure with the CTRL case (Fig. 9d). However, there are some differences, especially in the vertical motion field. Different mechanisms of releasing convective instability (grid resolvable vs subgrid processes) probably account for these differences.

In general, our results, as shown by the CTRL, NOCU, and NOCLD cases, are consistent with that reported by Hsu and Sun (1994). However, they concluded that the latent heat release by stratiform clouds, not the cumulus heating, is important in the formation of the LLJ. The difference in conclusions may be due to the fact that they define a “stratified” cloud as one that has a shallow cloud top (below 6 km) and is wide spread. Nevertheless, their cloud is still convective in nature.

To identify contributions made by convective processes, plots showing differences between the CTRL and NOCLD cases were constructed. Figure 18 shows the difference plot (CTRL minus NOCLD) of the pressure at a height of 1.5 km. The increase in the pressure gradient to the southwest of Taiwan is solely due to the development of the mesolow. The location of this mesolow coincides with that of the MCSs found in this area (i.e., the lowest pressure occurred in the region where the convection developed). Therefore, the latent heat released by the convection could be one of the mechanisms for the generation of this mesolow. Another reason could be the convergence of vorticity.

To reveal the mechanism responsible for the intensification of the LLJ, the horizontal wind speed differences were added to Fig. 18. Based on the distribution of the positive wind speed differences located to the southeast of the mesolow center, this excessive wind is generated by the Coriolis deflection of the flow moving toward the center of the mesolow where the deep vertical convection is located.

An xz difference plot (CTRL minus NOCLD, Fig. 19) for cross section AB was also constructed to show the flow structure induced by the convection. Because the NOCLD case begins to deviate from the CTRL only after 1200 UTC, there is no difference in wind vectors as shown in Fig. 19a. After 1200 UTC (Figs. 19b–d), the main difference between the CTRL and the NOCLD case is the enhanced development of the circulation to the south (right) side of the convection, which is characterized by a strong inflow and outflow at the lower and upper levels, respectively. The flow to the left side is generally weak except for that associated with the slantwise convection.

The reason for strong vertical motions inside a cloud can be easily understood by the release of latent heat from the condensation and deposition of hydrometeors. But why should there be a preferential development of the circulation to the right side of the convection? For typical moist convection developed in a weak vertical wind shear environment, the generation of the flow to the left side should be as strong as that of the right side, because the associated pressure perturbations produced at the upper and lower levels have a symmetric structure in the horizontal direction (Schlesinger 1973). Therefore, the reason for this preferential development will be discussed in the following section.

5. Physical processes that led to the intensification of the LLJ

The important physical processes that led to the intensification of the LLJ have been introduced. These processes explain the generation of ageostrophic flows at both upper and lower levels. These explanations include the upper-level jet, the LLJ, convection, a transverse circulation across the upper-level jet, the reversed Hadley circulation, convectively induced low-level mesoinflow, and slantwise convection.

a. Thermal wind adjustment

Hoskins et al. (1978) demonstrated that geostrophic motions tend to destroy thermal wind balance. However, it is known that large thermal wind imbalances seldom occur in the atmosphere. How can we reconcile these two facts? The mechanism that the atmosphere uses to restore thermal wind balance is its ability to produce ageostrophic wind. The Coriolis deflection of this newly created (ageostrophic) wind then restores thermal wind balance. The cause of this ageostrophic wind was also given by Uccellini and Johnson (1979), numerically simulated by Keyser and Pecnick (1985) and reviewed by Keyser and Shapiro (1986).

Based on our model results, the schematic representation of this ageostrophic secondary circulation is shown in Fig. 20a. This figure illustrates the relationship among the upper-level jet, ageostrophic transverse flow, and rising and sinking motion. Although the secondary circulation may be weak from low to middle levels, rising motions in this circulation can still play an important role in triggering convection if moisture is present. Therefore, the onset of convection occurs where rising motion (Fig. 11a) and moisture (Fig. 7) coexist at middle levels.

b. Convection processes, CSI, and the intensification of the LLJ

Once the initial cloud has formed, the potentially unstable air mass located in the boundary layer (see Fig. 7) lifts to form deep vertical convection (Fig. 8). The release of latent heat then drives a mesoinflow at low levels. The development of the LLJ results from the Coriolis deflection of this mesoinflow. Slantwise convection can also occur from low to middle levels. It signals the presence of conditional symmetric instability (CSI), which then intensifies the LLJ. A schematic representation of this flow structure is shown in Fig. 20b.

