manuscript submitted to
JGR: Atmospheres
Journal of Geophysical Research: Atmospheres
1
Supporting Information for
2
H
2
O
2
and CH
3
OOH (MHP) in the remote atmosphere.
3
II: Physical and chemical controls
4
Hannah M. Allen
1
, John D. Crounse
2
, Michelle J. Kim
2
, Alexander P. Teng
2
,
5
Kelvin H. Bates
3
, Eric A. Ray
4
,
5
, Paul O. Wennberg
2
,
6
6
1
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA
7
2
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA
8
3
School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA
9
4
Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado, Boulder,
10
CO, USA
11
5
Earth System Research Laboratory, National Oceanic and Atmospheric Administration, Boulder, CO,
12
USA
13
6
Division of Engineering and Applied Sciences, California Institute of Technology, Pasadena, CA
14
Contents of this file
15
•
GEOS-Chem Chemical and Physical Mechanisms
16
•
Heterogeneous HO
2
Loss in GEOS-Chem (Figure S1)
17
•
Comparison of CH
3
OO Reactions (Figures S2 and S3)
18
•
MHP Comparison to Photochemical Box Model (Figure S4)
19
•
Entrainment Velocity
20
•
Back Trajectory Analysis
21
•
CIT-CIMS and GEOS-Chem Correlations (Figures S5 and S6)
22
•
MHP Scavenging in GEOS-Chem (Figure S7)
23
•
Photochemical Box Model Convection Simulation (Figures S8 and S9)
24
•
Air Mass Age Using NO
x
/(NO
x
+HNO
3
) (Figure S10)
25
Introduction
26
This supporting information provides further details on the analytical methods used
27
to derive data and to support conclusions from this study. The GEOS-Chem chemical
28
and physical mechanism and discussion of heterogeneous HO
2
loss in GEOS-Chem sup-
29
plement the GEOS-Chem section by providing specific details on how the model works
30
and considers a potential systematic discrepancy in the model (HO
2
loss on aerosol sur-
31
faces). The MHP comparison to the photochemical box model supports Figure 4 in the
32
main text. The discussion on entrainment velocity supplements the analysis of H
2
O
2
de-
33
position by providing consideration of a potential interference in these results. The back
34
trajectory analysis discussion provides further details on the methodology used to gen-
35
erate Figure 10 in the main text. The CIT-CIMS and GEOS-Chem correlation (Figures
36
S5 and S6) and MHP scavenging analysis (Figure S7) provide further support for anal-
37
ysis described in Figure 9 and Section 3.3 in the main text. The photochemical box model
38
convection simulation discussion (Figures S8 and S9) of convective influence during ATom
39
by providing additional insight into the methods used to derive the results presented in
40
Table 1. The air mass age provides a point of comparison for Table 1 by using an alter-
41
native method to derive the presented results.
42
Corresponding author: Hannah M. Allen,
hallen@caltech.edu
–1–
manuscript submitted to
JGR: Atmospheres
S1. GEOS-Chem Chemical and Physical Mechanism
43
The chemistry simulation in GEOS-Chem includes coupled HO
x
-NO
x
-VOC-O
3
-
44
halogen-aerosol tropospheric and stratospheric chemistry. The chemical mechanisms and
45
rate constants are consistent with recommendations from JPL and IUPAC (Atkinson et
46
al., 2006; Burkholder et al., 2015). The model includes further mechanisms of PAN, iso-
47
prene, halogen and Criegee chemistry (Harvard, 2019). In addition to the gas-phase chem-
48
istry, GEOS-Chem includes gas-aerosol interactions through the effect of aerosol extinc-
49
tion on photolysis rates, heterogeneous chemistry, and gas-aerosol partitioning of semi-
50
volatile compounds, including HO
2
uptake onto aerosol surfaces using parameters listed
51
in Mao et al. (2010). Photolysis frequencies and rates are calculated using the Fast-JX
52
scheme (Bian & Prather, 2002) and implemented according to Mao et al. (2010). Emis-
53
sions in GEOS-Chem are from the Harvard-NASA Emission Component (HEMCO) mod-
54
ule, which allows users to select from a variety of local and global emissions inventories
55
(Keller et al., 2014).
