Comparing atmospheric [HO<subscr>2</subscr>]/[OH] to modeled [HO<subscr>2</subscr>]/[OH]: Identifying discrepancies with reaction rates

GEOPHYSICAL

RESEARCH

LETTERS,

VOL.

28,

NO.

6, PAGES

967-970,

MARCH

15,2001

Comparing

atmospheric

[HOz]/[OH]

to

modeled

[HOz]/[OH]'

Identifying

discrepancies

with

reaction

rates

E. J. Lanzendorf

1 T. F. Hanisco

1 P. O Wennberg

2 R. C Cohen

3, R M.

Stimpfle

•

and

J. G.

Anderson

1

Abstract.

Reactions

that

inter-convert

OH

and

HO2

are

directly

involved

in

the

catalytic

removal

of

O3

in

the

lower

stratosphere

and

in

the

catalytic

production

of

O3

in

the

upper

troposphere.

The

agreement

between

the

measured

and

modeled

[HO2]/[OH]

tests

our

current

understanding

of

this

important

chemistry.

Recent

changes

to

the

recommended

rate

constants

for

OH+O3

and

HO2+O3

call

into

question

how

accurately

the

chemistry

of

the

stratosphere

is

understood.

[HO2]/[OH]

calculated

with

the

new

recommendations

is

48%

higher

than

the

observations

throughout

the

lower

stratosphere,

exceeding

the

uncertainty

limits

of

the

observations

(20%).

The

extensive

atmospheric

data

set

allows

tests

of

the

rates

of

the

individual

processes

that

couple

these

free

radicals.

This

work

shows

that

the

discrepancy

is

largest

when

the

ratio

is

controlled

by

the

reactions

of

OH

and

HO2

with

ozone.

Introduction

The

partitioning

of

HOx

(OH

and

HO2)

is controlled

by

fast

cycling

reactions

that

inter-convert

OH

and

HO2

(Table

1).

These

reactions

are

significantly

faster

than

the

primary

sources

and

sinks

of

HOx

so

that

the

relative

concentration

of

HO2

and

OH

can

be

described

accurately

in

terms

of

cycling

reactions

alone.

In

the

lower

stratosphere,

reactions

with

NO

and

03

control

the

partitioning,

while

reactions

with

CO

and

NO

dominate

in

the

troposphere.

The

major

role

of

these

reactions

in

the

catalytic

removal

of

ozone

in

the

lower

stratosphere

and

in the

production

of

ozone

in

the

upper

troposphere

makes

the

quantitative

understanding

of

these

processes

important

[Cohen

et al.,

1994].

An

expression

for

the

ratio

of

HO2/OH

is given

by

the

ratio

of

rates

that

convert

OH-->HO2

to

the

rates

that

convert

HO2-->OH.

This

description

includes

small

contributions

from

the

halogen

oxides

(C10

and

BrO)

and

CH

4 reactions

summarized

in

Table

1:

[SO2

] kOH+O

• [0.3]"{-

kOH+co[CO]"•-

1.7 h• kOH+CH4[CH4]..•_..

'

•=

(•)

[OH]

,o2+,o

[vo]

+

Throughout

most

of

the

lower

stratosphere,

the

effects

of halogen

reactions

contribute

<10%

of

the

total

inter-conversion

rate.

The

reaction

of

OH

+ CH4

produces

-1.7

HO2

radicals

[Hanisco

et

al.,

2000],

and

accounts

for

-8%

of

the

OH-->HO2

rate

in

the

upper

troposphere

and

-3%

in

the

lower

stratosphere.

1Department

of

Chemistry

and

Chemical

Biology,

Harvard

University,

Cambridge,

MA

2Division

of

Geological

and

Planetary

Sciences

and

Division

of

Engineering,

California

Institute

of Technology,

Pasadena,

CA

-Department

of

Chemistry,

UC

Berkeley,

CA

Copyright

2001

by

the

American

Geophysical

Union.

Paper

number

2000GL012264.

0094-8276/01/2000GL012264505.00

The

agreement

between

the

measured

and

modeled

ratios

reflects

our

current

understanding

of the

sum

of the

terms

in Eq

1

and

our

understanding

of

the

chemistry

involved

in

catalytic

ozone

loss

and

production

in the

atmosphere

[Cohen

et al.,

1994;

Wennberg

et

al.,

1998].

