poster - University of Wollongong

Using SEAC RS observations to improve modeling of isoprene chemistry in GEOS-Chem
4
Jenny A. Fisher ([email protected]), Daniel J. Jacob , Patrick S. Kim , Katherine R. Travis , Lei Zhu , Karen Yu , and the SEAC RS Science Team
1
2
2
2
2
4
University of Wollongong, Australia; 2Harvard University, Cambridge, Massachusetts
Large model overestimates are seen
for organic nitrates formed from
isoprene oxidation under high-NOx
conditions. Ongoing work will identify
the sources of these biases, with initial
emphasis on organic nitrate sinks
including photolysis, deposition, and
loss to aerosols.
Campaign-averaged statistics for
isoprene and its oxidation products
show model high biases above 1 km,
likely due to errors in boundary layer
heights and mixing. Below 1 km,
GEOS-Chem provides a reasonable
simulation of several species, including
in particular isoprene peroxides and
HCHO. Isoprene shows a mean low
bias but with significant variability.
Over the low-NOx Ozarks region,
GEOS-Chem shows large positive
biases in isoprene and isoprene
peroxides. Although coincident with
a temperature bias of up to +3K, a
sensitivity simulation indicates that
temperature alone cannot explain the
model isoprene overestimates.
1. GEOS-Chem SEAC RS simulation
4
The GEOS-Chem full chemistry simulation was run with GEOS-FP meteorology for AugustSeptember 2013. The model was run at high horizontal resolution (0.25°x0.3125°)
over North America, nested within the global GEOS-Chem simulation, with boundary
conditions from a 4°x5° run. Results shown here are preliminary output, revision RB.
Anthropogenic emissions are from NEI 2008 and biomass burning emissions are from
FINN. Biogenic emissions are from MEGAN v2.1. Mean biogenic isoprene emissions for
the SEAC4RS period (1 Aug - 22 Sep 2014) are shown below for the full high-resolution
domain (left), and for the Southeast U.S. (right), defined here by the orange box.
Isoprene
4
HCHO
4
3
3
3
3
2
2
2
2
1
1
1
0
0
1.50e+12
3.00e+12
4.50e+12
6.00e+12
atoms C/cm2/s
Isoprene oxidation chemistry is based
on Mao et al. (2013). The first steps of
the oxidation process are shown in the
figure to the right. The mechanism also
includes explicit treatment of second
generation nitrates, as well as nighttime
oxidation of isoprene by nitrate radicals.
Isoprene nitrates are subject to removal
via photolysis, deposition, and oxidation
by OH, with lifetimes of hours to days.
Acknowledgments:
This work was supported by the NASA Tropospheric Chemistry
Program and by a University of Wollongong Vice Chancellor’s
Postdoctoral Fellowship.
0.00e+00
1.50e+12
0
1000
500 10001500
4.50e+12
3000
0
0
5000
Alkyl Nitrates (ΣANs)
DC-8 Obs. Total ΣANs
GEOS-Chem Total ΣANs
ISOPN
MVKN+MACRN
PROPNN
ETHLN
R4N2
3
2
1
atoms C/cm /s
2
0
200
400
ΣANs, pptv
600
800
Data: Cohen (PI)
20
200 400 600
1200
GEOS-Chem: slope=0.13, r=0.93
1000
800
600
400
200
0
0
2000
4000
6000
HCHO, pptv
Alkyl nitrates (ΣANs, left) are significantly
overestimated in GEOS-Chem at altitudes
<3 km. The lumped alkyl nitrate R4N2
dominates the simulated ΣANs budget.
Approximately 60% of simulated R4N2
comes from nighttime isoprene oxidation.
The ΣANs overestimate is driven by second
generation oxidation products; first
generation isoprene nitrates (ISOPN) are
unbiased in the lowest 1 km. Propanone
nitrate (PROPNN), responsible for on
average 20% of modeled ΣANs in surface
air, is a factor of 5 too high.
There is a strong observed correlation
between HCHO and ΣANs. The slope of
the relationship (0.14) is slightly lower
than during ICARTT (0.17, Mao et al.,
2013) and SENEX (0.25, Mao, pers. comm.)
1400 DC-8 Obs.: slope=0.14, r=0.74
Mao, J., F. Paulot, D.J. Jacob, R.C. Cohen, J.D. Crounse, P.O.
Wennberg, C.A. Keller, R.C. Hudman, M.P. Barkley, and L.W.