56
GEOS-Chem includes various physical mechanisms to simulate transport, depo-
57
sition, and convection of chemical species. The model implements chemical transport us-
58
ing the advection algorithm of Lin and Rood (1996), a multidimensional semi-Lagrangian
59
transport scheme, along with the latitude-longitude grid of GEOS meteorological data.
60
Dry deposition (loss due to gravitational settling and impaction) is based on a resistance-
61
in-series model and includes aerosol loss to snow and ice surfaces. Wet deposition is treated
62
in two cases: scavenging in wet convective updrafts and wash-out due to precipitating
63
columns (Jacob et al., 2000). The wet scavenging due to convective updrafts depends
64
upon several factors, including the conversion rate of cloud condensate to precipitate,
65
the fraction of the compound in the liquid phase or cloud ice, and the retention efficiency
66
of the compound in cloud condensation. Wet deposition primarily affects HNO
3
, H
2
O
2
,
67
CH
3
OOH, and HCHO, among others, and the retention efficiencies of these compounds
68
depends upon their Henry’s law coefficient. Including wet scavenging in GEOS-Chem
69
prevents soluble compounds from being transported and dispersed in the upper atmo-
70
sphere due to convection.
71
S2. Heterogeneous HO
2
Loss in GEOS-Chem
72
Studies have suggested that HO
2
may be lost via heterogeneous uptake onto aerosols,
73
thereby decreasing the HO
x
availability and reducing the formation of H
2
O
2
and other
74
peroxides. Comparison of HO
2
mixing ratios measured at a variety of sites with those
75
predicted from models show consistently higher HO
2
in models than was observed, par-
76
ticularly in the marine boundary layer, and potentially indicates missing HO
2
loss pro-
77
cesses in these models (Stone et al., 2012). Inclusion of HO
2
loss to aerosol surfaces was
78
able to significantly improve agreement in model-measurement correlations (Jacob et al.,
79
2000; Mao et al., 2010). However, the extent to which heterogeneous loss occurs remains
80
uncertain due to the complexity of this process and the uncertainty in potentially com-
81
peting loss mechanisms. A comparison of modeled HO
2
and H
2
O
2
mixing ratios over the
82
Pacific and Atlantic with the latitude and altitude range afforded by the ATom data set
83
may shed some light into the extent of HO
2
heterogenous uptake and the factors that
84
affect this chemistry.
85
The efficacy of HO
2
heterogeneous loss is evaluated using
γ
, the reactive uptake
86
coefficient, which is defined as the fraction of HO
2
collisions with aerosol surfaces that
87
irreversibly react or are permanently trapped by the aerosol surface. There exist a range
88
of purported
γ
values for HO
2
that span two orders of magnitude from 0.01–1, which have
89
been derived from a variety of field measurements and laboratory studies (Jacob et al.,
90
2000; Mao et al., 2010; George et al., 2013). In addition, the environment surrounding
91
the aerosol can have a large impact on how conducive that particle is to chemical up-
92
take. (J. A. Thornton et al., 2008), for example, found that
γ
depends strongly on the
93
particle phase, size, pH, and temperature. Using GEOS-Chem they report
γ
values of
94
–2–
manuscript submitted to
JGR: Atmospheres
Figure S1.
Effect of altering the HO
2
heterogeneous uptake coefficient (
γ
) on the H
2
O
2
bud-
get in GEOS-Chem. The GEOS-Chem model was run with
γ
= 0.2 (standard run) and with
γ
= 0 (no uptake), with the change in H
2
O
2
mixing ratios shown as a function of latitude and alti-
tude. An uptake coefficient of 0 leads to higher H
2
O
2
mixing ratios in the polar marine boundary
layer in summer but very little effect during the winter. Note the difference in coloring scaling
factors between February and August.
