• Brewer, A. W., 1949: Evidence for a world circulation provided by the measurements of helium and water vapour distribution in the stratosphere. Quart. J. Roy. Meteor. Soc.,75, 351–363.

  • Danielsen, E. F., 1982: A dehydration mechanism for the stratosphere. Geophys. Res. Lett.,9, 605–608.

  • ——, 1993: In situ evidence of rapid, vertical, irreversible transport of lower tropospheric air into the lower tropical stratosphere by convective cloud turrets and by larger-scale upwelling in tropical cyclones. J. Geophys. Res.,98, 8665–8682.

  • Dobson, G. M. B., 1956: Origin and distribution of the polyatomic molecules in the atmosphere. Proc. Roy. Soc. London,A236, 187–193.

  • Evans, M., and Coauthors, 2000: Evaluation of a Lagrangian box model using field measurements from EASE 1996. Atmos. Environ.,36, 3843–3863.

  • Evans, S. J., R. Toumi, J. E. Harries, M. P. Chipperfield, and J. M. Russell III, 1998: Trends in stratospheric humidity and the sensitivity of ozone to these trends. J. Geophys. Res.,103, 8715–8725.

  • Gage, K., and G. Reid, 1987: Longitudinal variations in tropical tropopause properties in relation to tropical convection and ENSO events. J. Geophys. Res.,92, 14 197–14 203.

  • Gibson, J. K, P. Kallberg, S. Uppala, A. Hernandez, A. Nomura, and E. Serrano, 1997: ECMWF Re-analysis Project Report Series, 1: ERA description. ECMWF, Reading, United Kingdom, 71 pp.

  • Highwood, E. J., and B. J. Hoskins, 1998: The tropical tropopause. Quart. J. Roy. Meteor. Soc.,124, 1579–1604.

  • Hodges, K. I., 1996: Spherical nonparametric estimators applied to the UGAMP model integration for AMIP. Mon. Wea. Rev.,124, 2914–2932.

  • Hoinka, K., 1998: Statistics of the global tropopause pressure. Mon. Wea. Rev.,126, 3303–3325.

  • ——, 1999: Temperature, humidity, and wind at the global tropopause. Mon. Wea. Rev.,127, 2248–2265.

  • Jackson, D. R., S. J. Driscoll, E. J. Highwood, J. E. Harries, and J. M. Russel III, 1998: Troposphere to stratosphere transport at low latitudes as studied using HALOE observations of water vapour 1992–1997. Quart. J. Roy. Meteor. Soc.,124, 169–192.

  • Kley, D., A. L. Schmeltekopf, K. Kelly, R. H. Winkler, T. L. Thompson, and M. McFarland, 1982: Transport of water through the tropical tropopause. Geophys. Res. Lett,9, 617–620.

  • ——, P. J. Crutzen, H. G. J. Smit, H. Vömel, S. J. Oltmans, H. Grassl, V. Ramanathan, 1996: Observations of near-zero ozone concentrations over the convective Pacific: Effects on air chemistry. Science,274, 230–233.

  • Kousky, V. E., and A. Leetmaa, 1989: The 1986/87 Pacific warm episode: Evolution of oceanic and atmospheric anomaly fields. J. Climate,2, 254–267.

  • Kritz, M., S. Rosner, K. Kelly, M. Lowenstein, and K. Chan, 1993: Radon measurements in the lower tropical stratosphere: Evidence for rapid vertical transport and dehydration of tropospheric air. J. Geophys. Res.,98, 8725–8736.

  • Mastenbrook, H. J., 1974: Water vapor measurements in the lower stratosphere. Can. J. Chem.,52, 1527–1531.

  • Methven, J., 1997: Offline trajectories: Calculation and accuracy. UGAMP Tech. Rep. 44, 18 pp. [Available from CGAM, University of Reading, Earley Gate, Reading RG6 6BB, United Kingdom.].

  • ——, P. Berrisford, and B. J. Hoskins, 1999: A Lagrangian climatology for the North Atlantic storm track. UKMO Hadley Centre Tech. Note 9, 99 pp. [available from Hadley Centre for Climate Prediction and Research, The Met. Office, London Road, Bracknell, Berkshire RG12 2SY, United Kingdom.].

  • Mote, P. W., K. H. Rosenlof, J. R. Holton, R. S. Harwood, and J. W. Waters, 1995: Seasonal variations of water vapor in the tropical lower stratosphere. Geophys. Res. Lett.,22, 1093–1096.

  • ——, and Coauthors, 1996: An atmospheric tape recorder: The imprint of tropical tropopause temperatures on stratospheric water vapour. J. Geophys. Res.,101, 3989–4006.

  • Newell, R. E., and S. Gould-Stewart, 1981: A stratospheric fountain? J. Atmos. Sci.,38, 2789–2796.

  • Pawson, S., and M. Fiorino, 1998: A comparison of reanalyses in the tropical stratosphere. Part 1: Thermal structure and the annual cycle. Climate Dyn.,14, 631–644.

  • Pickering, K. E., and Coauthors, 1996: Convective-transport of biomass burning emissions over Brazil during TRACE-A. J. Geophys, Res.,101, 23 993–24 012.

  • Reid, G. C., and K. S. Gage, 1996: The tropical tropopause over the western Pacific: Wave driving, convection and the annual cycle. J. Geophys. Res.,101, 21 233–21 241.

  • Russell, P. B., L. Pfister, and H. B. Selkirk, 1993: The tropical experiment of the Stratosphere–Troposphere Exchange Project (STEP): Science objectives, operations, and summary findings. J. Geophys. Res.,98, 8563–8589.

  • Schumann, U., 1994: On the effect of emissions from aircraft engines on the state of the atmosphere. Ann. Geophys.,12, 365–384.

  • Selkirk, H. B., 1993: The tropopause cold trap during STEP/AMAX 1987. J. Geophys. Res.,98, 8591–8610.

  • Simmons, A. J., A. Untch, C. Jakob, P. Kallberg, and P. Unden, 1999:Stratospheric water vapour and tropical tropopause temperatures in ECMWF analyses and multi-year simulations. Quart. J. Roy. Meteor. Soc.,125, 353–386.

  • Sparling, L. C., J. A. Kettleborough, P. H. Haynes, M. E. McIntyre, J. E. Rosenfield, M. R. Schoeberl, and P. A. Newman, 1997: Diabatic cross-isentropic dispersion in the lower stratosphere. J. Geophys. Res.,102, 25 817–25 829.

  • Thuburn, T., and G. Craig, 1997: The height of the tropopause. J. Atmos. Sci.,54, 869–882.

  • Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev.,117, 1779–1800.

  • Tompkins, A. M., and G. C. Craig, 1998: Radiative–convective equilibrium in a three-dimensional cloud-ensemble model. Quart. J. Roy. Meteor. Soc.,124, 2073–2097.

  • Toumi, R., J. D. Haigh, and K. S. Law, 1996: Tropospheric ozone-lightning feedback. Geophys. Res. Lett.,23, 1037–1040.

  • Webster, S., J. Thuburn, B. Hoskins, and M. Rodwell, 1999: Further development of a hybrid-isentropic GCM. Quart. J. Roy. Meteor. Soc.,125, 2305–2331.

  • Wong, T., G. L. Stephens, P. W. Stackhouse, and F. P. J. Valero, 1993:The radiative budgets of a tropical mesoscale convective system during EMEX-STEP-AMEX experiment, 1: Observations. J. Geophys. Res.,98, 8683–8694.

  • World Meteorological Organization, 1998: Scientific assessment of ozone depletion: 1998. World Meteorological Organization Global Ozone Research and Monitoring Project Rep. 44, 669 pp.

  • View in gallery

    The release positions of air particles advected by the trajectory model.

  • View in gallery

    ERA particles released over central Pacific from the 150-hPa level, 10-yr mean of DJF. (a) Mean pressure 5 days after release; (b) particle density 5 days after release. Contour interval: 10 hPa for (a) and 0.1 for (b). The box indicates the region from where the particles were released.

  • View in gallery

    As in Fig. 2, except that locations of particles released over Indonesia are shown.

  • View in gallery

    As in Fig. 2, except that locations of particles released over South America are shown.

  • View in gallery

    As in Fig. 2, except that locations of particles released over Africa are shown.

  • View in gallery

    Particle density 5 days after release over the central Pacific from the 150-hPa level for DJF 1982/83. Contour interval: 0.1.

  • View in gallery

    (a) Density overlap for particles 5 days after release; (b) mean pressure (weighted by particle density) for particles 5 days after release. Solid line: Indonesian particles; dotted line: central Pacific particles; dashed line: South American particles; dashed–dotted line: African particles. Years are indicated by “80” = DJF 1980/81, “82” = DJF 1982/83, etc.

  • View in gallery

    Histogram of time after release at which the lapse rate tropopause is crossed, for run P150_FORW. Only crossing events between 20°S and 20°N are included. Bin size is 1 day. (a) Indonesian particles. Sample size is 1795. (b) Central Pacific particles. Sample size is 1316. (c) African particles. Sample size is 498. (d) South American particles. Sample size is 510.

  • View in gallery

    As in Fig. 8, except only central Pacific particles that cross the tropopause north of 20°N or south of 20°S are included. Sample size is 825.

  • View in gallery

    Densities of tropopause crossing events for run P150_FORW, DJF 1985/86. (a) Indonesian particles; (b) central Pacific particles; (c) African particles; (d) South American particles. Contour interval: 0.1.

  • View in gallery

    As in Fig. 10, except fields for run TH_FORW are shown.

  • View in gallery

    As in Fig. 10, except fields for run TH_BACK are shown. The contour interval in (a) is 0.2.

  • View in gallery

    Histogram of central Pacific particles for DJF 1985/86. Bin size is 2.0 K. (a) For run P150_FORW at time of release (thin line), 10 days after release (medium thin line), 20 days after release (medium thick line), and 30 days after release (thick line); (b) for run TH_FORW 10 days after release (thin line), 20 days after release (medium line), 30 days after release (thick line). The dotted lines indicate the tropopause zone used in runs TH_FORW and TH_BACK.

