Comparison of TES and MLS Tropospheric Ozone and CO Retrievals with Results from GEOSChem

1
Comparison of TES and MLS Tropospheric Ozone and
CO Retrievals with Results from GEOS-Chem
Qinbin Li, Line Jourdain, Nigel Richards Jet
Propulsion Laboratory, California Institute of
Technology, CA 91109 Susan Kulawik, Greg
Osterman, Kevin Bowman, John Worden Helen
Worden, Mingzhao Lu, Reinhard Beer and the TES
Team Mark Filipiak, Jonathan Jiang, Brian
Knosp, Joe Waters and the MLS Team
Aura Validation Workshop September 22 CSC
2
The GEOS-Chem Global 3-D CTM
NASA /GMAO Assimilated Meteorological data 2o x
2.5o x 48 1o x 1o x 55
Harvard 1x 1x 30 (48, 55) 2o x 2.5o x 30 (48,
55) 4o x 5o x 30 (48, 55)
GEOS-Chem CTM 120 species 150 reactions 53
tracers
Harvard Chemistry modules Deposition
modules Radiation modules
Harvard transport shells

• Horizontal resolutions 4x5, 2x2.5, or 1x1
  over continental-scale region (one-way nesting).
• Vertical resolution 48 sigma levels (GEOS-3) or
  55 hybrid eta levels (GEOS-4) up to 0.01 hPa.
• Gas-phase chemical mechanism resolves about 120
  species, including 10 primary VOCs.
• Aerosol simulation includes sulfate, ammonia,
  nitrate, dust, BC, OC, sea salt, and SOAs. Will
  include aerosol microphysics (Peter Adams).
• Linked with the CMAQ regional air quality model
  (boundary condition) and the GISS GCM
  (chemistry-climate).

3
MLS vs MOPITT CO

• November 2004 monthly mean MLS and MOPITT CO
  fields calculated.
• MOPITT and MLS CO gridded horizontally onto a 5x5
  degree grid.
• MLS vertical profiles interpolated to MOPITT
  pressure levels.

MLS vs GEOS-Chem CO and O3

• All data examined for November 2004
• GEOS-Chem profiles sampled along MLS orbit track.
• GEOS-Chem profiles interpolated onto MLS pressure
  levels for comparison.

TES vs GEOS-Chem CO and O3

• All data examined for 1st 17th November 2004
  (10 step stares, 7 global surveys)
• GEOS-Chem profiles sampled along TES orbit track.
• GEOS-Chem profiles interpolated onto TES pressure
  levels and TES averaging kernels applied.

4
Monthly mean MLS CO at 150 mb for Nov 2004
MOPITT CO at 150 mb
GEOS-Chem CO at 150 mb
5
Monthly mean MLS CO at 250 mb for Nov 2004
MOPITT CO at 250 mb
GEOS-Chem CO at 250 mb
6
Monthly mean MLS CO at 350 mb for Nov 2004
MOPITT CO at 350 mb
GEOS-Chem CO at 350 mb
7
MLS vs. GEOS-Chem CO for Nov 2004
8
TES vs. GEOS-Chem CO for Nov 1st -17th 2004
9
May 8-12, 2005
MLS CO
GEOS-CHEM CO
The magnitude of CO at 200 hPa is much higher in
the observations than in the model. Consistency
is better at 150 and 100 hPa. It appears that
5-day mean MLS observations of CO can be very
useful for tracking pollution transport in the
upper troposphere.
10
Long-range Transport of Pollution Analysis of
MLS Upper Tropospheric Observations Using
GEOS-Chem
11
MLS Observations Show Dense High Clouds
Collocated with the Elevated CO
12
Wide-spread Pollution in the Upper Troposphere
over South Asia Trapping of Pollution by the
Tibetan Anticyclone
MOPITT CO at 300 hPa
GEOS-CHEM O3 and CO
Kar et al. GRL, 2004
South Asia in summer is influenced by the Asian
Summer Monsoon. Convection associated with the
ASM and orographic lifting transport Asian
pollution to the upper troposphere.
Li et al. GRL, 2005
13
November
TES CO (ppbv)
GEOS-CHEM CO (ppbv)
350 hPa
500 hPa
700 hPa
14
Scatterplot CO TES vs CO GEOS-CHEM (Tropics)
350 hPa
Data 6 GS global surveys November 2004 RMS
ppbv BIAS (TES-GEOS-CHEM)/GEOS-CHEM
500 hPa
700 hPa
15
November
350 hPa
500 hPa
700 hPa
16
Scatterplot CO TES vs CO GEOS-CHEM
350 hPa
Data 6 GS global surveys November 2004 RMS
ppbv BIAS (TES-GEOS-CHEM)/GEOS-CHEM
700 hPa
500 hPa
17
TES vs GEOS-Chem CO