0.1–0.3 in the tropical upper troposphere to
<
0.01 in the extra-polar lower troposphere;
95
similarly, (Mao et al., 2010) find
γ
values that range from 0.02 at 275 K to 0.5 at 220
96
K. The lowest
γ
are associated with solid surfaces, suggesting that only aqueous-aerosol
97
plays a major role in the atmosphere (Mao et al., 2010; George et al., 2013); while the
98
highest values of
γ
are associated with aqueous aerosols that contain transition metal
99
ions, particularly copper and iron, which convert HO
2
to H
2
O (J. Thornton & Abbatt,
100
2005; Mao et al., 2013). Despite these variations in
γ
, following (Jacob et al., 2000) and
101
(Mao et al., 2013) current recommendations for GEOS-Chem and other chemical trans-
102
port models utilize a static
γ
of 0.2 and full conversion of HO
2
on aerosol surfaces (Harvard,
103
2019).
104
Output from the GEOS-Chem model was compared to the ATom measurements
105
to evaluate how different estimations of the reactive uptake coefficient affect HO
x
and
106
therefore the HO
2
budget. The model was run at
γ
values of 0.2 (standard), 0.07 (mod-
107
erate), and 0 (no HO
2
heterogeneous loss). When the uptake coefficient is set to 0, more
108
HO
2
is available leading to higher H
2
O
2
mixing ratios. However, the simulations pro-
109
duce only a small effect on the H
2
O
2
budget when the HO
2
heterogeneous update rate
110
is varied. Compared to the standard simulation with
γ
=0.2, the production of H
2
O
2
in-
111
creases by very little, except in the polar regions during the summer months (Figure S1).
112
In February and May, H
2
O
2
increases by only about 0.1% in the equatorial region and
113
decreases by up to 0.3% in the polar regions. Most of the observed effect of altering
γ
114
occurs in the polar boundary layer of the south pole during August and in the polar bound-
115
ary layer of the north pole during October. In both these cases, H
2
O
2
increases by 5–
116
10% when
γ
is 0 compared with the standard simulation. However, these are also areas
117
–3–
manuscript submitted to
JGR: Atmospheres
in which the absolute H
2
O
2
mixing ratio is fairly low; hence the bulk of the data, which
118
samples in the tropics and subtropics, shows very little change.
119
S3. Comparison of CH
3
OO Reactions
120
Figure S2.
GEOS-Chem modeled fraction of CH
3
OO that reacts with NO (top), HO
2
(mid-
dle), or OH (bottom) across latitude and altitude for the Atlantic Ocean basin during the Octo-
ber deployment (ATom-3). Regions with high CH
3
OO + HO
2
produce MHP and are net oxidant
consuming. Note the different color bar scaling factors in each panel.
Figure S3.
GEOS-Chem modeled fraction of CH
3
OO that reacts with NO (top), HO
2
(mid-
dle), or OH (bottom) across latitude and altitude for the Pacific Ocean basin during the May
deployment (ATom-4). Regions with high CH
3
OO + HO
2
produce MHP and are net oxidant
consuming. Note the different color bar scaling factors in each panel.
–4–
manuscript submitted to
JGR: Atmospheres
Note one of the major differences between the photochemical box model and GEOS-
121
Chem in terms of the hydroperoxide budget is the treatment of CH
3
OO chemistry in cer-
122
tain regions as a result of two factors. The first is that the box model uses a temperature-
123
dependent reaction rate for the CH
3
OO + OH reaction, resulting in up to a 20% longer
124
lifetime at high altitudes and up to 25% shorter lifetime at low altitudes in the box model
125
compared to GEOS-Chem. The second is that the fraction of CH
3
OO lost to OH, NO,
126
and NO
2
is up to 10%–80% (depending on season) lower in the polar LS (
>
12 km al-
127
titude) in GEOS-Chem compared to the box model because GEOS-Chem includes CH
3
OO
128
+ halogen chemistry, while the box model does not.
129
S4. MHP Comparison to Box Model
130
Figure S4.
