TROPOSPHERIC OZONE AS A CLIMATE GAS, TROPOSPHERIC OZONE AS A CLIMATE GAS, AND SATELLITE OBSERVATIONS OF ITS PRECURSORSAND SATELLITE OBSERVATIONS OF ITS PRECURSORS

Daniel J. Jacob

GISS GCM

Atmosphericchemistry

Aerosolmicrophysics

emissions land use climate forcing

CACTUS model

climate chemistry

CHEMISTRY, AEROSOLS, AND CLIMATE:CHEMISTRY, AEROSOLS, AND CLIMATE:TROPOSPHERIC UNIFIED SIMULATION (CACTUS)TROPOSPHERIC UNIFIED SIMULATION (CACTUS)

A NASA Interdisciplinary Science (IDS) investigation:Harvard (Jacob, P.I.), Caltech (Seinfeld), NASA/GISS (Rind), UC Irvine (Prather)

“…Kyoto also failed to address two major pollutants that have an impact on warming:  black soot and tropospheric ozone.  Both are proven health hazards.  Reducing both would not only address climate change, but also dramatically improve people's health.” (George W. Bush, June 11 2001 Rose Garden speech)

GLOBAL BUDGET OF TROPOSPHERIC OZONEGLOBAL BUDGET OF TROPOSPHERIC OZONE

O3

O2 h

O3

OH HO2

h, H2O

Deposition

NO

H2O2

CO, CH4, VOC

NO2

h

STRATOSPHERE

TROPOSPHERE

8-18 km

Chem prod in troposphere

4900 Chem loss in troposphere

4200

Transport from stratosphere

500 Deposition 1200

Global sources and sinks, Tg O3 yr-1 (GEOS-CHEM model)

Climatology of observed ozone at 400 hPa in July from ozonesondes and MOZAIC aircraft (circles) and corresponding GEOS-CHEM model results for 1997 (contours).

GEOS-CHEM tropospheric ozone columns for July 1997

GLOBAL DISTRIBUTION OF TROPOSPHERIC OZONEGLOBAL DISTRIBUTION OF TROPOSPHERIC OZONE

Li et al. [2001]

Lifetime is ~ weeks

FOSSIL FUEL 23.1AIRCRAFT

0.5BIOFUEL 2.2

BIOMASSBURNING 5.2

SOILS 5.1

LIGHTNING 5.8

STRATOSPHERE 0.2 ANIMALS

90

LANDFILLS50

GAS60

COAL40TERMITES

25

RICE85

WETLANDS180

BIOMASSBURNING20

NOx (Tg N yr-1) METHANE (Tg C yr-1)

ANTHROPOGENIC INCREASE IN TROPOSPHERIC OZONE ANTHROPOGENIC INCREASE IN TROPOSPHERIC OZONE DRIVEN BY NODRIVEN BY NOxx AND METHANE EMISSIONS AND METHANE EMISSIONS

PRESENT-DAY EMISSIONS

RISE IN TROPOSPHERIC OZONE OVER 20RISE IN TROPOSPHERIC OZONE OVER 20thth CENTURY CENTURY

Preindustrialozone models

}Observations at mountain sites in Europe [Marenco et al., 1994]

RADIATIVE FORCING FROM TROPOSPHERIC OZONERADIATIVE FORCING FROM TROPOSPHERIC OZONE

But how good is radiative forcing as an indicator of climate change, when this forcing is so heterogeneous?

Annual mean values (0.49 W m-2 globally)

(9.6 m)

Mickley et al. [1999] 0.46

IPCC [2001] range 0.3-0.5

From 19th century obs

[Mickley et al., 2001]

0.80

Global radiative forcing F (W m-2)

GISS GCM ANALYSIS OF CLIMATE RESPONSE TO GISS GCM ANALYSIS OF CLIMATE RESPONSE TO TROPOSPHERIC OZONE CHANGE OVER 20TROPOSPHERIC OZONE CHANGE OVER 20thth CENTURY CENTURY