• TES shows a positive mean bias compared with
  GEOS-Chem at all altitudes.
• The bias is smaller near the surface where TES
  has little sensitivity.
• In the mid and upper troposphere where TES is
  most sensitive the bias is larger and shows much
  more variability.
• Typical retrieval errors on TES CO profiles range
  from 10 to 25 , with larger errors seen in the
  lower and uppermost troposphere.

18
GEOS-Chem CO (No Assimilation)
Assimilated GEOS-Chem CO
TES Observation Locations
19
GEOS-Chem CO (No Assimilation)
Assimilated GEOS-Chem CO
TES Observation Locations
20
GEOS-Chem CO (No Assimilation)
Assimilated GEOS-Chem CO
TES Observation Locations
21
Middle East Ozone Maximum in Summer
GEOS-CHEM Tropospheric Ozone
Middle East Ozone maximum in the middle to upper
troposphere, resulting from complex interplay of
dynamical and chemical factors, and of
anthropogenic and natural influences
Li et al. GRL, 2001
SAGE II Tropospheric Ozone (7 km)
An important test of our understanding of
tropospheric chemistry. An anthropogenic origin
for this maximum would have important
implications for radiative forcing of climate.
Kar et al. GRL, 2002
22
Global Air Pollution Crossroads over the
Mediterranean
Simulated CO time-height cross sections
Schematic air mass trajectories
Fractional contributions
East Europe
Boundary layer
West Europe
North America
Middle troposphere
Asia
Upper troposphere
Lelieveld et al. Science, 2002
23
IS THERE A UT OZONE MAXIMUM OVER SOUTHERN U.S.
IN SUMMER?
GEOS-CHEM monthly means at 300 hPa, July 2000 Li
et al., 2004
July 2000 ozonesonde data Newchurch et al.,
2003
GEOS- CHEM
sondes
24
Pollution Train Long-range Transport of
Southeast Asia Pollution to over the South
Atlantic
February-March
Adding another factor to the South Atlantic
Ozone Paradox
Pathway of Asian ozone pollution
TTO (Tropical Tropospheric Ozone)
Chatfield et al. GRL, 2003
25
GEOS-Chem CO at 200 hPa Long-range Transport of
Tropical Biomass Burning Emissions in the Upper
Troposphere
February 2005
April 2005
26
July 1 flight high ozone (80-120 ppb) in Asian
plumeinterleaved with stratospheric influence
Asian plume
DIAL O3
Asian plume 1
Asian plume 2
Strattrop Air
Troposph.
Mixing?
Lower strat
27
(No Transcript)
28
Sub-Optimal Kalman Filter

• The relationship between the observed quantity,
  y, and the model concentration, x, can be written
  as
• Where I represents the horizontal interpolation
  of the model grid to the locations and A
  represents the averaging kernel for the
  observations.
• The problem is then to find the optimal value of
  x, which inverts (1) for a given y allowing for
  observation and model errors.
• At each time step the solution of (1) is given by
• Where K is called the Kalman Gain matrix and is
  represented as
• Bt is the forecast error covariance at time t, O
  is the error covariance matrix of the
  observations and R is the representativeness
  error covariance.
• The analysis error covariance matrix is given by
• After the inversion xta is used as the initial
  condition for the chemistry transport model M to
  predict constituent concentrations for the next
  assimilation window

y A(I(x)) H(x)
(1)
xta xt K(y Hxt)
(2)
K BtHT(HBtHT O R)-1
(3)
Ba B BHT(HBHT O R)-1HB
(4)
xt?t M(t, xta)
(5)
29
Preliminary Assimilation Results