Comparison of MHP mixing ratios from measurements (CIT-CIMS) and those fol-
lowing chemical relaxation over 5 days after the measurements calculated using a photochemical
box model. The model more accurately captures MHP in the lower troposphere because MHP
is much less influenced by loss to wet and dry deposition. The results are averaged over 1 km
altitude bins and shaded region represent one standard deviation of the mean.
S5. Entrainment Velocity
131
To assess other factors that might convolute the analysis of H
2
O
2
deposition ve-
132
locity, the entrainment velocity of HCN was calculated and compared to the H
2
O
2
de-
133
position velocity. Entrainment is the mixing of an air mass into a second preexisting one;
134
i.e. the movement of free tropospheric air into the marine boundary layer. This move-
135
ment brings chemical compounds circulating aloft into the boundary layer and therefore
136
may affect the deposition calculation by providing an unaccounted source of H
2
O
2
that
137
masks the true H
2
O
2
loss due to deposition. In order to assess the potential extent of
138
this influence on H
2
O
2
mixing ratios, the entrainment rate of HCN was calculated. HCN
139
is primarily lost from the atmosphere via deposition and does not have significant pho-
140
tochemical loss, giving HCN a long atmospheric lifetime of 6 months. Because of this
141
long lifetime, HCN is considered well-mixed in the atmosphere, particularly above the
142
marine boundary layer. An estimate of the entrainment velocity (
V
e
) can be made by
143
comparing the flux of HCN to the ocean surface from deposition to the flux of free tro-
144
pospheric HCN entrained into the boundary layer from above:
145
V
e
=
V
d
×
[HCN]
BL
[HCN]
FT
−
[HCN]
BL
(1)
146
–5–
manuscript submitted to
JGR: Atmospheres
The deposition velocity (
V
d
) is assumed to be 0.12 cm s
−
1
(Singh et al., 2003; Li et al.,
147
2003). This calculation gives an average entrainment velocity of 0.08, 0.03, 0.04, and 0.08
148
cm s
−
1
for February, May, August, and October, respectively. Assuming that the entrain-
149
ment rate for H
2
O
2
is similar to that of HCN, these values are low enough that entrain-
150
ment is likely not a major factor in the H
2
O
2
budget. Instead, the transition between
151
the upper troposphere and the boundary layer is likely a region in which H
2
O
2
is lost
152
from the atmosphere due to in-cloud scavenging rather than a source of H
2
O
2
due to en-
153
trainment.
154
S6. Back Trajectory Analysis
155
Back trajectories were calculated using the Traj3D model (Bowman, 1993; Bow-
156
man & Carrie, 2002) run with the National Centers for Environmental Protections (NCEP)
157
Global Forecast System (GFS) 0.5
◦
by 0.5
◦
resolution meteorology. A cluster of 245 tra-
158
jectories were initialized in a cube with dimensions of 0.3
◦
longitude, 0.3
◦
latitude, and
159
20 hPa pressure (altitude) centered on one minute intervals along the flight track and
160
run backwards for 10 days. The latitude, longitude, and pressure altitude for each of the
161
245 trajectories were than averaged to a single latitude, longitude, and pressure for each
162
one minute point along the flight track. The probability of convective influence for each
163
parcel was calculated based on the coincidence of the parcel with clouds, high RH (above
164
50%), and cloud water. Cloud depth and height were obtained from the NASA Lang-
165
ley global gridded cloud products.
166
S7. CIT-CIMS and GEOS-Chem Correlations
167
Figure S5 and S6 indicate correlations between the GEOS-Chem model and mea-
168
surements made by the CIT-CIMS for H
2
O
2
and MHP, respectively. The GEOS-Chem
169
model overestimates H
2
O
2
and underestimates MHP across all four seasons. These cor-
170
relations exclude data collected over land.
171
Figure S5.
Correlation between CIT-CIMS measurements of H
2
O
2
mixing ratios in the re-
mote troposphere and those predicted by GEOS-Chem. Slopes of the correlations are 1.13, 1.05,
1.03, and 1.18 for February (ATom-2), May (Atom-4), August (ATom-1), and October (ATom-3),
respectively, with R
2
values of 0.66, 0.72, 0.69, and 0.56, respectively.