Horowitz. (2013), Ozone and organic nitrates over the
eastern United States: Sensitivity to isoprene chemistry, J.
Geophys. Res., 118, doi:10.1002/jgrd.50817.
0
GEOS-Chem provides a reasonable estimate of isoprene and its standard
oxidation products in the boundary layer. There is on average a small bias in
surface-level isoprene, although there is large variability in the structure of the bias.
Model overestimates above the boundary layer likely reflect problems with boundary
layer height and mixing driven by the underlying meteorological fields.
0
References:
2000
35
50
65
80 %
DC-8 ISOPOOH (PI Wennberg): 8/30, 9/06, 9/11
GEOS-Chem ISOPOOH: 8/30, 9/06, 9/11
The figure above compares median profiles from GEOS-Chem sampled along the
flight track to DC-8 observations over the Southeast U.S. for isoprene, formaldehyde
(HCHO), methyl vinyl ketone (MVK) + methacrolein (MACR) and isoprene peroxides
(ISOPOOH). For MVK+MACR, the dotted line shows the model concentrations of
MVK+MACR+ISOPOOH to bracket the uncertainty in the measurement.
6.00e+12
Mao et al. (2013)
1000
0
The lowest values during the campaign were
observed in the boundary layer over the
Ozarks during these three flights, indicating
very low NOx conditions. In this region
GEOS-Chem displays significant ISOPOOH
overestimates, shown below.
Data: Wisthaler, Hanisco, Wennberg (PIs)
Altitude, km
3.00e+12
1
Concentration, pptv
ΣANS, pptv
0.00e+00
4
GEOS-Chem
DC-8 Obs.
4
Southeast Region
High- vs. low-NOx regimes can be distinguished in the GEOS-Chem simulation based on
model chemistry. The map below shows the simulated percentage of isoprene peroxy
radicals (ISOPOO) reacting with NO along the DC-8 tracks for altitudes < 1 km during
flights on 8/30, 9/06, and 9/11, 2013.
High values (e.g., near Houston) imply most
% ISOPOO+NO: 8/30, 9/06, 9/11
ISOPOO is reacting with NO, indicating
high-NOx environments, whereas low values
indicate low-NOx environments.
ISOPOOH
4
5
GEOS-Chem nested domain
MVK+MACR*
3. Large model biases in the low-NOx
Ozarks region
8000
10000
Data: Cohen, Hanisco (PIs)
Despite model biases in the ΣANs
simulation, the HCHO-ΣANs relationship
is well represented in GEOS-Chem, with
a modeled slope of 0.13.
0
750
1500
2250
3000
0
pptv
750
1500
2250
3000
pptv
The ISOPOOH bias over the Ozarks is coincident with a high bias in isoprene of ~3 ppbv
and in boundary layer air temperature of ~3K. The temperature bias is pervasive
throughout the Southeast, with GEOS-FP temperatures up to 9K too high, and may help
explain the isoprene and ISOPOOH biases in the Ozarks. Distributions of observed and
modeled boundary layer air temperature in the Southeast region are shown below.
DC-8 Obs.
0.025
GEOS-Chem
310
305
GEOS-FP Tmod, K
We evaluate model representation
of isoprene sources and chemistry
in the Southeast U.S. using SEAC4RS
observations and a high-resolution
version of the GEOS-Chem model with
biogenic emissions from MEGAN v2.1.
2. Isoprene and its oxidation products in
the Southeast U.S.
Relative Frequency
SUMMARY
Altitude, km
1
2
0.020
300
0.015
295
0.010
290
0.005
285
0.000
285
290
295
Temperature, K
300
305
1:1
280
280
Best Fit (T > 293K):
Tobs = 0.79(Tmod) + 60K
290
300
DC-8 Tobs, K
310
We use the best fit line between DC-8 and GEOS-FP 0-1 km temperatures (for Tobs > 293
K) to correct surface air temperature for use in computing MEGAN isoprene emissions.
We find that this correction decreases isoprene over the Ozarks by <1 ppbv and ISOPOOH
by <200 pptv, but is insufficient to entirely correct the Ozarks biases.
Comparisons between the SEAC4RS observations and the nested
GEOS-Chem simulation highlight problems with current model
representation of isoprene sources and chemistry in both low- and
high-NOx environments. Isoprene and ISOPOOH are overestimated
in the low-NOx Ozarks region, while isoprene-derived nitrates are
overestimated across the Southeast U.S. Ongoing work will focus on
identifying the causes of these discrepancies.

Download Report