• Assimilation run using one global survey of TES
  CO, TES run 2286 (Nov 4/5th 2004)
• TES CO profiles and averaging kernels mapped to a
  reduced 14 level grid prior to assimilation
• TES data was filtered according to quality flags,
  profiles at latitudes greater than 60 degrees and
  those with missing levels were removed
• GEOS-Chem version v7-01-01 tagged CO simulation
  with 4x5 degree resolution
• Model error fixed globally to 200 ( gt TES error)
• Horizontal correlation length 500 km (1 grid
  box)
• Vertical correlation length 0.1 atmospheric
  scale height
• Assimilation frequency 1 hour (same as
  chemistry time step)

30
MODIS Active Fires 2004-11-05

• MODIS fires counts suggest active biomass burning
  in regions with increased CO.
• MOPITT 500 mb CO also suggests enhanced CO over
  the Indian and tropical South Atlantic and Pacific

31
(No Transcript)
32
Monthly mean MLS O3 at 316 mb for Nov 2004
GEOS-Chem O3 at 316 mb
Tropics only
33
Monthly mean MLS O3 at 215 mb for Nov 2004
GEOS-Chem O3 at 215 mb
Tropics only
34
Monthly mean MLS O3 at 147 mb for Nov 2004
GEOS-Chem O3 at 147 mb
Tropics only
35
TES vs. GEOS-Chem O3 for Nov 1st -17th 2004
Global
Tropics Only
36
TES vs. GEOS-Chem O3 for Nov 1st -17th 2004
Global
Tropics Only
37
May 8-12, 2005
MLS O3
GEOS-CHEM O3
There are still large discrepancies in the
magnitude of observed v.s. simulated O3 (focus on
the troipcs model values at high latitudes at
these pressure levels are likely not
reliable). It seems difficult to detect
signatures of pollution transport in the upper
troposphere from O3.
38
Degree of Freedom of TES Tropospheric Ozone
Retrievals
1.2-2.2 DOF
Run 2151 Sep. 20, 2004
Information mainly from the middle and the upper
troposphere but also at some places from the
lower troposphere
300
700
39
Difference between GEOS-CHEM and TES is
significant in the upper troposphere compared
against TES retrieval error
TES Retrieval Error Run 2151
( TES GC ) / Retrieval Error
200
350
500
700
Typical TES retrieval error in the troposphere is
5-35 .
40
Global Survey on September 21, 2004 Improved
Calibration
OLD
NEW
100 hPa
100 hPa
150 hPa
150 hPa
41
Global Survey on September 21, 2004 Improved
Calibration
OLD
NEW
200 hPa
200 hPa
350 hPa
350 hPa
42
Global Survey on September 21, 2004 Improved
Calibration
OLD
NEW
500 hPa
500 hPa
700 hPa
700 hPa
43
Special Observations of Tropical Ozone over South
Atlantic Runs 2151, 2199, 2210, 2219, 2228
(Sep.-Oct. 2004 Step Stare)
( TES GC ) / GC
TES

GC
Bias in
RMS in ppbv
44
November
350 hPa
500 hPa
700 hPa
45
Scatterplot O3 TES vs O3 GEOS-CHEM (Tropics)
350 hPa
Data 6 GS global surveys November 2004 RMS
ppbv BIAS (TES-GEOS-CHEM)/GEOS-CHEM
500 hPa
700 hPa
46
November
November
350 hPa
500 hPa
700 hPa
47
Scatterplot O3 TES vs O3 GEOS-CHEM
350 hPa
Data 6 GS global surveys November 2004 RMS
ppbv BIAS (TES-GEOS-CHEM)/GEOS-CHEM
500 hPa
700 hPa
48
Sep./Oct.
350 hPa
500 hPa
700 hPa
49
November
350 hPa
500 hPa
700 hPa
50
January
350 hPa
500 hPa
700 hPa
51
Ozone evolution between October and January
comparison with SHADOZ at Ascension and Natal
Ascension 7.98 S, 14.42 W
Comparison
SHADOZ (2000-2004)
TES /-1.5 lat,/-1.5 lon
SHADOZ TES Averaging Kernels
TES overestimates ozone in the UT (plt300 hPa)
and underestimates ozone in the MT and LT (p
gt300 hPa) in Oct. and Nov . January values are
in better agreement. TES captures the decrease of
ozone between October and January but only above
500 hPa.
52
Natal 5.42 S, 35.38 W
Comparison
(2000-2004)
SHADOZ
/-1.5 lat,/-1.5 lon
TES
SHADOZ TES Averaging Kernels
Overestimation of ozone in the UT and
underestimation of ozone in the LT TES captures
the ozone variation between September and January

