Multi-model estimates for

1
Multi-model estimates for intercontinental
transport of ozone pollution in the northern
hemisphere (and uncertainties therein)
Arlene M. Fiore (arlene.fiore_at_noaa.gov)
F. Dentener, O. Wild, C. Cuvelier, M. Schultz, D.
Reidmiller, C. Textor, M. Schulz, C. Atherton, D.
Bergmann, I. Bey, G. Carmichael, W. Collins, R.
Doherty, B. Duncan, G. Faluvegi, G. Folberth, M.
Garcia Vivanco, M. Gauss, S. Gong, D.
Hauglustaine, P. Hess, T. Holloway, L. Horowitz,
I. Isaksen, D. Jacob, D. Jaffe, J. Jonson, J.
Kaminski, T. Keating, A. Lupu, I. MacKenzie, E.
Marmer, V. Montanaro, R. Park, K. Pringle, J.
Pyle, M. Sanderson, S. Schroeder, D. Shindell, D.
Stevenson, S. Szopa, R. Van Dingenen, P. Wind, G.
Wojcik, S. Wu, G. Zeng, A. Zuber
GFDL lunchtime seminar, January 21, 2009
2
Evidence of intercontinental transport at
northern midlatitudes 2001 Asian dust event
Dust leaving the Asian coast in April 2001
Image c/o NASA SeaWiFS Project and ORBIMAGE
Reduced Visibility from Transpacific Transport of
Asian Dust
Glen Canyon, Arizona, USA
3
Hemispheric scale contribution of major source
regions to summertime surface ozone difficult
(impossible?) to observe directly
North America
Estimated with model simulations that zero out
anthropogenic emissions of O3 precursors within
the source region
Europe
GEOS-Chem Model 4x5 horizontal resolution Li et
al., JGR, 2002
Asia
4
U.N. Economic Commission for EuropeConvention on
Long-Range Transboundary Air Pollution (CLRTAP
established 1979)
51 parties in Europe, North America, and Central
Asia
Co-chairs Terry Keating (U.S. EPA) and André
Zuber (EC)
TF HTAP Mission Develop a fuller understanding
of hemispheric transport of air
pollution to inform future negotiations under
CLRTAP
5
Multi-model assessment involving gt25 modeling
groups
OBJECTIVES Quantify S-R relationships for HTAP
regions and assess uncertainties
in these estimates
HTAP S-R Regions

• Focus species
• Ozone and precursors
• Aerosols and precursors
• Mercury
• Persistent Organic Pollutants
• Idealized Tracers
• Oxidized Nitrogen

PRODUCTS 2007 Interim Report to inform the
review of the 1999 CLRTAP Gothenburg Protocol
to abate acidification, eutrophication, and
tropospheric ozone (www.htap.org). 2010
Assessment Report to inform the CLRTAP on
hemispheric air pollution and S-R relationships.
6
Wide range in prior estimates of
intercontinentalsurface ozone source-receptor
(S-R) relationships
NORTH AMERICA ? EUROPE
ASIA ? NORTH AMERICA
annual mean
events
events
seasonal mean
Surface O3 contribution from foreign region
(ppbv)
seasonal mean
annual mean
Studies in TF HTAP 2007 Holloway et al.,
2008 Duncan et al., 2008 Lin et al., 2008
Assessment hindered by different (1) methods, (2)
reported metrics, (3) meteorological years, (4)
regional definitions Few studies examined all
seasons
7
Objective Quantify assess uncertainties in
N. mid-latitude S-R relationships for ozone
TF HTAP REGIONS

• BASE SIMULATION (21 models)
• ? horizontal resolution of 5x5 or finer
• ? 2001 meteorology
• ? each groups best estimate for 2001 emissions
• ? methane set to 1760 ppb
• SENSITIVITY SIMULATIONS (13-18 models)
• ? -20 regional anthrop. NOx, CO, NMVOC
  emissions,
• individually all together (16
  simulations)
• ? -20 global methane (to 1408 ppb)

