Relationship Between NOx Emissions and NO2 Columns

1
Space-based insight into the global sources of
nitrogen oxides with implications for tropical
tropospheric ozone
Randall Martin Dalhousie University
With contributions from Bastien Sauvage, Neil
Moore, Thomas Walker Dalhousie
University Christopher Sioris Environment
Canada Christopher Boone and Peter Bernath
University of Waterloo Jerry Ziemke NASA
Goddard Lyatt Jaegle University of
Washington Xiong Liu, Kelly Chance
Harvard-Smithsonian Center for Astrophysics
2
Tropospheric Ozone is a Key Species in Climate
and Air Quality

• Major greenhouse gas
• Largely controls atmospheric oxidation
• Primary constituent of smog

Mesosphere
Stratosphere
Ozone layer
Half of all Americans live in regions that exceed
the surface ozone standard
Troposphere
3
Global Budget of Tropospheric Ozone Driven By
Production in the Troposphere
Ozone Production is Largely NOx-Limited
hv
hv,H2O
Nitrogen oxides (NOx) CO, Volatile Organic
Compounds (VOCs)
Ozone (O3)
Hydroxyl (OH)
Fires
Biosphere
Human activity
4
How Do We Evaluate and Improve A Priori Bottom-up
Inventories?
Bottom-up Estimates for Global NOx Emissions
(Range) in Tg N yr-1 for 2000
Fossil Fuel 24 (20-33)
Biomass Burning 6 (3-13) Soils 7 (4-21)
Lightning 6 (1-20)
5
Top-Down Information from Satellite Observations

• Nadir-viewing solar backscatter instruments
  including ultraviolet and visible wavelengths
• GOME 1995-2003
• Spatial resolution 320x40 km2
• Global coverage in 3 days
• SCIAMACHY 2002-present
• Spatial resolution 60x30 km2
• Global coverage in 6 days
• OMI 2004-present
• Spatial resolution up to 13x24 km2
• Daily global coverage
• GOME-2 2006-present
• Spatial resolution up to 40x80 km2
• Daily global coverage

6
Retrieve NO2 Columns To Map Surface NOx Emissions
NOx NO NO2
Tropospheric NO2 column ENOx
BOUNDARY LAYER
hv
NO2
NO
NO/NO2 ? ? W ALTITUDE
O3
lifetime lt1 day
HNO3
Emission
NITROGEN OXIDES (NOx)
7
Spectral Fit of NO2
Solar Io
Distinct NO2 Spectrum
Ozone
Backscattered intensity IB
NO2
Scattering by Earth surface and by atmosphere
Albedo A
O2-O2
Nonlinear least-squares fitting
Fitting Uncertainty 5-10x1014 molec cm-2
Martin et al., JGR, 2002, 2006
8
Total NO2 Slant Columns Observed from SCIAMACHY
Dominant stratospheric background (where NO2 is
produced from N2O oxidation)Also see
tropospheric hot spots (fossil fuel and biomass
burning)
May-October 2004
Uncertainty in Stratospheric Removal 2-10x1014
molec cm-2
9
Perform an Air Mass Factor (AMF) Calculation to
Account for Viewing Geometry and Scattering
Cloud Radiance Fraction IB,c / (IB,o IB,c)
IB,c
IB,o

• GOMECAT (Kurosu) FRESCO Clouds Fields
  Koelemeijer et al., 2002
• Surface Reflectivity Koelemeijer et al., 2003
• LIDORT Radiative Transfer Model Spurr et al.,
  2002
• GEOS-CHEM NO2 aerosol profiles

Io
q
Rc
Ro
Pc
AMF Uncertainty 40
dt
Rs
Martin et al., JGR, 2002, 2003, 2006
10
Cloud-filtered Tropospheric NO2 Columns Retrieved
from SCIAMACHY
Mean Uncertainty (5x1014 30)
NO/NO2 ? ? W ALTITUDE
May 2004 Apr 2005
Martin et al., JGR, 2006
11
ICARTT Campaign Over and Downwind of Eastern
North America in Summer 2004 Aircraft Flight
Tracks and Validation Locations Overlaid on
SCIAMACHY Tropospheric NO2 Columns
NASA DC-8
NOAA WP-3D
12
GEOS-Chem Chemical Transport Model

• Assimilated Meteorology (NASA GMAO)
• 2ox2.5o horizontal resolution, 30 vertical layers
• O3-NOx-VOC chemistry
• SO42--NO3--NH4-H2O, dust, sea-salt, organic
  elemental carbon aerosols
• Interactive aerosol-chemistry

41 tracers 90 species 300 reactions
Solve continuity equation for individual gridboxes

• Sources
• emissions
• chemical prod.