  • View in gallery

    Histogram of particles for run TH_BACK for DJF 1985/86 10 days before arrival at the release point (thin line), 20 days before arrival (medium line), and 30 days before arrival (thick line). Bin size is 2.0 K. (a) Indonesia; (b) South America. The dotted lines indicate the tropopause zone.

  • View in gallery

    As in Fig. 10, except densities for run TH_BACK, DJF 1982/83, are shown. (a) Indonesian particles; (b) South American particles. The contour interval is 0.2.

  • View in gallery

    As in Fig. 14, except particles for DJF 1982/83 are shown. (a) Indonesia; (b) South America.

  • View in gallery

    As in Fig. 10, except densities for run TH_BACK, DJF 1988/89, are shown. (a) Indonesian particles; (b) African particles. The contour interval is 0.2.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 13 13 6
PDF Downloads 1 1 0

Transport in the Low-Latitude Tropopause Zone Diagnosed Using Particle Trajectories

View More View Less
  • 1 The Met. Office, Bracknell, Berkshire, United Kingdom
  • 2 Department of Meteorology, University of Reading, Reading, Berkshire, United Kingdom
  • 3 The Met. Office, Bracknell, Berkshire, United Kingdom
© Get Permissions
Full access

Abstract

Recent literature has described a “transition zone” between the average top of deep convection in the Tropics and the stratosphere. Here transport across this zone is investigated using an offline trajectory model. Particles were advected by the resolved winds from the European Centre for Medium-Range Weather Forecasts reanalyses. For each boreal winter clusters of particles were released in the upper troposphere over the four main regions of tropical deep convection (Indonesia, central Pacific, South America, and Africa). Most particles remain in the troposphere, descending on average for every cluster. The horizontal components of 5-day trajectories are strongly influenced by the El Niño–Southern Oscillation (ENSO), but the Lagrangian average descent does not have a clear ENSO signature.

Tropopause crossing locations are first identified by recording events when trajectories from the same release regions cross the World Meteorological Organization lapse rate tropopause. Most crossing events occur 5–15 days after release, and 30-day trajectories are sufficiently long to estimate crossing number densities. In a further two experiments slight excursions across the lapse rate tropopause are differentiated from the drift deeper into the stratosphere by defining the “tropopause zone” as a layer bounded by the average potential temperature of the lapse rate tropopause and the profile temperature minimum. Transport upward across this zone is studied using forward trajectories released from the lower bound and back trajectories arriving at the upper bound. Histograms of particle potential temperature (θ) show marked differences between the transition zone, where there is a slow spread in θ values about a peak that shifts slowly upward, and the troposphere below 350 K. There forward trajectories experience slow radiative cooling interspersed with bursts of convective heating resulting in a well-mixed distribution. In contrast θ histograms for back trajectories arriving in the stratosphere have two distinct peaks just above 300 and 350 K, indicating the sharp change from rapid convective heating in the well-mixed troposphere to slow ascent in the transition zone. Although trajectories slowly cross the tropopause zone throughout the Tropics, all three experiments show that most trajectories reaching the stratosphere from the lower troposphere within 30 days do so over the west Pacific warm pool. This preferred location moves about 30°–50° farther east in an El Niño year (1982/83) and about 30° farther west in a La Niña year (1988/89). These results could have important implications for upper-troposphere–lower-stratosphere pollution and chemistry studies.

Corresponding author address: Dr. David Jackson, Middle Atmosphere Group, The Met. Office, Room 251, London Road, Bracknell, Berkshire RG12 2SZ, United Kingdom.

Email: drjackson@meto.gov.uk

Abstract

Recent literature has described a “transition zone” between the average top of deep convection in the Tropics and the stratosphere. Here transport across this zone is investigated using an offline trajectory model. Particles were advected by the resolved winds from the European Centre for Medium-Range Weather Forecasts reanalyses. For each boreal winter clusters of particles were released in the upper troposphere over the four main regions of tropical deep convection (Indonesia, central Pacific, South America, and Africa). Most particles remain in the troposphere, descending on average for every cluster. The horizontal components of 5-day trajectories are strongly influenced by the El Niño–Southern Oscillation (ENSO), but the Lagrangian average descent does not have a clear ENSO signature.

Tropopause crossing locations are first identified by recording events when trajectories from the same release regions cross the World Meteorological Organization lapse rate tropopause. Most crossing events occur 5–15 days after release, and 30-day trajectories are sufficiently long to estimate crossing number densities. In a further two experiments slight excursions across the lapse rate tropopause are differentiated from the drift deeper into the stratosphere by defining the “tropopause zone” as a layer bounded by the average potential temperature of the lapse rate tropopause and the profile temperature minimum. Transport upward across this zone is studied using forward trajectories released from the lower bound and back trajectories arriving at the upper bound. Histograms of particle potential temperature (θ) show marked differences between the transition zone, where there is a slow spread in θ values about a peak that shifts slowly upward, and the troposphere below 350 K. There forward trajectories experience slow radiative cooling interspersed with bursts of convective heating resulting in a well-mixed distribution. In contrast θ histograms for back trajectories arriving in the stratosphere have two distinct peaks just above 300 and 350 K, indicating the sharp change from rapid convective heating in the well-mixed troposphere to slow ascent in the transition zone. Although trajectories slowly cross the tropopause zone throughout the Tropics, all three experiments show that most trajectories reaching the stratosphere from the lower troposphere within 30 days do so over the west Pacific warm pool. This preferred location moves about 30°–50° farther east in an El Niño year (1982/83) and about 30° farther west in a La Niña year (1988/89). These results could have important implications for upper-troposphere–lower-stratosphere pollution and chemistry studies.

Corresponding author address: Dr. David Jackson, Middle Atmosphere Group, The Met. Office, Room 251, London Road, Bracknell, Berkshire RG12 2SZ, United Kingdom.

Email: drjackson@meto.gov.uk

1. Introduction

It is important to characterize stratosphere–troposphere exchange for a variety of reasons, not least for the proper representation of atmospheric constituents in the upper troposphere and lower stratosphere. In the upper troposphere, radiation from water vapor contributes significantly to the planetary radiation balance, and water vapor provides the principal source of OH radical in the stratosphere. Upper-tropospheric ozone is also an important greenhouse gas, and the issue of stratospheric ozone depletion, and the potential recovery of stratospheric ozone to their 1979 values, is a major area of research (see, e.g., World Meteorological Organization 1998). Furthermore, changes to ozone in the lower stratosphere can have a large impact on the surface ultraviolet flux. Global change and man’s activities can significantly affect the budgets of many constituents. For example, the budget of ozone in the upper troposphere and lower stratosphere can be greatly affected by subsonic aircraft emissions (e.g., Schumann 1994), biomass burning (e.g., Pickering et al. 1996) and the sensitivity of lightning to surface temperature (e.g., Toumi et al. 1996). In addition, recent analysis of Halogen Occultation Experiment observations of water vapor has confirmed that a global hydration of the stratosphere is occurring (Evans et al. 1998). Reasons for this hydration are unclear, though it could be related to the possible warming of the tropical tropopause. However, despite the large impact that changes to ozone and water vapor may have on climate, the upper-troposphere–lower-stratosphere zone is poorly understood. Accordingly, a better understanding of this zone, and in particular stratosphere–troposphere exchange, is a prerequisite for understanding the observed trend in these gases, and for understanding the impact that factors such as biomass burning and aircraft emissions will have on the future budgets of these gases.

The prevailing model of mean mass circulation in the tropical stratosphere, based on the work of Brewer (1949) and Dobson (1956), postulates a mean rising motion of air across the tropical tropopause, poleward drift, and slow descent in the extratropics. The tropical tropopause acts as a cold trap to freeze-dry air in this passage from troposphere to stratosphere. The strong influence of the temperature of the tropical tropopause on the distribution of water vapor in the lower stratosphere has guided much of the research into low-latitude troposphere to stratosphere exchange. Mastenbrook (1974) showed that air with lower water vapor mixing ratios than could be explained by the zonal mean tropical tropopause temperature was present in the lower stratosphere. This prompted Newell and Gould-Stewart (1981) to propose the hypothesis that air enters the stratosphere only in restricted areas and at certain times of the year. These regions, termed “stratospheric fountains,” were said to be located over the Indonesian region between November and March and over the Bay of Bengal region during boreal summer. Newell and Gould-Stewart suggested that a slow mean ascent of 5 mm s−1 would bring sufficient amounts of dry air (defined by them as air with a water vapor mixing ratio less than 3.5 ppmv) into the lower stratosphere. However, recent research has questioned the temporal and spatial restrictions of the stratospheric fountain. Satellite observations of water vapor have shown that air enters the stratosphere during the entire year and that there is a near-annual cycle in the zonal mean water vapor distribution in the equatorial stratosphere, which is linked to the annual cycle in tropical tropopause temperatures (Mote et al. 1995, 1996), and that there is evidence of dry air entering the low-latitude lower stratosphere in both March–April–May and September–October–November, which is contrary to Newell and Gould-Stewart’s findings (Jackson et al. 1998).

Other research has questioned Newell and Gould-Stewart’s assertion that slow large-scale ascent is the means by which air enters the lower stratosphere. For example, Kley et al.(1982) and Danielsen (1982) suggest the important role that tropopause-penetrating deep convective clouds, with associated rapid ascent rates, can play in this transport. Most tropical deep convection extends to the level of maximum outflow in the Hadley cell (about 200 hPa), and the top of the convective outflow is at about 150 hPa. However, the tropical tropopause, as determined by the World Meteorological Organization (WMO) lapse rate definition, is located at around 100 hPa. The minimum in tropical temperature profiles (the cold point) lies higher still at around 80–90 hPa. Most deep convection that reaches the 150–100-hPa zone or above is sporadic. Thus, recent work has introduced the idea of a substratospheric “transition zone” between the top of the Hadley cell and the lower-stratospheric Brewer–Dobson circulation, rather than a tropopause surface (e.g., Thuburn and Craig 1997; Highwood and Hoskins 1998).