GCM equilibrium simulation for present-day climate with present vs. preindustrial tropospheric ozone; includes “Q-flux” ocean

equilibriumclimate

T = 0.3oC

F = 0.49 W m-2

L.J. Mickley, Harvard

present-day ozone

Preindustrial ozone

Also sensitivity simulations with O3 = 18 ppb, CO2 = 25 ppm, giving same F

INHOMOGENEITY OF CLIMATE RESPONSE INHOMOGENEITY OF CLIMATE RESPONSE TO OZONE CHANGE OVER 20TO OZONE CHANGE OVER 20thth CENTURY CENTURY

• Greater warming in northern hemisphere

•Strong cooling in stratosphere:

Surface

Troposphericozone

9.6 m

Stratosphericozone

L.J. Mickley, Harvard

LOWER STRATOSPHERIC COOLING FROM TROPOSPHERIC LOWER STRATOSPHERIC COOLING FROM TROPOSPHERIC OZONE IS STRONGEST IN ARCTIC WINTER OZONE IS STRONGEST IN ARCTIC WINTER

GCM temperature change in lower stratosphere in DJF (oC) from increasing tropospheric ozone over 20th century

particularly sensitive region for recovery of ozone layer!

L.J. Mickley, Harvard

CLIMATE RESPONSE EXPERIMENTS WITH IDENTICAL CLIMATE RESPONSE EXPERIMENTS WITH IDENTICAL GLOBAL RADIATIVE FORCINGS (0.49 W mGLOBAL RADIATIVE FORCINGS (0.49 W m-2-2) FROM:) FROM:

1. tropospheric ozone2. uniform tropospheric ozone (18 ppv)3. carbon dioxide (25 ppmv)

• CO2 is a more effective warming agent at surface

• In lower stratosphere, CO2 causes warming while tropospheric ozone causes cooling

L.J. Mickley, Harvard

WHY IS COWHY IS CO22 MORE EFFECTIVE THAN OZONE MORE EFFECTIVE THAN OZONE

FOR SURFACE WARMING AT SAME RADIATIVE FORCING?FOR SURFACE WARMING AT SAME RADIATIVE FORCING?Correlation of forcing with 500 hPa humidity in tropics (25N-25S)

Overlap of CO2 and H2O bands causes CO2 forcing to shift poleward where ice feedback enhances warming

FCO2 – FO3

L.J. Mickley, Harvard

Ozone CO2

GCM SURFACE WARMING PATTERNS (GCM SURFACE WARMING PATTERNS (ooC) FROM INCREASING C) FROM INCREASING TROPOSPHERIC OZONE OVER 20TROPOSPHERIC OZONE OVER 20thth CENTURY – JJA SURFACE CENTURY – JJA SURFACE

Difference

Tropospheric ozone Equivalent uniform CO2

(white = insignificant or high altitude)

Largest warmings downwind of ozone source regions – ozone there is more effective than CO2

L.J. Mickley, Harvard

USING SATELLITE MEASUREMENTS OF NOUSING SATELLITE MEASUREMENTS OF NO22 AND HCHO AND HCHO

COLUMNS (SOLAR BACKSCATTER) TO MAP NOCOLUMNS (SOLAR BACKSCATTER) TO MAP NOx x AND AND

VOC EMISSIONSVOC EMISSIONS

Emission

NOh (420 nm)

O3, RO2

NO2

HNO3

1 day

NITROGEN OXIDES (NOx) VOLATILE ORGANIC CARBON (VOC)

Emission

VOC

OHHCHOh (340 nm)

hoursCO

hours

BOUNDARYLAYER

~ 2 km

Tropospheric NO2 column ~ ENOx

Tropospheric HCHO column ~ EVOC

Deposition

GOME satellite instrument

Instrumentsensitivity w()(“scattering weight”)

Vertical shapefactor S()(normalized mixing ratio)

what GOMEsees

AMFG = 2.08actual AMF = 0.71

IN SCATTERING ATMOSPHERE, THE AIR MASS FACTOR IN SCATTERING ATMOSPHERE, THE AIR MASS FACTOR (AMF) OF A SOLAR BACKSCATTER MEASUREMENT (AMF) OF A SOLAR BACKSCATTER MEASUREMENT DEPENDS ON VERTICAL DISTRIBUTION OF THE GASDEPENDS ON VERTICAL DISTRIBUTION OF THE GAS