Sinks - chemical loss - deposition
Transport flux divergence
Accumulation
Dx 200 km Dz 1 km
13
Air Mass Factor Calculation in SCIAMACHY
Retrieval Needs External Info on Shape of
Vertical Profile Increased Lightning NOx
Emissions Improves GEOS-CHEM Simulation of
Midlatitude NO2 Profiles Remaining Discrepancy
In Vertical Profile of NOx Emissions
In Situ 0.4 Tg N yr-1 1.6 Tg N yr-1
Midlatitude lightning Mean Bias in AMF 0.4
Tg N yr-1 12 9 3 1.6 Tg N
yr-1 1 5 3
Martin et al., JGR, 2006
14
Enhanced Midlatitude Lightning Reduces
Discrepancy with SCIAMACHY over North Atlantic
Profile of NOx Emissions (lifetime) Contributes
to Remaining Discrepancy
SCIAMACHY NO2 (1015 molec cm-2)
GEOS-Chem NO2 (1015 molec cm-2) 1.6 Tg N in Midlat
GEOS-Chem NO2 (1015 molec cm-2) 0.4 Tg N in Midlat
Martin et al., JGR, 2006
May-Oct 2004
15
Significant Agreement Between Coincident
Cloud-Filtered SCIAMACHY and In-Situ Measurements
r 0.77 slope 0.82
11 line
Cloud-radiance fraction lt 0.5 In-situ
measurements below 1 km above 3 km Assume
constant mixing ratio below lowest measurement
Add upper tropospheric profile from mean obs
Cohen (DC-8)
Ryerson (WP-3D)
Horizontal bars show 17th 83rd percentiles
Martin et al., JGR, 2006
16
Conduct a Chemical Inversion For NOx Emissions
min cost function
A Priori NOx Emissions (xa)
SCIAMACHY NO2 Columns (y)
1011 molec N cm-2 s-1
1015 molec N cm-2
GEOS-CHEM model F(x)
Sa
A posteriori emissions x
Error weighting
Top-Down Emissions
Sy
17
Significant Agreement Between A Priori and A
PosterioriLargest Discrepancy in East Asia and
Major Urban Centers
(2000)
r20.82
Martin et al., JGR, 2006
18
A Posteriori NOx Emissions from East Asia Exceed
Those from Either North America or
EuropeImplications for North American Air Quality
A posteriori (46 Tg N/yr)
A priori (38 Tg N/yr)
Martin et al., JGR, 2006
19
INTEX-B Long-Range Transport to North
AmericaAverage over April May 2006
Ozone Column (Dobson Units)
?Ozone Column (Dobson Units)
Sensitivity to Asian Emissions
Sensitivity to Lightning
Whistler, BC
Sensitivity at 750 hPa to PAN
?Ozone (ppbv)
standard
No Asian NOx
No lightning
Thomas Walker
20
Direct Retrieval of Tropospheric Ozone from
GOMEUsing Optimal Estimation in Ultraviolet with
TOMS V8 a priori
GOME
GEOS-CHEM
Tropospheric Ozone Column (Dobson Units)
Liu et al., JGR, 2006
21
In Situ Data Used for Tropical Evaluation
1.MOZAIC programme 1994-2005