The

variability

in

the

atmospheric

constituents

that

control

the

partitioning

of

OH

and

HO2

(03,

NO,

CO

....

) allows

errors

in

the

individual

terms

of

Eq

1 to

be

isolated.

In

this

letter

we

compare

the

calculated

to

measured

[HO2]/[OH]

using

the

most

recent

atmospheric

rate

constants

(JPL-00)

[Sander

et al.,

2000]

and

the

earlier

evaluation

(JPL-97)

[DeMote

et

al.,

1997].

Discrepancies

between

the

two

are

then

discussed

to

highlight

our

understanding

of

the

chemistry

in

the

atmosphere

and

to

determine

the

areas

where

further

detailed

studies

would

be

most

beneficial.

Measurements

The

in situ

measurements

presented

here

were

obtained

during

the

1994-1997

NASA

ER-2

field

campaigns.

The

ASHOE/MAESA

mission

deployed

from

Christchurch,

New

Zealand

(44øS,

172øE)

in

1994.

The

STRAT

campaign

deployed

from

Barbers

Point,

HI

(21øN,

155øW)

and

Moffett

Field,

CA

(37øN,

122øW)

in

1995-1996.

The

POLARIS

mission

deployed

l¾om

Fairbanks,

Alaska

(65øN,

148øW)

in

1997.

Collectively,

these

observations

span

large

variations

in

atmospheric

conditions,

having

been

obtained

during

all

four

seasons

and

over

a wide

range

of

altitude

(10-21

km)

and

latitude

(70øS

to

90øN).

The

instrument

used

for

the

measurement

of

OH

and

HO2

on

the

ER-2

research

aircraft

is described

in

detail

by

Wennberg

et

al.

[1994].

OH

is

measured

by

laser

induced

fluorescence

(LIF)

with

an

accuracy

of

+25%

(26),

and

an

instrument

precision

of

+lx104

molecules/cm

3 (typically

-1%)

for

a 1 min.

averaging

period.

HO

2 is

measured

by

chemical

conversion

to

OH

using

NO,

with

the

OH

subsequently

detected

by

LIF.

The

measurement

accuracy

of

HO2

is +30%

(213),

with

an

instrument

precision

of +2x104

molecules/cm

3 (typically-0.5%)

for

1 min.

averaged

data.

Because

HO

2 is measured

by

chemical

conversion

to

OH,

the

accuracy

of

the

measured

[HO2]/[OH]

is insensitive

to

the

calibration

of

the

instrument

to

OH

and

depends

primarily

upon

the

uncertainty

in

the

conversion

efficiency

of

HO2

to

OH

within

the

instrument

(approximately

+20%)

[Cohen

et al.,

1994;

Table

1.

HOx

Partitioning

Reactions

Conversion

of

OH-->HO2

Conversion

of

HO2-->OH

OH

+ O3-->HO2

+ 02

OH

+ CO

02

) HO2

+ CO2

OH

+ C10-->HO2

+ C1

OH

+ CH4-->-->CH20

+ HO2

CH20

+ hv-->-->

0.7

HO2

+ O3-->OH

+ 202

HO2

+ NO-•OH

+ NO2

HO2

+ C10-•HOC1

+ 02

HOC1

+

hv-->OH

+

C1

HO2

+ BrO-->HOBr

+ 02

HOBr

+

hv-->OH

+ Br

967

968

LANZENDORF

ET

AL.:

COMPARING

ATMOSPHERIC

HO2/OH

TO

MODELED

HO2/OH

Table

2.

ER-2

Measurements

for

HO2/OH

Analysis

Species

Uncertainty

Reference

OH

+25%

+.01pptv

(2(5)

Wennberg

et al.,

[1994]

HO2

+30%

+.02pptv

(2(5)

Wennberg

et al.,

[ 1994]

03

+ 5%

Proffitt

et aL,

[1989]

NO

_+ 6%+4pptv

Fahey

et aL,

[ 1989]

CO

+10%

Webster

et al.,

[ 1994]

CH4

-t- 5%

Webster

et

al.,

[ 1994]

C10

+ 15%

Brune

et al.,

[1989b]

BrO

+ 15%

Brune

et

al.,

[1989a]

Pressure

+ 0.25

mbar

Chart

et aL,

[ 1989]

Temp.

