The El Chichon volcanic cloud in the stratosphere: lidar observation at Fukuoka and numerical simulation

The El Chichon volcanic cloud in the stratosphere: lidar observation at Fukuoka and numerical simulation

The El Chichon volcanic cloud in the stratosphere : lidar observation at Fukuoka and numerical simulation TARASHI Department of Electrical MOTOWO De...

2MB Sizes 0 Downloads 2 Views

The El Chichon volcanic cloud in the stratosphere : lidar observation at Fukuoka and numerical simulation TARASHI Department

of Electrical

MOTOWO Department

Engineering,

FUJIWARA

SHIBATA Kyushu University.

Fukuoka

812, Japan

and MOT~KAZU HIRONO

of Physics, Kyushu

University,

(Received injindfirm

Fukuoka

812, Japan

4 June 1984)

Abstract-The stratospheric volcanic cloud from the eruptionof El Chichon, Mexico, on 4 April 1982 was observed routinely by a Nd : YAG lidar system from 18April 1982at Kyushu University, Fukuoka,Japan. The observed layers of the cloud above 20 km were in the easterly wind region and those below 20 km were in the westerly region. The main part of the cloud mass was in the upper layer. This upper layer broadened slowly until September 1982, then broadened rapidly and merged with the lower layer as the easterly wind changed IO the westerly wind. The vertical eddy diffusion coefficient estimated from the broadening of the upper layer was much smaller than the value usually used in the one-dimensional model calculation of chemical components until September and subsequently remained at about the same value. The increase of the integrated backscattering coefficient (IBC) was about two orders of magnitude larger than the largest increase after volcanic injections for the last IOyears. The IBC reached a maximum value on 3 May and gradually decreased until August 1982, then re-increased until December 1982. The IBC between December 1982 and February 1983 was about the same value as in May 1982. Using theonedimensional stratospheric sulfate aerosol model simulations it was concluded that to explain the broadening of the upper layer an eddy diffusion coefficient of about IO’ cm2s-’ would be needed in the easterly wind region in summer. II was also concluded that the IBC re-increase was caused after advective horizontal transport from lower to higher latitudes hy chemical reactions within the upper layer without meridional diffusion during summer and that the transport was controlled by nucleation, which gives rise to small particles, a decreasing settling velocity of the volcanic cloud and then the cloud being less affected by horizontal transport.

1. INTRODUCTION

thevolcaniccloud by usingthe lidar observation results after the eruption of volcan de Fuego. Since the 18th century volcanic materials in the The eruption of El Chichon, Mexico (17.3’N, stratosphere have attracted the attention of scientists 93.O”W), on 4 April 1982 injected an enormous amount because of their e&cts on climate (LAMB, 1970). Since of volcanic materials into the stratosphere. The then such materials have also been considered increase of the lidar backscattering ratio. which was important as indicators of stratospheric transport. observed by the lidar at Kyushu University. was about For the past twenty years direct measurements of the two orders of magnitude larger than the previous stratospheric aerosol layer have been made, and more increases for the past 10 years (HIKOYO and SHIBATA, precise discussions about these properties of volcanic 1983). materials were made after the eruptions of Agung There were two notable phenomena detected try lidar (MOSSOP.1964 ; GRAMS and FIOCCO, 1967 ; etc.), volcan observation of the El Chichon volcanic cloud. de Fuego ( FUJIWARA et al., 1975; HOFMANN and ROSEN, (1) The main increased layer was formed in the 1977; etc), Sierra Negra (FUJIWARAet al., 1982) and, summer easterly wind region over Fukuoka (33.5’N, especially, Mt. St. Helens (POLLACK.1981; HOFMANN 130.4’E). The layer had two high gradient top and and RCJSES, 1982). bottom boundaries. The layer width slowly broadened Lidar observation of the stratospheric aerosol layer until September 1982, when the easterly wind changed was begun about one year after the eruption of Agung to the westerly wind. A similar phenomenon was (F~occo and GRAMS, 1964). Lidar observation of the observed after the eruption of Mt. St. Helens in May stratospheric aerosol layer at Kyushu University was 1980 (HIRONO cr al.. 1984). begun just before the eruption of volcan de Fuego (2) The stratospheric integrated backscattering (HIROSO et al.. 1972). H~~osoer ul. (1977) and ITABEr/ coefficient (IBC) reached a peak value on 3 May 1987.1 t 01. (1980) discussed the gas to particle conversion then gradually decreased until August. but re-increased process oft he volcanic sulfur gases and the transport of significantly from September to December 19x2. After I121

1122

TAKASHISHIBATA.M~TOWO FUJIWARAand MOTOKA’ZUHIRONO

of aerosols (/i’,) is calculated as fl, = fiw(R - I), and is nearly proportional to the mass concentration of aerosols. As stated by HIROKO and SHIBATA(1983) extinction oflaser light in a dense aerosol layer must be taken into account in order to obtain the exact values of PA and also their integrated values. However. since at the fundamental wavelength the elfects of extinction for fi,, and its integrated value are less than about IO?,, the extinction of laser light by aerosols is neglected when @, is calculated in this paper. This neglection does not affect the results and the discussion below.

January 1983 IBC decreased slowly. This remarkable re-increase had not been seen in past increases which had been observed by the lidar at Kyushu University. There is a sulfur aerosol layer in the lower stratosphere even during volcanically quiet periods. The formation mechanism of this background stratospheric sulfate aerosol layer has recently been well explained utilizing the sulfur cycle of the troposphere and the stratosphere. Tropospheric sulfur gases, such as OCS, S02, etc.. are transported into the stratosphere and oxidized to HISO.,. By some nucleation processes fine H,SO-H,O solution droplet particles are formed. These fine particles grow into what are called Mie particles with the condensation of H,SO, (and HSO,) and through coagulation and then form the Junge layer in the lower stratosphere. These particles are finally transported back to the troposphere by sedimentation and atmospheric motion and are washed out there. A comprehensive review of observations on and theories about the stratospheric aerosol layer has recently been made by TURCO er nl. (1982). Volcanic eruptions inject large amounts of SO1 directly into the stratosphere. This injection can be thought of as a disturbance of the above mentioned sulfur cycle The vertical one-dimensional fully interactive model of Tu~co cr nl. (1979) (Ames model) simulates very well the stratospheric aerosol layer and sulfuric gas

400.

components.

greatly.