2.SHADOZ ozone sonde network (Thompson et al.,
2003ab) 1998-2004
MOZAIC SHADOZ sites used for model evaluation
gt 9000 vertical profiles within the Tropics
(30N-30S)
22
Northern Tropics Remain a Challenge for
Satellites and ModelsScan-Angle Method (Kim et
al., 2005) UV Method That Best Captures In Situ
Seasonal Variation
Comparison with MOZAIC Ozone Measurements
GOME GOME GEOS-CHEM GEOS-CHEM
R Bias R Bias
Caracas 0.57 0.8 0.54 8.7
Dakar -0.37 -3.8 0.81 5.2
Tel Aviv 0.96 -1.5 0.94 1.4
Bangkok 0.83 -2.4 0.94 7.2
Liu et al., JGR, 2006
23
Algorithm for partitioning top-down NOx inventory
(2000)
GOME NOx emissions
8.9
Algorithm tested using synthetic retrieval
Jaeglé et al., 2005
24
Speciated Inventory for Soil emissions
A posteriori 70 larger than a priori!
A priori
A posteriori
r2 0.62
(90)
(200)
North Eq. Africa
East Asia
Soils
Soils
Onset of rainy season Pulsing of soil NOx!
Jaeglé et al., 2005
25
Improved Bottom-up Inventory for Soil NOx
Emissions
Developments of soil temp/soil moisture, pulsing,
fertilizer application
Change in NOx Emissions
Soil NOx Emissions
molec cm-2 s-1
? molec cm-2 s-1
? Global Total 1.9 Tg N/yr
Global Total 7.8 Tg N/yr
Neil Moore
26
Top-down Constraint on Biomass Burning NOx
Emissions
GOME
Model original
Model constrained
DJF
MAM
NO2 Column (1015 molec cm-2)
Observed
Improved simulation of lower tropospheric O3
versus aircraft measurements
Pressure (hPa)
Top-down
Bottom-up
Sauvage et al., ACP, 2007
O3 Mixing Ratio (ppbv)
27
Global Lightning NOx Source Remains
UncertainConstrain with Top-down Satellite
Observations
SCIAMACHY Tropospheric NO2 Columns
ACE-FTS Limb HNO3 Measurements in
the Upper Troposphere
OMI MLS Tropospheric O3
10-year Mean Flash Rate from the OTD LIS
Satellite Instruments
Global rate 445 flash/sec Christian et al. 2003
30 500 moles NO per flash
Flashes km-2 min-1
28
Current Estimate of Annual Global NOx SourcesAs
Used In GEOS-Chem
Lightning Global 6.0 Tg N yr-1 Tropics 4.4 Tg
N yr-1
Other NOx sources (fossil fuel, biofuel, biomass
burning, soils) 39 Tg N yr-1
1010 molecules N cm-2 s-1
29
Simplified Chemistry of Nitrogen OxidesExploit
Longer Lifetimes in Upper Troposphere
Upper Troposphere
hv
NO
NO2
Ozone (O3)
O3, RO2
lifetime month
NOx lifetime week
HNO3
lifetime weeks
NO/NO2 ? ? with altitude
Boundary Layer
hv
NO2
Ozone (O3)
NO
O3, RO2
lifetime days
NOx lifetime lt day
HNO3
Nitrogen Oxides (NOx)
30
Strategy