In this study we investigate transport near the low-latitude tropopause using an offline trajectory model. Particles are initialized over regions of deep tropical convection and are then advected by winds from the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalyses (ERA) (Gibson et al. 1997). The aim is to simulate the statistics of trajectory destinations with particular emphasis on crossing events into the stratosphere. The flow is 3D so that the vertical motion following the trajectories has an influence on the horizontal distribution of trajectory end points and moreover the wind field is unsteady so the trajectories do not follow streamlines. Even the Lagrangian average end point would be difficult to infer quantitatively from the time mean winds. Furthermore, many trajectories from the top of the convective region remain within the troposphere so that it is not possible to infer regions where trajectories cross into the stratosphere from their Lagrangian average. In order to relate crossing events with trajectory origins, it is necessary to perform trajectory calculations. Such events necessarily involve cross-isentropic flow and the statistics of heating along trajectories will be used to characterize the transport. For example, Sparling et al. (1997) described diabatic dispersion in the stratosphere and showed that the behavior on timescales longer than a month resembles diffusion across isentropic surfaces. Can heating histories and temperatures particular to trajectories crossing the tropopause zone be identified?

Forward trajectories from 150 hPa are calculated for 10 boreal winter seasons. The climatological average of the destinations of all the particles are presented in section 2, together with a description of interannual variations, which are principally associated with El Niño and La Niña events. In section 3 the cross-tropopause transport is diagnosed in detail using two definitions of the tropopause: the WMO definition, which is based on lapse rate, and the minimum temperature definition, which is thought to be more physically meaningful in the Tropics (Highwood and Hoskins 1998). In the first experiment, transport from 150 hPa across the lapse rate–defined tropopause surface is examined. In a further two experiments the lapse rate and minimum temperature definitions are used to specify a “tropopause zone” (the upper portion of the “transition zone” described above). Transport across this zone is then investigated for a wide range of equatorial longitudes, enhancing the findings of observational campaigns, such as the Tropical Experiment of the Stratosphere–Troposphere Exchange Project (STEP) (Russell et al. 1993 and references therein), which focused on smaller geographical areas. Through this representation of the geographical distribution of cross-tropopause transport the spatial restrictions of Newell and Gould-Stewart’s stratospheric fountain can be reevaluated.

ERA is a high quality global dataset, but it is nevertheless important to assess how well it relates to the real atmosphere, and to discuss the implications for our results. This is done in section 4 and conclusions appear in section 5.

2. Description of the trajectory experiments with ERA winds

The trajectory scheme used here is described by Methven (1997) and has already proved to be a highly effective tool for the investigation of transport in the troposphere (Evans et al. 2000; Methven et al. 1999). It interpolates three-dimensional gridded winds to the trajectory locations using bilinear interpolation (in time and in the horizontal) and cubic Lagrange interpolation (in the vertical), and then the trajectories are integrated forward using a 4th-order Runge–Kutta scheme with a 0.6-h step size. Here, we show the mean of the pressure level that the particles reach 5 days after release. In addition, the geographical locations of the particles after 5 days are summarized using particle densities. The particle densities are calculated using a kernel estimation method (Hodges 1996), which has the advantages over grid box counting methods that there is no area bias due to box definition and that the estimation is much more robust and smooth than “binning.” The procedure is similar to finding a weighted average of a set of points, where a larger kernel spread is used where uncertainty is greater. The density function is normalized so that its integral over the sphere equals one.

Six-hourly ERA winds were used to drive the trajectory scheme. These fields are available on a 2.5° × 2.5° horizontal grid at 17 levels from 1000 to 10 hPa. Since the trajectories are calculated from the grid-scale winds, and do not take account of subgrid-scale transport, they are not the same as air parcel trajectories. Hence, throughout the paper we refer to them as particle, rather than parcel, trajectories. A discussion of the implications of these differences appears in section 4. Experiments were carried out for 10 December–January–February (DJF) seasons between 1979/80 and 1988/89. For each season, particles were released every 12 h from the 150-hPa level from the four regions shown in Fig. 1, namely, South America, Africa, Indonesia–western Pacific, and central Pacific. Within each region, the particles are spaced every 4° in latitude and longitude. These regions correspond to the regions of strongest equatorial convection in DJF. The trajectory for each particle was then calculated forward in time for 5 days. In section 2a we show particle locations averaged over 10 yr, while in section 2b interannual variability is discussed.

a. Climatology of particle destinations

Figure 2 shows mean pressure and normalized particle density for 5 days after release, for particles released over the central Pacific. To save computer time, only particles released at 24-h intervals, at 1200 UTC, were used to calculate these fields. Tests (see later) imply no detectable change in the results when the particle release interval is increased from 12 to 24 h. Many of the particles are rapidly transported eastward by the Walker circulation, as indicated by the maximum in the density distribution. However, the spread in the density is wide, highlighting the variability in trajectories from different days or points within the release area. This spread could not be attained from advection by the time mean winds, although some spreading would result from advection by steady 3D winds due to the size of the release area relative to typical horizontal scales in the wind field. Those trajectories ending over the western Pacific (the central and eastern Pacific) have ascended (descended) on average. After 2 days many of the particles have already reached the equatorial eastern Pacific (not shown here), and after 5 days they are transported poleward to the subtropics where they subside, reaching mean pressures up to 180 hPa.

The mean pressure and particle density fields for the Indonesian particles are shown in Fig. 3. Much of the initial transport is characteristic of the Hadley circulation, namely, low-latitude ascent, poleward motion, and subtropical descent. The eastward velocity of the subtropical jets results in a C-shaped density distribution centered on the equator for 2-day trajectories (not shown here). After 5 days, the density functions are nearly ring-shaped, since air that gets transported the fastest to the subtropics is transported eastward by the subtropical jets then equatorward again by the convergent meridional winds in the equatorial central Pacific. For particles that remain close to the Indonesian region, mean particle pressures are generally 130–140 hPa, but are in the 120–130-hPa range in some places, indicating the influence of regions of strong convection on the resolved vertical velocity. Note the asymmetry between hemispheres associated with the stronger subtropical jet in winter.

The transport of particles away from the South American and African release regions has a broadly similar pattern to that over Indonesia, except that the transport is less rapid. Five days after release, many South American particles are still located over the convective region, but other particles have been advected away from the continent, either by the weak tropical Atlantic westerlies or by the stronger subtropical westerlies (Fig. 4). Mean particle pressures in the convective region are 130–150 hPa. Five days after release many of the African particles reach the subtropics, and some undergo further advection eastward by the subtropical jets (Fig. 5). Mean particle pressures over the convective region are typically in the 140–160-hPa range.

b. Interannual variability

The 10 DJF seasons for which the trajectories were calculated include 2 El Niño years (1982/83 and 1986/87) and 1 year where there is a strong La Niña (1988/89). In El Niño years the Walker circulation in the equatorial Pacific moves eastward, with the ascending branch being located near the date line. The upper-tropospheric flow is less westerly in the equatorial central Pacific and anomalously poleward (equatorward) in the central (eastern) Pacific. The impact of these anomalies is seen by comparing the particle density 5 days after release from the central Pacific for 1982/83 (Fig. 6) with the 10-yr mean densities (Fig. 2). The effect of the La Niña of 1988/89 is to strengthen the Walker circulation, and at the 150-hPa level there are stronger westerlies in the equatorial central Pacific and stronger equatorward (poleward) winds in the central (eastern) equatorial Pacific in 1988/89 compared to the 10-yr mean field.

It would be possible to examine plots similar to Fig. 6 to describe the impact of El Niño and La Niña in other years and for other initial particle locations, but it is more concise to identify years with anomalous transport by the use of density overlaps and mean pressures weighted by particle density. The density overlap is an integral measure, over solid angle on the earth’s sphere Ω, that varies between zero, when there is no overlap of a particle density field and a control run density (P and Pc, respectively), and one, when the densities from the two runs are identical. It is defined by the equation
i1520-0469-58-2-173-e1

Subsampling experiments were carried out to verify the impact on the overlap measure of changes to the locations and frequency of particle release. Four such experiments were carried out for the DJF 1981/82 season: “2Δt” and “4Δt,” where particles were released every 2 and every 4 days, respectively, instead of every day; “compact,” where particles were only released from the central quarter of each cluster’s area; and “sparse,” where particles were only released at every other latitude and longitude point shown in Fig. 1.

For the 2Δt,t, and sparse experiments the density overlaps for 5 days after release were greater than or equal to 0.90 for all clusters. Therefore, if the density for a particular year has an overlap with the 10-yr climatology that is less than 0.9 it is possible to say that it is more different from the climatology than would be possible from uncertainties associated with sampling errors alone. For the compact particles, however, the density overlaps are smaller, being between 0.73 and 0.79. This indicates that the flow within the cluster regions shown in Fig. 1 has considerable spatial variations.

In the following investigation of the interannual variability, P and Pc are the densities from the individual years and the 10-yr means, respectively. The mean pressure weighted by particle density is defined by the integral
i1520-0469-58-2-173-e2
where p is the pressure at each particle’s location for the same interval after release as used to calculate the particle density, P.

Figure 7 shows the density overlap between the 10-yr mean and individual years and the mean pressure (weighted by particle density) for 5 days after release. The density overlap highlights the anomalous horizontal transport of both central Pacific and Indonesian particles in El Niño and La Niña years. For these years, the anomalies for South American and African particles are usually considerably smaller than for the central Pacific and Indonesian particles. An exception is 1982/83, where the anomaly for the South American particles is about the same as for the central Pacific particles. This is because of an eastward shift of the upper-tropospheric circulation over South America that is associated with the eastward shift of the Walker circulation. A comparison of the two El Niño years shows 1982/83 to be more anomalous than 1986/87. This is because the 1982/83 El Niño event was considerably stronger than the one in 1986/87, and on top of that the largest equatorial Pacific sea surface temperature anomalies for the 1986/87 event were observed in July 1987, whereas for the 1982/83 event they were observed in DJF 1982/83 (see Kousky and Leetma 1989).

