1. A multi-platform analysis of the North American reactive nitrogen budget during the ICARTT summer intensive
R. C. Hudman, D. J. Jacob, S. Turquety, Lee Murray, ITCT-2K4 science team, INTEX-NA science team
Measurements: M. Avery, R. Cohen, J. Dibb, F. Flocke, J. Holloway, A. Neuman,
T. Ryerson, G. Sachse, H. Singh, P. Wennberg
Modeling studies suggest eventual ozone production due to exported NOy is comparable to direct export of ozone pollution in terms of contribution to the global tropospheric ozone reservoir. The ICARTT campaign, by means of its considerable coverage and inter-compared aircraft, provides the opportunity to synthesize the results of smaller regional studies of NOy export from North America available in the literature. In this study, we use GEOS-CHEM, a global 3-D model of tropospheric chemistry to examine NOy composition within the boundary layer, its export and its evolution into the free troposphere over the North Atlantic using the ensemble of INTEX and ITCT observations.
GEOS-CHEM
NEI-99 emissions (VOC modifications)
2x2.5 resolution
NEAQS/ITCT2k4
ICARTT science meeting

2. ICARTT 0-2 km NOy* partitioning:
Rapid Oxidation of NOx downwind of continent
Merged Observations
GEOS-CHEM
NOx/NOy
NE/MW: 25% South: 18% OffShore: 15%
Large NOx sources
PAN/NOy
NE/MW: 19% South: 20% OffShore: 10%
Here I combine both ITCT-2K4 and INTEX observations and grid them to 2x25 averages The conbination of the data sets truly provides remarkable coverage of the eastern U.S. Doing this we can examine some interesting regional differences in NOy partitioning.
On the left is the observed fractions and on the right the modeled fractions.
The high values of HNO3 and relatively low concentrations of PAN offshore reflect the rapid oxidation of NOx and the short lifetime of PAN within the boundary layer as these airmasses move away from their sources. The elevated NOx fraction in the North can be attributed to the high concentration of sources in this region. The South favors HNO3 production likely due to its longer boundary layer residence time due to stagnation.
Daniel Jacob:
Make sure you draw attention to the fact that you merged the DC8 and P3 data for your analysis. You will be the only person in the meeting having done that, yet it’s at the heart of what ICARTT is about, so play it up. “Multi-platform” in the title is not enough notice – too vague.
Are you ignoring aerosol nitrate? If so tell us why (orally OK) There are other minor species PPN, HNO4, etc. that you should tell us are negligible if that’s the case – though HNO4 may not be negligible in the UT but that’s not your focus.
Need to say that the model is GC – here and elsewhere I would replace “Modeled” with “GEOS-Chem”. Also tell us that GC simulation was conducted for the ICARTT conditions, at the resolution shown here, using NEI99 emissions etc. No need for a dedicated slide, but the audience needs to know what is going into the model
HNO3/NOy
NE/MW: 56% South: 62% OffShore: 74%
Rapid NOx oxidation
*“NOy”=NOx + HNO3 + PAN

3. ICARTT 4-8 km NOy partitioning:
Role of PAN increases
Observed
Modeled
NOx/NOy
PAN/NOy
PAN in outflow
As we rise in the atmosphere to 4-8 km we see the relative contribution of PAN increasing due to its longer lifetime and HNO3 scavenging. This is particularly evident in the NA outflow corridor along the east coast. We also begin to see the influence of lightning and other uplift in the observations as the NOx fraction increases. A similar pattern is seen in the model but is underestimated and will be discussed later.
HNO3/NOy

4. ICARTT 8+ km NOy partitioning: Large Lightning Presence
Observed
Modeled
NOx/NOy
PAN/NOy
PAN dominates outflow
At 8km and above we see the increased importance of lightning both in the model and in the observations. The model captures well but underestimates this lightning impact. PAN dominates the outflow increasing to ~50% of NOy composition. HNO3 becomes much less important due to scavenging.
HNO3/NOy

5. GEOS-CHEM vs. INTEX-NA and ITCT2k4:
Comparison along the flight tracks*
NO2 underestimate above 6 km
Overestimate of CO & NOx anthro. source
Similar PAN underestimate
O3 FT bias
Shown here is GEOS-CHEM sampled along the INTEX and ITCT-2K4 flight tracks at the time of the flights. In the BL we see an overestimate of both CO and NOx in the current NEI 99 EPA inventories (Co bias is ~40 ppbv in BL, NOx really depends on location). Rising in the atmosphere…the middle tropopshere NOx and CO are simulated well, however, the model does not capture the drastic NO2 increase above 6km. A similar bias is seen in PAN. Ozone is also underestimated and may be related to the missing uppertropospheric NOx source. PAN is so important to the composition of NOy,The first question is if this PAN is coming in from outside the U.S. and a background problem or the result of a bias in pollution enhancement.
local, strat,
BB filtered
Gridded tp 2x25

