Materials Science &Technology

Improvement and validation of OMI Improvement and validation of OMI NONO22 observations over complex terrain observations over complex terrain

A contribution to ACCENT-TROPOSAT-2, Task Group 3A contribution to ACCENT-TROPOSAT-2, Task Group 3

Yipin Zhou, Yipin Zhou, Dominik BrunnerDominik Brunner, and Brigitte Buchmann, and Brigitte Buchmann Empa, Swiss Federal Laboratories for Materials Testing and Research Empa, Swiss Federal Laboratories for Materials Testing and Research

Dübendorf, SwitzerlandDübendorf, Switzerland

Folkert Boersma and Ruud DirksenFolkert Boersma and Ruud Dirksen Royal Netherlands Meteorological InstituteRoyal Netherlands Meteorological Institute

De Bilt, The NetherlandsDe Bilt, The Netherlands

2Materials Science&Technology

Goal: Study air quality and pollution sources over domain of a small country with complex topography (focusing on NO2)

GOME

SCIA

OMI

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

3Materials Science&Technology

Evolution of sensor resolution

PhD D. Schaub

GOMEGOME(launch 1995)(launch 1995)

320 320 40 km 40 km22SCIAMACHYSCIAMACHY(launch 2002)(launch 2002)

60 60 40 km 40 km22

Schaub, D. et al., Atmos. Chem. Phys. 5, 23-37, 2005 Schaub, D. et al., Atmos. Chem. Phys. 6, 3211-3229, 2006 Schaub, D. et al., Atmos. Chem. Phys. 7, 5971-5987, 2007

PhD candidate Y. Zhou

OMIOMI(launch 2004)(launch 2004)

up to 24 up to 24 13 km 13 km22

Jan 2004 – Dec 2007 meanJan 2004 – Dec 2007 mean

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

4Materials Science&Technology

xa = a priori tropospheric NO2 profileb = forward model parameters

- cloud fraction- cloud pressure- surface pressure- surface albedo- aerosols- viewing geometry

xa = a priori tropospheric NO2 profileb = forward model parameters

- cloud fraction- cloud pressure- surface pressure- surface albedo- aerosols- viewing geometry

Tropospheric NO2 from satellites

Tropospheric air mass factor AMFtrop computed with radiative transfer model:

AMFtrop = f(xa,b)

Retrieval methodRetrieval method

trop

trop

trop

straobstrop AMF

SCD

AMF

SCDSCDVTC

Sources of uncertainty in AMFtrop calculation

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

5Materials Science&Technology

Representation of surface topography

Topography as used inTopography as used inDOMINO retrievalDOMINO retrieval

Topography at fullTopography at fullresolution of OMI pixelsresolution of OMI pixels

TM4/ECMWF topography,consistent with a prioriNO2 profiles

GTOPO-30 topography reduced to OMI pixel resolution

Too loo low over Alps

Too high over adjacent plains

m m

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

6Materials Science&Technology

Possible implications of inaccurate pixel surface height

Schaub et al., ACP (2007)

„„old picture“old picture“

Averaging kernel:sensitivity decreasingtowards surface

Strong sensitivityto altitude change

„„new picture“new picture“

Averaging kernel:sensitivity decreasingtowards surface

Modest sensitivityto altitude change

AK profile changes when surface height changes

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

7Materials Science&Technology

Box air mass factor profile (Box air mass factor profile (mmll)) A priori NOA priori NO22 subcolumn profile ( subcolumn profile (xxa,la,l))

Possible implications of inaccurate pixel surface height

psurf = 788 hPapsurf = 928 hPa (ref.)psurf = 1008 hPa

lla

llal

trop x

xm

AMF,

,

Note: NO2 volume mixing ratio profile preservedby vertical scaling

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

8Materials Science&Technology

Effect of cloudsEffect of clouds

Possible implications of inaccurate pixel surface height

NO2 profile

AK profilecloudy part

cloud

z z

TM4/ECMWFtopography

GTOPOtopography

Illustration of effect of inaccurate Illustration of effect of inaccurate topography for cloudy pixelstopography for cloudy pixels