The mean pressure weighted by particle density is generally greatest for the Indonesian particles. This is because many of the Indonesian particles are rapidly transported north and then subside in the northern subtropics close to the date line (see Fig. 3). Signatures of El Niño include the very large mean pressure over Indonesia in 1986/87 and the small mean pressure over South America in 1982/83. However, the mean pressures show clear year-to-year variability in non–El Niño and non–La Niña years as well. The strong El Niño and La Niña signals in the particle density overlaps indicate the high sensitivity of the horizontal component of trajectories to changes in the Hadley and Walker circulations in El Niño and La Niña years. The weaker signal in the pressure weighted by particle density (the El Niño and La Niña signals are within the range of interannual variability) probably arises because the particles are released near the top of large-scale ascent regions. Five days after release, particles tend to have descended in the tropospheric circulation on average, irrespective of their detailed path in the horizontal.

3. Cross tropopause transport

a. Trajectories from 150 hPa over convective regions

In this section the particle trajectories shown in section 2 are examined for evidence of cross-tropopause transport. The results from the 5-day trajectory experiments provide an indication of transport in the tropical upper troposphere. Although the mean pressures reached by trajectories are always higher than the tropopause pressure, the ERA vertical velocity fields indicate that at least some of the particles released at 150 hPa will reach the tropopause within 5 days. However, the ascent is slow and it is possible that the number of particles crossing the tropopause will increase if the length of the trajectories is also increased. Therefore, it is necessary to extend the length of the trajectories in order to determine the distribution of transit times from the 150-hPa release level to the tropopause. The original experiment is repeated for DJF 1985/86, but using 30-day trajectories instead of 5-day ones (run P150_FORW). The period 1985/86 was chosen because the density overlaps in that year are large (see Fig. 7) and it can thus be considered to be a “typical,” or nonanomalous, year.

It is first necessary to decide which definition of tropical tropopause should be used in run P150_FORW. Two definitions that are widely used are those based on the lapse rate (the WMO definition) and on minimum temperature. The ERA data have a relatively low vertical resolution, with pressure levels near the tropical tropopause only available at 70, 100, and 150 hPa, and since the tropical tropopause will usually lie between these levels, some interpolation or extrapolation of the data is desirable. The lapse rate tropopause is found by calculating lapse rate on the ERA levels and linearly interpolating lapse rate onto intermediate levels at 5-hPa spacing in order to facilitate the search for the lowest level at which the average lapse rate between this level and all higher levels within 2 km does not exceed 2 K km−1. The temperature minimum tropopause is calculated in two steps: the ERA level with lowest temperature is found, and the temperature profiles for the two points above and the two points below are extrapolated and their intersection sought. These methods serve to provide two surfaces with average levels in accord with temperature profiles from radiosonde ascents, but the minimum temperature tropopause was sometimes erroneously diagnosed as a result of the occasional failure of the intersection method, particularly when the lower stratosphere is nearly isothermal. Accordingly, in run P150_FORW the lapse rate definition of the tropopause is used. The WMO lapse rate tropopause does not have a clear physical significance, as will be discussed in the next section, and some particles fluctuate across the surface (possibly because the lapse rate tropopause just grazes the top of the tropospheric circulations). In order to eliminate most short term excursions into the stratosphere, crossing events are defined as those times when a trajectory enters the stratosphere and stays there for the remainder of the 30-day trajectory.

Figure 8 shows the distribution of the number of particles crossing the lapse rate tropopause in run P150_FORW between 20°S and 20°N against time after release. It can be seen that the number of particles crossing the tropopause increases for the first 5–10 days (Indonesian and central Pacific particles), or for the first 15–20 days (South American and African particles), then decreases with time afterward. An interpretation of these results is that particles leaving the top of convective regions either stay within the tropospheric Walker and Hadley circulations (the majority of particles as seen in the average pressure picture of section 2) or drift upward across the tropopause. During the first 10 days after release, the majority of particles drifting upward from 150 hPa have had sufficient time to reach the tropopause. The flux of particles across the tropopause does not stop because some trajectories will make excursions throughout the troposphere and return to the tropopause zone. These results indicate that 5-day trajectories are not long enough to study the favored crossing locations for particles leaving the tops of the convective regions, but that 30-day trajectories will represent most of this transport, especially for particles released over Indonesia and the central Pacific. Outside the 20°S–20°N band the number of particles that cross the tropopause increases with time up to around days 20–25. Figure 9 shows this distribution for the central Pacific particles, but corresponding plots for the other three clusters are broadly similar. Particle densities of the locations of tropopause crossing (see later) show that many of the extratropical crossing locations are situated in the storm track regions, and thus 20–25 days may indicate typical transport times to there from the tropical upper troposphere. A more complete representation of extratropical cross-tropopause transport of particles released from the tropical upper troposphere would require the trajectories to be run for longer than 30 days. However, this is beyond the scope of this work.

Figure 10 shows the density of points where trajectories cross the lapse rate tropopause for all four clusters in run P150_FORW. Many of the Indonesian particles cross into the stratosphere over the western Pacific region (Fig. 10a), and the corresponding plots for the other release regions also show some cross-tropopause transport there. Similar plots to Figs. 10a and 10b (central Pacific cluster), but for 15-day trajectories, show that in the first 15 days after release most of the cross-tropopause transport takes place over the western Pacific. For the trajectories released from the central Pacific, a significant amount of cross-tropopause transport in the first 15 days also takes place in the southern subtropics east of about 180° over the South Pacific convergence zone (SPCZ). This crossing region does not stand out for 30-day trajectories because later crossing events occur farther from the release region and some of the trajectories diagnosed as “crossing” at 15 days may have returned to the troposphere subsequently. Most of the cross-tropopause transport of the African and South American particles (Figs. 10c and 10d, respectively) occurs at locations distant from the release points, chiefly at other tropical longitudes and at the edge of the southern storm track region where the tropopause slopes down toward the Pole. The subtropical crossing events are likely to be associated with quasi-isentropic motion across the tropopause and chiefly take place 15–30 days after release.

The possibility of the return of trajectories to the troposphere soon after crossing into the stratosphere is high when describing the tropopause as a single surface. Although many of these events have been ruled out by requiring that the trajectories remain in the stratosphere for the remainder of the 30 days, it is possible that a number of these trajectories would return if they were doubled in length, for example. The use of a single surface that undulates in time to diagnose crossing events may highlight transient transport rather than movement of particles into the stratospheric residual circulation. In the following section this problem is addressed by defining a tropopause zone and examining transport across the zone.

b. The tropopause as a transition zone

Tropopause crossing events and the location of those events were diagnosed in run P150_FORW assuming that the tropopause is a surface, whereas it may be better considered to be a tropopause zone with finite thickness (e.g., Thuburn and Craig 1997; Highwood and Hoskins 1998). This viewpoint is particularly apt when using the ERA dataset, which has insufficient vertical resolution to adequately resolve a tropopause surface.

It has been suggested (e.g., Gage and Reid 1987) that the lapse rate definition of the tropopause (used in run P150_FORW) is rather arbitrary and has limited physical meaning. Another possible definition for the tropopause is the temperature minimum (T-min), or cold point. Highwood and Hoskins (1998) concluded that this is a more appropriate definition for the tropopause in the deep Tropics, and it has been deemed important for stratosphere–troposphere exchange by, for example, Selkirk (1993), at least partially because the potential temperature at the cold point is negatively correlated, in both space and time, with mesoscale convection below. As shown in Fig. 13 of Highwood and Hoskins (1998), there is a transition zone between the top of the average convective outflow (around 150 hPa) and the lower stratosphere, where the Brewer–Dobson circulation predominates. The temperature minimum located at around 80–90 hPa and the lapse rate tropopause at 95–105 hPa divide the transition zone into two: the zone between the average top of convection and the lapse rate tropopause, and a layer above up to the temperature minimum. Here the upper layer will be described as the tropopause zone.

The tropopause zone is defined using seasonal mean potential temperature values averaged over each cluster. The reasoning for this is as follows.

First, the spatial variation in seasonal mean tropopause height, pressure, and potential temperature (for both lapse rate and minimum temperature definitions) within each cluster shown in Fig. 1 is small. Maps such as those shown by Hoinka (1998, 1999) show that the changes in tropopause θ within each cluster region are usually less than 5 K, which is small compared to the 40–45 K drop in tropopause θ seen between the subtropics and midlatitudes. There are only small-amplitude variations across the Tropics; the main zonal anomaly is over the western Pacific warm pool where temperature and potential temperature are lower (Δθ ≈ −5 K). Highwood and Hoskins (1998) showed that the large-scale dynamical response to convective heating would involve a negative tropopause θ anomaly of this magnitude directly above the heating.

Second, the daily fluctuation in the lapse rate tropopause altitude is small in ERA compared to the regional anomalies (Δθ ≈ 1 K). The daily fluctuation in the T-min tropopause is larger and may result from the difficulty of diagnosing its altitude with coarse vertical resolution. In order to circumvent erroneous fluctuations in the height of the diagnosed minimum temperature tropopause, we define the tropopause zone using seasonal mean potential temperature values, rather than temporally varying values. The lower bound of the tropopause zone is the potential temperature of the lapse rate tropopause calculated using seasonal mean ERA data (averaged over each cluster area). Calculations of lapse rate and minimum temperature tropopause isentropic surfaces over a number years showed that their separation is almost constant; the mean difference in potential temperature for DJF was 9.9 ± 1.8 K averaging over every particle release region and year. Accordingly, the potential temperature of the upper bound of the tropopause zone was chosen to be exactly 10 K greater than the lower bound. This definition has the advantage that a fixed amount of heating is required to cross the tropopause zone for all locations and years and the transit time indicates the average heating rate along a trajectory.