In the following

Section 2 the results of the lidar

3, the vertical eddy diffusivity will be determined from phenomenon (1) and thecasue ofthe re-increase (2) will also be suggested, using a modified Ames model. observation

will be examined.

Next. in Section

2.2. Imreasc

after the eruption of El Chichorl

Some increases in the stratospherical aerosol scattering ratio have been observed with the Nd : YAG lidar system since December 1979 (FUJWARA et al., 1982; HIROXO et al., 1981, 1983). The eruption of El Chichon (Mexico, 17.3”N, 93.2”W) on 4 April 1982 injected a large amount of volcanic materials into the stratosphere and an enormous increase in the scattering ratio was observed over Fukuoka (HIROSO and SHIBATA, 1983). The first increase was observed on 18 April at about 16 km and 26 km in altitude. At the previous observation on 15 April there were no such increased layers. On 3 May the peak of the scattering ratio was at 24 km and the ratio had increased to about The

peak

scattering

gradually

December

ratio,

decreased.

1982 and January

The

while

fluctuating

ratio

throughout

1983 varied between 20

and 40 (Fig. 1).

Until September 1982 thelayerdisplayed a two layer structure, which was divided by a clean region at an altitude of about 20 km. The upper layer was in the easterlywindregion,which

wasalsoaboveabout

2Okm

in summer. The peak of the upper layer was at about

2. OBSERvATION

24 km. The lower layer, whose peak was at about 18 km. was in the tvesterly region below 20 km. Figure 2 shows

2.1. Lidar

heights ofthe peaks in scattering ratio for the upper and

Since October 1979 Nd : Y AG Lidar has been used to observe the stratospheric aerosol layer. The characteristics of the lidar system and the methods of data analysis have been described by SHIRATA CI ul. (1980). The results at the fundamental wavelength only (1.06 pm) are discussed in this paper. The lidar backscattering ratio R is defined as R

=

PA-+/L II

lower layers and of the minimum

I’M

where p, and fl,u are the backscattering coefficient of aerosols and of atmospheric molecules respectively. R- 1 = BA,‘/i’nris roughly proportional to the mass mixing ratio of aerosols. The backscattering coefficient

and

the zero zonal wind velocity over Fukuoka.

The wind is

easterly above this height and westerly below. The upper layer has top and bottom boundaries with high gradients at about 27 km and 21 km. respectively. As will be shown in the following section, the width of the

(2.1)

between April

August 1982. Figure 3 shows the variations of height in

upper

September

layers 1982.

broadened The

upper

very layer

slowly also

until

had

fine

structures and the ratio of the minimum

IO maximum

values of the scattering

of these were

ratio-minus-one

sometimes more than ten. After

September

the two layers gradually

merged

into one layer and at the same time the wind system

The El Chichon volcanic cloud m the stratosphere changed from the summer

type IO the winter type, i.e. the summer easterly changed to the westerly throughout the stratosphere. Up until November the layer became broad and smooth (Fig. I b). The time variation in the integrated backscattering coefficient (IBC hereafter) between altitudes of 13.5 km and 28.5 km from April 1982 to April 1983 is shown in Fig. 4. The IBC reached a peak value on 3 May and greatly fluctuated until July, although it tended to

1123

decrease during this time. After August the IBC reincreased, and reached the same level as in May during November. Then the IBC gradually decreased from December 1982 to April 1983,Theperiod when the IBC re-increased just overlaps the period when the IWO layers merged. 2.2. I. Diflitsion O/ rite upper /a_wr. FLJILVAR.~ TVol. (1982) examined the source of the increased aerosol layer in December 1979 using diffusion theor!. The)

23

i

;,,

;

(a)

;i

Scattermg ratlo

i t i i

7

Scattemg

ratio

k’i_e I (a, The scarwring ratlo profiles a~ 1.06 ,rm. I5 April-14 June 1982. ch) The scaltsring rtiilo prt~lilr* .:I 1.06pn. 3 July-3 December 19X2.

,cfmri,llw,; (!I,‘I-I

TARASHI S’HIBATA.MOT~WO FUJIWAKA and MOTWAZU

1124

JOtl 31

i 30

HIRONO

-?

t

-

25 -

2

-

5

% 3 a-

3 a

I5

‘0 -

L

IO

I

.c

13

Scottertng

(cf

I

Fig. t. continued. (c) The scattering ratio profiles at 1.06 pm, 4 January-g

?% -=LI(r) i:r 2;

1

I

/

(2.2)

.

g (.I

where z is the number mixing ratio of the aerosol particles. !I(-_)is the vertical diffusion coeflicient, I is

{

Greotej:

mox:mo

0

Smclie~ Min:~c

mox~mo

+ /

May f983.

If the eflzct of particle sedimentation is neglected the change in the vertical profile of the layer can be expressed by the diffusion equation

calculated rhe time of the volcanic eruption after measuring the layer width and utilizing an assumed diffusion coefficient value near the layer altitude. Knowing that the volcanic materials were injected into the stratosphere by the El Chichon event, it was now possible to roughly estimate the diffusion coefficient around the increased upper layer using the diffusion equation and the layer width.

3c

IQ

:ctio

1

Fig. 2,Thehei~htsofthepealisin~~atteringratiofortheupprrandl~twer ISISL to Februar! i383.

on0

half - width of

R

J---r--?----

1a)ersandofthrminimum

from April

The El Chichon volcanic cloud

in the stratosphere

Fig. 3. The variation in height of the zero zonai

time and z is height. As described in Section 2.1. since R - 1 is roughly proportional to the mass mixing ratio of aerosols and the mass mixing ratio is also roughly proportional to the number mixing ratio, R - f can be used instead of ): in equation (2.2). Thus

wind velocity over Fukuoka.

the layer as o,, then rro = 3.33(Dt,)’ ‘. The half width

G

at time t is given by 02 = o;+ll.IDr.

