Xiong Liu ( [email protected]) [email protected] Harvard-Smithsonian Center for Astrophysics Collaborators: Kelly Chance, Christopher Sioris, Robert.

Xiong Liu ([email protected])

Harvard-Smithsonian Center for Astrophysics

Collaborators: Kelly Chance, Christopher Sioris, Robert Spurr, Thomas Kurosu (CFA)

Randall Martin (Dalhousie Univ., Canada)Mike Newchurch (University of Alabama in Huntsville)

May Fu, Jennifer Logan, Daniel Jacob, Paul Palmer (Harvard University)PK Bhartia (Atmospheric Chemistry and Dynamics Branch, NASA GSFC )

Rob Chatfield (Atmospheric Chemistry and Dynamics Branch, NASA AMES )

Beijing, China

May 26, 2004

First Directly-Retrieved Global Distribution of Tropospheric Column Ozone from GOME

2

Outline

Introduction to Ozone and Tropospheric Ozone Satellite-based Tropospheric Ozone Retrievals Algorithm Description Intercomparison with TOMS, Dobson, and

Ozonesonde observations Examples of Daily Retrievals Global distribution of Tropospheric Column Ozone

and comparison with the 3D GEOS-CHEM model Summary and Future work Atmospheric Measurements and Studies at Harvard-

Smithsonian CFA (Time allowed)

3

Ozone

Stratopause

Ozonelayer

Tropopause

Courtesy of Randall Martin

First discovered by Schönbein [1840], a reactive oxidant in the atmosphere.

First quantitative observation: early 20th century in Europe (e.g., Dobson, Götz)

First discovery of ozone hole by Farman et al. [1985]

First attempts to understand ozone in the 1930s. Until the last 30-40 years, the stratospheric ozone chemistry (NOx, HOx, ClOx) is well understood.

Noble chemistry prizes were awarded to Paul Crutzen, Mario Molina, and Sherwood Rowland in 1995.

4

Tropospheric Ozone

CO, VOCs, NOx

HO2OH

NO2

H2O2

O3

hv, H2O

HNO3

OH

CO, VOCs

NOx

HOx

Simplified Tropospheric O3 Chemistry

Courtesy of Randall Martin

Key species in climate, air quality, and tropospheric chemistry Major Greenhouse gas, 15-20% of climate radiative forcing Primary constituents of photochemical smog Largely controls tropospheric oxidizing capacity

NO

5

Residual-based Satellite Tropospheric Ozone Retrievals Why satellite observations: global coverage

Challenge: only 10% of Total column Ozone (TO)

Residual-based approaches: TO – Strat. Column Ozone (SCO)

Tropospheric ozone residual: TOMS minus SAGE/SBUV/HALOE/MLS [Fishman et al., 1990, 2003, Ziemke et al., 1998, Chandra et al., 2003]

Cloud/clear difference techniques: [Ziemke et al., 1998; Newchurch et al., 2003, Valks et al., 2003]

Modified residual method [Kim et al., 1996; Hudson et al., 1998]

Topographic contrast method [Jiang et al., 1996, Kim and Newchurch et al., 1996, Newchurch et al., 2003]

Scan-angle method (Special) [Kim et al., 2001]

Limits: poor spatiotemporal resolution and large spatiotemporal variability of SCO mostly climatological or tropics

Limb/Nadir matching: TO/SCO from same instrument or satellite

6

Launched April 1995 on ERS-2

Nadir-viewing UV/vis/NIR

240-400 nm @ 0.2 nm

400-790 nm @ 0.4 nm

Footprint 320 x 40 km2

10:30 am cross-equator time

Global coverage in 3 days

ESA Global Ozone Monitoring Experiment

7

GOME Radiance Spectrum and Trace Gases Absorption

1E+08

1E+10

1E+12

1E+14

1E+16

200 300 400 500 600 700 800Wavelength [nm]

Inte

nsity [a

rbitra

ry u

nits]

O3 UV

O3 vis

HCHO

OClO

O4

O2

H2O

SO2 NO2

BrO

Satellite group: http://giger.iup.uni-heidelberg.de/

Atmospheric trace gas absorptions detected in satellite spectra

8

Physical Principles of Ozone Profile Retrieval (UV/Vis.)

Wavelength-dependent O3 absorption & dependence of Rayleigh scattering provide discrimination of O3 at different altitudes from backscattered measurements.

Temperature-dependent ozone absorption in the Huggins bands provides additional tropospheric ozone information.

Hartley & Huggins bands (245-355 nm) Huggins bands (318-340 nm)

Chappuis bands (400-800 nm)

Chance et al., 1997

9

Ozone Profile Retrieval from GOME

Direct tropospheric ozone retrieval: daily global distribution of tropospheric ozone without other SCO measurements or deriving SCO

Several groups [Munro et al., 1998; Hoogen et al., 1999; Hasekamp et al., 2001; van der A et al, 2002; Muller et al., 2003] have developed ozone profile retrieval algorithms from GOME: each of them demonstrates that limited tropospheric ozone information can be detected.