In a parallel work (C. Chen et al. 1997, manuscript submitted to J. Atmos. Sci.), we used a two-dimensional, nonhydrostatic model to demonstrate how ageostrophic circulations associated with the upper-level front can trigger convection and lead to the intensification of the LLJ. We found that once convection developed, internal inertia–gravity waves were generated and propagated away from the center of convection. The development of a low-level mesoinflow is the manifestation of this outwardly propagating internal inertia–gravity waves. This may explain why the horizontal length-scale of the mesoinflow and the LLJ in our study is about 400–500 km, because this length is consistent with the Rossby radius of deformation λR = C/f, where C ≃ 20 m s−1 and f = 5 × 10−5 s−1.

In the same two-dimensional idealized model simulation, we also found that quasi-steady-state convection can be produced with a slantwise structure similar to that shown in Fig. 6a, if an alongfrontal moisture gradient is present at low levels. Furthermore, the intensity of the mesoinflow and the LLJ increases as slantwise convection develops from low to middle levels.

The LLJ does not simply advect moisture. Its (as well as the upper-level jet’s) strength is important in determining inertial stability. The crucial parameter is absolute momentum M (Hoskins 1974; Emanuel 1983), where M is defined as M = fx + υ, f is the Coriolis parameter, x is the distance from the left boundary of the cross section, and υ is the wind normal to the cross section. In a neutral and dry atmosphere, air parcels move freely along constant M surface. Just as ∂θ/∂z > 0 denotes thermal stability in convection, so does the presence of a positive ∂M/∂x denote inertial stability. The opposite is true for the unstable case. The presence of inertial stability can be used to explain whether the right (south) branch of the circulation is enhanced while that of the left (north) side is suppressed (Fig. 19, difference plot showing the suppression).

Figure 21 shows contours of the M (solid) and the cloud structure (thick solid). At low levels, for the right branch, the slope of the M surface is flat, especially at later simulation times. Therefore, the low-level inflow from the right side can reach the center of the convection by following a constant-M surface. On the other hand, the inflow of the left branch has to cross a zone with packed M surfaces before it can reach the convection. Therefore, the atmospheric conditions are not favorable for the development of the left branch of the low-level inflow. At upper levels, for the right branch, the environment is essential inertially unstable. Therefore, right outflow is enhanced. The development of the so-called reversed Hadley circulation (Chou et al. 1990) on the right side of the convection appears to be the result of both convective and inertial instability.

At upper levels, on the left side, the outflow encounters an environment with inertial stability ranging from weakly unstable to neutral to strongly stable. This may explain why the speed of the northward flow at upper levels is reduced in the region where ∂M/∂x > 0. However, caution is necessary because the difference plot may fail to reflect the fact that the left branch of the cloud-top outflow can work coherently with the ageostrophic flow across the upper-level jet.

The distribution of equivalent potential temperature θe is also overlayed in Fig. 21. At low levels, high θe represents the abundance of moisture that is an essential element for fueling convection. In this case, high θe is located on the bottom-right side of the cross section. Because the slope of the M surface is flat in this high-θe area, a parcel will encounter CSI when it is perturbed toward the center of the convection. This is a classical CSI example as explained by Emanuel (1983). The slantwise cloud structure extending from the surface to a height of 6 km (see heavy solid line, Fig. 21) is the evidence for the occurrence of CSI.

6. Summary

The Penn State–NCAR MM5 mesoscale model was used to conduct numerical simulations that explored the development of the LLJ in a mei-yu front. This front produced heavy precipitation to the south and southwest of Taiwan during the period 23–24 June 1991. This weather was characterized by a deep northeast–south-southwest midlatitude trough over China. The initial development of the LLJ along the western Pacific rim from the South China Sea to Taiwan to Japan was attributed to the westward migration of the Pacific high and the onset of the East Asian monsoon over the South China Sea.

Numerical experiments showed that the intensification of the LLJ in the region to the south and southwest of Taiwan is sensitive to convective processes. Once mesoscale convection began, the intensity of the LLJ increased. This allowed moisture-laden air to be advected into the region and fuel further convection. There is a positive feedback between the LLJ and convection.

Despite the importance of the LLJ in advecting warm moist air into the frontal zone (Ninomiya 1984) for the development of cloud, deep convection did not occur atop the LLJ. Rather, convection usually occurred to the southeast of the upper-level jet and northwest of the LLJ (i.e., in an area between the upper- and low-level jets). The ample supply of moisture is tapped by a low-level transverse ageostrophic circulation (mesoinflow) that flows from the LLJ to the convective system. This inflow is originally generated by vertical convection and further enhanced by the slantwise convection. Subsequently, the development of this mesoinflow can lead to the intensification of the LLJ through geostrophic adjustment.

