1. Aura Science and Validation Team Meeting
Continental outflow and intercontinental transport of ozone pollution as determined by O3-CO correlations from TES
Lin Zhang, Daniel J. Jacob, Kevin W. Bowman, Jennifer A. Logan, Solène Turquety, Rynda C. Hudman, Qinbin Li, Reinhard Beer, Helen M. Worden, John R. Worden, Curtis P. Rinsland, Susan S. Kulawik, Michael C. Lampel, Mark W. Shephard, Brendan M. Fisher, Annmarie Eldering, Melody A. Avery
I am supposed to focus my talk on the O3-CO correlations in continental outflow, but after Hanwant Singh’s talk, I would also like to talk about integrating satellite observations and INTEX-B measurements to study transpacific transport of Asian pollution. So my talk will have two parts.
Aura Science and Validation Team Meeting
Sep 11, 2006

2. O3-CO correlation: Indicator of ozone production
Parrish et al., JGR1998
Black: observed
Red: GEOS-Chem
First, let’s look at the O3-CO correlation. The O3-CO correlations measured from surface sites or aircraft campaigns have been used to infer ozone production and characterize tropospheric ozone sources.
The left panel here is a study from Parrish in It shows the O3-CO correlation observed at Sable Island. This is a surface site off the east coast of Canada. In August, it has a significant positive correlation with a typical slope of 0.3 mol/mol. The positive slope indicates strong summertime ozone production.
In the winter month, ozone and CO are negative correlated. Negative slopes are often observed when downward mixing with air masses from aloft or in the absence of photochemical activity, ozone is depleted by anthropogenic emissions. Parrish used the slope of 0.3 mol/mol combined with CO inventory over North America to estimate the amount of O3 exported in the boundary layer from NA to south Atlantic.
O3-CO correlations were also used to test model simulation of ozone production. The right panel shows the monthly variation of the observed ozone-CO correlation and also the GEOS-Chem model simulated correlation.
The correlation coefficient (R) and slope can provide different information. The correlation coefficient characterizes the ozone sources. It tells us to which extent ozone originates from combustion. The slope is a diagnose of the ozone production efficiency. GEOS-Chem model captures the observed positive slope in the summer time and negative slope in the winter months.
Li et al., JGR2002
O3-CO correlations in surface and aircraft data have been used to test understanding of ozone production but the data are sparse.

3. TES ozone and CO observations in July 2005 at 618 hPa
Eastern US
East Asia
TES allows us to study the relationship between ozone and CO on a regional or global scale.
This slide shows the mean TES ozone and CO concentrations at 618 hPa in July when large signatures of ozone pollution at northern mid-latitudes are expected. High ozone and CO concentrations are observed over polluted regions of northern mid-latitudes and biomass burning regions of Africa.
We look at the two box regions representing the outflow from Eastern US and East Asia. The right panel shows the scatterplot of all observations in July over the two regions. TES observes strong positive ozone-CO correlations in the two regions, indicating ozone export from these polluted regions. The red circles are GEOS-Chem model simulations with TES averaging kernels applied, and it captures the observed correlations and slopes.
TES and model show positive O3-CO correlations in continental outflow suggesting ozone export from polluted regions.

4. Deriving ozone-CO correlations from TES data
Divide the globe into 10ox10o cells
In each cell, calculate the O3-CO correlation and regression coefficients for each month. Each cell has data points for July 2005.
We can construct a global view of the O3-CO correlation patterns from the TES observations. Here we divide the global into 10x10 degree boxes. Each cell has data points for the July We did reduced major axis regression analysis in each cell.
Here shows an example over the red grid box. TES and GEOS-Chem model display similar positive correlations.
For another test, we looked at the aircraft measured correlations. We don’t have measurements in July 2005, but we used aircraft measurements from ICARTT took place in July-August 2004 to validate the correlations seen by TES. The right panel shows aircraft measured correlation over the same region in a narrow pressure band. Model simulations were sampled along the flight track. Similar correlations were observed compared with model and TES.
Zhang et al. GRL, 2006