The transport across the tropopause zone is investigated in two experiments: one in which particles are released at the bottom of the zone and run forward for 30 days (run TH_FORW), and one in which particles are released at the top of the zone and 30-day back trajectories are calculated (run TH_BACK). The experiments are run for DJF 1985/86 and, as in run P150_FORW, the particles are released from the locations shown in Fig. 1. The lower bound of the tropopause zone is shown in Table 1. Crossing events in run TH_FORW are defined as times when a particle that has been released from the lower bound and traversed the tropopause zone upward crosses the upper bound and is still in the stratosphere at t = 30 days. Crossing events in run TH_BACK are defined as times when a trajectory originating in the troposphere (at t = −30 days) crosses the lower bound en route to the upper bound.

The locations for crossing events for run TH_FORW and run TH_BACK, respectively, are shown in Figs. 11 and 12. These should be compared with crossing events for the run P150_FORW in Fig. 10. There are very few crossing events diagnosed in midlatitudes, since the isentropic surfaces bounding the tropopause zone only approximate the tropopause position over the particle release regions and thus lie far above the tropopause in the extratropics. About 50% of the Indonesian and central Pacific particles cross the tropopause in run TH_FORW compared to around 15% in run P150_FORW. However, Figs. 11a and 11b show that the densities have some similarities with those from run P150_FORW, with high densities over the west Pacific warm pool and parts of Africa and South America, which suggests that a large proportion of the particles crossing the top of the tropopause zone in run TH_FORW may have originated at the 150-hPa level or below.

The vast majority of the Indonesian particles in run TH_BACK enter the tropopause zone over or near the west Pacific warm pool region (Fig. 12a) at locations close to the upward branch of the Walker circulation. The density distribution for the central Pacific particles (Fig. 12b) also indicates enhanced tropopause crossing in this region. Almost all particles (65%–85%) arriving at the upper bound of the tropopause zone have originated from below the lower bound in the troposphere. These results highlight the west Pacific region as a strongly preferred location for diabatic transport across the tropopause zone.

Most trajectories released from the African and South American regions in run TH_FORW cross the upper bound of the tropopause zone in regions close to the release grids. The maximum crossing density lies to the west of release for the African case and to the east of release for the South American case (Figs. 11c and 11d). The number of particles that cross the tropopause zone is 45%–60%, compared to about 10% crossing the lapse rate tropopause from 150 hPa in run P150_FORW. The proximity of crossing events to the release grids in run TH_FORW arises because the horizontal wind is weaker in the tropopause zone and the transit time across the tropopause zone is slightly less than that from 150 hPa to the lapse rate tropopause. The low proportion of trajectories crossing the lapse rate tropopause in run P150_FORW arises because most particles leaving 150 hPa are caught within the tropospheric circulations. In contrast, the much higher proportion of trajectories crossing the tropopause zone indicates that there is a high probability that a trajectory reaching the lapse rate tropopause will subsequently cross the cold point into the stratosphere.

A comparison with back trajectories arriving at the upper bound of the tropopause zone over Africa and South America is very interesting. Most trajectories (about 60%) have originated from the troposphere (below the lower bound) as for the Indonesian and central Pacific arrival regions. However, the maximum crossing event density occurs to the west of the South American arrival grid, in opposition to the crossing events for run TH_FORW, and to both the west and east of the African arrival grid. Thus, there is no clear indication of a preferred location for diabatic transport in these cases. The heating history along trajectories is examined in more detail in the next section.

c. Particle potential temperature distributions

By investigating the distribution of particle potential temperatures at selected intervals after release (or before arrival for back trajectories) we can deduce the cross-isentropic transport, and in particular determine the diabatic heating rate along each trajectory. Figure 13a shows the distribution of potential temperature for all particles in run P150_FORW 0, 10, 20, and 30 days after release from the central Pacific cluster. Histograms for the other three release regions are fairly similar. The slow increase with time of values greater than about 370 K, which is close to the level of the ERA lapse rate tropopause, is suggestive of particles that have entered the lower stratosphere and are subsequently undergoing slow diabatic ascent associated with the Brewer–Dobson circulation. There is also a slightly faster spreading of particles to lower isentropic surfaces with time, and by day 30 the tail of the distribution has extended down to around 300 K. A similar histogram, in which only particles that cross into the stratosphere are included, shows that very few of these particles have potential temperatures of less than 350 K at any point during their history. Thus the much larger frequencies seen in the θ < 350 K range in Fig. 13a must indicate particles that have remained in the troposphere throughout, on average descending in the Hadley and Walker circulations. The distribution becomes remarkably flat in the range 325–360 K indicating that the spread is not diffusive in nature, which would result in a Gaussian distribution. A possible explanation is that the particles in the troposphere leaving the convective regions experience fairly constant radiative cooling and descent with occasional sudden increases in θ associated with latent heating in convective regions, resulting a well-mixed distribution of particles in isentropic coordinates. The heating history is far from the symmetric, small-step random walk found in the stratosphere and described as diabatic diffusion by Sparling et al. (1997).

The θ histogram for run TH_FORW shows a slow spread in θ values with time (Fig. 13b) about the initial value and a slight drift in the peak to higher θ. This behavior does resemble the diabatic dispersion in the stratosphere described by Sparling et al. (1997). They showed that for trajectories longer than the Lagrangian decorrelation timescale for heating, the cross-isentropic motion is like a small-step random walk resulting in a diffusive spread in θ. Differences from run P150_FORW are that the peak is around 380 K instead of 350–360 K, since these particles are initialized at a higher level, and even after 30 days there are very few particles with a potential temperature of less than 350 K. An explanation for this is as follows. The particles in run P150_FORW are initialized closer to the level of maximum outflow from the tropospheric circulations (i.e., around 200 hPa), and thus a relatively large number will undergo rapid poleward and subsequent downward transport. On the other hand, the run TH_FORW particles are initialized higher up (in the transition zone) at levels where such strong poleward and downward transport is less likely to occur.

The θ histograms for run TH_BACK are quite different. For the Indonesian particles (Fig. 14a) there is a clear bimodal structure both 20 and 30 days prior to arrival at the release point, with peaks near 300 and 350–360 K. The histogram for the central Pacific particles has a similar structure. The peak near 300 K indicates particles that undergo rapid diabatic transport en route to the upper bound of the tropopause zone. Further examination shows that most of the particles with potential temperatures below 330 K originate near the west Pacific warm pool region near the upwelling branch of the Walker circulation, between 500 hPa and the surface. Only a small fraction of these particles, chiefly those with potential temperatures below about 295 K, originate outside this region. The diabatic heating rate implied by this transport is around 2.5–3.0 K day−1, although for some particles the heating rate exceeds 4.0 K day−1. This is larger than that implied by the zonal mean Hadley circulation (around 1.0 K day−1 in the upper troposphere) but is closer to typical values for large-scale heating averaged over regions of deep tropical convection.

The peak at 350–360 K in Fig. 14a shows that many particles that reach the top of of the tropopause zone do so by slower ascent than the ascent in deep convective regions discussed above. This peak implies a heating rate of around 0.5–1.5 K day−1, which means that these particles originated from near the bottom of the transition zone, and are slowly transported upward. Histograms of heating rate along trajectories released from all regions in run TH_FORW (not shown here) are more Gaussian-like with a single peak at about 0.3 K day−1 but a wide range (−1.5–2.0 K day−1). Strong heating rates (≫1 K day−1) are not found within the transition zone. The heating rate magnitudes are similar to those calculated by Sparling et al. (1997) for the 500 K isentropic surface using U.K. Met. Office stratospheric analyses of temperature and a radiation code. It is not possible to attribute the heating experienced along trajectories to particular physical processes using ERA data because the output from the physical parameterizations of the ECMWF model is not available. The slow resolved heating experienced in the transition zone may be associated with sporadic convective heating above the average convective tops (150 hPa) or radiative heating associated with the displacement from radiative equilibrium by the mean meridional circulation, or a combination.

The histogram of the potential temperature of South American (Fig. 14b) and African (not shown) particles in run TH_BACK shows a predominantly single peak structure, and the values near the peak imply diabatic heating rates of 0.5–1.0 K day−1. This suggests that most particles that reach the top of the tropopause zone from these clusters do so by slow ascent and very few do so by rapid ascent from the tropical lower or middle troposphere. Rapid large-scale ascent, associated with the strongest convection, into the transition zone followed immediately by slow ascent across the tropopause zone is largely restricted to the west Pacific region.

d. Influence of El Niño and La Niña

Experiment TH_BACK was also run for DJF 1982/83 and DJF 1988/89 to investigate possible impacts of El Niño and La Niña on transport through the tropopause zone. The lower boundary of the tropopause zone was calculated from seasonal mean ERA fields for these years and is shown in Table 1. The potential temperature variation between years is at most 5 K, around half the width of the tropopause zone.

Figure 15 shows particle densities of tropopause crossing events in DJF 1982/83. The chief difference from 1985/86 (Fig. 12) is that for the Indonesian and central Pacific (not shown) particles, the region where most cross-tropopause transport takes place has moved around 30°–50° farther east in 1982/83, and far more particles enter the tropopause zone over or near the central Pacific and South American particle release regions. Histograms of Indonesian (Fig. 16) and central Pacific (not shown) particle potential temperature for 1982/83 10, 20, and 30 days prior to arrival have a bimodal structure at 20 and 30 days with peaks near 300 and 350 K. This structure is similar to that in 1985/86, except that there are relatively more central Pacific particles near the 300 K peak in 1982/83, which is simply due to the fact that the central Pacific cluster is closer to the upward branch of the Walker circulation in 1982/83 than in 1985/86. This is confirmed by the fact that most particles with a potential temperature close to 300 K originate from the middle or lower tropical troposphere in both years, but from a longitude of around 180°E–20°W in 1982/83 compared to around 150°E in 1985/86.