The creation of an unstable environment by symmetric instability is directly associated with the presence of the LLJ. This environment results from the flattening of the absolute momentum (M) surfaces (reducing inertial stability) and by advecting moisture into the vicinity of the frontal zone (drawing in high-θe air). Therefore, air can move freely along constant-M surfaces in the region between the LLJ and the convective system and becomes unstable because ∂θe/∂z < 0. Obviously, there is a positive feedback between the LLJ and CSI. Consequently, the LLJ, low-level mesoinflow and heavy precipitation will intensify as long as processes to maintain the LLJ and convection are not interrupted.

The dynamics of the upper-level jet was also found to be important. At the jet confluent entrance, the generation of an ageostrophic flow, moving from the warm side to the cold side of the jet, can be explained by thermal wind adjustment. It is speculated that the divergence of this ageostrophic flow on the warm side of the upper-level jet provides a triggering mechanism for the onset of underlying convection.

In contrast to conjectures made by Chen et al. (1994), our results indicate that the intensification of the lee cyclone is not necessary for the intensification of the LLJ because the lee cyclone in our study was neither strong nor intensifying. However, in our case, mesoscale cyclones are generated toward the later stages, and their generation is closely related to the development of mesoscale convective systems.

In summary, our study shows that the LLJ may form to the southeast and beneath the entrance of an upper-level jet. Deep convection occurs between the upper-level jet and the LLJ. The divergence of the ageostrophic wind associated with the upper-level jet appears to be an important trigger for the onset of deep convection. The intensification of the LLJ is caused by geostrophic adjustment of the mesoinflow, which is induced by the development of the mesoscale convective system. Furthermore, the generation of the slantwise convection and CSI are important factors that can strengthen the mesoinflow and lead to further intensification of the LLJ.

Acknowledgments

We are very grateful to three anonymous reviewers for their constructive comments and suggestions to improve this paper. We are also grateful to Sue Chen, Jimy Dudhia, and Bill Kuo (National Center for Atmospheric Research) for their support and help in making the MM5 model available for researchers at Goddard Space Flight Center (GSFC). Our sincere appreciation also goes to Yi-Leng Chen (University of Hawaii) and Craig Bishop (Pennsylvania State University) for their important contributions in this study. Special thanks also go to Stephen Lang and Dean Duffy for their reading and revising of the manuscript.

Chaing Chen, Wei-Kuo Tao, and George S. Lai are supported by NASA Headquarters Physical Climate Program, the Interdisciplinary Investigation of the Earth Observing System (EOS). We would like to show our appreciation to Drs. R. Kakar, R. Adler, and K. M. Lau for their support of this research. Additional acknowledgement is made to NASA/GSFC for computer resources used in this research.

REFERENCES

  • Akiyama, T., 1973: Frequent occurrence of the heavy rainfalls along the north side of the low-level jet stream. Pap. Meteor. Geophys.,24, 379–388.

  • Arakawa, A., and W. H. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large scale environment. Part I. J. Atmos. Sci.,31, 674–701.

  • Blackadar, A. K., 1957: Boundary layer wind maxima and their significance for the growth of nocturnal inversions. Bull. Amer. Meteor. Soc.,38, 283–290.

  • Bonner, W. D., 1968: Climatology of the low level jet. Mon. Wea. Rev.,96, 833–850.

  • Chang, C.-P., and G. T. J. Chen, 1995: Tropical circulations associated with southwest monsoon onset and westerly surges over the South China Sea. Mon. Wea. Rev.,123, 3254–3267.

  • Chen, C., C. H. Bishop, G. S. Lai, and W.-K. Tao, 1997: Numerical simulations of an observed narrow cold frontal rainband. Mon. Wea. Rev.,125, 1027–1045.

  • Chen, G. T. J., 1983: Observational aspects of the mei-yu phenomenon in subtropical China. J. Meteor. Soc. Japan,61, 306–312.

  • ——, and C. C. Yu, 1988: Study of low-level jet and extremely heavy rainfall over northern Taiwan in the mei-yu season. Mon. Wea. Rev.,116, 884–891.

  • Chen, Q., 1982: The instability of the gravity–inertial wave and its relation to low-level jet and heavy rainfall. J. Meteor. Soc. Japan,60, 1041–1057.

  • Chen, X. A., and Y.-L. Chen, 1995: Development of low-level jets during TAMEX. Mon. Wea. Rev.,123, 1695–1719.

  • Chen, Y.-L., X. A. Chen, and Y.-X. Zhang, 1994: A diagnostic study of the low-level jet during TAMEX IOP 5. Mon. Wea. Rev.,122, 2257–2284.

