The Impact of Convection on CO Distribution in the Tropical Troposphere

1
The Impact of Convection on CO Distribution in
the Tropical Troposphere M. Manyin1, M.
Schoeberl2, A. Douglass2
1 Science Systems and Applications, Inc. 2 NASA
Goddard Space Flight Center
Abstract Convective transport plays a important
role in the vertical redistribution of
atmospheric constituents. To better characterize
this role we use CO as a tracer, evaluating
satellite data, numerical model output, and in
situ measurements. Observations from MLS and AIRS
are analyzed, revealing complementary
distributions of CO in the upper and lower
troposphere respectively. Convective regions are
identified using TRMM-calibrated rainrates. Model
results from GEOS-5 (global) and WRF-Chem
(mesoscale) are used to interpret the observed
distributions. In situ data from the TC4
campaign are included in the analysis.
Preliminary results are presented.
Primary Concept CO was selected as a tracer in
this study, because of its tropospheric lifetime
(1 to 3 months) and this initial observation In
July 2005, AIRS observed plumes of CO traveling
westward from Africa to South America, at a
relatively low altitude (500 hPa). In subsequent
days, MLS detected a trail of CO at a higher
level (215 hPa) covering Central America and
points west. A possible explanation lies
in the mass flux associated with deep convective
systems entrainment at low levels, and
detrainment at higher levels. Using high
rainrates as a proxy for convection, the
TRMM-calibrated 3B43 monthly product shows
significant convective activity in the same
region where AIRS CO arrives from the east, and
MLS CO begins to head west.
Approach In order to incorporate in situ
aircraft data from the Tropical Composition,
Cloud and Climate Coupling (TC4) Mission, initial
work has focused on the July/August 2007
timeframe. CO retrievals from MLS and AIRS were
studied, along with TRMM-calibrated 3B42
(3-hourly product), for evidence of convective
lofting. Once likely cases were identified,
global model data were consulted, to see if they
corroborated. The model provided a more complete
spatial picture and finer temporal resolution.
Then aircraft data were compared against both
satellite measurements and model output,
primarily to confirm upper level CO patterns.
Lastly, a mesoscale model was brought to bear,
allowing higher resolution simulations.

• Motivation
• How important is convection in the transport of
  trace gases in the troposphere?
• Does convection lead to characteristic patterns
  of vertical distribution, particularly in the
  ITCZ region near Central America?
• How well do global models simulate this
  convective transport?
• Do mesoscale models with parameterized convection
  permit more accurate characterization?

3B43 Average Rainrate July 2005
MLS CO
Convective area in TRMM 3B43
AIRS CO
Using Satellite-based Data Starting with
TRMM-calibrated rainfall A, classify areas of
high rainrate as candidate convective regions.
For each region, consult lower tropospheric
measurements of CO from AIRS B. Where
significant, search the same region directly
above (MLS 215 hPa) for evidence of increased CO
burden C. In this example, there are several
areas to pursue in the ITCZ. Where matches are
found in MLS we must determine if the CO appeared
at that level during the time of heaviest
rainfall, or if it already resided at that level,
upwind of its current location. Where CO is not
found at 215 hPa, CO may have risen to an
intermediate level, or beyond. AIRS measurements
alone do not distinguish between different
vertical levels, as they only provide a single
independent measurement in the troposphere.
Including Numerical Simulations We turn to an
AGCM to help compensate for the sparse MLS
measurements and the lack of vertical definition
in the AIRS data. The Goddard Earth Observing
System (GEOS-5) global model was run in support
of the TC4 campaign. The models convective
precipitation patterns D tend to follow those
of the TRMM product, particularly in the tropics.
The low level CO E is aligned with the AIRS
retrievals, although concentrations over water
are somewhat lower. At upper levels F pattern
in the ITCZ region are similar to MLS but the CO
burden in the model tends to be greater. To
effectively use the GEOS-5 output we must
systematically characterize its correlation to
the satellite data, and then gauge the convective
effects on CO.
A
B
C
D
E
F
All images show daily values for July 29, 2007
GEOS-5 Convective Precipitation
GEOS-5 CO at 800 hPa
GROS-5 CO at 200 hPa
3B42 Rainfall
AIRS lower tropospheric CO
MLS CO at 215 hPa (Scaled by 0.6)
G
Global vs. Regional Model The GEOS-5 global model
is run at 0.5x0.625 degree resolution, while
convective storms occur on much smaller scales.
We can turn to mesoscale models such as the
Weather Research and Forecasting (WRF) model to
achieve higher resolution. We have begun to use
WRF-Chem in tracer mode to run at GEOS-5
resolution on an outer grid, and 18 km on a
nested grid. Convection must still be
parameterized, but higher accuracy is possible,
as well as flexibility in parameterization
settings. The figures to the right show upper
level CO concentrations for the nested domain run
with and without convective effects. Convection
clearly enhanced the CO levels in three areas.
Again, a thorough analysis is required, as well
as comparison with in situ data.
Using Aircraft Data Data measured from a DC8
during TC4 are useful for gauging accuracy of the
MLS retrievals as well as providing vertical
profiles against which model output can be
compared. CO measurements from one day are shown
here both as a time series G and as a spatial
plot of upper level segments H. Using the
Goddard Trajectory Model, MLS products with
enhanced spatial resolution were generated at
144-minute intervals. Three timesteps are shown,
with the corresponding in situ measurements
plotted over them. These segments show a
convincing visual correlation a statistical
evaluation remains to be done, and the aircraft
data should also be compared to model output.
H
CO data from DC8 7/29/07
CO from DC8 and MLS 1800-2024 UTC