Figure 15 also shows that far more South American particles enter the tropopause zone over or near the arrival region in 1982/83 than in 1985/86. The African particle density distribution (not shown) for 1982/83 is fairly similar to that for 1985/86. Figure 16 shows that the potential temperature of the South American particles in 1982/83 (for 20 and 30 days before arrival time) has a bimodal structure that is not seen in 1985/86. However, the distribution of potential temperatures (and thus diabatic heating rates) 10 days before arrival time is similar in both years. Closer examination shows that in 1982/83 particles with potential temperatures near 300 K undergo rapid ascent (heating rates typically 2–4 K day−1) near the central Pacific upward branch of the Walker circulation 20 to 30 days before arrival, and then are slowly advected eastward and upward until they arrive over South America. Thus there is little enhanced ascent associated with El Niño over South America.

The distributions of particle potential temperatures, and thus diabatic heating rates, for 1988/89 are generally similar to those in 1985/86. However, there are differences in the particle density distributions, especially for the Indonesian and African clusters. Figure 17 shows that most Indonesian particles cross the tropopause zone at similar longitudes in both 1985/86 and 1988/89, but in 1988/89 there appear to be more crossing events in the subtropics, especially in the Southern Hemisphere. This is because the upper-tropospheric circulation in the Tropics has shifted about 30° to the west in 1988/89 compared to 1985/86. Thus, within the Indonesian cluster more ascent takes place over the SPCZ and in the northern subtropics in 1988/89. Most of the particles that cross the tropopause zone in the subtropics in 1988/89 do so within about 10 days prior to arrival at the upper bound. These particles originate (t = −30 days) in the lower or middle equatorial troposphere near 120°E, rather than near 150°E, because of the westward shift of the Walker circulation. Note that the subtropical crossing events here are not related to the sloping tropopause and quasi-isentropic transport seen for run P150_FORW because the tropopause zone is bounded by constant isentropic surfaces.

Figure 17 also shows that many more African particles enter the tropopause zone over the warm pool region in 1988/89 than in 1985/86. These particles enter 20–30 days prior to arrival at the top of the zone, at locations near the upwelling branch of the Walker circulation, and are then slowly advected westward to Africa. The horizontal transport in this area in 1988/89 is not greatly different to that in 1985/86, but since most such particles enter the tropopause zone about 30°E farther west in 1988/89, they are thus more likely to be subsequently transported into the African region than in 1985/86.

4. Discussion on the representation of the tropopause zone in ERA

ERA is highly suited to trajectory studies because it provides global, 6-hourly fields over a long time period. In addition, ERA benefits from using the same assimilation technique throughout the entire dataset period. The dataset enables troposphere to stratosphere transport to be investigated over much of the Tropics and for many years; many other studies have instead utilized the results of intensive observational campaigns [e.g., STEP (Russell et al. 1993)], which tend to be of short duration and concentrate on relatively small regions. However, it should be noted that the accuracy of the ERA fields near the tropical tropopause may be questionable, since this zone suffers from a dearth of reliable observations, and it is also where atmospheric models often have their largest errors. In addition, many studies have indicated that troposphere-to-stratosphere transport, and dehydration of the lower stratosphere, is strongly associated with deep tropical convection (e.g., Danielsen 1982; Kley et al. 1982). However, convective cloud scales are far too small to be resolved by ERA, and no ERA subgrid-scale convective mass fluxes have been made publicly available (although convection is parameterized in the model used by ERA; see Tiedtke, 1989). Accordingly, it is important to summarize the accuracy of the ERA dataset and how well ERA represents the real atmosphere, and to discuss the implications for the results we have presented in section 3.

Pawson and Fiorino (1998) compared monthly means of ERA equatorial temperatures at 100 hPa with radiosonde observations and analyzed temperatures from the National Centers for Environmental Prediction (NCEP). They found that the ERA temperatures had a small cold bias compared to radiosonde observations, and they were considerably closer to observations than temperatures from the NCEP analyses. Diagnostics of the ERA lapse rate tropopause (P. Berrisford 1999, personal communication) show that the tropical tropopause temperature is less than 196 K at most locations and less than 190 K over the warm pool. The latter figure compares well with radiosonde observations (e.g., Reid and Gage 1996). Hoinka (1998, 1999) has presented statistics of ERA tropopause pressure, height, temperature, and water vapor. His lapse rate tropopause pressure in the Tropics has a similar pattern to the field Berrisford calculated, but is around 5–10 hPa greater. The effect of this is to make his calculated tropical tropopause around 2 K warmer than Berrisford’s values. This is probably because of the different methods used to estimate tropopause pressure from the data on ERA model levels. Highwood and Hoskins (1988) have commented on the sensitivity of the representation of the tropopause to vertical resolution. Such sensitivity suggests that in this study it is more appropriate to think of the tropopause as a zone (runs TH_FORW and TH_BACK) than as a surface (run P150_FORW).

Since the transport of air into the tropical lower stratosphere is likely to be nonuniform in time, it is also necessary to assess how well the ERA dataset represents the temporal variability of the atmosphere. Simmons et al. (1999) compared time series of ECMWF operational analyzed temperatures at 90 hPa from DJF 1996/97 with twice-daily radiosonde observations at Kota Kinabalu (6°N, 116°E), Malaysia, and Truk (7°N, 152°E). They showed that the analyses reasonably reproduced the temporal variability observed by the radiosondes. It should be noted that the model used for these analyses has some differences from the one used for ERA, notably a higher horizontal resolution (T213 instead of T106). On the other hand, the models also have many similarities, including the same vertical resolution and the same convection scheme. One may speculate from this information that the ERA dataset can adequately represent the temporal variability near the tropical tropopause, but a rigorous investigation is required to confirm this (as suggested in Pawson and Fiorino 1998).

Observations made over the northern coast of Australia by Wong et al. (1993) indicated that upward velocities in the upper troposphere during convection can be as great as 7 m s−1. In addition, observations of chemical tracers made by Kritz et al. (1993) at similar locations to Wong et al. show that air can be transported from the surface to the tropopause in a few days. However, most particles in run TH_BACK take 10–30 days to travel from the lower troposphere, across the tropopause zone, and into the stratosphere. This serves to highlight the difference between the trajectories of particles following the resolved flow and the trajectories of air parcels. The ECMWF model’s mass-flux parameterization of convection (Tiedtke 1989) provides representations of shallow, midlevel, and deep convection and computes subgrid-scale vertical fluxes of mass, heat, momentum, and water vapor at each model level with the help of a simple model of convective plumes interacting with their environment. The plume model represents the effects of an ensemble of updrafts with varying detrainment levels and also allows for convective downdrafts. Convective updrafts have horizontal scales much smaller than the resolution of the analyses, and their fractional area coverage is also very small. The large-scale ascent is an average over convective updrafts, downdrafts, and their descending environment. Since ascent wins in the average, but the updraft area fraction is small (<10%); this means that the vertical velocity within convective updrafts is at least an order of magnitude greater than the area-averaged vertical velocity in convective regions. For example, seasonal mean vertical upper-tropospheric velocities resolved by ERA in the warm pool region are 0.5–1.0 m s−1.

Based on the above considerations it is clear that the particle trajectories below the 360 K level are not the trajectories of air parcels in the troposphere. However, they do serve to illustrate the heating history following the resolved flow, in which convection is treated as a subgrid-scale heat flux into a fixed mass of air surrounding the particle. The surface of this “air mass” is clearly not a material surface. In the transition zone above the average top of convection but below the temperature minimum, it is not known how much of the heating following the resolved flow arises from subgrid-scale heat fluxes. If the subgrid-scale contribution is small, then the particle trajectories will be similar to air parcel trajectories, as is usually assumed for trajectories in the stratosphere. Subgrid-scale heating may arise through occasional penetration of parameterized convection into the transition zone and the representation of subgrid-scale radiative heating in the ECMWF model used for ERA. Tompkins and Craig (1998) simulated the tropical radiative–convective equilibrium with a cloud-resolving model and found that occasionally convective turrets reach the cold point well above the average top of convection (about 160 hPa). Also they noted that convective scale differences in the radiative heating were important for convective organization and the approach to radiative–convective equilibrium. The importance of these effects in the tropical transition zone relative to the large-scale radiative heating associated with the the wave-driven meridional circulation is not known.

Danielsen (1982) argued that the deepest convective turrets penetrating the tropopause zone result in the establishment of a higher tropopause by the turbulent entrainment of stratospheric air from above and ice crystals from below, then spreading out to form an extensive cirrus anvil. Radiative cooling above and heating below the anvil produce turbulent mixing and a pseudoadiabatic lapse rate across the layer. Based on this argument, Reid and Gage (1996) hypothesized that the coldest tropopause temperatures correspond to recent convective overshooting and that mass exchange from the troposphere to the stratosphere also occurs during these events. In this scenario the dryness of the lower stratosphere will depend on the details of the fluctuations in tropopause temperature and their correlation with crossing events. Further results from run P150_FORW show that the histograms of tropopause pressure and temperature collated for tropopause crossing events are similar to the corresponding distributions for the tropopause at the release regions (irrespective of crossing times). Their similarity shows that transient fluctuations in the tropopause diagnosed from ERA are not correlated with crossing events and provides extra justification for the use of a seasonally fixed (in θ) tropopause zone in runs TH_FORW and TH_BACK. Furthermore, this implies that the distribution of saturation water vapor mixing ratio for the crossing events will be indistinguishable from the distribution calculated from the seasonally average temperature minimum at each crossing location. Since most of the air crossing the tropical tropopause is close to saturation, this suggests that the lower-stratospheric water vapor field at low latitudes could be reasonably estimated from the seasonal mean saturation mixing ratio at the tropopause, given the density map of crossing events (e.g., Fig. 12). However, the above mechanism for the correlation between the temperature minimum and crossing events is unlikely to be represented by the ECMWF model; the lack of correlation for ERA data does not rule out the possibility that such a mechanism is important for constituent transport in the atmosphere.