(2.3)

Ifit isassumed that D(z) isconstant around the layer and that the initial distribution of the IaTer is a Gaussian distribution then it follows that R-l

= t(nD(~+f~))-~‘~

exp

where K is a constant and z. is the peak altitude of the layer. Here to is determined so that 3.33 (Dt,)’ ’ is the halfwidth of the initial layer. Putting the half width of

II25

(2.5)

Then cr*is a linear function oft. Now. D can be derived from thehalfwidthoftheupperlayer. FigureSshowso’ for each observed layer. After October Q* rapidly increased in value. The regression lines are also shown in the figure. The lines are determined by the least-square method for the period from April to September and from October to March. The regression coefficients, which correspond to I 1.1 D in equation (2.5), before September and after October are very different. while the derived D value before September is 3.1 x i02 cm2 s-’ and after October is

I-- ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ i ’ ’ ’ ’ ’ ’ : 135-285

km

Fig. 4. The observed IBC between 133 and 28.5 km altitude from April 1982 IO July lYR3

Fig. 5. The observed variation

of ~5 from April 1982 IO March

1983 and the regression

2.3 x IO3 cm’ se

’ . This value of 3.1 x IO2 cm’ s-’ be-

which

tween altitudes

of approximately

increases.

The

before September is nearly one or two orders of magni-

November

1980 is not clear in Fig. 6, as it overlaps the

tude smaller

fluctuations

after the eruption of Mt. St. Helens in Ma>

than the vertical

which are deriled

20 km and 30 km eddy dilTusion values

to explain the observed distribution

of chemical components

or radioactive

MAWE and HUSTEN. After

the eruption

eruptions IBC

er al., 1976:

the IBC

had not returned

of Mt. St. Helens

in May

1980 a

eruption

of El Chichon

decreased

almost

variation

after April

18km

layer had not merged into the lower layer around

1980. when the wind changed to the

type, i.e. the wind direction

westerly throughout ROSES, 1982;

changed

the stratosphere

to the

(HOFSIASS

and

HIROSO er al. 1984).

the easterly region would be smaller than IO’ cm’ s

‘.

taking into account the effect of wind shear \vhich thins LEFRERE et al. (198 1) also derived

similar values from the behavior of the same layer. The layer in the easterly region was dispersed as the wind system changed into the winter type in both the (HIROW

(‘r al.. 1984) and El Chichon

in

1982 value

before the

the increased IBC values had

uniformI!,

if compared

with

the

I983 (Figs. 4 and 6).

As shown in Fig. 4, the IBC reached it’s maximum value in the beginning fluctuating

of May

largely. gradually

1982. In August IBCvaluesin

estimated that the vertical eddy diffusion coefficient in

Xlt. St. Helen\

to the background

1982 and. although

decreased until August

1982 the IBC

reached it’s minimum

value and then re-increased until December

Usingthedispersionofthislayer,HtRo~orra/.(l984)

the width of the layer.

of these

occurred

before the increase caused b) the El Chichon eruption

where the easterlq wind region forms in summer. The

winter

which

began. In spite of these large fluctuations.

I98 1).

thin layer was formed around 21 km, which is exactly

until October

were the sources

increase

1980. After the increase which began in January

debris in the

stratosphere in this height range(JoHWroK

volcanic

December 1981and Januar!

the same magnitude February

as those in May

I982. The

19S3wereof 1982. Since

1983 the IBC has been slowly decreasing.

Figure 7 shows the time variations the 19.5 28.5 km and I35

of the IBC from

19.5 km altitude

regions.

These regions are above and below the clean level mentioned

above. Since the IBC ofthe upper region is

much larger than that of the lower region. The time variation

for the IBC of the upper region is about the

events. Though rhe la)ers in the easterly region did not

satucastl~atoftheIBCinF1g.4fr~~mMa~

merge into the louver layer for five months from April to

IY83. The IBC‘ ofthc lower region Increased uniform]!

toDrcembcr

.August, thesame layers merged or smoothed out within

from August lYS1 to Fehruar)

a month after the wind system changed to the winter

that ofthe upper region from Januar)

19X3. The IBC of

t!‘pe In October.

the

decreased

The D value 2.3 x IO3 cm2 s coincides with the vertical usually

’ after October roughly

eddy diffusion coefficient

used in one-dimensional

modeling

of the

chemical components in this altitude range. _._._. ’ 7 3 R+i~l(.r(~~.seo(r/lr /BC. Figure 6sho\vs the IBC

upper

November

region

since

of the IBC

the effects of sedimentation

atmospherictranspnrt Figure

uniformly

1982. This interchange

demonstrates the lower

has

19X.7 and is larger than

ofthe particles

values

and

the

from the upper to

region. 8 shwvs

the variations

in the peah altitude of

between altitudes of 13.5 km and 28.5 km from October

the scattering ratio untJl January

I979 to May lYY3. There were six increases in the IBC.

lY82

inclusive of the El Chichon event. Figure 6 also show?

velocitv ofahout

the peak

altitude

1983 .\fter

uniforml!

I km month

‘.This

August

descended

31 ;I

period coincldcb

The El Chichon volcanic cloud in the stratosphere

Integrated 135-285

10-5’

,

1979

back5catterlng

1127

coefflclent

km

I



8

,

,

1981

1980



,

1



,

1

1982

Fig. 6. The observed IBC between 13.5 and 28.5 km altitude from October 1979 to August 1983

with the period when the IBC re-increased. As mentioned in Section 2.1, the scattering ratio peak nearly coincides with the mixing ratio peak. Since vertical diffusion cannot influence the peak altitude of the mixing ratio, this descent of the layer is caused by sedimentaion. The velocity ofdescent of 1 km month-’ or 0.04 cm s-r coincides with the sedimentation velocity of the particles of about 1 pm radius in the 2025 km altitude region.

