Evaluation of transport in the lower tropical stratosphere in a global
chemistry and transport model
Anne R. Douglass, Mark R. Schoeberl, and Richard B. Rood
NASA Goddard Space Flight Center, Greenbelt, Maryland, USA
Steven Pawson
Goddard Earth Sciences and Technology Center, University of Maryland, Baltimore County, Baltimore, Maryland, USA
Received 25 June 2002; revised 13 September 2002; accepted 9 January 2003; published 2 May 2003.
[1] Off-line chemistry and transport models (CTMs) use meteorological information from
a general circulation model (GCM) or from a data assimilation system (DAS) to calculate
the evolution of stratospheric constituents. Here constituent fields from two CTM
simulations are compared with each other and with observations from satellite, aircraft,
and sondes to judge the realism of the tropical transport. One simulation uses winds from
a GCM and the second uses winds from a DAS that has the same GCM at its core. A
simulation using the GCM fields reproduces many observed features for O
3
,CH
4
, and the
age of air. The same comparisons for a simulation using DAS fields show rapid upward
tropical transport and excessive mixing between the tropics and middle latitudes. The
assimilation system changes the temperature and wind fields to produce consistency
between a GCM forecast and observations, behaving like an additional forcing has been
added to the equations of motion and possibly leading to the unrealistic transport
produced by the DAS fields. These comparisons highlight aspects of the transport in the
lower tropical stratosphere, and suggest that while a CTM driven by DAS fields provides
good short-term simulations when event-by-event comparisons with observations are
desired, a CTM driven by GCM fields may be more appropriate for long-term
calculations such as required to assess the impact of changes in stratospheric
composition. INDEX TERMS: 0340 Atmospheric Composition and Structure: Middle atmosphere—
composition and chemistry; 0341 Atmospheric Composition and Structure: Middle atmosphere—constituent
transport and chemistry (3334); 3334 Meteorology and Atmospheric Dynamics: Middle atmosphere dynamics
(0341, 0342); KEYWORDS: stratospheric transport, tropical barrier, age-of-air
Citation: Douglass, A. R., M. R. Schoeberl, R. B. Rood, and S. Pawson, Evaluation of transport in the lower tropical stratosphere in a
global chemistry and transport model, J. Geophys. Res., 108(D9), 4259, doi:10.1029/2002JD002696, 2003.
1. Introduction
[2] As first discussed by Rood et al. [1989], constituent
evolution calculated using an off-line chemistry and trans-
port model (CTM) that is forced by meteorological fields
from a data assimilation system (DAS) will reproduce
observed constituent variability if the advection scheme is
sufficiently accurate that scheme numerics have little
impact, the assimilation fields reflect the actual atmospheric
state and the photochemistry in the CTM is realistic. Prob-
lems in each of these areas have been a dominant source
of error. Allen et al. [1991] showed that a monotonic,
upstream-biased differencing transport scheme gave supe-
rior performance in a CTM compared with a spectral trans-
port scheme in common use at that time. Particularly notable
was the ability of such a scheme to maintain correlations
observed between long-lived constituents. In older versions
of the Goddard Earth Observing System (GEOS) assimila-
tion systems, the residual circulation diagnosed directly
from the wind fields bore little resemblance to the residual
circulation calculated from the heating rates, and the trans-
port produced using winds from these systems was similarly
deficient [Weaver et al., 1993]. The fields produced by
improved assimilation systems have resulted in more real-
istic atmospheric transport [Coy and Swinbank, 1997]. ER-2
measurements of nitrogen species showed the importance of
heterogeneous reactions on sulfate aerosols and led to major
revisions in the CTM stratospheric photochemical mecha-
nism [Kawa et al., 1993]. Current applications rely on
fidelity between observed and calculated constituent evolu-
tion, and correct representation of both photochemistry and
transport within the CTM. For example, Chipperfield and
Jones [1999] used a CTM to quantify the relative contribu-
tions of transport and photochemistry to ozone changes in
the lower stratosphere on seasonal and longer timescales.
[3] The approach of using DAS fields in CTMs and
comparisons of simulated fields with observations to
address issues concerning photochemistry and transport
has become standard during the past decade. CTMs use
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. D9, 4259, doi:10.1029/2002JD002696, 2003
Copyright 2003 by the American Geophysical Union.
0148-0227/03/2002JD002696$09.00
ACH 2 - 1

winds from analyses produced by the United Kingdom
Meteorological Office (UKMO) [Chipperfield et al., 1994,
1996], by the European Centre for Medium-Range Weather
Forecasts (ECMWF) [Lefe`vre et al., 1994; Deniel et al.,
1998], and by the Goddard Earth Observing System Data
Assimilation System (GEOS DAS) [Rood et al.,1991;
Douglass et al., 1997; Kawa et al., 2003]. CTM simulations
have been used to interpret observations from different
platforms, including aircraft [Douglass et al., 1993; Lefe`vre
et al., 1994], satellite [Geller et al., 1995; Chipperfield et
al., 1996], balloon [Kondo et al., 1996] and ground-based
instruments [Goutail et al., 1999; Chipperfield and Pyle,
1998; Chipperfield, 1999; Sinnhuber et al., 2000]. CTMs
using DAS fields have been used to simulate transport and
buildup of pollutants from hypothetical supersonic aircraft
flying in the lower stratosphere [Weaver et al., 1996].