5. Conclusions

The destinations of particle trajectories released from the four main regions of deep convection in the Tropics, near the average top of convection, were investigated. Trajectories were integrated using winds obtained by interpolation of the 6-hourly 3D velocity field resolved in the ERA dataset. The release regions have a horizontal extent of about 4000 km so that a spread in trajectories from the horizontal divergence above convection would result in a spread in the destination of trajectories downstream, even if the winds were steady. Time dependence in the wind field enhances the rate of separation of neighbouring trajectories to the extent that the distribution of locations would be hard to estimate without explicit trajectory calculations.

In section 2 the density of particle destinations (projected onto the sphere) for a fixed trajectory length of 5 days was calculated. In addition, the mean pressure of trajectories as a function of destination longitude and latitude was estimated. Most trajectories from 150 hPa remain in the tropospheric Hadley and Walker circulations and descend away from the tropopause, particularly near the subtropical jets. This Lagrangian average behavior is readily diagnosed from the maxima in the particle density and the corresponding mean pressure there. Particle density overlaps and particle mean pressures were used to examine interannual variability. The results show that horizontal advection is sensitive to changes in the Hadley and Walker circulations during El Niño and La Niña years. However, the mean pressure (and thus the mean vertical advection) is just as variable between non–El Niño and non La Niña years.

The particle trajectories were further examined for troposphere to stratosphere transport. The histogram of transit times from 150 hPa to the WMO lapse rate tropopause showed a wide peak at around 10 days and much smaller amplitude by 30 days. The interpretation is that most trajectories from 150 hPa stay trapped within the tropospheric circulations and descend away from the tropopause. Some of the others ascend across the transition zone, and most of those that do so have crossed the tropopause by 15 days after release. Thus 30-day trajectories (run P150_FORW) are adequate to represent the low-latitude cross-tropopause transport of particles released from the average top of tropical convection. For a proper representation of the tropopause crossing in the extratropics from these release regions, an even longer trajectory length would be required. The density of crossing events for the Indonesian release (when trajectories leave 150 hPa, cross the lapse rate tropopause, and stay in the stratosphere) showed a preferred location for troposphere to stratosphere exchange over the west Pacific warm pool. For other release regions crossing occurs over South America, Africa, and in the southern subtropics where the tropopause slopes steeply down to the Pole.

Other authors have proposed that troposphere to stratosphere exchange events in the Tropics are often related to the penetration of overshooting deep convection across the tropopause (Danielsen 1982, 1993), and further have suggested that crossing events must correlate with a locally raised tropopause and the coldest tropopause temperatures (Reid and Gage 1996). However, it was found that crossing events for trajectories in run P150_FORW did not correlate with significantly colder temperatures. If this is true for the atmosphere, then the seasonally averaged temperature minimum at crossing locations could be used to estimate the lower-stratospheric water vapor mixing ratio. However, it is also possible that such correlations are important for water vapor transport but that ERA cannot accurately represent such behavior.

Short-term crossing events associated with adiabatic flow across a fluctuating tropopause surface are unlikely to be modeled accurately due to problems with defining the lapse rate and temperature minimum tropopauses from coarse-resolution ERA data. In order to circumvent this difficulty the DJF mean potential temperatures (θ) averaged over the release regions were calculated for the lapse rate tropopause surface and the temperature minimum. Other authors (e.g., Selkirk 1993 and references therein) have suggested that the altitude of the temperature minimum is more suitable for marking the bottom of the stratosphere, in particular associated with moisture content. Highwood and Hoskins (1998) describe a “transition zone” between the average top of convection and the temperature minimum. The lapse rate tropopause divides the transition zone into two, and here we describe the layer between the lapse rate and temperature minimum tropopauses as the “tropopause zone.” It was found that the θ difference between the two surfaces in ERA was 9.9 ± 1.8 K for all regions and years examined. Accordingly, the tropopause zone was defined as a layer with thickness Δθ = 10 K with a lower bound given by the average isentropic surface corresponding to the lapse rate tropopause. Any transport across the zone must correspond to the same integrated heating.

In a further two trajectory experiments the transport across the tropopause zone was investigated. In run TH_FORW, trajectories were released from the lower bound of the tropopause zone and integrated forward for 30 days. “Crossing events” were diagnosed when trajectories crossed the upper bound and did not return. In run TH_BACK, back trajectories were calculated arriving at the upper bound of the tropopause zone. Crossing events were diagnosed when trajectories originating in the troposphere (30 days before arrival) crossed the lower bound of the zone.

Histograms of potential temperature for both the forward trajectory experiments (runs P150_FORW and TH_FORW) show a very slow spread to high values (heating rates < 1.0 K day−1). The cross-isentropic transport within the transition zone features both ascent and descent so that the TH_FORW histogram spreads almost diffusively with a Gaussian-like shape and a slow drift to higher θ. This could arise through a quasi-random walk in θ as described by Sparling et al. (1997). Transience in the heating field may be responsible, but Webster et al. (1999) also show that the time-averaged heating field on the 360 K surface in an isentropic coordinate GCM has considerable structure that could account for changes from heating to cooling along trajectories. The area with heating within the Tropics (θ̇ ≥ 4 K day−1) is greater than the area with cooling (|θ̇| ⩽ 1 K day−1), which would be consistent with the shift in the θ histogram to higher values.

While hardly any particles drift below 350 K in run TH_FORW, after 30 days the distribution of trajectories is well mixed between 325 and 360 K in run P150_FORW and the tail extends down to 300 K. It is hypothesised that this is related to the dominance of slow radiative cooling interspersed with short bursts of strong heating in convective regions. The rate at which the well-mixed distribution in this region is achieved following the resolved flow will be a gross underestimate of the time taken to reach a similar state following the full flow including subgrid-scale convective transport.

In run TH_BACK back trajectories were used to investigate the origin of particles that arrive at the top of the tropopause zone. For the Indonesian and central Pacific arrival regions, the θ histogram becomes strongly bimodal for trajectory lengths greater than 10 days. The majority of the particles arriving at this level originate from below, and reach there either by slow ascent (heating rates of 0.5–1.5 K day−1) across the transition zone, or by rapid ascent (heating rates of 2.5–4.0 K day−1), suggesting heating related to deep convection. The strongest ascent is located at or near the upward branch of the Walker circulation, which is near the equator and 150°E in DJF 1985/86, and moves about 30°–50° farther east in an El Niño year (1982/83) and about 30° farther west in a La Niña year (1988/89). The rapid ascent takes place between the lower and middle troposphere and the bottom of the transition zone, and then is followed by slow ascent through this zone. The very rapid drop in the θ histograms between 350 and 360 K indicates the sharp distinction between the troposphere mixed rapidly by convection and the transition zone above. This distinction also explains the marked difference between the P150_FORW and TH_FORW histograms.

Trajectories arriving at the upper bound of the tropopause zone over South America and Africa only experience slow ascent through the transition zone in most years. However, bimodality in the θ histogram is seen in particular years when these regions are adjacent to the upwelling regions (South America in 1982/83, Africa in 1988/89). Rapid transport from the well-mixed portion of the troposphere to the stratosphere is largely restricted to the west Pacific region. However, it should be noted that on average trajectories arriving over all regions experience slow ascent across the tropopause zone, characterized by a spread in θ histograms about their peak and a slight shift to higher values. There is also notable quasi-isentropic transport across the lapse rate tropopause in the subtropics, particularly in the Southern Hemisphere (in DJF).

The preferred crossing region also matches the one that Newell and Gould-Stewart (1981) identified as a “stratospheric fountain” through a study of the water vapor distribution in the vicinity of the tropopause. However, Newell and Gould-Stewart suggested that slow mean ascent across the tropopause only occurs over this region whereas it also takes place in most other regions. It is only when the history of the air is examined using trajectories that the preferred pathway from the lower troposphere to the stratosphere over the west Pacific is identified, involving a combination of rapid ascent related to deep convection followed immediately by slow ascent across the transition zone. This may have important consequences for the study of the transport of tropospheric pollutants into the lower stratosphere and for atmospheric chemistry in the transition zone (e.g., Kley et al. 1996).

Acknowledgments

This work was carried out under the Public Meteorological Services Research Programme of The Met. Office. We thank ECMWF for the use of the ERA data, which were obtained through the British Atmospheric Data Centre, and Paul Berrisford for the use of his seasonal mean ERA tropopause diagnostics. J. Methven would like to thank The Met. Office Hadley Centre for Climate Prediction and Research for funding during the course of this research. Thanks to the reviewers for motivating a more thorough study of the diabatic transport.

REFERENCES

  • Brewer, A. W., 1949: Evidence for a world circulation provided by the measurements of helium and water vapour distribution in the stratosphere. Quart. J. Roy. Meteor. Soc.,75, 351–363.

  • Danielsen, E. F., 1982: A dehydration mechanism for the stratosphere. Geophys. Res. Lett.,9, 605–608.

  • ——, 1993: In situ evidence of rapid, vertical, irreversible transport of lower tropospheric air into the lower tropical stratosphere by convective cloud turrets and by larger-scale upwelling in tropical cyclones. J. Geophys. Res.,98, 8665–8682.

  • Dobson, G. M. B., 1956: Origin and distribution of the polyatomic molecules in the atmosphere. Proc. Roy. Soc. London,A236, 187–193.

  • Evans, M., and Coauthors, 2000: Evaluation of a Lagrangian box model using field measurements from EASE 1996. Atmos. Environ.,36, 3843–3863.

  • Evans, S. J., R. Toumi, J. E. Harries, M. P. Chipperfield, and J. M. Russell III, 1998: Trends in stratospheric humidity and the sensitivity of ozone to these trends. J. Geophys. Res.,103, 8715–8725.

  • Gage, K., and G. Reid, 1987: Longitudinal variations in tropical tropopause properties in relation to tropical convection and ENSO events. J. Geophys. Res.,92, 14 197–14 203.