1

t I

1982

1983

Fig. 7. The observed time variations of IBC between the 13.5 19.5 km and 19.5-28.5 km altitude regions from April 1982 to March 1983.

Two explanations are possible to account for the reincrease of the IBC. (a) Similar to the situation following the eruption of Agung in 1963, the main part of the volcanic cloud was trapped in the tropical region. The cloud was mainly transported to higher latitudes in winter by general circulation (DYERand HICKS, 1968). In the northern hemisphere, the winter westerly wind in the stratosphere is replaced by a summer easterly wind in early spring (MATSUKOand SHIMAZAKI,1981). El Chichon erupted on 4 April 1982 when the wind system had not completely changed into the summer type and the zonal wind stream was highly meandered. This meandering brought the cloud to the latitude of Fukuoka and caused the IBC to reach a maximum valuein May 1982. In themid-summer this meandering was so small that the main part of the cloud was well trapped in the tropical region and only a small part of the cloud was at the latitude of Fukuoka or higher. In September and October the summer easterly wind was again replaced by the winter westerly wind. Then the above mentioned general circulation occurred and the cloud was again transported to Fukuoka and higher latitudes. Thesechanges in horizontal transport caused the observed variation of the IBC. (b) An explosive volcanic eruption injects ash particles and gas components directly into the stratosphere. The large ash particles are rapidly removed from the stratosphere because of their relatively large settling velocities. The gas components, especiallySO,,areoxidizedtoH,SO,andconverted to

TAKASHISHIBATA,MOTOWO

1128

I

I

I

I

FUJIWARAand MOTCXAZUHIRONO

I

r

I

I

I

I

I

25 z 2

24

4 Y 2

23

“0 8 d

22

I A

I

I

I M

J

I A

.I

0

I

I

I

I

s

r:

0

I J

-I F

1983

I982

Fig. 8. The observed varation in peak altitude of the scattering ratio at 1.06 m. sulfuric

acid droplet particles by the processes of nucleation and condensation. The most important reaction which oxidizes SO2 is SO1 +OH + M + HSO, + M.

(R8)

and SO, +H,O

+ HISO,.

t s=exp (-5 >

(R7)

Then USOs is converted to HISO, by an unknown reaction. The assumed reactions are often field to be HSO, +OH + SO, + H,O,

where r, and zz are the oxidation time of SO, and the residence time of the materials in the stratosphere, respectively. Equation (2.6) is solved as

(RIO)

(2.8)

with !=l+L. 7 11

f2

Then, the solution of equation (2.7) is

The second reactions (R8, RlO) are assumed to be n = exp( - i)-exp((2.91 much faster than the first reaction (R7). From the reaction rate of reaction R7 the oxidation time of SO2 in the lower stratosphere was estimated to ~ereitisassum~thatsisl.OandnisOatt =O.Figureg be about 3 x 106-8 x 10’ s (MOORTGAT and JUNGE, shows the solution of equation (2.9) for some 1977). The residence time of the materials in the combinations of ri and TV.In Fig. 9 it is shown that the stratosphere was estimated to be - 1-2 years. If it is total amount of aerosols(n) in the stratosphere reaches assumed that SOz is converted to aerosol dropiets as its maximum value 5-8 months after the injection of soonas it isoxidized.The variations in the total amount SO, into the stratosphere and that the larger the ‘cl of SO, (s) and df sulfuric acid droplets (n) in the value the longer it takes for n to reach maximum vaiue. stratosphere can be expressed as HIRONO~~al. (1977) and CADLE(1980) pointed out that enormous amounts of SO2 consume OH during d.s ss -=-----I (2.6) reactions (R7) and (R8). It takes a longer time to oxidize J6 Tl T2 SO, in this case than when OH is not consumed. Since dn s n the tI values of the cases in Fig. 9 are within the values -=---t (2.7) which are estimated for normal OH concentration, if dr x1 z2

l).

The El Chichon volcanic cloud in the stratosphere

Days

Fig. 9. The solutions of equation (2.9). 7s is assumed lo be 1 year, and z, is assumed to be (a) 10’ s,(b) 10’ s and (c) 10s s.

OH consumption has occurred it also takes longer for n to reach maximum value than in case (c) in Fig. 9. Following the discussion above, it can be said that the IBCmaximum value in May 1982and the following decrease indicates the existence of larger ash particles and their fast removal from the stratosphere. It can also be said that there-increase in autumn 1982, followed by the second maximum value in December 1982 and January 1983, indicates fairly slow oxidation of SO, and its conversion to sulfuric acid droplets. The airborne Iidar observation which was conducted by NASA Langley Research Center revealed latitudinal variation of the stratospheric aerosol vertical profile. Until October 1982 the densest part of the volcanic cloud was trapped in the latitude region between the southern boundaries 6-10”s and the northern ones 30-37”N. The observation in early November 1982 showed that the northern boundary was at about 30”N and was at about the same region in July 1982 (SWIXS~NIAN INST~TWION, 1982). These results from the airborne lidar observations indicate that the volcanic cloud was trapped in the lower latitudes in the summer season and supports explanation (a) as a major strategy, but explanation (b) may still be ofminor importance. KRUEGER (1983) estimated the total amount of SO, injected into the stratosphere by the eruption of El ultraviolet light absorption Chichon through measured by the Nimbus 7 total ozone mapping spectrometer. The total mass was estimated to be 3.3 megatons on 6 April 1982. Concerning the observed profile at Fukuoka, the lidar observation at Mauna Loa, Hawaii, (f9.S”N, 155.6”W) showed that the main aerosol layer was between 20-22 km and 27-30 km in altitude(CouLsoN,