53
Tropospheric Ozone columns (TOC) Nov. 2004 (6
Global Surveys)
54
Latitudinale variation of TOC in the tropics
TES zonal mean
GEOS-CHEM zonal mean
Mean TOC Runs November
mean of the TES error (max of the error not
representative)
55
Tropical Tropospheric Ozone Columns (TTOCs) over
Atlantic
TES
GEOS-Chem
56
Comparison with a previous simulation of GEOS-CHEM
TOC (DU)
NOV 2004
(GEOS 4)
Liu et al. (2005)
NOV 1997
(GEOS-STRAT)
TOC (DU)
57
Tropical Tropospheric Ozone Columns
(TTOCs) GEOS-4 v.s. GEOS-STRAT Simulations
GEOS-STRAT
GEOS-4
Liu et al. 2005
SON 2004
O3 column DU
Trop. O3 column DU
SON 1997
Trop. O3 column DU
Trop. O3 column DU Martin et al. 2002
(GEOS-STRAT)
58
GEOS-Chem v.s. Ozonesonde at Ascension Island
Red SHADOZ ozonesonde Blue NRT GEOS-Chem
simulation from Solene Black GEOS-Chem
simulation with scaled up lightning NOx emissions
GEOS-CHEM underestimates free troposphere ozone
but captures the enhancement in the middle and
upper troposphere.
59
Comparison between GEOS-CHEM and SHADOZ ozone
SHADOZ
GEOS-CHEM mean sept.,oct.,nov., 2004 SHADOZ mean
2000-2004
GEOS-CHEM
200
500
200
500
ppbv
60
Comparison between GEOS-CHEM and SHADOZ ozone
GEOS-CHEM mean sept.,oct.,nov., 2004 SHADOZ mean
2000-2004
SHADOZ
GEOS-CHEM
200
500
200
500
Ozone (ppbv)
61
Comparison between GEOS-CHEM and SHADOZ ozone
SHADOZ
GEOS-CHEM mean sept.,oct.,nov., 2004 SHADOZ mean
2000-2004
GEOS-CHEM
200
500
200
500
Ozone (ppbv)
62
Comparison between GEOS-CHEM and SHADOZ ozone
SHADOZ
GEOS-CHEM mean sept.,oct.,nov., 2004 SHADOZ mean
2000-2004
GEOS-CHEM
200
200
Ozone (ppbv)
63
Issues with GEOS4
Tropical ozone in UT too low compared to SHADOZ
for continental stations or close to continents.
linked to the altitude of the convection top
too low in particular in the tropics compared to
GEOS 3 (Rynda Hudman )
Rynda comparison
20
tropopause height
10
-50
50
-50
50
Latitude
64
2001 Sept.
2001 Sept.
Lightning NOx distributed too low in altitude
/GEOS3 Decrease of the emission in the tropics
between July and October (-20 )
/GEOS3 Additionnal potential problems with the
convective transport of ozone and precurseurs.
65
Summary Conclusions

• High correlation between TES retrieved and
  GEOS-Chem simulated tropospheric ozone.
• Largest difference in the upper tropopshere
  systematic high bias in TES retrieval in the
  upper troposphere and a low in GEOS-Chem values
  in the corresponding region.
• New calibration improves the comparison in the
  upper troposphere with no significant impact in
  the lower troposphere.
• Issues with GEOS-Chem (GEOS-4) simulation
  Tropical ozone (over the Atlantic) in the middle
  and upper troposphere too low compared with
  SHADOZ ozonesonde data problem with deep
  convection in GEOS-4.

66
(No Transcript)
67
Comparison of the Total Ozone Column TES vs
(SHADOZ SBUV)
Total Ozone column
Total Ozone column
TES /- 5 SHADOZ stations
TES /- 1.5 SHADOZ stations