5. Global distribution of the O3-CO correlation
July 2005
The figure shows the resulting correlation coefficient and regression slopes after interpolation. TES observes strong positive correlations over the three regions, eastern Pacific, East Asia and South Africa. The three regions are known to have large ozone production in the northern hemispheric summer.
We used it to test the anthropogenic influence on global ozone pollution and intercontinental transport in the GEOS-Chem model. The second row shows the corresponding model simulated O3-CO correlations patterns. Model captures the strong positive correlations over the three regions, but it shows stronger correlation coefficients and extends further downwind of the continents, displaying a pollution transport belt at northern mid-latitudes.
O3-CO correlations from TES provide an valuable test of anthropogenic influence on ozone in global models.

6. Effect of retrieval error
July 2005
Why does the model predict much stronger correlations? We find the retrieval measurement error could degrade the satellite observed correlations. When applying TES averaging kernels, the measurement error can be averaged out, but it can also increase the variability of observed values and then affect the correlation.
Here we applied random numbers with the magnitude of the measurement errors to the model fields. Now we can see the model simulated ozone-CO correlation is decreased. It shows remarkably similar correlations and slopes as those observed by TES.
New TES data have largely reduced the magnitude of retrieval error, and will provide more valuable information on the O3-CO correlation.
Apply random retrieval error to each model vertical profile

7. INTEX-B Mean CO and ozone vertical profiles
April 23- May 15, 2006 CO
Now let’s move to the INTEX-B. This slide show the mean vertical profiles of CO and ozone measured by aircraft and simulated by GEOS-Chem model. For CO, model underestimate the observations by 20 ppb all thought the troposphere. This CO underestimation is a complex problem. We didn’t see it in previous campaigns such as TRACE-P. It is probably due to old emissions used in the model or switching meteorological fields from GEOS-3 to GEOS-4. We find it is a hemispheric problem when comparing with MOPITT measurements. And we are looking into this problem. But here we add 20 ppbv to the model CO vertical profiles to correct the bias.
Ozone simulation compares well with the aircraft measurements. The overestimation in the upper troposphere at south of 30N may reflect a convection problem and the underestimation over north of 45N is likely due to stratospheric input since GEOS-Chem doesn’t simulated the stratospheric chemistry.
Ozone
Black: INTEX-B measurements
Red: GEOS-Chem

8. Transpacific Transport observed from TES
Asian Outflow
CO: R = 0.81
O3: R = 0.71
During the INTEX-B period, TES conduct extensive observations over the North Pacific. It has global surveys every other day and special observations in the alternative day.
Here we pick up the two regions over Northwest and Northeast Pacific representing Asian outflow and NA inflow. We average TES CO and ozone observations in the two regions for each day. The two panels show time series of tropospheric CO column and ozone concentration at 510 hPa during the INTEX-B period.
TES observations and model simulations are high correlated. We see TES observations over northwestern pacific show outflow events every 3-5 days. At least two big transpacific transport events are detected from TES as indicated by the blue arrows.
Time series of ozone shows some different features. There are periods that ozone is well correlated with CO but also periods lack of correlation. This needs further investigation. As a case study, we look into the transport event arriving Northeast Pacific on May 8.
NA Inflow
CO: R = 0.85
O3: R = 0.9
TES: solid
Model: dash

9. Transpacific transport of CO pollution
Tropospheric CO column
TES
GEOS-Chem with TES AK
GEOS-Chem
May 5
May 6
May 7
This slide shows TES CO measurements from May 5-May 9. The data are averaged into 2x2.5 grid cells. GEOS-Chem simulation is sampled along the TES orbit and applied with TES averaging kernels. The third column shows the original model field. We clear see that the plume comes out of Asian, spreads and arrives at North America. Although TES daily measurements are sparse, it observes these hot spots coming across the Pacific.
INTEX-B campaign predicted the transport event and has flights on May 8 and 9 over the northeastern pacific. This transport event provides a great case for integrating the two data sets.
May 8
May 9