1129

1983~.~eiatitndeoftheo~rvatoryat Hawaiiisin the densest latitudes mentioned above. From the latitudinal distribution of the volcanic cloud and the vertical profile described above, if we assume that 3.3 megatons of SO2 were distributed uniformly between O”N and 3O”N and between 21 km and 27 km in altitude, then the concentration of SO, in these altitudes would be about 3.8 x 10” cme3, or about 40 ppbv at an altitude of 24 km. CADLE (1980) estimated that SO, depletes OH at 18 km altitudes when its mixing ratio exceeds 4 ppbm. With the same estimation it can be shown that the mixing ratio of SO1 at which the OH depletion begins is about 8 ppbm at the central altitude of the main layer, i.e. 24 km, since the rate of formation of OH at 24 km altitudeisabout twiceaslargeas that at 18 kmaltitude. This estimated SO1 mixing ratio of 40 ppbv at 24 km is sufficiently above 8 ppbm and depletion of OH would occur. Then r t in equation (2.6) would be larger than the usual value 3 x 106-8 x 10’s and it would take longer for the sulfuric acid aerosol content in the stratosphere to reach maximum value than when OH depletion does not occur. This supports explanation (b), which could be seen if the dynamic factor is eIiminated.

3. MODEL CALCULATION To examine the observed results in Section 2 a vertical one-dimensionai model calculation of volcanically injected stratospheric SO, and the evolution of aerosol particles will be presented. The existance of ash particles in the ejecta is not taken into consideration for the present since they would fall into the troposphere at an early period. Horizontal transport cannot be treated by a vertical one-dimensional model. However, since the injected volcanic cloud was trapped in the Iatitudinal region between about 5”s and 35”N until November 1982, or for several months after the eruption, we can approximately neglect the effect ofhorizontal transport in these latitudes during this period. The effects of horizontal transport are discussed in the next section. The model is mostly based on the Ames model (T~~coet ai., 1979 ;TOONet al., 1979). Some parameters are changed from the Ames model due to the results of recent observations and theoretical calculations. In the one-dimensional model the concentrations of gaseous components and aerosol particles for each radius werecalculated. ~econtiRuity equations which describe the time variations of the concentrations are solved by a finite difference method. The computed altitude range is between 10 km and 50 km with a grid size of 1 km. The grid size of the particle radius is taken

TAKASHI SHIBATA, Morowo

1130

FUJIWARA and

as the volume of the particles differentiated by factors of 2. Vertical transport in the model is expressed as eddy diffusion. As shown in Section 2.2.1, theestimated eddy diffusion coefficient is very small in the summer easterly wind region, but has about the same value as usually used in one-dimensional models for the winter westerly wind region. In winter (October-March) the eddy diffusion coefficient value of MAssEand HUNTEN(198 1, Fig. lo), which was obtained from the distributions of nitrous oxide, methane, ozone and carbon 14, is used. Massie and Hunten’s value is within the values determined by other investigators. The vertical profile of eddy diffusion cannot be deduced only from the discussion in Section 2.2.1. The coefficient profile (S) in Fig. 10 is assumed for the summer easterly region. The value between 20 km and 30 km is smaller than the value deduced in Section 2.2.1, but coincides with the value estimated by HIRONOet al. (1984). This value is taken as the extreme case. 3.1. Chemical and physical processes in the model 3.1.1. Chemistry of the gaseous component. The calculated gaseous components are the sulfur constituents OCS, S, SO, SOZ, SO,, HSO, and H,SO, and the OH radical. The chemical reactions calculated

Diffuston Fig.

JC JZ

OCS+hu-s+co HzSO, + ho - SO, + products

J3

SO,+hc~-so+O

Rl

ocs+o-so+co

R2 R3 R4 RS R6 Rl R8 R9 RlO RI1 R12

RI3 R14 Rl5 R16

s+o, -so+0 so+o, - so,+0 so+o, - so,+o, SO +NO, -SO,+NO SO,+O+M-so,+A4 SO,+OH+M-HSO,+M HSO,+OH-SO,+H,O HSO, - aerosols (-H2S04) SO, + Hz0 - HzSO, HzSO, - aerosols SO, - washout

HSO,

-washout

H zS0. - washout O(‘D)+H,O - 20H OH +X - products

(cm’ s-‘I

10. The assumed vertical eddy diffusion coefficient profiles in winter (MH) and in summer (S).

reactions

Rate coefficient 6.9x 2.8 x 8.0x 1.1 x 3.1 x 2.7x 3.0x

coefftcient

here are listed in Table 1. For the other constituents, 0(3P), O(‘D), Os, NO2 and H,O, the fixed vertical profiles are assumed as in Fig. 11. The profiles of 0 and O(‘D) are the approximate mean values from Fig. 1 of SHIMAZAKIand WHITTEN(1976) and those of O3 and NO, are from Fig.4ofTuncoet a/.(1979).Theprofileof H,O is the linearly approximated values from Fig. 2 of

Table 1. The chemical Reaction

MOTOKAZU HIRONO

10m9at 15km lo-’ at 30 km IO-‘at 15 km lo-’ at 30 km LO-” at 15 km IO-‘jat 30km IO-” exp(-2270/T)

2.2x lo-‘2 3.0x IO-l3 exp(-2800/T) 2.5x IO-‘* exp(-1050/T) 1.5 x lo-” 3.4~ lO-)z exp(-1130/T) 8.2 x 10-‘3/(7.0x lO”+[M]) 1.0x lo-” see text 9.1 x lo-t3 see text 13-Z 3.8 x lO-6zCl3km 13 0 z> 13km 13-z 2.3 x 10-5z 13km 3.5x IO-‘0 see text

Reference

YAMAMURAet a[.