  • Gibson, J. K, P. Kallberg, S. Uppala, A. Hernandez, A. Nomura, and E. Serrano, 1997: ECMWF Re-analysis Project Report Series, 1: ERA description. ECMWF, Reading, United Kingdom, 71 pp.

  • Highwood, E. J., and B. J. Hoskins, 1998: The tropical tropopause. Quart. J. Roy. Meteor. Soc.,124, 1579–1604.

  • Hodges, K. I., 1996: Spherical nonparametric estimators applied to the UGAMP model integration for AMIP. Mon. Wea. Rev.,124, 2914–2932.

  • Hoinka, K., 1998: Statistics of the global tropopause pressure. Mon. Wea. Rev.,126, 3303–3325.

  • ——, 1999: Temperature, humidity, and wind at the global tropopause. Mon. Wea. Rev.,127, 2248–2265.

  • Jackson, D. R., S. J. Driscoll, E. J. Highwood, J. E. Harries, and J. M. Russel III, 1998: Troposphere to stratosphere transport at low latitudes as studied using HALOE observations of water vapour 1992–1997. Quart. J. Roy. Meteor. Soc.,124, 169–192.

  • Kley, D., A. L. Schmeltekopf, K. Kelly, R. H. Winkler, T. L. Thompson, and M. McFarland, 1982: Transport of water through the tropical tropopause. Geophys. Res. Lett,9, 617–620.

  • ——, P. J. Crutzen, H. G. J. Smit, H. Vömel, S. J. Oltmans, H. Grassl, V. Ramanathan, 1996: Observations of near-zero ozone concentrations over the convective Pacific: Effects on air chemistry. Science,274, 230–233.

  • Kousky, V. E., and A. Leetmaa, 1989: The 1986/87 Pacific warm episode: Evolution of oceanic and atmospheric anomaly fields. J. Climate,2, 254–267.

  • Kritz, M., S. Rosner, K. Kelly, M. Lowenstein, and K. Chan, 1993: Radon measurements in the lower tropical stratosphere: Evidence for rapid vertical transport and dehydration of tropospheric air. J. Geophys. Res.,98, 8725–8736.

  • Mastenbrook, H. J., 1974: Water vapor measurements in the lower stratosphere. Can. J. Chem.,52, 1527–1531.

  • Methven, J., 1997: Offline trajectories: Calculation and accuracy. UGAMP Tech. Rep. 44, 18 pp. [Available from CGAM, University of Reading, Earley Gate, Reading RG6 6BB, United Kingdom.].

  • ——, P. Berrisford, and B. J. Hoskins, 1999: A Lagrangian climatology for the North Atlantic storm track. UKMO Hadley Centre Tech. Note 9, 99 pp. [available from Hadley Centre for Climate Prediction and Research, The Met. Office, London Road, Bracknell, Berkshire RG12 2SY, United Kingdom.].

  • Mote, P. W., K. H. Rosenlof, J. R. Holton, R. S. Harwood, and J. W. Waters, 1995: Seasonal variations of water vapor in the tropical lower stratosphere. Geophys. Res. Lett.,22, 1093–1096.

  • ——, and Coauthors, 1996: An atmospheric tape recorder: The imprint of tropical tropopause temperatures on stratospheric water vapour. J. Geophys. Res.,101, 3989–4006.

  • Newell, R. E., and S. Gould-Stewart, 1981: A stratospheric fountain? J. Atmos. Sci.,38, 2789–2796.

  • Pawson, S., and M. Fiorino, 1998: A comparison of reanalyses in the tropical stratosphere. Part 1: Thermal structure and the annual cycle. Climate Dyn.,14, 631–644.

  • Pickering, K. E., and Coauthors, 1996: Convective-transport of biomass burning emissions over Brazil during TRACE-A. J. Geophys, Res.,101, 23 993–24 012.

  • Reid, G. C., and K. S. Gage, 1996: The tropical tropopause over the western Pacific: Wave driving, convection and the annual cycle. J. Geophys. Res.,101, 21 233–21 241.

  • Russell, P. B., L. Pfister, and H. B. Selkirk, 1993: The tropical experiment of the Stratosphere–Troposphere Exchange Project (STEP): Science objectives, operations, and summary findings. J. Geophys. Res.,98, 8563–8589.

  • Schumann, U., 1994: On the effect of emissions from aircraft engines on the state of the atmosphere. Ann. Geophys.,12, 365–384.

  • Selkirk, H. B., 1993: The tropopause cold trap during STEP/AMAX 1987. J. Geophys. Res.,98, 8591–8610.

  • Simmons, A. J., A. Untch, C. Jakob, P. Kallberg, and P. Unden, 1999:Stratospheric water vapour and tropical tropopause temperatures in ECMWF analyses and multi-year simulations. Quart. J. Roy. Meteor. Soc.,125, 353–386.

  • Sparling, L. C., J. A. Kettleborough, P. H. Haynes, M. E. McIntyre, J. E. Rosenfield, M. R. Schoeberl, and P. A. Newman, 1997: Diabatic cross-isentropic dispersion in the lower stratosphere. J. Geophys. Res.,102, 25 817–25 829.

  • Thuburn, T., and G. Craig, 1997: The height of the tropopause. J. Atmos. Sci.,54, 869–882.

  • Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev.,117, 1779–1800.

  • Tompkins, A. M., and G. C. Craig, 1998: Radiative–convective equilibrium in a three-dimensional cloud-ensemble model. Quart. J. Roy. Meteor. Soc.,124, 2073–2097.

  • Toumi, R., J. D. Haigh, and K. S. Law, 1996: Tropospheric ozone-lightning feedback. Geophys. Res. Lett.,23, 1037–1040.

  • Webster, S., J. Thuburn, B. Hoskins, and M. Rodwell, 1999: Further development of a hybrid-isentropic GCM. Quart. J. Roy. Meteor. Soc.,125, 2305–2331.

  • Wong, T., G. L. Stephens, P. W. Stackhouse, and F. P. J. Valero, 1993:The radiative budgets of a tropical mesoscale convective system during EMEX-STEP-AMEX experiment, 1: Observations. J. Geophys. Res.,98, 8683–8694.

  • World Meteorological Organization, 1998: Scientific assessment of ozone depletion: 1998. World Meteorological Organization Global Ozone Research and Monitoring Project Rep. 44, 669 pp.

Fig. 1.
Fig. 1.

The release positions of air particles advected by the trajectory model.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 2.
Fig. 2.

ERA particles released over central Pacific from the 150-hPa level, 10-yr mean of DJF. (a) Mean pressure 5 days after release; (b) particle density 5 days after release. Contour interval: 10 hPa for (a) and 0.1 for (b). The box indicates the region from where the particles were released.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 3.
Fig. 3.

As in Fig. 2, except that locations of particles released over Indonesia are shown.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 4.
Fig. 4.

As in Fig. 2, except that locations of particles released over South America are shown.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 5.
Fig. 5.

As in Fig. 2, except that locations of particles released over Africa are shown.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 6.
Fig. 6.

Particle density 5 days after release over the central Pacific from the 150-hPa level for DJF 1982/83. Contour interval: 0.1.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 7.
Fig. 7.

(a) Density overlap for particles 5 days after release; (b) mean pressure (weighted by particle density) for particles 5 days after release. Solid line: Indonesian particles; dotted line: central Pacific particles; dashed line: South American particles; dashed–dotted line: African particles. Years are indicated by “80” = DJF 1980/81, “82” = DJF 1982/83, etc.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 8.
Fig. 8.

Histogram of time after release at which the lapse rate tropopause is crossed, for run P150_FORW. Only crossing events between 20°S and 20°N are included. Bin size is 1 day. (a) Indonesian particles. Sample size is 1795. (b) Central Pacific particles. Sample size is 1316. (c) African particles. Sample size is 498. (d) South American particles. Sample size is 510.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 9.
Fig. 9.

As in Fig. 8, except only central Pacific particles that cross the tropopause north of 20°N or south of 20°S are included. Sample size is 825.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 10.
Fig. 10.

Densities of tropopause crossing events for run P150_FORW, DJF 1985/86. (a) Indonesian particles; (b) central Pacific particles; (c) African particles; (d) South American particles. Contour interval: 0.1.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 11.
Fig. 11.

As in Fig. 10, except fields for run TH_FORW are shown.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 12.
Fig. 12.

As in Fig. 10, except fields for run TH_BACK are shown. The contour interval in (a) is 0.2.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 13.
Fig. 13.

Histogram of central Pacific particles for DJF 1985/86. Bin size is 2.0 K. (a) For run P150_FORW at time of release (thin line), 10 days after release (medium thin line), 20 days after release (medium thick line), and 30 days after release (thick line); (b) for run TH_FORW 10 days after release (thin line), 20 days after release (medium line), 30 days after release (thick line). The dotted lines indicate the tropopause zone used in runs TH_FORW and TH_BACK.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 14.
Fig. 14.

Histogram of particles for run TH_BACK for DJF 1985/86 10 days before arrival at the release point (thin line), 20 days before arrival (medium line), and 30 days before arrival (thick line). Bin size is 2.0 K. (a) Indonesia; (b) South America. The dotted lines indicate the tropopause zone.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 15.
Fig. 15.

As in Fig. 10, except densities for run TH_BACK, DJF 1982/83, are shown. (a) Indonesian particles; (b) South American particles. The contour interval is 0.2.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 16.
Fig. 16.

As in Fig. 14, except particles for DJF 1982/83 are shown. (a) Indonesia; (b) South America.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Fig. 17.
Fig. 17.

As in Fig. 10, except densities for run TH_BACK, DJF 1988/89, are shown. (a) Indonesian particles; (b) African particles. The contour interval is 0.2.

Citation: Journal of the Atmospheric Sciences 58, 2; 10.1175/1520-0469(2001)058<0173:TITLLT>2.0.CO;2

Table 1.

Potential temperature (K) of the lapse rate tropopause for DJF averaged over the cluster release regions. This forms the lower bound of the tropopause zone for runs TH_FORW and TH_BACK.

Table 1.
Save