manuscript submitted to
JGR: Atmospheres
H
2
O
2
and CH
3
OOH (MHP) in the remote atmosphere.
1
II: Physical and chemical controls
2
Hannah M. Allen
1
, Kelvin H. Bates
2
, John D. Crounse
3
, Michelle J. Kim
3
,
3
Alexander P. Teng
3
, Eric A. Ray
4
,
5
, Paul O. Wennberg
3
,
6
4
1
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA
5
2
School of Engineering and Applied Sciences, Harvard University, Cambridge, MA, USA
6
3
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA
7
4
Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado, Boulder,
8
CO, USA
9
5
Earth System Research Laboratory, National Oceanic and Atmospheric Administration, Boulder, CO,
10
USA
11
6
Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA
12
Key Points:
13
•
The global distribution of H
2
O
2
and MHP reflects the influence of photochemistry,
14
convective transport, and wet and dry deposition.
15
•
Deposition of H
2
O
2
plays a key role in removing HO
x
within the marine bound-
16
ary layer.
17
•
MHP in the middle and upper troposphere is highly sensitive to how convective
18
transport is parameterized and the GEOS-Chem model substantially underesti-
19
mates the advective mass fluxes of MHP.
20
Corresponding author: Hannah M. Allen,
hallen@caltech.edu
Corresponding author: Paul O. Wennberg,
wennberg@caltech.edu
–1–
Accepted
Article
This article has been accepted for publication and undergone full peer review but has not been through
the copyediting, typesetting, pagination and proofreading process, which may lead to differences between
this version and the
Version of Record
. Please cite this article as
doi: 10.1029/2021JD035702
.
This article is protected by copyright. All rights reserved.
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
Abstract
21
Hydrogen peroxide (H
2
O
2
) and methyl hydroperoxide (MHP, CH
3
OOH) serve as
22
HO
x
(OH and HO
2
radicals) reservoirs and therefore as useful tracers of HO
x
chemistry.
23
Both hydroperoxides were measured during the 2016–2018 Atmospheric Tomography Mis-
24
sion (ATom) as part of a global survey of the remote troposphere over the Pacific and
25
Atlantic Ocean basins conducted using the NASA DC-8 aircraft. To assess the relative
26
contributions of chemical and physical processes to the global hydroperoxide budget and
27
their impact on atmospheric oxidation potential, we compare the observations with two
28
models, a diurnal steady-state photochemical box model and the global chemical trans-
29
port model GEOS-Chem. We find that the models systematically under-predict H
2
O
2
30
by 5–20% and over-predict MHP by 40–50% relative to measurements. In the marine
31
boundary layer, over-predictions of H
2
O
2
in a photochemical box model are used to es-
32
timate H
2
O
2
boundary-layer mean deposition velocities of 1.0–1.32 cm s
−
1
, depending
33
on season; this process contributes to up to 5–10% of HO
x
loss in this region. In the up-
34
per troposphere and lower stratosphere (UTLS), MHP is under-predicted and H
2
O
2
is
35
over-predicted by a factor of 2–3 on average. The differences between the observations
36
and predictions are associated with recent convection: MHP is under-estimated and H
2
O
2
37
is over-estimated in air parcels that have experienced recent convective influence.
38
Plain Language Summary
39
Hydrogen peroxide (H
2
O
2
) and methyl hydroperoxide (MHP, CH
3
OOH) in the at-
40
mosphere can act as reservoirs for one of the main drivers of atmospheric chemistry, HO
x
41
(HO
x
= OH and HO
2
). Both H
2
O
2
and MHP were measured during the 2016–2018 At-
42
mospheric Tomography Mission (ATom), which investigated the atmosphere over the oceans
43
far from direct human influence. The measurements are compared to two types of mod-
44
els to assess our understanding of the chemical and physical processes that control their
45
abundance. We find that these models consistently predict H
2
O
2
to be lower and MHP
46
to be higher than was measured during ATom. We use the discrepancy between the model
47
and the measurements to investigate the role of deposition (removal of compounds from
48
the Earth’s atmosphere due to interactions with surfaces and with liquid water) on H
2
O
2
49
in the lowest portion of atmosphere and the role of convection (vertical transport dur-
50
ing storms and other meteorological events) on MHP between 6 and 12 km altitudes.
51
1 Introduction
52
Hydrogen peroxide (H
2
O
2
) and methyl hydroperoxide (MHP, CH
3
OOH) are of key
53
importance in the atmosphere because they reside at the center of the cycling of the at-
54
mosphere’s main oxidant HO
x
(OH and HO
2
radicals). They are both reservoirs of HO
x
55
due to their formation from HO
x
chemistry and reformation of HO
x
upon further pho-
56
tooxidation. H
2
O
2
is formed in the atmosphere primarily via the HO
2
self-reaction:
57
HO
2
+ HO
2
−−→
H
2
O
2
+ O
2
(1)
58
Whereas MHP arises primarily via the reaction of HO
2
with the methyl peroxy radical
59
(MPR, CH
3
OO). MPR is formed via the oxidation of methane (CH
4
) by OH:
60
CH
4
+ OH
−−→
CH
3
OO + H
2
O
(2)
61
CH
3
OO + HO
2
−−→
CH
3
OOH + O
2
(3)
62
The photochemistry of other larger organic molecules, such as acetone, can also lead to
63
MPR and subsequently MHP formation. The formation of H
2
O
2
and MHP is inversely
64
related to the abundance of NO and mostly occurs in low to moderate NO
x
(NO and
65
NO
2
) environments. High NO
x
limits hydroperoxide production because NO competes
66
with HO
2
for reaction with the peroxy radical precursors (HO
2
and MPR) to instead
67
–2–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
form either OH (for HO
2
) or CH
3
O (which decomposes to HCHO and HO
2
in the pres-
68
ence of O
2
). The abundance of H
2
O
2
and MHP is thus indicative of a key branching in
69
the oxidative chemistry of the troposphere: whether peroxy radicals react with HO
2
lead-
70
ing to radical termination or react with NO leading to radical propagation. This branch-
71
ing has particular consequences for atmospheric odd oxygen (O
x
, comprising O
3
and the
72
compounds with which it rapidly cycles such as NO
2
, NO
3
, etc.) as the former leads to
73
loss of O
x
whereas the later leads to production of O
x
.