(K)

+ 0.3

K

Chart

et aL,

[1989]

Wennberg

et

al.,

1994].

Nothing

has

changed

in

the

instrument

configuration

over

the

three

missions

that

would

bias

the

conversion

efficiency

over

time.

All

of

the

molecular

species

in

Eq

1 that

govern

HO2/OH

are

measured

on

the

ER-2.

These

are

summarized

with

pressure

and

temperature

measurements

in

Table

2 (experimental

uncertainties

are

included).

BrO

is inferred

from

the

empirical

BrO-N20

relationship

[Wamsley

et al.,

1998]

determined

during

the

ASHOE/MAESA

campaign.

Results

The

contributions

of

each

individual

rate

to

the

total

error

in

the

calculated

[HO2]/[OH]

can

be

isolated

by

restricting

data

so

that

[HO2]/[OH]

is most

sensitive

to

the

terms

in

question.

Figure

2 shows

the

fractional

error

of

the

ratio

calculated

from

Eq

1

plotted

versus

the

fractional

contribution

of

the

rate

of

OH+O3

(romo3)

in (a, b) and

the

rate

of HO2+O3

(rHO2+O

3) in (c, d) to the

total

inter-conversion

rate

of

OH

and

HO2

(rtotat).

The

JPL-97

rate

constants

are

used

in

all

panels

with

the

exception

of

JPL-00

komo3

in (b)

and

JPL-00

kilo2+03

in (d)

(the

upper

trace

in (d)

utilizes

both

JPL-00

koH+03

and

JPL-00

kilo2+03).

The

fractional

error

is

expressed

as

E'=

([HO2]/[OH]cak.-'[HO2]/[OH]

........

)

/[HO2]/[OH]meas

and

the

fractional

contribution

of

each

rate

is

expressed

as X = roH+o3/rtota

t or X = rHO2+o3/rtota

!. The

results

of

each

regression

are

used

to

estimate

the

systematic

error

in

each

ratio

of

rates

(see

Cohen

et

al.,

[2000]).

The

measurement

uncertainty

is determined

by

adding

the

weighted

uncertainties

in

quadrature,

i.e. 0 -2 = o-2no2/on

+ a12(52NO

+ a22(52co

+ .... where

ai

are

the

fractional

weighting

terms.

In panels

(a,

b)

the

data

are

restricted

so

that

HO2+O3

is <10%

and

HO2+C10

and

HO2+BrO

combined

are

<5%

of

the

total

conversion

rate

of

HO2-->OH.

The

total

uncertainties

(weighted

error

from

the

measurement

uncertainties

and

rate

constants)

from

these

terms

are

included

with

the

measurement

uncertainty.

With

this

restriction,

the

ratio

is

dominated

by

the

flux

in

three

reactions:

The

comparisons

between

the

calculated

and

measured

[HO2]/[OH]

in the

troposphere

(a)

and

stratosphere

(b)

are

shown

in

Figure

1 using

both

the

JPL-97

and

JPL-00

evaluations.

In

the

upper

troposphere,

the

difference

between

the

calculations

using

the

two

recommendations

is

small

(<3%)

because

the

conversion

of

OH-->HO2

in

the

upper

troposphere

is dominated

by

OH+CO

ß r•

15

(Rou+co)

and

the

conversion

of

HO2-->OH

depends

almost

entirely

upon

HO2+NO

(RHo2+NO).

The

rate

constants

of these

•

•0

reactions

are

the

same

in

JPL-97

and

JPL-00.

The

mean

value

of

the

calculated

ratio

using

either

the

JPL-97

or

JPL-00

evaluations

is

-7%

greater

than

the

measured

ratio

in

the

troposphere.

The

5

mean

uncertainty

of

the

calculated

ratio

is -110%,

mostly

due

to

the

large

uncertainty

of komco

(-100%).