1

0

( ) ( )GAMF AMF w S d

Illustrative retrieval of HCHO column at 340 nm

Palmer et al. [2001]

geometric

USE GLOBAL 3-D MODEL DRIVEN BY ASSIMILATED USE GLOBAL 3-D MODEL DRIVEN BY ASSIMILATED METEOROLOGICAL DATA TO PROVIDE AMFsMETEOROLOGICAL DATA TO PROVIDE AMFs

FOR EVERY SATELLITE VIEWING SCENE FOR EVERY SATELLITE VIEWING SCENE

SATELLITE DATA

SLANTCOLUMN

GEOS-CHEM GLOBAL 3-D MODEL OF TROPOSPHERIC CHEMISTRY:

provides S(

AMF

VERTICALCOLUMN

VERTICALCOLUMN

• Best information applied to each scene• Consistency in comparing model and observed columns• Apply with any 3-D model (recalculate AMFs using tabulated scattering weights)

ADVANTAGES OF 3-D MODEL APPROACHFOR COMPUTING AMFs

spectralfit

LIDORT RAD.TRANSFER MODEL:

provides w()

HCHO COLUMNS FROM GOME OVER U.S.:HCHO COLUMNS FROM GOME OVER U.S.:July 1996 meansJuly 1996 means

BIOGENIC ISOPRENE IS THE MAIN SOURCE OF HCHO IN U.S. IN SUMMER

Palmer et al. [2001]

GEIAisopreneemissions

R = 0.83Bias 14%

Precision:4x1015 cm-2

GOMEGOMEslantslant

GOMEGOMEverticalvertical

GEOS-CHEMGEOS-CHEMverticalvertical

DifferenceDifference

MAPPING OF ISOPRENE MAPPING OF ISOPRENE EMISSIONS FOR JULY 1996 EMISSIONS FOR JULY 1996 BY SCALING OF GOME BY SCALING OF GOME FORMALDEHYDE COLUMNS FORMALDEHYDE COLUMNS [Palmer et al., 2002][Palmer et al., 2002]

GEIA (IGAC inventory)

BEIS2(official EPA inventory)

GOME

COMPARE TO…

SEASONAL VARIABILITY OF HCHO COLUMNS (9/96-8/97)SEASONAL VARIABILITY OF HCHO COLUMNS (9/96-8/97)- proxy for isoprene emissions -- proxy for isoprene emissions -

SEP

AUG

JUL

OCT

MAR

JUN

MAY

APR

Dorian Abbott (Harvard)

GOME GEOS-CHEM GOME GEOS-CHEM

GOME Tropospheric NO2 GEOS-CHEM Tropospheric NO2

1015 molecules cm-2

DJF 96-97

MAM 1997

JJA 1997

SON 1996

r = 0.75 bias=5%

R.V. Martin

with a priori emissions (scaled GEIA)

OPTIMIZED NOOPTIMIZED NOxx EMISSION INVENTORY FROM GOME EMISSION INVENTORY FROM GOME

A POSTERIORI A PRIORIR.V. Martin, Harvard

TRACE-P AIRCRAFT MISSION (March-April 2001):TRACE-P AIRCRAFT MISSION (March-April 2001):top-down constraints on Asian emissionstop-down constraints on Asian emissions

CO fossil and biofuel bottom-up emissions (D.R. Streets, ANL)

Daily CO biomass burning emissions from AVHRR (C.L. Heald, Harvard)

Chemical forecastsCTMs

MOPITT COObservations(IR emission)

CO aircraft observations(G.W. Sachse)

Jacob et al. [2002]

MOPITT VALIDATION PROFILES DURING TRACE-PMOPITT VALIDATION PROFILES DURING TRACE-P

aircraftAircraft w/MOPITT av kernels

MOPITT

Averagingkernels

MEAN MOPITT MEAN MOPITT CO COLUMN DATA CO COLUMN DATA DURING TRACE-P DURING TRACE-P (Mar-Apr 2001)(Mar-Apr 2001)

MOPITT

GEOS-CHEMmodel w/av kernels

DifferenceC.L. Heald, Harvard

• FTIR spectrometer (3.3 - 15.4 m) to be launched on Aura satellite in 2004 (P.I. Reinhard Beer, JPL)