74
Both H
2
O
2
and MHP undergo photochemical loss via photolysis or reaction with
75
OH that return HO
x
to the atmosphere. For H
2
O
2
, these losses are:
76
H
2
O
2
+
hv
−−→
2OH
(4)
77
H
2
O
2
+ OH
−−→
HO
2
+ H
2
O
(5)
78
For MHP, these losses directly return HO
x
as well as form formaldehyde (HCHO) which
79
may further react to return HO
x
to the atmosphere.
80
CH
3
OOH +
hv
−−→
CH
3
O + OH
(6)
81
CH
3
O + O
2
−−→
HCHO + HO
2
(7)
82
CH
3
OOH + OH
∼
0
.
7
−−−→
CH
3
OO + H
2
O
(8)
83
∼
0
.
3
−−−→
HCHO + OH + H
2
O
(9)
84
The branching ratio of the MHP + OH reaction varies between 0.65–0.83 in favor of ab-
85
straction at the hydroxy hydroperoxide leading to CH
3
OO formation, with a recommended
86
average of 0.70 (Niki et al., 1983; Vaghjiani & Ravishankara, 1989; Atkinson et al., 2006;
87
Anglada et al., 2017). For both hydroperoxides, photolysis recycles HO
x
and results in
88
a net of no change to total HO
x
while reaction with OH is net oxidant consuming. For-
89
mation and subsequent photolysis of H
2
O
2
converts HO
2
to OH, similar to the reaction
90
of HO
2
with NO, with an important difference being that the former does not lead to
91
O
3
production. However, hydroperoxide photochemical loss may not occur in the same
92
region as their formation, resulting in transport of HO
x
to areas that may have very dif-
93
ferent chemical regimes. For example, Jaegl ́e et al. (2000) found that convective trans-
94
port moves MHP from a region of generally low NO to the mid and upper troposphere
95
where NO levels are higher, leading to higher O
x
production.
96
H
2
O
2
and MHP are also subject to loss through wet and dry deposition that re-
97
moves these HO
x
reservoirs from the atmosphere, likely permanently. Deposition is pa-
98
rameterized as two distinct processes: dry deposition, which is the removal of gases or
99
particles from the atmosphere due to impaction onto land and ocean surfaces following
100
turbulent transfer; and wet deposition, which occurs when gases are incorporated into
101
suspended liquid water either by in-cloud scavenging or by washout from falling precip-
102
itation. Depositional loss depends not only upon the chemical properties of the gas, such
103
as solubility, but also upon a variety of factors including the planetary boundary layer
104
height, surface properties (e.g. area, roughness, moisture content, etc.), cloud liquid wa-
105
ter content, and meteorological parameters such as vertical wind speed (Walcek, 1987;
106
Jobson et al., 1998; B. D. Hall & Claiborn, 1997; Chang et al., 2004; Nguyen et al., 2015).
107
Due to its high solubility, H
2
O
2
is particularly susceptible to depositional losses while
108
the less soluble MHP is affected significantly less (Lee et al., 2000). Hydroperoxide loss
109
by deposition represents a net loss of oxidant as H
2
O
2
and MHP are removed with no
110
return of HO
x
to the atmosphere.
111
In addition to photochemical and depositional loss, hydroperoxides alter the atmo-
112
sphere’s oxidative potential via their transport in, for example, convective activity (Fig-
113
ure 1). Convection occurs when parcels of air become unstable with respect to vertical
114
transport; with strong enough convection, large towering cumulus clouds form that can
115
penetrate deep into the upper troposphere and lower stratosphere (UTLS, typically 8–
116
12 km). Because MHP and H
2
O
2
have different solubilities, the ratio of these two com-
117
pounds can be used as a metric to identify areas with recent convective activity (see, e.g.,
118
–3–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
H
2
O
2
CH
3
OOH
HO
x
CH
3
OOH
HO
x
NO
Time since convective influence
H
2
O
2
CH
3
OOH
Concentration
Time since convective influence
Concentration
O
3
HNO
3
NO
x
O
3
HNO
3
H
2
O
2
NO
2
Figure 1.
Simplified schematic of hydroperoxide cycling in the remote atmosphere. (Left) In
the lower troposphere, generation of HO
x
forms H
2
O
2
and MHP that cycle back to HO
x
with
photochemical reactions. H
2
O
2
readily undergoes deposition, removing it from the atmosphere
under both wet and dry conditions. MHP is less soluble and therefore may be lofted to the UTLS
during convection events, where it participates in HO
x
and NO
x
(e.g. from lightning) chemistry.
