Transcript Aura NO2 talk

Improving Retrievals of Tropospheric NO2
Randall Martin, Dalhousie and Harvard-Smithsonian
Lok Lamsal, Gray O’Byrne, Aaron van Donkelaar, Dalhousie
Ed Celarier, Eric Bucsela, Joanna Joiner, NASA
Folkert Boersma, Ruud Dirksen, KNMI
Chao Luo, Yuhang Wang, Georgia Tech
September 14, 2009
Air Quality Working Group
Aura Meeting
Leiden, Netherlands
Seasonal Differences Between OMI NO2 Products
Direct Validation Has Not Arbitrated
Standard (SP)
DOMINO (DP)
DP-SP
DJF
2005
JJA
2005
0.1 1 2 3 4 5 6 7 8 9 10
Tropospheric NO2 Column (1015 molecules cm-2)
-5 -3 -1 1 3 5
Δ(1015 molecules cm-2)
Lamsal et al., JGR, submitted
Indirect Validation of OMI
(A) In-situ surface NO2 measurements from the SEARCH (photolytic) and
EPA/AQS (molybdenum) networks at rural sites in Eastern US
Use GEOS-Chem NO2 profiles to estimate surface-level NO2 from
OMI (Lamsal et al., JGR, 2008)
(B) Updated bottom-up emission
inventories for 2005-2006
Apply GEOS-Chem to infer top-down emissions from OMI by mass
balance (Martin et al., JGR, 2003)
Multiple Approaches Yield Similar Results
SEARCH “True” NO2”, Southeast U.S.
NOx Emissions, SEARCH domain
AQS/EPA “Corrected” NO2, Eastern U.S.
NOx Emissions, US + Canada
Lamsal et al., JGR, submitted
Stratosphere-troposphere Separation and AMF Together
Explain Difference Between DP and SP
Air mass factor
Strat-trop separation
Combined
ΔTropospheric NO2 Column DP – SP (1015 molecules cm-2)
Lamsal et al., JGR, submitted
Produce DP_GC From DP Averaging Kernels and GEOS-Chem NO2 Profiles
SEARCH “True” NO2”, Southeast U.S.
NOx Emissions, SEARCH domain
AQS/EPA “Corrected” NO2, Eastern U.S.
NOx Emissions, US + Canada
Lamsal et al., JGR, submitted
Surface Reflectivity
Lambertian Equivalent Reflectivity (LER)
OMI LER (Kleipool et al. 2008) Best Represents Surface LER
Cloud-, Snow-, and Aerosol- Free LER
(2005-2007)
Use MODIS/Aqua to Eliminate Cloud
and Aerosol from OMI Scenes
Use NISE Snow Flag to Eliminate
Snow
Global Annual
Mean Difference
x100 (unitless)
Standard Deviation
x100 (unitless)
TOMS MinLER
-0.8
2.2
GOME MinLER
1.2
2.6
OMI MinLER
-0.2
3.3
OMI LER (if snow-free)
0.02
1.1
LER Difference of 2%  15-30% Bias in NO2
(Martin et al., 2002; Boersma et al., 2004)
O’Byrne et al., JGR, submitted
Unrealistic Relation in OMI NO2 versus Cloud & Snow
(In situ NO2 data show variation < 15%)
Winter Mean Trop. NO2 (molec/cm2)
Winter OMI NO2 over Calgary & Edmonton
≥ 5cm of snow
0 > snow < 5cm
no snow
OMI Reported Cloud Fraction
O’Byrne et al., JGR, submitted
Large Spatial Variation in Snow-Covered Surface LER
Current Algorithms Assume Snow Reflectivity = 0.6
0
0.2
0.4
0.6
0.8
1
Snow-covered Surface LER (unitless)
Snow Weakly Represented in Previous Climatologies
Leads to Ambiguity in Accounting for Snow
OMI LER
-0.8
-0.6
-0.4
-0.2
0
0.2
Snow-Covered LER Difference (Previous Climatology – Snow-Covered Surface LER)
O’Byrne et al., JGR, submitted
Spatially-Varying Biases in OMI NO2 over Snow
To correct NO2 retrieval for snow
• Use snow-covered surface reflectivity
• Use MODIS-determined cloud-free scenes to correct clouds
NO2 bias for MODIS-determined cloud-free scenes
•Positive (negative) bias from underestimated (overestimated) surface LER
•OMI reports clouds when surface LER is underestimated
With All Cloud
Fractions
With Cloud
Fraction
Threshold (f < 0.3)
-50
0
 original  corrected 
Relative NO2 Bias 100* 

corrected


100
O’Byrne et al., JGR, submitted
Recommendations
Remote Sensing Community:
• Use two reflectivity databases: one snow-free, one for snow
•Switch from TOMS or GOME reflectivity databases to OMI
•Switch from annual mean to monthly mean NO2 profiles for SP
• Evaluate Stratosphere-Troposphere Separation
• Develop instrumentation with finer spatial resolution
(more cloud-free scenes reduces dependence on assumed profile )
• Following DP, include Averaging Kernels (or Scattering Weights) in trace
gas products so the user can remove the effect of the assumed profile
Ground-based Measurement Needs:
•span satellite footprint
•full year
•research quality (e.g. NO2)
•vertical profile
Modeling Community:
Continue develop representation of vertical profile