AMFcloud significantlyoverestimated overplanes when usingTM4/ECMWF topo

zc: cloud top heightz0: surface terrain heightw: cloud radiance fraction

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

9Materials Science&Technology

Results: Effect on retrieved NO2

Difference in terrain heightDifference in terrain height(GTOPO – TM4 topography)(GTOPO – TM4 topography)

+40

+15

0

-15

Difference (%) in retrieved NODifference (%) in retrieved NO22

(GTOPO – TM4)/TM4(GTOPO – TM4)/TM4

December 2006 mean difference,cloud radiance fraction < 50%

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

10Materials Science&Technology

Results: Effect on retrieved NO2 Relative difference (%) in retrieved NO2 VTC, December 2006

(GTOPO – TM4)/TM4

Cloud radiance fraction < 50%Cloud radiance fraction < 50% Cloud radiance fraction < 10%Cloud radiance fraction < 10%

+40

+15

0

-15

+40

+15

0

-15

Differences of about 30%over parts of Swiss plateauand over Po basin

Differences of about 10%

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

11Materials Science&Technology

Monthly mean difference in retrieved NOMonthly mean difference in retrieved NO22 for for

selected point over Po basin (cloud rad. fraction < 50%)selected point over Po basin (cloud rad. fraction < 50%)

Dif

fere

nce

in

ret

riev

ed N

O2 c

olu

mn

(%

)absolute differenceNO2 with TM4 topoNO2 with GTOPOrelative difference (%)

Results: Effect on retrieved NO2

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

12Materials Science&Technology

Validation with ground-based in situ NO2

Approach:

- Data obtained from about 60 stations in Lombardyregion (Po basin)

- Select OMI pixels withcenters close to station

- Construct „ground-basedvertical tropospheric columns“ using TM4 a priori NO2 profile for each individual OMI pixel

Example: OMI pixels for station MottaExample: OMI pixels for station Motta(rural station south of Milano)(rural station south of Milano)

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

13Materials Science&Technology

Validation with ground-based in situ NO2

Motta: rural station, moderately polluted(crfrac < 50%)

211 pixels

Ground based columnOMI TM4 topo columnOMI GTOPO column

Ground based versus Ground based versus OMI columnsOMI columns

Comparison between meanComparison between meanannual cyclesannual cycles

Improved agreement in winterwith GTOPO

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

14Materials Science&Technology

Validation with ground-based in situ NO2

Agrate: moderately polluted(crfrac < 50%)

164 pixels

Ground based versus Ground based versus OMI columnsOMI columns

Comparison between meanComparison between meanannual cyclesannual cycles

Improved agreement in winterwith GTOPO

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

Ground based columnOMI TM4 topo columnOMI GTOPO column

15Materials Science&Technology

Validation with ground-based in situ NO2

Ground based versus Ground based versus OMI columnsOMI columns

Comparison between meanComparison between meanannual cyclesannual cycles

Abbiategrasso: Abbiategrasso: urban, heavily urban, heavily pollutedpolluted

Bergamo: rural, Bergamo: rural, least pollutedleast polluted

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions

16Materials Science&Technology

Using correct surface pressure is important, but less important than previously thought

Using correct surface pressure more important for cloudy part of pixel than for clear part

NO2 over Po basin underestimated by current retrieval by about 5% in summer and up to 20% in winter assuming a cloud radiance fraction threshold of 50% (equivalent to a cloud fraction of about 20%)

Agreement with in situ surface observations improves with new retrieval using accurate surface pressure(seasonal cycle is more pronounced in better agreement with observations)

Next steps after improving surface pressure: Use a high resolution albedo data set (e.g. snow cover) Use high resolution NO2 a priori profiles (e.g. from a regional

air quality model) Reprocess OMI data over Europe

Summary and outlook

Goal

Retrieval basics

Potentialtopography effects cloud free cloudy

Results NO2 change cloud effect seasonal

Validation

Conclusions