(Right) Schematic reflecting the changes in peroxide, NO
x
, HNO
3
, and O
3
relative mixing ratios
following convection to the UTLS.
Snow et al. (2007)). H
2
O
2
and MHP have similar mixing ratios in the boundary layer
119
(H
2
O
2
/MHP
∼
1–3), but H
2
O
2
is preferentially removed by cloud water (high scaveng-
120
ing efficiency) and precipitation that forms during convection while MHP is lofted with
121
minimal loss (low scavenging efficiency) (Heikes et al., 1996; O’Sullivan et al., 1999; Barth
122
et al., 2016; Bela et al., 2018; Cuchiara et al., 2020). The scavenging efficiency of these
123
hydroperoxides depends upon the interactions of these species with the environment as
124
they are lofted: interactions with the freezing and/or evaporation of cloud particles may
125
lead to less efficient scavenging of H
2
O
2
and/or more efficient scavenging of MHP (Bela
126
et al., 2016; Bozem et al., 2017; Y. Li et al., 2019). Following convection, MHP in the
127
UTLS may be enhanced by 3–6 times background levels (Cohan et al., 1999; Jaegl ́e et
128
al., 2000; Ravetta et al., 2001).
129
Overall, the influence of these compounds on the UTLS due to convective trans-
130
port lasts on order of 3–10 days based on the lifetime of MHP and H
2
O
2
and the sub-
131
sequent relaxation to local steady-state (Jaegl ́e et al., 1997; Bertram et al., 2007). How-
132
ever, in that time MHP may be photolyzed or oxidized to produce HO
x
and therefore
133
the net influence of convective transport of MHP may be to increase HO
x
levels in the
134
UT. For example, MHP photolysis may contribute 20–40% of HO
x
production in con-
135
vective outflow, compared with just 3–10% in background UTLS air (Prather & Jacob,
136
1997; Cohan et al., 1999; Jaegl ́e et al., 2000; Ravetta et al., 2001). Through transport
137
via convection and subsequent photochemistry, hydroperoxides can facilitate the effec-
138
tive transport of HO
x
from the lower atmosphere to the upper atmosphere and thereby
139
significantly increase the HO
x
abundance in the latter. Accurate parameterization of con-
140
vective transport of trace gases is a well-known challenge in models as the efficiency of
141
such transport depends on the details of many poorly quantified aspects of convection,
142
such as entrainment and detrainment rates, cloud water size distribution, vertical veloc-
143
ities, etc (Lawrence & Rasch, 2005; Zhang et al., 2021).
144
–4–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
In this study, the chemical and physical controls on global hydroperoxide mixing
145
ratios are assessed through comparisons between global seasonal measurements and pho-
146
tochemical models. The data collection methodology and global hydroperoxide distri-
147
bution are outlined in a companion paper (Allen et al., 2021). Here, we discuss the rel-
148
ative importance of photochemistry in setting hydroperoxide distributions from nearly
149
pole-to-pole over the Atlantic and Pacific oceans. We investigate the role of physical pro-
150
cesses on the distribution of H
2
O
2
and MHP, including estimating the rate of H
2
O
2
de-
151
position in the marine boundary layer needed to reconcile observations with box model
152
predictions. Finally, we use GEOS-Chem, a global chemical transport model, to inves-
153
tigate the role of convection in lofting hydroperoxides and their impact on the UTLS.
154
2 Methods
155
2.1 Field Deployment: Atmospheric Tomography Mission
156
Global observations of H
2
O
2
and MHP were made during the Atmospheric Tomog-
157
raphy (ATom) Mission, which used the NASA DC-8 to collect atmospheric vertical pro-
158
files of trace gases and aerosols in the remote atmosphere. The deployments were sched-
159
uled to sample each season: ATom-1 in August 2016 (7/29/16–8/23/16), ATom-2 in Febru-
160
ary 2017 (1/26/17–2/21/18), ATom-3 in October 2017 (9/28/17–10/27/17), and ATom-
161
4 in May 2018 (4/24/18–5/21/18). Each deployment consisted of 11–13 flights that fol-
162
lowed a prescribed flight track spanning latitudes between -85
◦
to 85
◦
by first traveling
163
southbound over the Pacific Ocean and then traveling northbound over the Atlantic Ocean.
164
During each flight, the aircraft underwent continuous ascents and descents to gather ver-
165
tical profiles ranging from altitudes of about 180 m above the ocean to just under 13,500
166
m. Hydroperoxides were measured using the CIT-CIMS, which combines a time-of-flight
167
and a triple quadrupole chemical ionization mass spectrometer using CF
3
O
–
ion chem-
168
istry to sensitively detect gas-phase atmospheric hydroperoxides. ATom primarily resulted
169
in data collected over the remote ocean, but did include periods over land due to flight
170
requirements; the data presented here have been filtered to exclude the measurements
171
collected over land. The ATom Mission and CIT-CIMS technique are discussed in much
172
further detail in the companion paper (Allen et al., 2021).