10. MOPITT and AIRS observations of the Asian plume
Total CO column
MOPITT (AM)
AIRS
May 5
May 6
May 7
MOPITT and AIRS have a better daily coverage. This figure shows MOPITT and AIRS observations of the same transpacific transport event. Only daytime MOPITT measurements are shown here. Both MOPITT and AIRS see the pollution coming out of Asia on May 5 and arriving NA on May 8 and 9.
May 8
May 9

11. Transpacific transport of ozone pollution
Ozone concentration at 500 hPa
GEOS-Chem
with TES AK
GEOS-Chem simulated Asian ozone enhancement
TES
May 5
May 6
May 7
May 8
May 9
TES also observes ozone. This slide shows the transport of ozone pollution during this event. TES sees these high ozone values transporting across the pacific. Model simulations, although are lower, capture the same pattern. The third column shows model simulated ozone enhancement from Asian anthropogenic emissions. Unlike the CO plume spreading over the Pacific, the ozone plume separates over the East Pacific suggesting ozone photochemical production.

12. The May 5-9 transport event measured from aircraft
C130 flight on May 8
DC8 flight on May 9
Now from the aircraft, the C130 flight on May 8 observed surprising high ozone values at 2km along the coast. From the DC8 flight on May 9, a spiral over Eastern Pacific observed the pollution layer arriving US in the free troposphere.

13. Ozone production in the transpacific plume
Simulated Asian CO Enhancement
Simulated Asian Ozone Enhancement
May 5 – May 11
TES observation and simulated Asian enhancement at 618 hPa
R=0.64
R=0.60
R=0.45
R=0.62
TES CO at 618 hPa
TES Ozone at 618 hPa
Simulated Asian CO enhancement
Simulated Asian ozone enhancement 17 11 3 48
109 93 53
Northern box
Then does TES see the ozone production in the Northeastern Pacific?
The top two panels here show model simulated Asian CO and ozone enhancements below 6km. CO diffuses quickly from West to East, but ozone has another maximum over the eastern pacific suggesting ozone production.
To test the model prediction, we analyze TES CO and ozone observations at 618 hPa in the two regions indicated by the red boxes. We compare TES CO and ozone observations with model simulated Asian CO and ozone enhancement. TES observations are well correlated with simulated Asian enhancements.
When the plumes transport to the southern region, TES CO decreases from 103 ppb to 93 ppb, but ozone increases from 48 to 53. The model simulated Asian CO enhancement also decrease but ozone stays the same. Making a rough estimation from TES observations, if ozone diffused at the same rate of CO, there would be 12 ppb of ozone produced.
Southern box

14. Summary and future work
TES observed significant positive O3-CO correlations in continental outflow at North mid-latitude in July.
TES CO and ozone measurements provide useful insights in tracking transpacific transport and understanding ozone production through the transport.
Combination with INTEX-B reactive nitrogen species will allow us to look into the ozone production efficiency in the transpacific plumes.
Further study to assess the implication of Asian pollution plumes on US air quality.
To summarize, we see TES observes significant positive O3-CO correlations in continental outflow at northern mid-latitude. This confirms the large continental source ozone in the summertime.
We see transpacific transport of Asian pollution and ozone production over eastern Pacific from TES.
Combination with INTEX-B reactive nitrogen species will allow us to look into the ozone production efficiency in the transpacific plumes
As our bottom line in term of its implication on US air quality, whether transpacific transport events are detectable at the surface? Previous studies shows there were no indication of episodic enhancements in Asian influence in the models or observations. There is 10 times dilution during the entrainment. This curtain plot shows GEOS-Chem simulated Asian ozone enhancement during the transport event. The transport takes place mostly around 600 hPa. More work is needed to see whether we can detect the plume at the surface and how it affects air quality over US.