(1983);see text

TURC~ et al. (1979) WESTENBERGand DE HAAS (1969) DAVIS et al. (1972) SCHOFIELD ( 1973) SCHO~ELD (1973) SCHOFIELD (1973) DAVIS ( 1974) MO~RTGAT and JUNGE (1977) TURCO et al. (1979) CASTLEMANet al. (1975) TURCO er al. ( 1979)

TURCII er al. (1979) TURCCIet ol. ( 1979) HAM~J~~Nand GARVIN (1975)

The

El Chichon

volcanic cloud in the stratosphere

Concentrotcon

Fig. 11.The profiles

of O(O(‘P)),

HARRIES(1976). Atmospheric temperature and pressure are taken from the values in mid-latitude spring/fall from the U.S. Standard Atmosphere Supplements (U.S. GOVERNMENTPRINTING OFFICE, 1966). The continuity equation which described the time variation in the concentration of chemical constituent A is expressed as 44 dr

= P,---L,[A,+

[email protected](z)m$(~)) -R,,CAl-CC&l,

(3.1)

where [A] is the concentration of constituent A, P, is thechemical production rate ofcomponent A, L,[A] is the loss rate, D(z) is the eddy diffusion coefficient, m is the density of the atmosphere, R,,[A] is the wash out rate and G[A] is the loss rate due to the attachment to the surface of the aerosol particles or by nucleation. OH is so active that it interacts with a number of atmospheric components. such as HO,, H,O,, H, H,, HNO,, NOz, NO, HNO,. N, CH4 and OH itself. In reaction R16, X represents all thesecomponents which react with OH, except SOZ and HSOJ. Though all reactions with these chemical components must be treated in the model calculation to simulate accurately the OH variation, here a simplified method is used to examine the influence of increased SO, on OH depletion. In the model reaction R 15 is assumed to be the only source of OH. k16[X] is determined so as to satisfy chemical equilibrium.

2k5CO(‘@l CH,Ol = k,eCXlCOHol

O(‘D),

1131

(cm“) 0,.

NO2 and

H20.

or

k,,CXl =

2k,

s[OC’Wl[Hz01 W-U ’

0.2)

where [OH,,] is the initially given OH radical concentration from Fig. 12. The profile of [OH,] is taken as that coinciding with the approximate mean value of the [OH] profiles in Fig. 7 of SHJMAZAKI and WHITIEN (1976) between altitudes of 10 km and 25 km and with the [OH] profile given in TURCOet al. (1979) above 25 km in altitude. The value below 25 km in SHIMAZAKIand WHI~N (1976) is near to the value estimated by TLRCO et al. ( 198 la). Reaction R 15 is the most important origin of stratospheric OH. If we neglect eddy diffusion, the equation that describes

IC4

.:

.F

r Loncentra:lon

_-

!cm-!.

Fig. 12. The profiles of OH0 and OH

TAKASHI

1132

[OH] variation is approximately

WHI

-

dt

SHIBATA,

state, [OH]

FUJIWARAand MOTOKAZLIHIRONO

as follows

= &CO(‘D)l CH,Ol-k,,CXl

In a stationary

MOTOWO

T_hecomputational expression of the finite difference form ofequation (3.4) is the same as that in the Ames model (TIJRCCIet al., 1979), although some differences from the Ames model are described below.

[OH].

exactly coincides with

WHoI.

3.2. Results of the numerical simulation

Figure 12 also shows the [OH] profile which was calculated from theequation that included thediffusion term and the term of depletion by SO, and HSO,, or

3.2.1. Model simulation for the background state. Formation of H2S0.,. In the period when volcanic activity is low the main source ofstratospheric sulfate is OCSand SO1, which arediffused from the troposphere. OCS is stable in the troposphere. In the stratosphere OCS is dissociated by reaction Jl, then the produced S is oxidized to SO1 by reactions R2, R3, R4 and RS. SOz is transformed to HSO, mainly by reaction R7, i.e.

dCOH1

-

=

2k,sCo(‘~)l C%Ol -MXl

dt+

PHI

-$(W;(~))

- k,[SOJ

[M] [OH] - k,[HSO,]

[OH].

(3.3)

[OH] nearly coincides with [OH,] in most of the range between 10 km and 40 km in altitude. This coincidence indicates that k16[X] can be used as the depletion rate for OH in the model whose basic equation includes the diffusion effect and that in the usual state OH depletion by SO2 and HSO, is negligible. From the above results, continuity equation (3.3) can simulate the stationary state of [OH] or the profile of [OH,] and the variation of [OH] can be approximately deduced when [SO,] is enormously increased by volcanic eruption. Following the suggestion by ARNOLD AND BUHRKE (1983)concemingthephotodissociationrateofH,SO,, the relevant rate for HNO, has been used instead of the rate for HCI, which was used by Tunco et al. (1979). 3.1.2. Sulfate aerosol model. The physical processes involved in this model are nucleation, condensation, water vapor growth, evaporation, coagulation, sedimentation, diffusion and washout. The continuity equation expressing these processes is

&I)

(3.4) where n du is the particle number density in the volume range (0, v + du) at I and altitude z, J is the nucleation rate, K(u,u) is the coagulation constant between particles of volume o and volume u, o, is the sedimentation velocity, D is the eddy diffusion coefficient, m is the density of the atmosphere R,, is the wash out rate in the troposphere.

SO,+OH+M-,HSO,+M.

(R7)

Though the reactions which convert HSOB to H2S0, are not well known, reactions such as R8 and R 10 would produce HISO,, i.e. HSOJfOH-,S0,+H20,

(R8)

and SOP + Hz0 -+ H,SO*.