However, global distribution of tropospheric column ozone has not been published from these algorithms Require accurate and consistent calibrations. Need to fit the Huggins bands to high precision. Tropospheric column ozone is only ~10% of total column ozone Limited Vertical Resolution

10

Algorithm Description Ill-posed problem: non-linear optimal estimation [Rodgers, 2000]

Y: Measurement vector (e.g., radiances)

X, Xi, Xi+1: State vector (e.g. ozone profile)

Xa: a priori state vector

K : Weighting function matrix, sensitivity of radiances to ozone

Sa: A priori covariance matrix

Sy: Measurement error covariance matrix

2 2

2

2 2

{ ( - ) -[ ( )]} ( - ) a

1 1- -2 2

y i i+1 i i i+1 aS K X X Y - R X S X X

( ) { [ - ( )] - ( )} a a

T -1 -1 -1 T -1 -1i+1 i i y i i y i i aX X K S K S K S Y R X S X - X

11

Algorithm Description — Radiative Transfer Simulation

RcRo

IB,o IB,c

Pc

Rs

d

Radiative transfer model: LIDORT [Spurr et al., 2001] Model Ring effect with a first-order single-scattering model Radiance polarization correction with a look-up table

Forward model inputs SAGE strat. [Bauman et al., 2003] & GOCART trop. aerosols [Chin et al., 2002]Daily ECMWF T profiles and NCEP PsClouds: Lambertian surfacesCloud-top pressure from GOMECAT [Kurosu et al., 1999] Cloud fraction derived at 370.2 nm with surface albedo database [Kolemeijer et al.,2003]Wavelength dependent albedo (2-order polynomial) from 326-339 nm to take account of residual aerosol and cloud effects

12

Algorithm Description Perform external wavelength and radiometric calibrations

Derive variable slit widths and shifts between radiances/irradiances Co-add adjacent pixels from 289-307 nm to reduce noise Perform undersampling correction with a high-resolution solar reference Fit degradation for 289-307 nm on line in the retrieval

Optimize fitting windows: 289-307 nm, 326-339 nm

Latitude/monthly dependent TOMS V8 climatology

Retrieval Grid: 11 layers, use daily NCEP tropopause to divide the troposphere and stratosphere, 2-3 tropospheric layers

Tropospheric column ozone: sums of tropospheric partial columns

State Vector: 47 variables (ozone, Ring, surface albedo, undersampling, degradation, wavelength shifts, NO2, SO2, BrO)

Spatial resolution: 960×80 km2

13

Averaging Kernels (DX’/X)

VR: 7-12 km (at 10-37 km)7-12 km (at 7-37 km)

8-12 km (at 20-38 km)

14

A Priori Influence

A Priori influence in TCO: 15% in the tropics, 50% at high-latitudes

15

Retrieval Errors

Precision: 2-8% (< 2DU) in the strat., <12%(5DU) in the troposphereSmoothing: 10% at 20-40 km, 15% at > 40 km, and 30% at <10 km

TO: <2 DU(0.5); 3 DU (1.0%)

SCO: <2 DU(1%); 2-5 DU (1-2%)

TCO: 1.5-3 DU(6-12%); 3-6 DU(12-25%)

16

An Orbit of Retrieved Ozone Profiles

Ozone Hole (120 DU)Biomass burning

17

Validation GOME data are collocated at 33 WOUDC ozonesonde stations during 96-99. Validate retrievals against TOMS V8, Dobson/Brewer total ozone, and ozonesonde TCO. Data mostly from WOUDC Collocation criteria:

Within ~8 hours, 1.5° latitude and ~500 km in longitude Average all TOMS points within GOME footprint

Number of comparisons: 4711, 1871, and 1989 with TOMS, Dobson, and ozonesonde, respectively.

http://www.woudc.orghttp://croc.gsfc.nasa.giv/shadozhttp://toms.gsfc.nasa.govhttp://www.cmdl.noaa.gov

18

Total Column Ozone Comparison

GOME-TOMS/Dobson: within retrieval uncertainties and saptiotemporal variability.

Means Biases: <6 DU (2%) at most stations

1: 2-4 DU (1.5%) in the tropics, <6.1 DU (2.4%) at higher latitudes

Means Biases: <5 DU (2%) at most stations

1: 3-6 DU (<3%) in the tropics, <8-16 DU (<5%) at higher latitudes

A Priori Retrieval Dobson TOMS

19A Priori Retrieval Ozonesonde

GOME-SONDE within retrieval uncertainties.