8
Large inter-model range multi-model mean
generally captures observed monthly mean surface
O3
Mediterranean
Central Europe gt 1km
Central Europe lt 1km
NE USA
SW USA
SE USA
Surface Ozone (ppb)
Japan
Mountainous W USA
Great Lakes USA
9
North America as a receptor of ozone pollution
Annual mean foreign vs. domestic influences
Annual mean surface O3 decrease from -20
NOxCONMVOC regional anthrop. emissions
(1.64)
Full range of 15 individual models
import sensitivity
ppbv
domestic
Sum 3 foreign
10
North America as a receptor of ozone pollution
Seasonality of response to -20 foreign anthrop.
emissions
SUM OF 3 FOREIGN REGIONS
15- MODEL MEAN SURFACE O3 DECREASE (PPBV)
EA
EU
SA
Spring (fall) max due to longer O3 lifetime,
efficient transport e.g., Wang et al., 1998
Wild and Akimoto, 2001 Stohl et al., 2002 TF
HTAP 2007
11
North America as a receptor of ozone
pollutionSeasonality of response to -20
foreign anthrop. emissions
NOxNMVOCCO
MODEL ENSEMBLE MEAN SURFACE O3 DECREASE (PPBV)
NMVOC
NOx
CO
Wide range in EU anthrop. NMVOC inventories ?
large uncertainty in the estimated response of NA
O3
12
North America as a receptor of ozone pollution
Seasonality in import sensitivity
Surface O3 response to -20 domestic (NA
NOxNMVOCCO) anthrop. emis.
Response to -20 Foreign gt Domestic (winter)
PPB RATIO
13
Surface O3 response to decreases in foreign
anthropogenic emissions of O3 precursors
Source region SUM3 NA EA EU SA
EU receptor
Surface O3 decrease (ppb)
EA receptor
SA receptor
14
Monthly mean import sensitivities
15
Surface ozone response to -20 global
CH4similar decrease over all regions
Full range of 18 models
ppb
1 ppbv O3 decrease over all regions Dentener
et al.,2005 Fiore et al., 2002, 2008 West et
al., 2007
EU NA E Asia S
Asia

• Estimate O3 response to -20 regional CH4
  anthrop. emissions to
• compare with O3 response to NOxNMVOCCO
• -20 global CH4 -25 global anthrop. CH4
  emissions
• Anthrop. CH4 emis. inventory Olivier et al.,
  2005 for regional emissions
• Scale O3 response diagnosed with change in
  global burden to
• obtain O3 response due to regional CH4 emission
  change

16
Tropospheric O3 responds approximately linearly
to anthropogenic CH4 emission changes across
models
Anthropogenic CH4 contributes 50 Tg (15) to
tropospheric O3 burden 5 ppbv to
surface O3
Fiore et al., JGR, 2008
17
Comparable annual mean surface O3 response to
-20 foreign anthropogenic emissions of CH4 vs.
NOxNMVOCCO
Sum of annual mean ozone decreases from 20
reductions of anthropogenic emissions in the 3
foreign regions
ppb
Receptor NA EU
E Asia S Asia
(Uses CH4 simulation anthrop. CH4 emission
inventory Olivier et al., 2005 to estimate O3
response to -20 regional anthrop. CH4 emissions)

18
Summary Hemispheric Transport of O3

• Benchmark for future Robust estimates key
  areas of uncertainty
• Import Sensitivities (D O3 from anthrop. emis.
  in the 3 foreign vs. domestic regions) 0.5-1.1
  during month of max response to foreign emis
  0.2-0.3 during month of max response to domestic
  emissions
• Comparable O3 decrease from reducing equivalent
  of CH4 and NOxNMVOCCO over foreign regions
  (0.4-0.6 ppb for 20 reductions)

More info at www.htap.org, 2007 TF HTAP Interim
Report TF HTAP multi-model publications
Shindell et al., ACP, 2008 Transport to the
Arctic Sanderson et al., GRL, 2008 NOy
transport Fiore et al., JGR, in press Surface
O3 Reidmiller et al., in prep U.S. surface O3
Jonson et al., in prep Evaluation with O3
sondes Casper et al., in prep Health impacts
(O3) Schultz et al., in prep Idealized tracers
19
Some Remaining Questions

• What is the contribution of hemispheric transport
  to metrics relevant
• to attainment of O3 air quality standards?
• How important is interannual variability?
• What is causing model spread? Can we tie to
  specific processes and use observational
  constraints to reduce uncertainty?
• Can we scale O3 responses to other combinations
  and magnitudes of emission changes?
• How will climate change affect hemispheric
  transport of air pollution?