6. PAN bias: Background or NA Enhancement?
Top:
PAN bias: Background or NA Enhancement?
Modeled
Observed
PAN - CO correlations in free tropopshere
PAN discrepency in enhancement
Bottom: Modeled Tagged tracer correlations 6-8 km
Looking at the PAN-CO relationship for the full chemistry model and the observations. The model is in blue and the observations are in black for 6-8, 8-10 and km. We see the model captures the lower background values…but underestimates in the enhancement.
At this elevation the enhancement could be due to NA convection, lightning, or Asian influence. The bottom figures show a tagged by region-CO simulation with the Asian and NA FF enhancement vs. PAN. We see this correlation is governed by the NAFF tracer. So this could be convection.lightning or a combination of the two.
PAN correlation dominated by NA tracer
[Q. Liang, UW]

7. Relative Importance of Upper trop NOx sources: July Lightning and Aircraft Totals
IPCC Aviation and Global Atm (1999)
1992 ~.5
2015(proj) ~1.0
Most we can increase aviation srce x2
NOT ENOUGH!
July Total (Tg N)
Global
Aircraft Source July (Tg N): .05, Yearly ~ .5
Lightning Source July (Tg N): .5, Yearly ~ 5
U.S.
Aircraft Source July (Tg N): .012
Lightning Source July (Tg N): .067
LI ~ 5 x Aircraft

8. Insufficient Convection? …not likely
Propane – NO2 correlations in BL and 8-10 km
To examine the simulation of convection we can look at the propane-NO2 correlation in the boundary layer and in the upper free troposphere. We see that the model overestimates C3H8 and NO2 in the boundary layer…but in the upper free troposphere we see the discrepency is due to missing NO2 not necessarily C3H8. So the answer must lie in the simulation of lightning.

9. OTD/LIS Climatology and GEOS-CHEM 2004
Mean July flash rate
GEOS-CHEM
OTD/LIS
Here we compare the GEOS –CHem July 2004 lightning flashrate in flashes/km2/s and OTD/LIS climatilogical flash rate over the U.S. We see that the gulf region is well simulated but there is about a factor of 2 underestimate north of 35N warranting a scaling up of NA lightning emissions.
[Flashes/km2/day]
 Factor of two underestimate above 35 N over U.S.

10. Convective Mass Flux GEOS-4 vs. GEOS-3:
Vertical Placement of NOx emissions greatly influences mean [NOx]
GEOS-CHEM uses a lightning parameterzation of (CTH)^5 over land. So it is highly dependent on the location of the cloud tops. The current version of GEOS4 has much different cloud top heights than previous versions of the assimilation. This is important because, although we scale the lightning total globally, NOx emissions are only emitted to the cloud top heights. I am still working on the corrections for these issues, but have done some sensitivity simulations to examine the effect of the vertical distribution of NOx on the concentrations of NOx observed in the upper troposphere.
Different assumptions in the vertical placement of lightning NOx  50% increase in mean [NOx]

11. Before we can compare NOy export with model simulation, I need to get the NOx lightning issue resolved, but here is a summary of results from the observations. Here are the NOy export efficiencies as calculated by Parrish 2005 for 4 regions: South, OffShore, NARE region, NE/MW. We should focus our attention on the region below 8 km where the NOy is not dominated by lightning NOx.
The NE and midwest show the highest export fraction around 10-15%. The south shows lower export efficiencies possibly due to less frequent ventilation reflected in the higher HNO3 fraction in the observations. The NARE region in the north atlantic shows the lowest f, likely due to the uplift of clean marine boundary layer air dominated by HNO3.
Longer residence time in the l

12. [Lee Murray] O3 vs. CO correlation 0-2 km: All INTEX and ITCT-2K2 data
All Data
Northeast
RMA slope = .47
R^2 = .5
Lower O3/CO correlation in South
South
RMA slope = .35
R^2 = .3
RMA slope = .47
R^2 = .49
Lee Murray, an undergraduate working in our research group is looking at is the O3/CO correlations in the various regions I mentioned as well as in NA anthropogenic pollution plumes in NA outflow. Above we see that O3/CO slope for the ensemble of ICARTT data is ~.43 –This may be an increase over previous studies (possibly due to the higher NOx/CO emissions ratios, due to the rapidly decreasing CO).
Looking at North South Differences in the O3/CO correlation – we would expect that the O3/CO correlation would be steeper in the south where the OPE is expected to be higher, but its seems that slope is steeper in the northeast and this lower slope is matched in th emodel. Lee is looking into this further.
[Lee Murray]