• Next Steps
• Proceed from anecdotal to statistical analysis as
  mentioned above
• Consider Ice Water Content from MLS
• Request that AIRS retrievals be recomputed with a
  priori profile derived from GEOS-5

CO from DC8 and MLS 1312-1536 UTC
CO from DC8 and MLS 1536-1800 UTC
Numerical Models
Satellite Data

• TRMM 3B42 Rainrate Product
• calibrated with TMI instrument aboard the
  Tropical Rainfall Measuring Mission (TRMM)
  satellite
• resolution 0.25 x 0.25 degree, 50 N-S
• product interval every 3 hours
• combination of IR and microwave precipitation
  estimates

• GEOS-5 (including GOCART)
• GCM / DAS
• resolution 0.5 x 0.625 deg.
• convective parameterization RAS
• vertical levels 72
• run in support of TC4

• AURA MLS Microwave Limb Sounder
• near polar/sun synchronous orbit
• resolution 300 km at limb tangent pt
• 16 days before repeating scan pattern
• processed with Goddard Trajectory Model

Acknowledgements CO is in GEOS-5 thanks to those
who coupled it with the GOCART aerosol model
Peter Colarco, Randy Kawa, Huisheng Bian and
Arlindo Dasilva (see http//gmao.gsfc.nasa.gov/hi
ghlights/tc4mission/ g5chem) Thanks to Joanna
Joiner for the AIRS CO retrievals and her advice
on their use. Plots of 3B42 were produced using
the Giovanni online data system, developed and
maintained by the NASA Goddard Earth Sciences
(GES) Data and Information Services Center (DISC).
References Bloom, S. et al, 2005. Documentation
and Validation of the Goddard Earth Observing
System (GEOS) Data Assimilation System - Version
4. Technical Report Series on Global Modeling and
Data Assimilation 104606, 26 Grell, G. A. et al,
2005 Fully coupled online chemistry within the
WRF model, Atmos. Environ., 39,
6957-6975. Schoeberl, M., and L. Sparling (1995),
Trajectory modelling, in Diagnostic tools in
Atmospheric Physics Varenna on Lake Como, Villa
Monastero, 22 June ? 2 July 1993, Int. School.
Phys. Enrico Fermi, vol. 124, edited by G. Fiocco
and G. Visconti, pp. 289 ? 305, IOS Press,
Amsterdam.

• WRF-Chem (run in tracer mode)
• mesoscale, nonhydrostatic atmos. model
• resolution outer grid54km, inner18km
• conv paramn Grell-Devenyi ensemble
• vertical levels 36
• initialization, boundaries and emissions from
  GEOS-5

• AQUA AIRS Atmospheric Infrared Sounder
• polar, sun synchronous orbit
• resolution 13.5 km at nadir
• coverage global, twice daily
• retrievals using Joiners algorithm

Aircraft data

• DC-8 flown during TC4
• diode laser spectrometer
• altitude up to 200 hPa
• references Sachse, Glen