.. And much more TF HTAP work ongoing to inform
2010 report!
20
Wide model range in AQ-relevant metrics MAM
average daily max 8-hour (MDA8) surface O3 over
the USA (HTAP models)
ppb
21
CASTNet sites used in model evaluation
Reidmiller, AGU, 2008 in prep for JGR
22
Model Evaluation

• Wide spread in individual models, but ensemble
  represents obs quite well

• Correlations are generally strongest in East
  weaker in West
• Models have large () biases in summer that are
  largest in the East

Reidmiller, AGU, 2008 in prep for JGR
23
10-model mean response of MDA8 ozone to 20
reductions of foreign emissions MAM average
Ensemble mean base case MDA8 O3
Ensemble mean MDA8 O3 decrease from -20 EA EU
SA anthrop. emissions
ppb
Ensemble mean MDA8 O3 decrease from -20 NA
anthrop. emissions
ppb
Variability in response to foreign emissions?
e.g., clean vs. polluted conditions?
(Models regridded to common 2x2.5 grid)
ppb
24
MDA8 O3 response to 20 emissions reductions 3
foreign regions vs. North America region

• Response to foreign emissions greatest in spring
• For most regions O3 response (to -20 foreign
  emissions ) is 0.5 ppbv in spring

• Response to domestic emissions greatest in
  summer at high end of O3 distribution
• For all regions, response to domestic emissions
  greater than to foreign emissions (2-10x varies
  with season)

Reidmiller, AGU, 2008 in prep for JGR
25
Some Remaining Questions

• What is the contribution of hemispheric transport
  to metrics relevant
• to attainment of O3 air quality standards?
• How important is interannual variability?
• What is causing model spread? Can we tie to
  specific processes and use observational
  constraints to reduce uncertainty?
• Can we scale O3 responses to other combinations
  and magnitudes of emission changes?
• How will climate change affect hemispheric
  transport of air pollution?

26
Image from http//www.cpc.noaa.gov/products/precip
/CWlink/pna/nao.timeseries.gif
MOZART2 simulations
GMI simulations
GMI simulations
MOZART runs
http//jisao.washington.edu/data_sets/pna/
27
Inter-model spread generally larger than that due
to interannual variability in meteorology
28
Some Remaining Questions

• What is the contribution of hemispheric transport
  to metrics relevant
• to attainment of O3 air quality standards?
• How important is interannual variability?
• What is causing model spread? Can we tie to
  specific processes and use observational
  constraints to reduce uncertainty?
• Can we scale O3 responses to other combinations
  and magnitudes of emission changes?
• How will climate change affect hemispheric
  transport of air pollution?

29
A measure of uncertainty in SR relationships (sfc
o3) Model spread can be gt factor of 2
Multi-model mean
Models do not always rank in consistent way!
Comparison with surface obs shows no relationship
btw base-case bias and source-receptor
relationship (e.g., NA-gtEU)
30
Individual models sometimes rank consistently in
terms of source region influence on multiple
receptors
Individual model
Possible explanations -- Emissions (not major
player similar across models, except EU
NMVOC) -- Export / transport / mixing processes
-- Chemistry
31
Idealized tracer experiments (CO tagged by
region all models use same emissions) suggest
some role for vertical mixing in source region
EXAMPLE for NA ? EU surface ozone
Springtime (MAM) r2
0.56
Surface O3 decrease over EU from -20 NA
emis. (SR1-SR6NA ppb)
Individual model
Ratio of NA CO tracer burden at 3-8km to 0-2km
over NA
Less vertical mixing More vertical mixing
Index adapted from Martin Schultz
32
Models likely differ in export of O3
precursors, downwind chemistry, and transport to
receptor region
PAN
O3
NOx
O3
NOy
N2O5
O3
other organic nitrates HNO3
Fires
Land biosphere
Human activity
Ocean
Ocean
Continent 1
NOy partitioning (e.g., PAN vs. HNO3) influences
O3 formation potential far from source region
33
PAN transport can lead to highly efficient O3
production downwind
Hudman et al., JGR, 2004
Emmerson Evans, ACPD, 2008 significant
variations in the calculated concentration of PAN
between the schemes model chemical mechanisms
Does surface O3 response to emission changes in
foreign regions correlate with -- O3 response
over source region? Yes EU (r20.3-0.5) NO in
other regions (r2lt0.2) -- Changes in PAN over
upwind and/or receptor region?
34
Change in PAN over upwind region (or receptor)
correlates with surface ozone change over
receptor region
EA?NA O3 vs. N. Pacific PAN r2 0.36
NA?EU O3 vs. N. Atlantic PAN r2 0.36
Surface o3 decrease over receptor region (ppb)
Individual model
Individual model
Change in PAN burden (trop column) over upwind
region in spring (Tg)