  • Chou, L. C., C. P. Chang, and R. T. Williams, 1990: A numerical simulation of the mei-yu front and the associated low level jet. Mon. Wea. Rev.,118, 1408–1428.

  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci.,46, 3077–3107.

  • ——, 1993: A nonhydrostatic version of the Penn State–NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev.,121, 1493–1513.

  • Emanuel, K. A., 1983: On assessing local conditional symmetric instability from atmospheric soundings. Mon. Wea. Rev.,111, 2016–2033.

  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev.,121, 764–787.

  • Hoskins, B. J., 1974: The role of potential vorticity in symmetric stability and instability. Quart. J. Roy. Meteor. Soc.,100, 480–482.

  • ——, I. Draghici, and H. C. Davies, 1978: A new look at the ω-equation. Quart. J. Roy. Meteor. Soc.,104, 31–38.

  • Hsu, W.-R., and W.-Y. Sun, 1994: A numerical study of a low-level jet and its accompanying secondary circulation in a mei-yu system. Mon. Wea. Rev.,122, 324–340.

  • Keyser, D., and M. J. Pecnick, 1985: A two-dimensional primitive equation model of frontogenesis forced by confluence and horizontal shear. J. Atmos. Sci.,42, 1259–1282.

  • ——, and M. A. Shapiro, 1986: A review of the structure and dynamics of upper-level frontal zones. Mon. Wea. Rev.,114, 452–499.

  • Matsumoto, S., 1972: Unbalanced low-level jet and solenoidal circulation associated with heavy rainfalls. J. Meteor. Soc. Japan,50, 194–203.

  • ——, 1973: Lower tropospheric wind speed and precipitation activity. J. Meteor. Soc. Japan,51, 101–107.

  • ——, K. Ninomiya, and S. Yoshizumi, 1971: Characteristic features of “baiu” front associated with heavy rainfall. J. Meteor. Soc. Japan,49, 267–281.

  • Ninomiya, K., 1984: Characteristics of baiu front as a predominant subtropical front in the summer Northern Hemisphere. J. Meteor. Soc. Japan,62, 880–894.

  • ——, and T. Akiyama, 1974: Band structure of mesoscale clusters associated with low-level jet stream. J. Meteor. Soc. Japan,52, 300–313.

  • Ray, P. S., Ed., 1986: Mesoscale Meteorology and Forecasting. Amer. Meteor. Soc., 793 pp.

  • Schlesinger, R. E., 1973: A numerical model of deep moist convection: Part I. Comparative experiments for variable ambient moisture and wind shear. J. Atmos. Sci.,30, 835–856.

  • Stauffer, D. R., and N. L. Seaman, 1990: Use of four-dimensional data assimilation in a limited-area mesoscale model. Part I: Experiments with synoptic-scale data. Mon. Wea. Rev.,118, 1250–1277.

  • Tao, S., and L. Chen, 1987: A review of recent research on the East Asian summer monsoon in China. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 60–92.

  • Tseng, S.-F., 1993: A case study of heavy rainfall in the mei-yu season (in Chinese). M. S. thesis, Department of Atmospheric Physics, National Central University, Chung-Li, Taiwan, 121 pp.

  • Uccellini, L. W., and D. R. Johnson, 1979: The coupling of upper and lower tropospheric jet streaks and implications for the development of severe convective storms. Mon. Wea. Rev.,107, 682–703.

  • Zhang, D.-L., and R. A. Anthes, 1982: A high-resolution model of the planetary boundary layer—Sensitivity tests and comparisons with SESAME-79 data. J. Appl. Meteor.,21, 1594–1609.

Fig. 1.
Fig. 1.

Contour analyses of 300-hPa geopotential (m, solid), wind vectors, and isotherms (K, thin dashed) for (a) 23 June 1991 and (b) 24 June 1991 (after Tseng 1993). The panels (c) and (d) show the corresponding analyses for 850 hPa. The isotherms at 300 hPa are for temperature, while those at 850 hPa are for equivalent potential temperature. The heavy dashed line represents the location of the trough.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 2.
Fig. 2.

Surface map analyses of surface pressure (hPa, solid) and temperature (K, dashed) for (a) 0000 UTC 23 June 1991 and (b) 0000 UTC 24 June 1991 (after Tseng 1993).

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 3.
Fig. 3.

Geostationary Meteorological Satellite (GMS) IR imagery at 2332 UTC 23 June 1991 (after Tseng 1993).

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 4.
Fig. 4.