The

uncertainties

in the

0

calculated

ratio

are

determined

by

adjusting

the

reaction

rate

constants

and

in

situ

measurements

for

the

relevant

terms

to

their

lo

uncertainty

limits

and

then

calculating

the

weighted

root

sum

20

of

the

squares

of

the

uncertainties.

Data

in

the

troposphere

show

more

scatter

than

in

the

stratosphere

because

of

decreased

precision

resulting

from

lower

mixing

ratios

of

OH

and

HO2

and

higher

background

noise

at lower

altitude.

In the

stratosphere,

the

ratio

differs

substantially

between

the

JPL-97

and

JPL-00

calculations.

The

mean

value

of

the

calculated

ratio

using

JPL-97

is 14%

higher

than

the

measured

5

ratio.

The

calculation

using

JPL-00

is

48%

higher

than

the

observations,

well

outside

the

20%

uncertainty

limits

of

the

[HO2]/[OH]

observations.

However,

all

data

are

within

the

0

uncertainties

of

the

calculated

ratio,-125%

(JPL-97)

and-90%

(JPL-00).

The

large

difference

between

the

JPL-97

and

JPL-00

cases

results

from

the

importance

that

komo3and

kHo2+o3

play

in

the

stratosphere,

where

concentrations

of

03

are

high.

The

OH+O3

reaction

(RoH+o3)

accounts

on

average

for-95%

of the

OH•>HO2

conversion

and

the

HO2+O

3 reaction

(RHo2+o3)

accounts

for

36%

(JPL-97)

and

31%

(JPL-00)

of

the

HO2-->OH

conversion

rate

respectively.

The

combined

effect

of

an

increase

of-25%

in the

numerator

term

kon+o3

and

decrease

of-25%

in

the

denominator

term

kHo2+o3

at stratospheric

temperatures

is a

shift

of

+33%

in

the

calculated

ratio

of

[HO2]

to

[OH].

I ' I ' I ' I • I ' I

2o

a) Troposphere

I

'

I

'

I

'

I

o

.

ß

JPL-97

'

•.-'•

ß

o JPL-00

-

..........

+20%

'

I

'

I

'

I

'

I

'

I

'

I

'

I

'

I

'

I

'

0

2

4

6

8

10

12

14

16

18

2O

Stratosphere

v-oo

o ...........

-

]:]

,.: ',-'-"',-

o

:':•,:::e'"':•;•

?':*'""'

ß

.- ......

'

_-

'-

+40

•

,;•,,:,:o;'•,•,

---'

---

0

2

4

Measured

Figure

1.

Measured

[HO2]/[OH]

versus

the

calculated

ratio

using

JPL-97

and

JPL-00

for

(a)

tropospheric

(STRAT)

and

(b)

stratospheric

(ASHOE/MAESA

and

POLARIS)

data

averaged

at

1 min.

intervals.

Tropospheric

data

were

restricted

with

the

criteria

tomco

>5x

romo•

and

NO

> 50

pptv.

Stratospheric

data

are

restricted

with

the

crit'eria

SZA

< 80 ø and

air number

density

<

2.5x10

•

molecules

cm

-3.

The

dashed

lines

show

the

20%

uncertainty

of

the

[HO2]/[OH]

measurement.

The

40%

error

bars

are

shown

for

reference

in

(b).

LANZENDORF

ET

AL.'

COMPARING

ATMOSPHERIC

HO2/OH

TO

MODELED

HO2/OH

969

!.0

0.5

0.0

0.5

0.0

a) JPL-97

ø

c) JPL-97

1.0

ø

E = 0.11

- 0.06X

E = 0.11

+ 0.08X

8

o

q,%

o ,• oO

o

0.5

•o

o

o• •o

o

o o

o ø

o

•

.... '•_. g• .... • .... •_ .... • ....

•'

•?•,

•

• .......

.......

. .......

& ....

•q•.o

...................

• ...................................................

0.0

0.2

0.4

0.6

x = ,'ou%/(ro,+o,,co.c•

h)

,

I

,

:

I

:

I

:

-0.5

.d) JPL-00

] ! .0

kno

+o

& komo

E = 0.34

+ 0.32X1

**d),

,':"

' ,c.,

.e".

r

.;

.:..