(RW

This produced H,SO, is converted into H,SO,Hz0 solution droplet aerosols by nucleation and condensation. As will be. seen below, HSO, also condenses on aerosol particles in our model. Background nucleation. HAMILL et al. (1977) estimated that the most probable nucleation process in the lower stratosphere is the heterogeneous heteromolecular nucleation of HrSO, and H,O onto preexisting CN particles which were transported into the stratosphere from the troposphere. The existence of core particles in the stratospheric sulfur aerosol particles supports this process. The impactor measurements indicate that these core particles are composed of ammonium sulfate (BIGG it al., 1970). Following this estimation TURCO et al. (1979) reconstructed the stratospheric aerosol layer in their model by using the parameterized heterogeneous nucleation onto core CN particles which have tropospheric origin. They also determined the evolution ofcore particles in the sulfur aerosol particles. In spite of these estimations, after careful sampling of the stratospheric aerosols HAYES et al. (1980) did not detect ammonium sulfate crystahne particles in the sulfuric acid particles. Ytnzand D~~~~~(1982)pointedout thatat theregion in the stratosphere where the temperature is lower than about -75°C the homogeneous nucleation rate is orders ofmagnitude larger than the rate which had been

The El Chichon

volcanic

for a temperature of about - 55°C (HAMILL 1977).Near theequatorial tropopause the temperature is lower than -75X, therefore the fine sulfuric acid particles are more easily formed by homogeneous nucleation than at other latitudes. It is possible that these fine particles are transported to higher latitudes and altitudes by the Hardley circulation. The estimated homogeneous nucleation rate profile by HAMILLet al. (1982) indicates that the rate is sulliciently larger than the value to maintain the steady-state particle population by in situ particle formation near the tropopause where the temperature is lowest in the stratosphere. In the mode1 calculation of HAMXLL et al. (1982) the stratospheric aerosol concentration was maintained by the diffusion of very small sulfuric acid particles from the upper troposphere, where the nucleation rates are so much larger. HAMILL et al. (1982) concluded from their calculations that several nucleation mechanisms lead to the same genera) properties of the aerosol layer and that particular nucleation mechanism(s) cannot be identified as responsible for the aerosol iayer with the present data base. Based on these results and the discussion above, only a much simplified nucleation process for background nucleation is used in this model in order to reconstruct the background aerosol layer. It was assumed in this model that aerosols are homogeneously nucleated between altitudes of 10 km and 17 km, that nucleation occurs at 0.001 pm and that the nucleation rate would be I x lo- 3 particles cm- 3 s. As described below, under these assumptions the aerosol layer can be well simulated. Additional microphysical processes. Produced fine particles grow into larger particles through the condensation of H2S0,, HSO, and Hz0 and by coagulation. Condensation occurs as the HZ0 vapor pressure of each particles comes into equilibrium with the local partial pressure of environmental H,O. Since the partial pressure of H,O and the temperature arc given (Fig. 1I), the H,SO, weight percentage and the vapor pressure of H,SO, can be determined (U.S. GOVERNMENT PRINTING OFFICE, 1966 ; GMITRO and VERMEULEN, 1964 ; Table 2). Gmitro and Vermeulen’s H,SO.+ vapor pressure is corrected to coincide with the vapor pressure measured by ROEDEL( 1979).The Kelvin effect, or the effect of sphericity on vapor pressure, is also taken into account. It was assumed that HS03 is converted into H,SO, as soon as it attaches to the aerosol droplets, through unknown reactions. As they are growing through condensation and coagulation, the particles are also transported to higher altitudes by eddy diffusion and to lower altitudes by diffusion and sedimentation. The particles which drop estimated

cloud in the stratosphere

1133

Table 2. The assumed temperature, H,SO, of the particles and vapor

Altitude (km)

Temperature WI

10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 SO

223.4 216.7 216.7 216.7 216.7 216.7 218.6 220.6 222.5 224.5 226.5 228.5 233.7 239.3 244.8 250.4 255.9 261.4 266.9 270.7 210.7

weight percentage pressure of H,SO,

Weight 0x of H,SC;,O -.___

of

Vapor pressure of H,SO, (mmH&

71.0 68.6 70.0 71.3 72.5 73.7 75.7 71.5 79.1 80.6 81.8 82.8 85.1 87.0 89.5 93.0 95.0 96.0 98.0 98.0 98.0

7.8E - 14 4.6E-15 9.2E- 15 1.8E- 14 4.4E - 14 6.2E - 14 3.1E- 13 1.3E- 12 S.OE- 12 1.9E- I1 5.8E-11 1.2E - 10 l.SE-9 9.OE - 9 5.8E-8 4.7E-7 9.3E-7 2.3E-6 8.OE-6 RSE-6 9.OE - 6

to lower than 13 km are washed out by the linearly height dependent time constant, R,:. RWAis assumed to be R WA

=

1.2x 10-5(:3.5-2)/13.5s-‘,

where z is the height in km (TURCOet al., 1979). Since tropospheric particles are neglected, in the lower boundary condition it is assumed that the particle concentration is zero at 10 km. Above about 32 km altitude evaporation occurs because the H,SO, vapour pressure of the particles exceeds the partial pressure of H,SO,. At 50 km altitude the particle concentration was fixed as zero, assuming that the sulfuric acid droplets are totally evaporated at this height. Evaporated H,SO, is thought to be the source of gaseous sulfuric acid. 3.2.2. Results for the background stale. To obtain a steady state solution, model simulations covering a period of several years were conducted. In order to examine the effect of HSOJ condensation the steady state solution is obtained when (a) only HISO is condensable to the aerosol particles, and (b) both H,SO, and HSOJ are condensable. using the Massie and Hunten’s eddy diffusion coefficient. Next, to examine the effect of small eddy

TAKASHI SHIBATA, MOTOWO FWW~RA

ii34

Concentrot~on

and Moro~~zr;

(cm-‘)

Fig. 13. The concentration pro&s of H,SO., NSO, and H,SO,+ HSOs in case (a). The measured vaiues of H,SO, + HSO, are within the broken lines (VIGGIANOand ARNOLD, 1983).

R (I 06pm)

Fig. 15.The scattering ratio profiles at [email protected] case(b). The observed values of R are shown by the bars (SHIBATA et al.,

diffusion in summer (Section X2.11, a simulation was made where (c)thetwoeddydiffusioncoefficientprofilesin were used semiannualiy.