173
2.2 GEOS-Chem
174
Observations of atmospheric hydroperoxide mixing ratios from ATom were com-
175
pared to those predicted by the global transport model GEOS-Chem. GEOS-Chem is
176
a three-dimensional atmospheric chemistry model driven by meteorological data from
177
radio sondes and satellite observations of the Earth’s land surface, atmosphere, ocean,
178
and biogenic parameters (Lin & Rood, 1996; Bey et al., 2001; Bian & Prather, 2002; Keller
179
et al., 2014). The GEOS-Chem chemical module simulates atmospheric concentrations
180
of various species taking into account emissions, transport, chemistry, aerosol microphysics,
181
and deposition (Harvard, 2019). Further details on the chemical and physical mechanisms
182
used in the GEOS-Chem simulations are given in the Supporting Information. The me-
183
teorological data are assimilated from the Goddard Earth Observing System (GEOS)
184
of the NASA Global Modeling and Assimilation Office (GMAO). GEOS-Chem integrates
185
the meteorological data using the GEOS Forward Processing (GEOS-FP) data archive
186
with a native resolution of 0.25
◦
latitude by 0.325
◦
longitude and 72 vertical atmospheric
187
layers and a 3-hour temporal resolution (1-hour for surface data).
188
In this study, GEOS-Chem simulations were conducted for 2016–2018, with a one-
189
year spin up, using GEOS-Chem v11-2d at 2
◦
x 2.5
◦
latitude-longitude grid resolution
190
using the GEOS-FP meteorology archive. The model was updated with CH
3
OO + OH
191
chemistry (k = 1.6
×
10
−
10
cm
3
s
−
1
), as well as with improvements to certain emissions
192
inventories, as described in Bates et al. (2021). Sensitivity studies were conducted on the
193
rate of HO
2
loss on heterogenous surfaces by altering the uptake coefficient (
γ
, Stone et
194
–5–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
al. (2012)), on MHP wet scavenging by altering the MHP Henry’s Law Coefficient, and
195
on the rate of the CH
3
OO + OH reaction by altering the rate coefficient to assess the
196
impact of this chemistry on the hydroperoxide budget (see below and the Supporting In-
197
formation). GEOS-Chem results are presented in two forms: one in which model times
198
and locations are matched to the flight campaign data at 2 minute temporal resolution
199
and one in which ocean basin curtains are generated using a monthly averaged output
200
of each deployment that is centered on either -170
◦
longitude (Pacific Ocean) or -25
◦
lon-
201
gitude (Atlantic Ocean).
202
2.3 Photochemical Box Model
203
A zero-dimensional diurnal photochemical box model is used to evaluate the mea-
204
surements of hydroperoxides against their concentrations as predicted at pseudo steady-
205
state. The box model contains a detailed mechanism for remote tropospheric HO
x
-NO
x
-
206
VOC chemistry that uses over 35 chemical species and 85 reactions. The model does not
207
include physical processes such as heterogeneous chemistry (e.g. HO
2
loss on aerosols
208
is not included), transport, or wet or dry deposition. Data used for analysis have been
209
filtered such that the rate of NO
2
photolysis at each point is greater than 1
×
10
−
3
s
−
1
,
210
ensuring only measurements collected in daylight are used. Compounds included in the
211
model are either initiated with measured values when available or calculated from steady-
212
state and parameters such as temperature, pressure, and H
2
O mixing ratio are constrained
213
to their observed values.
214
Data used to initiate the model includes measurements of OH and HO
2
(uncertain-
215
ties of
±
35%, Brune, Miller, and Thames (2019)); photolysis rates (uncertainties of
±
15%,
216
S. R. Hall and Ullmann (2019)); H
2
O
2
, MHP, HO
2
NO
2
, and HNO
3
(uncertainties of
±
30%,
217
±
30%,
±
30%, and
±
30%, respectively, Allen et al. (2019)); H
2
O (uncertainties of
±
5%,
218
Diskin and DiGangi (2019)); peroxyacyl nitrates (PAN, uncertainties of
±
20%, Huey et
219
al. (2019)); HCHO (uncertainties of
±
10%, Hanisco et al. (2019)); CH
4
and CO (uncer-
220
tainties of
±
0.7 ppb and
±
3.6 ppb, respectively, McKain and Sweeney (2018)); NO, NO
2
221
and O
3
(uncertainties of
±
0.03–100%,
±
0.06–100%, and
±
0.03%, respectively, Ryerson
222
et al. (2019)); PAN (uncertainties of
±
0.06–100%, Elkins et al. (2019)); and acetone (CH
3
C(O)CH
3
,
223
uncertainties of
±
20%, Apel et al. (2019)); as well as temperature and pressure. The model
224
data for August (ATom-1) is limited by the availability of peroxyacetic nitrate (PAN)
225
measurements, leading to high uncertainty at the most poleward extremes.
226
Using the observations as an initial point, the model calculates the diurnally vary-
227
ing production and loss of each chemical species over the course of 120 simulated hours.