• 1) Use GEOS-Chem model to identify species,
  regions, and time periods dominated by the
  effects of lightning NOx production
• 2) Constrain lightning NOx source by interpreting
  satellite observations in those regions and time
  periods

31
Simulated Monthly Contribution of Lightning,
Soils, and Biomass Burning to NO2 Column
32
Annual Mean NO2 Column at Locations Months with
gt60 from Lightning, lt25 from Surface Sources
SCIAMACHY (Uses 15 of Tropical Observations)
Meridional Average
GEOS-Chem with Lightning (62 Tg N yr-1)

SCIAMACHY
GEOS-Chem with Lightning (8 bias, r0.75)
GEOS-Chem without Lightning (-60 bias)
GEOS-Chem without Lightning
NO2 Retrieval Error 5x1014 molec cm-2
Tropospheric NO2 (1014 molec cm-2)
Martin et al., 2007
33
ACE HNO3 over 200-350 hPa for Feb 2004 Feb 2006
HNO3 Mixing Ratio (pptv)
Data from Boone et al., 2005
34
GEOS-Chem Calculation of Contribution of
Lightning to HNO3
HNO3 With Lightning (62 Tg N yr-1)
Focus on 200-350 hPa
Fraction of HNO3 from Lightning
No Lightning
HNO3 from Lightning
Fraction from Lightning
Jan
Jul
35
Annual Mean HNO3 Over 200-350 hPa at Locations
Months with gt 60 of HNO3 from Lightning
ACE (Uses 83 of Tropical Measurements)
Meridional Average
GEOS-Chem with Lightning (62 Tg N yr-1)
ACE-FTS
GEOS-Chem with Lightning (-12 bias, r0.75)
GEOS-Chem without Lightning
GEOS-Chem without Lightning (-80 bias)
HNO3 Retrieval Error 35 pptv
Martin et al., 2007
HNO3 Mixing Ratio (pptv)
36
OMI/MLS Tropospheric Ozone Column
Jan
Jul
Data from Ziemke et al. (2006)
37
Calculated Monthly Contribution of Lightning to
O3 Column
O3 Column from Lightning
Column Fraction from Lightning
Martin et al., 2007
38
Annual Mean Tropospheric O3 Columns at Locations
Months with gt 40 of Column from Lightning
OMI/MLS (Uses 15 of Tropical Measurements)
Meridional Average
GEOS-Chem with Lightning (62 Tg N yr-1)
GEOS-Chem with Lightning (-1 bias, r0.85)
OMI/MLS
GEOS-Chem without Lightning (-45 bias)
GEOS-Chem without Lightning
O3 Retrieval Error lt 5 Dobson Units
Martin et al., 2007
Tropospheric O3 (Dobson Units)
39
Spatial Distribution of GEOS-Chem Lightning NOx
Source
Local Scaling to Match 10-year HRAC Seasonal
OTD-LIS Climatology
Scaled version
Original version
DJF
DJF
JJA
JJA
Lightning NOx emissions (109 molec N cm-2 s)
Same intensity 6 Tg N yr-1
Sauvage et al., ACP, 2007
40
Ozone Sensitivity to Spatial Distribution of
Lightning NOx
Snapshot of the model evaluation
Original Modified In situ
Scaled
Pressure (hPa)
Pressure (hPa)
O3 (ppbv)
O3 (ppbv)
-O3 highly sensitive in the MT-UT -O3 simulations
improved by 5-15 ppbv versus In situ -Main
influence near subsidence areas South America
Middle East Atlantic
Sauvage et al., ACP, 2007
41
Ozone sensitivity to Lightning NOx 4 TgN/yr 6
TgN/yr 8 TgN/yr
Scaled
Scaled
Scaled
Pressure (hPa)
Pressure (hPa)
O3 (ppbv)
Evaluation for the Tropics 8Tg N/yr ? O3 over
estimation 4Tg N/yr ? O3 under estimation
62Tg N/yr ? general agreement
Sauvage et al., ACP, 2007
O3 (ppbv)
42
Lightning NOx Dominant Source for Tropical
Tropospheric Ozone
Sensitivity to decreasing NOx emissions by 1 for
each source
6 Tg N/yr
6 Tg N/yr
DJF
6 Tg N/yr
MAM
JJA
SON
?DU
Lightning Ozone Production Efficiency 3 times
OPE of each surface source
Atmospheric Oxidation Largely Controlled by
Lightning NOx Source
Sauvage et al., JGR, in press
43
Simulated Annual Mean Characteristics
O3
ppb
Sauvage et al., JGR, in press
44
Conclusions

• Growing confidence in top-down constraints on NOx
  emissions
• South Atlantic Maximum largely results from
  lightning NOx due to high ozone production
  efficiency
• Global lightning NOx source likely between 4 8
  Tg N / yr
• 6 Tg N / yr is a best estimate
• Further refinement of lightning source will
  require
• - stronger constraints on midlatitude source
• - improved satellite retrieval accuracy (e.g.
  NO2)
• - more observations (e.g. HNO3)
• - model development to better represent
  processes (e.g. lightning NOx representation,
  vertical transport)

Acknowledgements
Supported by NASA, CFCAS, and NSERC