0.8

0.

0.2

0.3

0.4

0.5

0.6

0.7

X = rHO•,

O•/(rHO•,O•,UO,½lO.mO)

Figure

2.

The

fractional

error

in the

calculated

[HO2]/[OH]

i•

plotted

versus

the

fractional

contribution

of

OH+O3

(a,

b)

and

HO2+O3

(c,

d)

to

the

total

inter-conversion

rate

between

OH

and

HO2.

JPL-97

rates

are

used

throughout,

except

in

(b)

JPL-00

kou+o

is used

and

in (d)

JPL-00

kuo•+o.•

is used.

The

data

in (a,

b) ar6

restricted

by

NO

> 50

pptv.

Acfdihonal

constraints

are

that

ruo•+o,

is

less

than

10%

of

the

total

inter-conversion

rate

of

HO}-->"OH

and

T = 205+10K.

The

data

in (c, d) are

restricted

as in

Figure

lb.

Additional

constraints

are

that

rHO2+C•

o and

rno2+B•O

each

contribute

less

than

10%

to

the

total

inter-conversion

rate

of

HO2-->OH

and

that

T = 225+10K.

The

temperature

in

each

panel

was

chosen

to

span

the

widest

range

of

the

rate

being

tested.

Data

are

averaged

at

l min.

intervals

(small

dots)

and

into

10

bins

of

the

x-coordinate

(large

dots).

The

lines

are

linear

fits

to

the

1 min.

data.

The

dashed

lines

represent

the

uncertainties

determined

from

all

of

the

measurement

uncertainties

as

well

as

the

weighted

uncertainties

of

kuo

+c•o

and

kuo•+B•o.

The

upper

trace

in

Figure

2

(d) shows

the fractional

error

widen

both

JPL-00

kou+o

3 and

JPL-

00 kuo2+o3

are

used.

[OH]

kHo•

+•vo [ NO

]

(2)

When

XOH+O

3 '- 0,

the

ratio

is essentially

determined

by

Rou+co

and

RHO2+NO

SO

that

the

error

(Ex=o)

can

be

ascribed

to

kou+co[CO]/kuo2+No[NO

]. Note

that

the

uncertainties

in

the

measurements

(CO,

NO,

and

03)

are

included

in

the

total

measurement

uncertainty,

which

is dominated

by

the

uncertainty

in HO2/OH.

When

Xou+o

3 = 1, the

ratio

is described

by Ron+o

3

and

RHO2+NO

and

the

error

(Ex=l)

is

from

koH+o310311kHo2+No[NO

]. The

slope

is equal

to the

difference

in

the

errors

of

koH+co[CO]

and

koH+o3103].

In

panel

(a)

E =

0.1 l(1)--0.06(3)XoH+O

3, where

the

values

in parenthesis

are

the

statistical

uncertainties

of

the

least

significant

figures

determined

from

the

regression.

With

the

uncertainty

from

the

in

situ

measurements

included,

the

errors

determined

using

the

JPL-97

evaluation

are:

Ex=o

= 0.11

+0.24,

and

Ex_-•

= 0.05+0.24.

In

panel

(b)

E = 0.11(1)+0.21(3)Xo•+o3.

The

errors

using

the

JPL-00

evaluation

for

kon+o

3 are:

Ex=o

= 0.11+0.24,

and

Ex=•

=

0.32+0.24.

The

absolute

difference

between

the

slopes

in

each

panel

(0.27+0.03)

is equal

to

the

difference

in

the

rate

constants

of koH+O

3 at 205

K (JPL-00

is 26%

higher

than

JPL-97).

In

panels

(c,

d)

the

data

are

restricted

so

that

OH+CO

contributes

less

than

5%

of

the

total

OH-->HO2

rate

and

HO2+C10

and

HO2+BrO

contribute

less

than

10%

of

the

total

HO2-->OH

rate.

The

weighted

uncertainties

from

these

terms

are

included

with

the

measurement

uncertainty.