Several groups are using this approach for interpretation
of observations of tropospheric aerosols [Chin et al., 2000;
Ginoux et al., 2001] and constituents [Bey et al., 2001].
[4] It is well known that the assimilation-driven CTMs
reproduce observed synoptic and planetary scale variability
of stratospheric ozone and other constituents at middle and
high latitudes. However, good agreement of observations
and simulation for a single tracer does not guarantee that the
CTM transport represents the actual atmospheric transport.
For example, Considine et al. [2003] demonstrated that
horizontal and vertical transport in the high-latitude polar
winter produce good agreement between observed and
simulated values for vortex N
2
O throughout the northern
winter 1999–2000 but poor agreement for NO
y
which has
different relative vertical and horizontal gradients. There are
additional nagging problems as well. Douglass et al. [1997]
and Chipperfield [1999] show poor representation of tracer
gradients particularly between tropics and middle latitudes
using the Goddard Space Flight Center (GSFC) CTM with
winds from GEOS DAS and the SLIMCAT CTM with
winds from UKMO, respectively. Both of these studies find
that CTM ozone generally compares better with observa-
tions than do long-lived tracers, with the exception of a high
bias between simulated and observed ozone in the summer
high-latitude lower stratosphere. The weak tracer gradients
between the tropics and middle latitudes are consistent with
the results of Weaver et al . [2000], who developed a
climatology for the production of laminae in ozone profiles
from ozonesonde profiles and found that the CTM driven by
DAS fields produced excessive lamination in the subtropics.
[5] A CTM driven by DAS fields can be used to interpret
observations of various constituents made from different
platforms (e.g., balloon, aircraft, and satellite) for the specific
meteorological conditions under which the measurements
are made. However, a primary application for atmospheric
models is to predict the future condition of the atmosphere,
and assess the importance of natural and anthropogenic
changes in atmospheric composition to stratospheric ozone.
Assessment calculations often require long integrations, and
cumulative errors in transport and chemistry impact the
calculation severely. Thus the requirements for model per-
formance are stringent. Ideally, the model ozone evolution
will match observations because the relative contributions of
transport and photochemical processes to the ozone tendency
are represented correctly. It is also necessary that the balance
among photochemical processes be represented correctly
[Wennberg et al., 1994]. Comparisons of output from long
integrations with observations of reactive constituents such
as ozone and long-lived constituents such as CH
4
may be
made to determine if model balances are realistic.
[6]Previously,Douglass et al. [1999] found that the
transport produced by a CTM driven by winds from a
version of the Middle Atmosphere Community Climate
Model was superior to that produced using the same CTM
with winds from a different GCM or using winds from the
STRAT version of the GEOS DAS (see Table 1). This
judgment relied on data based diagnostics that could be
applied to constituent fields from any three-dimensional
model. The uncertainties in the assessment calculations that
are introduced by poor agreement of various aspects of
constituent transport in the face of ‘‘good agreement’’ for
other aspects of transport are difficult to quantify.
[7] The mean age of stratospheric air, the mass weighted
average of the transit times from the tropical tropopause to
any given location, is a sensitive diagnostic of model
transport which reflects the quality of the global circulation
[Hall and Plumb, 1994; Hall et al., 1999]. Calculations of
mean age were compared with the age determined from
observations of SF
6
and CO
2
as part of Models and
Measurements Intercomparison II [Park et al., 1999]. Hall
et al. [1999] show that in most models, the age of air in the
middle and high-latitude lower stratosphere is too young,
indicating that the overall model circulation and mixing are
too rapid. Schoeberl et al. [2003] use trajectory calculations
with meteorological input from assimilation systems and a
GCM to show how horizontal mixing and vertical transport
characteristics of the meteorological fields impact the age
spectrum, i.e., the distribution of parcel transit times that
comprise the mean age. The age spectra computed using
winds from a GCM differ from spectra calculated using
DAS fields. The latter are too broad as a result of too much
exchange between the tropics and midlatitudes and too
much vertical dispersion. Schoeberl et al. [2003] also found
that despite their unrealistic age spectra, the DAS fields
produce an appropriate tropical mean age because excessive
vertical dispersion offsets midlatitude mixing.
Table 1. Goddard Earth Observing System Data Assimilation Systems
a
GEOS-1
(UARS; STRAT)
GEOS-2
(TRMM)
GOES-3
(TERRA)
GEOS-4
(FVDAS)
CORE GCM GEOS GEOS GEOS FVGCM
Dates 1/91–9/93 5/95–5/98 11/97–12/99 11/99–present In validation
Latitude
Longitude L levels 4
5L46 2
2.5L46 2
2.5L70 1
1L48 2
2.5L55
Analysis Scheme Optimal Interpolation Optimal Interpolation Physical-Space Statistical
Analysis System (PSAS)
PSAS PSAS
Incremental Analysis
Update (IAU)
IAU IAU IAU Intermittent Update
a
References to the various systems are provided in the text.
ACH 2 - 2 DOUGLASS ET AL.: EVALUATION OF TRANSPORT IN LOWER TROPICAL STRATOSPHERE