• Field of view: 0.5x8 km2

Std. products Nadir Limb

Temperature X X

Surface temp. X

Land surf. emissivity

X

O3 X X

CO X X

H2O X X

CH4 X X

NO X

HNO3 X

TROPOSPHERIC EMISSION SPECTROMETER (TES)TROPOSPHERIC EMISSION SPECTROMETER (TES)

Averaging kernels

“True” profile(GEOS-CHEM)

CO retrieval

a priori

retrieved

D.B. Jones, Harvard

About 3 pieces of information on vertical profile: ~ 1 more than MOPITT

TES NADIR RETRIEVAL OF COTES NADIR RETRIEVAL OF CO

WILLTES NADIR CO OBSERVATIONS IMPROVEWILLTES NADIR CO OBSERVATIONS IMPROVEOUR ABILITY TO QUANTIFY REGIONAL CO EMISSIONS?OUR ABILITY TO QUANTIFY REGIONAL CO EMISSIONS?

Assume that we know the true sources of CO

From these sources of CO.Use GEOS-CHEM model to simulate “true” CO concentration field

Sample model along TES orbit tracks, apply TES averaging kernels and noise to simulate retrieval of this CO field

Make a priori (deliberately wrong)

estimate of CO sourcesby applying errors

to the “true” source

Apply optimal estimation

inverse method

Obtain a posteriori sources and errors;

How successful are we at finding the true

source and reducing the error?

1 day GEOS-CHEM CO data (March 10, 2001) along TES orbit at 500 hPa

CONSTRUCTING THE MODEL ERROR COVARIANCE MATRIXCONSTRUCTING THE MODEL ERROR COVARIANCE MATRIXUSING TRACE-P OBSERVATIONS AND FORECASTSUSING TRACE-P OBSERVATIONS AND FORECASTS

• Assume that difference between successive GEOS-CHEM CO forecasts during TRACE-P (to+48h and to + 24 h) describes the covariant error structure (“NMC method”)

• Assume that mean bias in GEOS-CHEM model simulation of CO in TRACE-P is due to emission errors, and that residual relative error (RRE) is due to transport; use RRE to scale covariant error structure.

• Add representativeness error of 5% (small) based on subgrid variablity in TRACE-P aircraft data

D.B. Jones and P. I. Palmer, Harvard

CO TRACE-P dataGEOS-CHEM error,TRACE-P simulation

Forecast error (blue) and scaled model transport error (red)

MODEL TRANSPORT ERRORSMODEL TRANSPORT ERRORS(DIAGONALS OF ERROR COVARIANCE MATRIX)(DIAGONALS OF ERROR COVARIANCE MATRIX)

RELATIVE ERROR850 hPa,

March-April 2001

D.B. Jones, Harvard

CHEM

NAFF

RWFF RWBB

AFBB

EUFF

SABB

ASFFASBB

NAFF: 121.3 SABB: 96.5EUFF: 131.1 RWBB: 98.0ASFF: 258.3 RWFF: 149.8ASBB: 96.0 CHEM: 1125.0AFBB: 193.9

“True” Emissions (Tg/yr)

Total = 2270 Tg/yr

• FF = Fossil Fuel + Biofuel• All sources include contributions from oxidation of VOCs; assume 50% error• OH is specified• Use a “tagged CO” method to estimate contribution from each source (Jacobian matrix)• Use linear inverse method to solve for annually-averaged emissions

D. B. Jones, Harvard

REGIONAL CO SOURCES TO BE RETRIEVED REGIONAL CO SOURCES TO BE RETRIEVED IN TES SIMULATION EXERCISEIN TES SIMULATION EXERCISE

INVERSION RESULTS (5 days of TES data, Mar 10-15 2001)INVERSION RESULTS (5 days of TES data, Mar 10-15 2001)

a priori a posteriori true

• With Gaussian unbiased error statistics, TES is extremely powerful for constraining global sources of CO (because it provides lots of data!)• Still very powerful when using only data above 500 hPa• Realization of this potential rests on quantification of error statistics, particularly for the model transport error – a very difficult problem!

D. B. Jones, Harvard