With

these

restrictions

the

ratio

is determined

primarily

by:

k...+...[G]

(3)

When

JHO2+O3

= 0, the

ratio

is controlled

by RoH+o

3 and

RHO2+NO

and

the

error

(Ex=.)

is from

kOH+O310311kHo2+No[NO

]. When

JHO2+O3

= l, the

ratio

is described

by ROH+O

3 and

RHO2+O3

and

the

error

(Ex=•)

is from

koH+o3103]/kH02+03103].

The

inverse

slope

is

equal

to the

difference

in the

error

in both

kho2+o3103]

and

kHO2+No[NO

] (inverse

because

these

are

denominator

terms).

In

panel

(c) E = 0.11

(0)+0.06(1)__Xn02+o•.

The

errors

using

the

JPL-

+0.29

97

rate

constants

are:

Ex=o

= 0.1

1+0.21,

and

EX=l

= 0.17_0.24.

•n

panel

(d)

E = 0.11

(0)

+ 0.26(1

)XHo2+o3.

The

errors

using

the

JPL-

00

rate

constant

for

kno

+o

are:

Ex=o

= 0.11+0.21

and

Ex=l

=

+0.29

2

3

-

'

-

0.37-0.24'

The

absolute

difference

between

the

slopes

in panel

(c)

and

panel

(d)

(0.20_+0.01

) is equal

to the

difference

in the

rates

of kno•+o•

at 225

K (JPL-97

is 22%

higher

than

JPL-00).

When

JPL-O•

k•3H+O

3 is included

in panel

d (upper

trace),

E is higher

by

0.23,

and

the

corresponding

errors

increase

to:

Ex=o

= 0.34+0.21,

+0

29

and

Ex=•

= 0.66-0124'

For

this

case,

when

XH02+03

= 1 and

the

ratio

is controlled

solely

by

ROH+O3/RH02+03,

the

error

is 66%,

42%

beyond

the

combined

uncertainties

of

the

measurements

and

remaining

rate

constants

(+24%:

dashed

line

in

all

panels

of

Figure

2).

The

results

of Figure

2 are

summarized

in Table

3.

Discussion

and

conclusions

The

good

agreement

observed

in Figure

2(a)

and

2(c)

could

be

fortuitous.

The

NO,

CO,

and

03

measurements

could

have

offsetting

errors'

kon+co,

kon+o

3, and

kHO2+NO

could

all be high

by

the

same

amount;

or

some

combination

of

offsetting

errors

in

both

the

in

situ

measurements

and

rate

constants

could

be

present.

However,

the

ratio

of

rate

constants

in

Figure

2(a)

and

(c)

suggest

that

if

any

single

rate

constant

in

the

JPL-97

recommendation

is

accurate

then

the

other

rate

constants

are

accurate

to

within

the

uncertainties

of

the

in

situ

measurements.

When

the

JPL-00

recommendation

for

both

koH+o

3 and

kHo2+o3

are

used

(upper

trace

in Figure

2(d)),

and

when

Xno2+

% = 1, the

calculated

[HO2]/[OH]

disagrees

with

in

situ

observations

by

66%,

well

beyond

the

uncertainties

in

the

observations.

The

66%

Table

3.

Errors

in

the

rate

constants

determined

from

regressions

of

the

fractional

error

of

calculated

[HO2]/[OH]

versus

Xon+o

3 and

XHO2+O3

in Figure

2. The

uncertainties

include

the

measurement

uncertainty

of

[OH]/[HO2]

and

the

weighted

uncertainties

from

the

unregressed

rates

ie.

kH02+c•o[C10],

kHo2+Bro[BrO],

and

kOH+CH4[CH4].

Uncertainties

of

the

slope

errors,

(dE/dX),

eg.,

the

difference

in

the

errors

of

koH+03103]

and

kon+co[CO],

do not

include

the

uncertainty

of

the

measured

[HO21/[OH].

X=

roH+o3/rtota

I

X=

rHo2+o3/rtotal

JPL-97

JPL-00

JPL-97

JPL-00

dE//dX

-0.06+0.12

0.21+0.12

0.06+0.12

0.26+0.12

Ex=o

0.11+0.24

0.11+0.21

0.34+0.21

+0.29

Ex=•

0.05+0.24

0.32+_0.24

0.17

-0.24

0.66

-0.24