228
Photolysis rates for relevant species are calculated using actinic flux with cross sections
229
and quantum yields from Burkholder et al. (2015). The actinic flux is produced from the
230
Tropospheric Ultraviolet and Visible (TUV) radiation model (NCAR), which utilizes in-
231
puts of temperature, pressure, ozone column, and altitude to determine cloud-free ac-
232
tinic fluxes at the latitude, longitude, altitude, and time of year of the ATom measure-
233
ments. Comparisons of model-generated photolysis rates with those available from ac-
234
tinic flux measurements using the Charged-coupled device Actinic Flux Spectroradiome-
235
ters (CAFS) onboard indicate good agreement between the two and modeled photoly-
236
sis rates have been scaled to match CAFS observations where available. Chemical rates
237
are similar to those used in GEOS-Chem and calculated using temperature-dependent
238
rate constants from Burkholder et al. (2015), including a temperature dependent rate
239
constant of 3.7
×
10
−
11
exp(350/
T
) for CH
3
OO + OH chemistry (Jenkin et al., 2019). From
240
the TUV-generated actinic flux of a 24-hour solar cycle, the box model calculates a 5-
241
day diurnal pattern of compound mixing ratios at each point along the flight track. Five
242
days was chosen because the concentrations of most compounds have reached steady-
243
state (i.e. concentrations were invariant over multiple days) within this time frame.
244
–6–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
Figure 2.
Fraction of OH loss relative to photolysis (OH/(OH+
hv
)) for H
2
O
2
and MHP
across latitude bins (averaging all altitudes) and altitude bins (averaging all latitudes) for all
four deployments of ATom as predicted by a photochemical box model. A1–A4 refers to the four
different ATom deployments. Shading represents one sigma standard deviation of the mean.
3 Results and Discussion
245
3.1 Hydroperoxide Lifetime and Photochemistry
246
During ATom, H
2
O
2
mixing ratios were highest in the lower troposphere within
247
the tropical and subtropical latitudes, regions that typically exhibited high HO
x
-formation
248
and generally lower NO concentrations. Based on box model predictions, the highest pro-
249
duction of H
2
O
2
from HO
2
self-reaction occurs within latitudes of -10
◦
to 20
◦
degrees,
250
driven by the highest UV fluxes, and quickly falls off poleward. Similarly, the highest
251
production of H
2
O
2
from the HO
2
self-reaction occurs within the boundary layer and
252
lower troposphere (
<
2 km altitude) and quickly declines with increasing altitude. The
253
highest estimated rate of H
2
O
2
production from this chemistry occurred in the Octo-
254
ber deployment (ATom-3) with an average rate of 7.1
×
10
−
4
s
−
1
, or 1.3 ppb per day while
255
the lowest occurred during the February deployment (ATom-2) with an average produc-
256
tion rate of 4.3
×
10
−
4
s
−
1
or 0.6 ppb per day. However, H
2
O
2
also forms in regions where
257
other factors such as biomass burning drive high HO
x
and VOC concentrations that lead
258
to higher mixing ratios of this hydroperoxide (see Allen et al. (2021)).
259
Similarly, H
2
O
2
photochemical loss occurs in regions with strong photochemical
260
activity, primarily in the boundary layer of the tropical and subtropical latitudes. H
2
O
2
261
photolysis tends to comprise more than half the photochemical H
2
O
2
loss (loss due to
262
deposition is discussed in detail in the following section) with the remainder accounted
263
for by OH reaction (Figure 2). On average, reaction with OH is 30–35% of H
2
O
2
pho-
264
tochemical loss with global minimums of 2–6% and maximums of 63–75%, depending
265
on season, due to the slower average OH loss rate (calculated using the H
2
O
2
+ OH rate
266
constant from Burkholder et al. (2015) and measured [OH],
∼
3.6
×
10
−
6
s
−
1
) and more
267
than twice as fast average photolysis rate (calculated using the box model,
∼
9.0
×
10
−
6
268
s
−
1
). The relative contribution of OH to H
2
O
2
loss has a slight dependence on latitude
269
and season but mostly shows variation depending on altitude. The average contribution
270
–7–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
of OH reaction to H
2
O
2
loss is higher at lower altitudes (40-45% on average) and decreases
271
at higher altitudes (20-25% on average), although some variation does exist above 12 km
272
(e.g. ATom-3) likely due to overall less data collected at the highest altitudes. ATom ob-
273
servations show OH mixing ratios decline with altitude in the tropics but are fairly con-
274
sistent with altitude outside this region (Brune, Miller, Thames, Allen, et al., 2019), while
275
the increased radiation at higher altitudes leads to increasing photolysis rates in the up-
276
per atmosphere (Travis et al., 2020). These observations suggest the altitude dependence
277
of relative H
2
O
2
loss is likely due to the changes in radiation increasing photolysis rates.
278
Because H
2
O
2
photolysis conserves HO
x
while loss to OH represents a net loss of HO
x
,
279
areas with a high ratio of H
2
O
2
loss to OH indicate regions that are net oxidant consum-
280
ing.
281
The relative contribution of OH to the overall MHP photochemical loss (OH/(OH+
hv
))
282
exhibits latitudinal and altitudinal patterns very similar to that of H
2
O
2
. As shown in
283
Figure 2, MHP loss to OH comprises a higher percentage of MHP photochemical loss
284
than H
2
O
2
+ OH does of H
2
O
2
photochemical loss. During the ATom deployments, the
285
average rate of photolysis was 7.3
×
10
−
6
s
−
1
while the average rate of OH loss was nearly
286
twice as fast at 15
×
10
−
6
s
−
1
, leading to a much higher fraction of MHP loss to OH than
287
to photolysis. The average global value of MHP loss to OH varies from 66% (February,
288
ATom-2) to 72% (August, ATom-1), with minimums of 12–25% and maximums of 90–
289
95%, depending on season. This average percentage contribution of OH to photochem-
290
ical loss shows a very slight dependence on latitude and altitude. Loss to OH is typically
291
highest in the tropical and subtropical region and decreases moving poleward and typ-
292
ically highest at low altitudes (contributing about 80%) and decreases with increasing
293
altitude (to an average of 65%). Note that prior to running the model, points along the
294
flight track in which NO
2
photolysis was below 1
×
10
−
3
s
−
1
were excluded, leading to
295
some potential biases in the poleward extremes. MHP may undergo deposition as well,
296
but due to the relatively low Henry’s Law constant of MHP this loss is not nearly as im-
297
portant as it is for H
2
O
2
.