Biases: <4 DU (15%) except –5.5, 4.4, 5.6 DU (16-33%) at NyÅlesund, Naha, Tahiti

1 : 3-7 DU (13-28%)

A Priori Retrieval Ozonesonde

Capture most of the temporal variability

GOME-SONDE within retrieval uncertainties.

Biases: <3.3 DU (15%)

1 : 3-8 DU (12-27%)

Tropospheric Column Ozone Comparison

20

GEOS-CHEM global 3D tropospheric chemistry and

transport model

Driven by NSAA GEOS-STRAT GMAO met data [Bey et al., 2001]

22.5o resolution/26 vertical levels

O3-NOx-VOC chemistry

Recent anthropogenic, biogenic, natural emissions

Synoz flux: 475 Tg O3 yr-1 from stratosphere

A 18-month simulation (June 1996-Nov 1997)

21

Examples of Daily Retrievals

22

Examples of 3-Day Composite Global Maps

Biomass burning

Mid-latitudesHigh TCO Band

LOW TCO over the Pacific

High-latitudehigh TCO

Transport of mid-latitude high TCO air to the tropics

23

Monthly-mean Tropospheric Column Ozone (12/96-11/97)

24

GOME vs. GEOS-CHEM

Similar overall structuresGlobal biases:

<2±4 DU, r=0.82-0.9

SH:

<1±2 DU,r=0.94-0.98

NH:

<4.3±4.6 DU, r=0.6-0.8

25

GOME vs. GEOS-CHEM Usually within 5 DU.

Large positive bias of 5-15 DU at some northern tropical and subtropical regions: central America, tropical North Africa, Southeast Asia, Middle East

Usually >0.6.

Poor correlation: central America, equatorial remote Pacific, tropical North Africa and Atlantic, North high latitudes

26

GOME/GEOS-CHEM vs. MOZAIC (Central America)

20

30

40

50

MOZAIC: www.aero.obs-mip.fr/mozaicData: 1994-2004 (vary from location to location)Evaluate GOME/GEOS-CHEM TCO in seasonality

27

GOME/GEOS-CHEM vs. MOZAIC (Southeast Asia)

A Priori Retrieval GOES-CHEM MOZAIC

28

GOME/GEOS-CHEM vs. MOZAIC (Accra)

A Priori Retrieval GOES-CHEM MOZAIC

29

GOME/GEOS-CHEM vs. MOZAIC (Middle East)

A Priori Retrieval GOES-CHEM MOZAIC

30

Summary Ozone profiles and Tropospheric Column Ozone (TCO) are retrieved

from GOME using the optimal estimation approach. Retrieved TO and TCO compare very well with TOMS,

Dobson/Brewer, and ozonesonde measurements. The retrievals clearly show signals due to convection, biomass

burning, stratospheric influence, pollution, and transport, and are capable of capturing the spatiotemporal evolution of TCO in response to regional or short time-scale events.

The overall structures between GOME and GEOS-CHEM are similar, but some significant positive biases occur at some northern tropical and subtropical regions.

The GOME retrievals usually agree well with the MOZOAIC measurements, to within the monthly variability and some biases can be explained by the reduced sensitivity to lower tropospheric ozone, spatiotemporal variation, and the large spatial resolution of GOME retrievals.

31

Future Work

Improve retrieval algorithms (Chappuis bands, external degradation correction) and complete more than 8-year GOME data record.

Apply the algorithm to SCIMACHY, OMI, GOME-2, OMPS, or future geostationary satellite measurements.

Integrate with the GEOS-CHEM model and other in-situ data, improve our understanding of global/regional budget of tropospheric ozone

Tropospheric ozone radiative forcing

Acknowledgements This study is supported by the NASA and by the Smithsonian Institution. Thank all collaborators. We thank WOUDC and its data providers (e.g., SHADOZ, CMDL), TOMS, MOZAIC for providing correlative measurements. We are grateful to NCEP/NCAR ECMWF reanalysis projects. We appreciate the ongoing cooperation of the European Space Agency and the Germany Aerospace Center in the GOME program.

32

Variable Slit Widths and Shifts

33

Fitting Residuals

34

Aerosol Effects

35

Comparison with SAGE-II (>15 km)

Comparison with SAGE-II V6.2 ozone profiles above 15 km during 1996-1997 (5732)

Means Biases: <15%

1: <10% for the top seven layers and <15% for the bottom layer

Systematic biases in the retrievals but not in the a priori, suggesting residual measurement errors in the GOME level-1 data

36

A Priori Influence (06/7-9/1997)TOMS V8 A Priori

GEOS-CHEM A Priori

Retrieval with TOMS V8 A Priori

Retrieval with GEOS-CHEM A Priori

37

GOME TCO (Dec 96-Nov 97)

38

GOME/GEOS-CHEM vs. MOZAIC