–6–
manuscript submitted to
JGR: Atmospheres
Figure S6.
Correlation between CIT-CIMS measurements of MHP mixing ratios in the re-
mote troposphere and those predicted by GEOS-Chem. Slopes of the correlations are 0.58, 0.49,
0.57, and 0.58 for February (ATom-2), May (Atom-4), August (ATom-1), and October (ATom-3),
respectively, with R
2
values of 0.65, 0.58, 0.79, and 0.75, respectively.
S8. MHP Scavenging in GEOS-Chem
172
Figure S7.
Difference in predicted MHP mixing ratios across latitude and altitudes when
GEOS-Chem is run with no MHP wet deposition compared with the standard simulation. MHP
mixing ratios increase significantly in the polar regions (
>
50% increase) but only by 20–30% in
the equatorial UTLS. This increase in MHP mixing ratios brings the GEOS-Chem predicted mix-
ing ratios closer to those measured, but is not enough to offset the difference between the model
and the measurements.
–7–
manuscript submitted to
JGR: Atmospheres
S9. Photochemical Box Model Convection Simulation
173
In order to assess the effect of convection on the UTLS, the photochemical box model
174
was initialized with measurements collected during ATom and a high ratio of MHP frac-
175
tion [MHP/(MHP+H
2
O
2
] and of high NO
x
fraction [NO
x
/(NO
x
+HNO
3
)] to simulate
176
conditions immediately following convection. The model progressed until steady-state
177
was reached (
∼
10 days). An expression was fit to the change in MHP fraction and NO
x
178
fraction between initiation and steady-state for 1 km altitude bins from 6 to 12 km as
179
shown in Figures S8 and S9. This expression was then used to derive air mass ages for
180
measurements collected between -30
◦
and 30
◦
during ATom in the manner of (Bertram
181
et al., 2007).
182
Figure S8.
Modeled fraction of MHP/(MHP+H
2
O
2
) over time following convective activity
for each altitude bin in the UTLS for ATom-1. An expression based on the mean change in MHP
fraction (black) was fit based on model results for each data point in the altitude bin (gray).
Figure S9.
Modeled mean MHP/(MHP+H
2
O
2
) over time following convective activity for
each altitude bin in February (ATom-2) and August (ATom-1). This mean change in MHP frac-
tion for each 1 km altitude bin between 6 km and 12 m was fit to an expression and used to
estimate the age of air masses sampled during ATom.
–8–
manuscript submitted to
JGR: Atmospheres
S10. Air Mass Age Using NO
x
/(NO
x
+HNO
3
)
183
Figure S10.
Comparison of the derived mean age (hours) of air mass encountered at 1 km
altitude bins from 6–12 km, using measured MHP/(MHP+H
2
O
2
) (black) fraction and the mea-
sured NO
x
/(NO
x
+HNO
3
) (red) fraction. Error bars indicate the standard deviation for the mean
in each altitude bin. Data from
−
30
◦
to 30
◦
latitudes.
References
184
Atkinson, R., Baulch, D. L., Cox, R. A., Crowley, J. N., Hampson, R. F., Hynes,
185
R. G., . . . Troe, J. (2006). Evaluated kinetic and photochemical data for atmo-
186
spheric chemistry: Volume II — gas phase reaction of organic species.
Atmos.
187
Chem. Phys.
,
6
, 3625-4055.
188
Bertram, T. H., Perring, A. E., Wooldridge, P. J., Crounse, J. D., Kwan, A. J.,
189
Wennberg, P. O., . . . Cohen, R. C. (2007). Direct measurements of the convec-
190
tive recycling of the upper troposphere.
Science
,
315
, 816-820.
191
Bian, H., & Prather, M. J.
(2002).
Fast-j2: Accurate simulation of stratospheric
192
photolysis in global chemical models.
J. Atmos. Chem.
,
41
, 281-296.
193
Bowman, K. P. (1993). Large-scale isentropic mixing properties of the Antarctic po-
194
lar vortex from analyzed winds.
J. Geophys. Res. Atmos.
,
98
, 23013-23027.