298
The lifetime of H
2
O
2
with respect to photochemical loss is 21 hours (daytime) on
299
average and spans the range from just a few hours (4–8) to several hundred (
>
100) de-
300
pending on season and latitude. The H
2
O
2
lifetime shows little dependence on altitude
301
but a strong dependence on latitude due to the variation in UV actinic flux. The H
2
O
2
302
photochemical lifetime is shortest in the equatorial region and increases moving poleward.
303
Similarly, the global average photochemical lifetime of MHP in the atmosphere is around
304
11 hours (daytime) and varies considerably between 1–3 hours to much longer (
>
50 hours)
305
depending on atmospheric region. Like H
2
O
2
, MHP photochemical lifetime does not vary
306
significantly with altitude but does show a latitudinal dependence due to sunlight. The
307
MHP lifetime is shortest in the tropics and subtropics and increases moving poleward.
308
While the H
2
O
2
photochemical lifetime is longer than that of MHP, H
2
O
2
is subject to
309
much larger non-photochemical losses than MHP and thus the overall lifetime of these
310
two species in the atmosphere is similar when physical losses are taken into account.
311
In addition to H
2
O
2
and MHP lifetime, the GEOS-Chem simulation reveals the dis-
312
tribution of atmospheric regions that are dominated by either HO
x
or NO
x
chemistry.
313
Figure 3 compares the fraction of MPR that reacts with NO, HO
2
, or OH in the Atlantic
314
Ocean basin for the May (ATom-4) deployment. ATom-4 is fairly representative of re-
315
action patterns across the deployments and ocean basins with some enhancement in ei-
316
ther the tropics or mid-latitudes depending on season and basin (see the Supporting In-
317
formation). These reactions show some latitudinal dependence; MPR + OH is localized
318
to the tropical regions while MPR + HO
2
is prominent in the tropics but occurs in the
319
mid-latitude regions as well. However, these reactions have a stronger altitudinal depen-
320
dence. In the lowest portion of the atmosphere, HO
2
contributes up to 60% of CH
3
OO
321
loss while reaction with OH contributes up to 25% of CH
3
OO loss as this is a very pho-
322
tochemically active region with higher HO
x
production. The contribution of HO
2
to MPR
323
reactions decreases with increasing altitude and declines to
<
10% in the upper tropo-
324
sphere (
>
8 km) as HO
x
declines and NO becomes more prominent due to additions of
325
–8–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
Figure 3.
Fraction of CH
3
OO that reacts with NO (top), HO
2
(middle), or OH (bottom)
across latitude and altitude for the Atlantic Ocean basin during the May deployment (ATom-4)
as predicted by GEOS-Chem. 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.
NO
x
from lightening and the increase in the lifetime of NO
x
due to the lower O
3
mix-
326
ing ratios. Hence this strong gradient with altitude correlates well with the expected dis-
327
tribution of both HO
x
and NO
x
sources. The impact of CH
3
OO + OH on MHP pro-
328
duction will be further explored in Section 3.3.
329
3.2 H
2
O
2
Deposition in the Marine Boundary Layer
330
Non-photochemical loss of H
2
O
2
is estimated here by comparing measurements of
331
H
2
O
2
to predictions from a photochemical steady-state box model. The box model con-
332
tains all expected gas-phase chemistry affecting the hydroperoxide budget, but lacks any
333
physical parameters such as transport, dry deposition, or wet scavenging. The box model
334
severely over-predicts H
2
O
2
, particularly in the lower troposphere below 3–4 km altitude
335
where the model on average predicts 2–4 times higher mixing ratios of H
2
O
2
than are
336
measured (Figure 4) and this under-prediction is consistent across time of year. Assum-
337
ing the model captures H
2
O
2
photochemical production and loss correctly and that the
338
model has reached steady-state with respect to H
2
O
2
, the observed over-prediction by
339
the model is likely a result of a missing non-photochemical loss term. Given that depo-
340
sition is expected to comprise a significant portion of H
2
O
2
loss, this missing loss term
341
likely reflects the lack of this term in the model. In addition, MHP is less likely to un-
342
dergo depositional loss and does not exhibit the same measurement–model disparity at
343
low altitudes (Figure S2). Here, we use the difference between instantaneous daylight
344
measurements and the box model to infer the magnitude of the missing loss rate and there-
345
fore the expected rate of H
2
O
2
deposition. Assuming steady-state, the difference between
346
the model and the measurements can be expressed as
347
[H
2
O
2
]
mod
−
[H
2
O
2
]
meas
=
P
L
−
P
L
+ NPL
(10)
348
NPL =
L
(
[H
2
O
2
]
model
[H
2
O
2
]
meas
−
1
)
(11)
349
–9–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
Figure 4.