• Separation of lowest sensitivity models and
  higher

35
What observations would help discriminate among
models (and reduce uncertainties in O3 response
to foreign emission changes)?
Simulated O3 response not correlated with total
ozone or ozone bias w.r.t. obs.
? Potential for PAN measurements to help
reduce uncertainty? 2001 Observations
TRACE-P (Asian outflow), PHOBEA (inflow to N.
America)
36
What causes the 1 ppb spread across models in
surface ozone response to -20 global CH4?
Full range of 18 models
Annual mean surface O3 decrease (ppb )
EU NA E Asia S
Asia
Surface O3 decrease over EU vs. NA (ppb)

• Strong correlations for all regions
• EA vs. EU r2 0.75
• SA vs. EA r2 0.83
• NA vs. SA r2 0.86
• Some models more
• responsive to CH4 changes

r2 0.88
Individual model
37
Differences in lifetime against tropospheric OH
are a strong contributor to model spread in O3
response to CH4

Surface O3 decrease (ppb) over NA from -20 CH4
Total atmospheric methane lifetime
Normalizing by CH4 lifetime reduces inter-model
range from 1 to 0.5 ppb (similar in other
receptor regions)
38
HTAP Event Simulations Moving towards
process-based evaluation
I. Bey, M. Evans, K. Law, R. Park, E. Real, S.
Turquety
1. chemical signatures of air masses, 2.
chemical evolution in background vs. polluted
plume ensembles, 3. export efficiencies, 4.
injection heights on biomass burning plumes
Preliminary results from the French model
MOCAGE, courtesy of N. Bousserez and J.-L-
Attié, Laboratoire daérologie Toulouse, France
observations
model
clean lower trop.
biomass burning influenced air masses
middle-upper troposphere
polluted lower trop.
from Isabelle Beys presentation at DC June 2008
HTAP meeting
39
Some Remaining Questions

• What is the contribution of hemispheric transport
  to metrics relevant
• to attainment of O3 air quality standards?
• How important is interannual variability?
• What is causing model spread? Can we tie to
  specific processes and use observational
  constraints to reduce uncertainty?
• Can we scale O3 responses to other combinations
  and magnitudes of emission changes?
• How will climate change affect hemispheric
  transport of air pollution?

40
Intercontinental O3 response to changes in NMVOC
emissions scales linearly not so for NOx (except
summer)
O3 response to -20 vs. -100 EU emissions NOx
vs. NMVOC in spring (GMI model)
Change of scale (5x gives same height bars)
Receptor regions
Receptor regions

• Relative benefit of decreasing NOx vs. NMVOC
  emissions increases
• with the magnitude of emission reductions

Wu et al., submitted to GRL
41
Some Remaining Questions

• What is the contribution of hemispheric transport
  to metrics relevant
• to attainment of O3 air quality standards?
• How important is interannual variability?
• What is causing model spread? Can we tie to
  specific processes and use observational
  constraints to reduce uncertainty?
• Can we scale O3 responses to other combinations
  and magnitudes of emission changes?
• How will climate change affect hemispheric
  transport of air pollution?

1/16/09 update on future climate HTAP
simulations The RCP process is expected to
publish scenarios in the February/March
timeframe.  Given the central role that these
scenarios are likely to play in the science and
policy processes over the next several years, we
believe that it is worth waiting to make some
final decisions until this work is completed.
Terry Keating (US EPA) and Andre Zuber (EC),
co-chairs of TF HTAP
42
Conclusions Hemispheric Transport of O3
More info at www.htap.org, 2007 TF HTAP Interim
Report, Fiore et al., JGR, in press

• Benchmark for future Robust estimates key
  areas of uncertainty
• Import Sensitivities (D O3 from anthrop. emis.
  in the 3 foreign vs. domestic regions) 0.5-1.1
  during month of max response to foreign emis
  0.2-0.3 during month of max response to domestic
  emissions
• Comparable O3 decrease from reducing equivalent
  of CH4 and NOxNMVOCCO over foreign regions
  (0.4-0.6 ppb for 20 reductions)
• Variability of O3 response within large HTAP
  regions U.S. example
• -- multi-model mean captures much of
  day-to-day variability
• Inter-model differences gtgt year-to-year
  variability
• Different model responses seem partially due to
  vertical mixing, PAN
• -- potential for PAN measurements to help
  constrain O3 response?
• Non-linearity in O3 response to foreign NOx
  emissions impacts relative benefit of NOx vs.
  NMVOC emission controls