195
Bowman, K. P., & Carrie, G. D. (2002). The mean-meridional transport circulation
196
of the troposphere in an idealized GCM.
J. Atmos. Sci.
,
59
, 1502-1514.
197
Burkholder, J. B., Sander, S. P., Abbatt, J., Barker, J. R., Huie, R. E., Kolb, C. E.,
198
. . . Wine, P. H. (2015).
Chemical kinetics and photochemical data for use in
199
atmospheric studies, evaluation no. 18
.
Pasadena: Jet Propulsion Laboratory.
200
(JPL publication 15-10)
201
George, I. J., Matthews, P. S. J., Whalley, L. K., Brooks, B., Goddard, A., Baeza-
202
Romero, M. T., & Heard, D. E.
(2013).
Measurements of uptake coefficients
203
for heterogeneous loss of HO
2
onto submicron inorganic salt aerosols.
Phys.
204
Chem. Chem. Phys.
,
15
, 12829-12845.
205
Harvard, A. C. M. G. (2019).
GEOS-Chem.
Retrieved from
http://www.geos-chem
206
.org
(online)
207
Jacob, D. J., Liu, H., Mari, C., & Yantosca, R. M. (2000). Harvard wet deposition
208
scheme for gmi.
Harvard University Atmospheric Chemistry Modeling Group
,
209
online
, 1-6.
210
Keller, C. A., Long, M. S., Yantosca, R. M., Da Silva, A. M., Pawson, S., & Jacob,
211
D. J. (2014). HEMCO v1.0: a versatile, esmf-compliant component for calcu-
212
lating emissions in atmospheric models.
Geosci. Model Dev.
,
7
, 1409-1417.
213
Li, Q., Jacob, D. J., Yantosca, R. M., Heald, C. L., Singh, H. B., Koike, M., . . .
214
Streets, D. G.
(2003).
A global three-dimensional model analysis of the at-
215
mospheric budgets of HCN and CH
3
CN: Constraints from aircraft and ground
216
measurements.
J. Geophys. Res.
,
108
, 8827.
217
–9–
manuscript submitted to
JGR: Atmospheres
Lin, S.-J., & Rood, R. B. (1996). Multidimensional flux-form semi-lagrangian trans-
218
port schemes.
Mon. Wea. Rev.
,
124
, 2046-2070.
219
Mao, J., Fan, S., Jacob, D. J., & Travis, K. R.
(2013).
Radical loss in the atmo-
220
sphere from Cu-Fe redox coupling in aerosols.
Atmos. Chem. Phys.
,
13
, 509-
221
519.
222
Mao, J., Jacob, D. J., Evans, M. J., Olson, J. R., Ren, X., Brune, W. H., . . .
223
Carouge, C.
(2010).
Chemistry of hydrogen oxide radicals (HO
x
) in the
224
arctic troposphere in spring.
Atmos. Chem. Phys.
,
10
, 5823-5838.
225
Singh, H. B., Salas, L., Herlth, D., Kolyer, R., Czech, E., Viezee, W., . . . Kondo,
226
Y. (2003). In situ measurements of HCN and CH
3
CN over the pacific ocean:
227
Sources, sinks, and budgets.
J. Geophys. Res.
,
108
, 8795.
228
Stone, D., Whalley, L. K., & Heard, D. E. (2012). Tropospheric OH and HO
2
rad-
229
icals: field measurements and model comparisons.
Chem. Soc. Rev.
,
41
, 6348-
230
6404.
231
Thornton, J., & Abbatt, P. D.
(2005).
Measurements of HO
2
uptake to aqueous
232
aerosol: Mass accommodation coefficients and net reactive loss.
J. Geophys.
233
Res.
,
110
, D08309.
234
Thornton, J. A., Jaegl ́e, L., & McNeill, V. F.
(2008).
Assessing known pathways
235
for HO
2
loss in aqueous atmospheric aerosols: Regional and global impacts on
236
tropospheric oxidants.
J. Geophys. Res.
,
113
, D05303.
237
–10–