Comparison of H
2
O
2
mixing ratios from measurements (CIT-CIMS) and those fol-
lowing chemical relaxation over 5 days after the measurements calculated using a photochemical
box model. Throughout the lower troposphere, H
2
O
2
mixing ratios are less than half of their
steady state values, reflecting the importance of loss via wet and dry deposition. The results
are averaged over 1 km altitude bins and shaded region represent one standard deviation of the
mean.
where NPL (Non-Photochemical Loss) is the missing loss rate (s
−
1
) needed to reconcile
350
the model with the measurements,
L
is the H
2
O
2
photochemical loss term, and
P
is the
351
H
2
O
2
production term (from HO
2
+ HO
2
chemistry). For the marine boundary layer,
352
in assigning all of the NPL to dry deposition, the deposition velocity can be estimated
353
as:
354
V
d
= NPL
×
BLH
(12)
355
where BLH is the marine boundary layer height.
356
Figure 5 indicates the estimated non-photochemical first-order loss for each deploy-
357
ment averaged over altitude and latitude calculated from Eq. 11. As expected from Fig-
358
ure 4, the loss rate is highest at low altitudes and decreases with increasing altitude. Within
359
the boundary layer, the average NPL is (11
±
3)
×
10
−
6
s
−
1
and varies considerably de-
360
pending on the month sampled (e.g. highest in August and lowest in October). From
361
the model, the average total photochemical loss rate is on order of (12
±
6)
×
10
−
6
s
−
1
362
in the boundary layer, hence physical loss is highly competitive in the lower atmosphere
363
and is estimated to result in the majority of H
2
O
2
loss. Above 8 km the NPL rate de-
364
clines to close to zero, indicating that the loss at these altitudes is primarily photochem-
365
ical and that the UTLS is closer to photochemical steady-state. The NPL rate also shows
366
some latitudinal dependence (Figure 5). The loss is highest in the tropics and subtrop-
367
ical latitudes and declines moving poleward. A low NPL rate in the subtropics (20–30
◦
)
368
suggests the influence of dry downwelling air in this region that is much closer to steady-
369
state. Similarly, the poles show an average NPL rate that is close to zero, suggesting that,
370
on average, physical losses are not as important as photochemistry in these regions.
371
Assuming that deposition to the ocean dominates the derived NPL term, we can
372
estimate the depositional velocity in the lower atmosphere, which depends upon the H
2
O
2
373
loss rate and the height of the marine boundary layer. Because the regions in which NPL
374
is close to zero, such as occurs at high latitudes (Figure 5), likely have other processes
375
beyond dry deposition contributing to peroxide loss, the deposition velocity is only cal-
376
culated using data from -30
◦
to 30
◦
latitudes and for altitudes less than the estimated
377
MBL (modeled via GEOS5). Eq. 12 gives median (mean
±
standard deviation) depo-
378
–10–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
Figure 5.
Calculation of the H
2
O
2
non-photochemical loss (NPL) rate averaged over altitude
(left, includes all latitudes) and latitude (right, includes all altitudes) for each deployment. The
apparent loss was found by comparing the ATom measurements to the predictions by a photo-
chemical box model and attributing the difference to a missing deposition loss term in the model.
A1–A4 refer to the four different ATom deployments. Shading represents one sigma standard
deviation of the mean.
sitional velocities of 1.2 (1.7
±
2.0), 1.3 (1.3
±
2.2), 1.2 (2.5
±
5.4), and 1.0 (1.5
±
2.0)
379
cm s
−
1
for the marine boundary layer average in February, May, August, and October,
380
respectively. These velocities correspond to median (mean
±
standard deviation) wind
381
speeds of 11 (17
±
17), 8.4 (8.5
±
3.3), 8.6 (8.3
±
2.5), and 6.4 (11.4
±
9.3) m s
−
1
, re-
382
spectively, within the same latitude and altitude region. Previous estimates, conducted
383
by comparing airborne or ship-based measurements with Lagrangian, chemical box, or
384
global circulation models (EMAC), found a rate between 0.5–1.8 cm s
−
1
at wind speeds
385
of 5–10 m s
−
1
(Stickler et al., 2007; Fischer et al., 2015). Hence the calculated deposi-
386
tion velocities in this study are within the range of previously estimated values. These
387
studies note that the deposition rate primarily depends upon the transfer velocity of H
2
O
2
388
to the ocean surface, which is determined by wind speed, rather than other parameters
389
such as ocean uptake resistance. However, other factors not accounted for in this anal-
390
ysis may impact the calculated deposition velocity. Entrainment of H
2
O
2
from aloft, for
391
example, will lead to an underestimation of the H
2
O
2
deposition velocity by providing
392
an unaccounted source of H
2
O
2
in the observations (for example, (Q. Li et al., 2003; Singh
393
et al., 2003)). The effect of entrainment on the deposition calculation presented here is
394
evaluated in the Supporting Information.