Schematic diagram showing the second nested grid domain. The grid resolution in this domain is 45 km. The solid lines AB and CD represent the orientation of the xz cross sections for the following analyses.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 5.
Fig. 5.

Cross sections (xy) of the accumulated precipitation (cm, shaded) at the surface and contours of potential temperature θ (solid, every 1 K) at the height of 500 m (MSL). The time sequence of the plots is (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The time interval used to calculate the accumulation is shown in each panel. The dotted line represents the 301-K isentrope.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 6.
Fig. 6.

Vertical cross sections (xz) of θ (solid lines, every 4 K) and hydrometeor content (g kg−1, shaded) for 2400 UTC 23 June 1991 along the lines (a) AB and (b) CD. The horizontal distance between the tick marks represent a spatial scale of 45 km. The dotted line represents the 308-K isentrope.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 7.
Fig. 7.

Atmospheric sounding sampled at the intersection of lines AB and CD at 1200 UTC.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 8.
Fig. 8.

Vertical cross sections (xz) of ageostrophic wind vectors and hydrometeor content (g kg−1, shaded) along the line AB for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The horizontal spatial scale is the same as in Fig. 6a.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 9.
Fig. 9.

Vertical cross sections (xz) of ageostrophic wind vectors and normal wind speed (m s−1, shaded) along the line AB for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The character J denotes the LLJ position. The horizontal spatial scale is the same as in Fig. 6a.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 10.
Fig. 10.

Cross sections (xy) of the wind speed (m s−1, shaded) and streamlines for 1200 UTC 23 June 1991 at the height of z = (a) 13 km and (b) 10 km.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 11.
Fig. 11.

Same as in Fig. 8 except for the vertical velocity (every 2 m s−1, shaded) and overlayed cloud outline.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 12.
Fig. 12.

Same as in Fig. 10 except for the ageostrophic wind vectors and the divergence of the ageostrophic wind (s−1, shaded).

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 13.
Fig. 13.

Schematic diagram showing the flow structure of an observed mei-yu front (after Chen et al. 1994). The thin solid line depicts the direct (D) circulation while the thin dashed line depicts the indirect (I) circulation. The heavy solid line shows the frontal position. The character J denotes the jet positions. The thick heavy line represents the tropopause boundary. Regions with relative humidity greater than 70% are shaded.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 14.
Fig. 14.

Cross sections (xy) of the wind speed (m s−1, shaded) and streamlines at the height of z = 1.5 km for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991. The horizontal extent of the LLJ is represented by the dark shaded area with wind speeds greater than 20 m s−1.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 15.
Fig. 15.

Cross sections (xy) of the accumulated precipitation (cm, shaded) at the surface from 1200 to 2400 UTC 23 June 1991 for cases (a) NOPBL, (b) NOCU, (b) NOEXP, and (d) CTRL.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 16.
Fig. 16.

Same as in Fig. 14 except at 2400 UTC for cases (a) NOPBL, (b) NOCU, (b) NOEXP, and (d) NOCLD.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 17.
Fig. 17.

Same as in Fig. 9 except at 2400 UTC for cases (a) NOPBL, (b) NOCU, (b) NOEXP, and (d) NOCLD.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 18.
Fig. 18.

Cross section (xy) of wind speed (m s−1, solid line) and pressure (hPa, shaded) at z = 1.5 km showing the difference between two cases (CTRL minus NOCLD).

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 19.
Fig. 19.

Same as in Fig. 8 except for showing the difference between two cases (CTRL minus NOCLD).

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 20.
Fig. 20.

Conceptual schematic diagram illustrating the role of (a) the upper-level jet, and (b) the vertical and slantwise convection in the development of the LLJ. The heavy solid line with triangles shows the frontal position. The character J denotes the jet position, and the boldness of the J represents the jet strength. The thick heavy line represents the tropopause boundary. The dashed line depicts the weaker circulation, and the thin sold line represents the stronger circulation. Regions with cloud are shaded.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Fig. 21.
Fig. 21.

Vertical cross sections (xz) of M surfaces (m s−1, solid line), equivalent potential temperature (348 K, shaded), and cloud profile (thick solid line) along the line AB for (a) 1200, (b) 1600, (c) 2000, and (d) 2400 UTC 23 June 1991.

Citation: Monthly Weather Review 126, 2; 10.1175/1520-0493(1998)126<0349:TIOTLL>2.0.CO;2

Table 1.

Summary of numerical experiments.

Table 1.

1

An example of the vertical and horizontal structure of the LLJ can be seen in Figs. 9d and 14d, respectively.

2

The details of this scheme can be found in an NCAR technical note (NCAR/TN-398+1A) entitled “A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5).”

Save