395
Because H
2
O
2
deposition represents a permanent loss from the atmosphere, this
396
loss is net oxidant consuming. A H
2
O
2
deposition rate of (8–12)
×
10
−
6
s
−
1
results in an
397
average net loss of 80 ppt H
2
O
2
per day for median boundary layer H
2
O
2
mixing ratios
398
(400 ppt). Combined with H
2
O
2
loss due to OH, this results in an average loss of 300
399
ppt HO
x
per day in the remote marine boundary layer. To assess the total magnitude
400
of this H
2
O
2
deposition on HO
x
, GEOS-Chem was run with zero H
2
O
2
deposition and
401
with the current (“standard”, see the Supporting Information) H
2
O
2
deposition rate dou-
402
bled. The standard run predicts H
2
O
2
H
2
O
2
dry deposition velocities of 0.5–1.5 cm s
−
1
,
403
with an average of 1.18 cm s
−
1
that gives a predicted H
2
O
2
lifetime of 23.5 hours against
404
dry deposition (assuming 1 km MBL height). Doubling the standard H
2
O
2
deposition
405
rate decreases boundary layer H
2
O
2
by 10–40% and provides a closer match to observed
406
H
2
O
2
mixing ratios at lower altitudes (
<
1 km altitude) for latitudes between -60
◦
and
407
60
◦
(Figure 6). However, note that GEOS-Chem does generally over-predict H
2
O
2
at
408
–11–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
Figure 6.
Correlation between CIT-CIMS measured and GEOS-Chem simulated H
2
O
2
mixing
ratios for different model deposition velocities in the non-polar remote marine boundary layer (al-
titudes
<
1km and latitudes between
−
60
◦
and 60
◦
). Doubling the H
2
O
2
deposition rate (right)
in the model provides a closer match to observed H
2
O
2
mixing ratios below 1 km altitude. The
RMS error averaged across deployments for each simulation type is 0.78 for no deposition, 0.39
for the standard deposition, and 0.30 for 2x deposition. The dashed line indicates a 1:1 (perfect)
comparison.
Figure 7.
Effect of H
2
O
2
deposition on HO
x
in the Atlantic remote marine boundary layer
during the October (ATom-3) deployment. Total HO
x
in the boundary layer declines by 1–5% in
the high deposition simulation compared to when deposition is not included, and particularly af-
fects the equatorial and mid-latitudes (40–60
◦
) suggesting this is where H
2
O
2
deposition is most
important.
all altitudes, not just within the MBL (see Section 3.3 and Figure S5), and thus increas-
409
ing the deposition rate may be helping compensate for an issue in the photochemical pro-
410
duction or loss of H
2
O
2
in GEOS-Chem that exists at all altitudes. With this increased
411
deposition, H
2
O
2
mixing ratios in the boundary layer in GEOS-Chem are up to 2.5–4
412
times lower than their value in the no deposition run and result in a 5–10% decrease (de-
413
pending on season) in total HO
x
, indicating the importance of H
2
O
2
deposition as a HO
x
414
sink in the marine boundary layer (Figure 7). These losses are especially important at
415
the equator and in the southern mid-latitudes (40–60
◦
) in February and October and
416
prevalent in the equatorial to northern mid-latitudes (40–60
◦
) and northern pole (
>
80
◦
)
417
in May and August, following the pattern in seasonal distributions of sunlight and rain.
418
Both the Atlantic and Pacific ocean basins have a very similar distribution in the change
419
in HO
x
.
420
–12–
Accepted
Article
This article is protected by copyright. All rights reserved.
manuscript submitted to
JGR: Atmospheres
3.3 MHP Transport via Convective Activity
421
Figure 8.
Ratio of measured (obs) H
2
O
2
(top row) and MHP (bottom row) with that pre-
dicted by GEOS-Chem (model) as a function of altitude. Lighter points were collected in the
troposphere and darker points were collected in the lower stratosphere, as determined by O
3
mea-
surements above 100 ppbv for altitudes above 7 km. The solid line indicates the median value for
all points in 1 km altitude bins and the dashed line represents 1:1 or prefect correlation between
the model and the measurements. GEOS-Chem systematically over-predicts H
2
O
2
and under-
predicts MHP relative to the measurements at all altitudes, with the discrepancy most severe at
altitudes above 8 km.
Correlations between GEOS-Chem and the measurements across the whole deploy-
422
ment (shown in the Supporting Information) for H
2
O
2
indicate generally good agreement
423
between the two, although the model does systematically over-predict H
2
O
2
. In partic-
424
ular, the months of August (ATom-1) and May (ATom-4) produce correlations between
425
the model and the measurements with slopes of 1.03 and 1.05 with R
2
values of 0.69 and
426
0.72, respectively; the bias increases in February and October, with slopes of 1.13 and
427
1.18, respectively (measurement uncertainty is 30%). However, this agreement worsens
428
in the UTLS as indicated in Figure 8, which depicts the ratio of measured H
2
O
2
and MHP
429
to that predicted by GEOS-Chem. Above 8 km, the average ratio of the model to mea-
430
surements ranges between 2–4, depending on season and altitude, and the model may
431
be as much as 10 times higher than the measurements in the lower stratosphere. This
432
ratio corresponds to an absolute difference of several tens of pptv: observed H
2
O
2
above
433
8 km altitude is in the range of
<
1–600 pptv with a mean 95
±
100 pptv) while GEOS-
434
Chem predicts 10–900 pptv with a mean of 210
±
170 (averaged across all four deploy-
435
ments). GEOS-Chem more accurately captures the hydroperoxide precursor, HO
2
, though
436
does under-predict HO
2
at the highest (above 10 km) and lowest altitudes (below 2 km),
437
relative to measurements. Above an altitude of 10 km, GOES-Chem over-predicts HO
2
438
by a factor of about 2, corresponding to an absolute difference of about 5–10 pptv (see,
439
–13–