Fig. 10

In calculation (c), the solution to (bf was used as the initial condition. (1) Results of Case (a). Figure 13 shows the concentrations ofH,SO,, HSO, and H,SO, + HSO,. The simulated concentration of H,SO,+ HSOr is larger than the concentration of H,SO,+ HSO, observed by VICCIANO~~~ ARNOLD(1983)below 25 km altitude. The profiles of the other components coincide with the profiles in case (b). (2) Results of Case (b). Figure 14 shows the

IO”

IO5

Concentrotlon

198% concentrations of HrSO,, HSO, and HzS04 + HSOs. The calculate-d values are we11 within the observed values and display a better coincidence with the profile observed by VIGGIANO and ARNOLD (1983). The resulting scattering ratio profile at 1.06 pm is shown in Fig. 15. The profile of the particle number density N(r > 0.1Spm) whose radius r is larger than 0.15 m, is shown in Fig. 16(a). Figure 16(b) shows the profile of channel ratio, i.e. the ratio N(r > 0.15 pm)/N(r > 0.25 pm). Figure 17 shows the concentration profiles of SO2 and OCS. Ail these calculated profiles reproduce well the observed profiles of aerosols and sulfur gases.

i8 (0~~‘)

Fig. 14. The concentration profiles of H$O,, HSO, and H,SO, + HSO, in case fb). The measured values of HISO, + HSO, are within the broken lines (VIGGIANO and ARNOLD,

1983).

HIRONO

Portiles

mg-’ err

%o :s’%o

25

Fig. 16. (a) N(r > 0.15 pm) profile and (b) the channel ratio profile in casefb). The observed values are shown by the bars (HOFMANN and ROSEN,1981).

The El Chichon volcanic cloud in the stratosphere

1135

a ZO-

IO 10-4



I

‘111*1’



’ *

11111’

Mlxing

ratlo

I



10-2

10-3

5

(ppbv)

FIG. 17. The mixing ratio profiles of SO1 and OCS in case(b). The measured SO, values are shown by-and OCS values by H. (a) INN et al. (1981). (b) INN et crl. (1979). (c) MEKNER et 01. (1981).

(3) Results of Case(c). Figures 18,19,20 and 21 show the profiles of the scattering ratios at 1.06 pm, the channel ratios, the mixing ratio of SO, and OCS and the concentrations of H,SO,+HSO,. Each profile shows the values at the end of summer (S) (i.e. at the end of the period of small eddy ditrusivity) and at the end of winter (IV) i.e. at the end of the period of larger eddy diffusivity) in the fourth year. The profiles at (IV’)are nearly thesameasthoseincase(b).Thisis because ofthe rapid transport by larger diffusivity in winter. Since the sulfur gases are of tropospheric origin, they do not easily reach higher altitudes through small atmospheric eddy diffusivity. The OCS mixing ratios at (S) are smaller than those at (IV) above 20 km altitude, where the dilTusion coefficient values are smaller in

I

to ’ 11-1’

IO--'

IO



’ ’ ’

50

tnqt’

N(r>O.l5~m)/N(r>O25~m) Fig. 19. The channel ratio profiles in case(c). S is at the end of summer and W is at the end of winter.

summer. The mixing ratio of SO2 and the concentration of H,SO, + HSOJ is also smaller at (S) than at (IV) between about 20 km and 32 km altitude. Since aerosol particles are nucleated below 18 km, fewer particles reach higher altitudes at (S), especially the larger particles due to their larger sedimentation velocity. Because of this, at (S) above about 25 km the scattering ratios have smaller and the channel ratios have larger values (Figs. 18, 19). Although the calculated values of the sulfur gas concentrations are smaller at (S) than (W), they are within both the errors and dispersions of the observed concentrations. The assumed variation of the eddy diffusion coefficient does not conflict with the observed results. 3.2.3. Results ojthe disturbed stare. As a disturbance by volcanic eruption, SO2 gas injected into the stratosphere is considered. The effects of other disturbances are discussed in the next section.

40

-

30 E r 4 < f: 4

zo-

loMixing Fig. 18. The scattering ratio profiles at 1.06 pm in case(c). S is al the end of summer and W’ is at the end of winter.

ratio

(ppbv)

Fig. 20. The mixing ratio profiles of SO, and OCS in case(c). S is at the end of summer and W’ is at the end ol winter.

1136

TAKASHI SHIBATA, MOTOWO FUJIWARA and MOTOKAZU

HIRONO

l&O, and H2SOI which were produced by oxidation of the increased SO1. In&eased H,SO, also increases the ion nucleation ratios at each altitude. Four cases werecalculated again toexamine whether or not the estimations of the diflusion coefficient in Section 2.2.1 were appropriate and to examine the effects of the ion nucleation.

(d) Massie and Hunten’s eddy diffusion profile was used from the beginning and the model calculated without ion nucleation. El Chichon erupted in March and at the beginning of April, about when the summer easterly wind begins. Concentration (cm-3) Fig. 21. The concentration profiles of H,SO, + HSO,. S is at the end of summer and W is at the end of winter.

The increase of SOz and the resulting increase of H,SO, would cause additional particle formation to the background aerosol nucleation. As described in Section 3.2.1, the aerosol nucleation processes in the stratosphere are not well understood even now. Though increased particle nucleation has been inferred by observations of CN particles following the eruption of El Chichon and Mt. St. Helens (HOFMANN and ROSEK,1983), the nucleation mechanism has not been explained for these disturbed states as well. We are now discussing the vertical scattering ratio profile and the total sulfuric acid mass. Active nucleation after volcanic eruption would form numerous numbers of fine aerosol particles and the existence of these fine particles has some effects on the scattering ratio profile and the total mass. The average descending velocity of the layer would be decreased by these small particles. This is because of their relatively smaller sedimentation velocities. The concentration of sulfur gases is decreased by the condensation of sulfur gases into these small particles. Although the mechanisms of nucleation are not very well understood, the ion nucleation process which was recently proposed by ARNOLD(1982) is here used to estimate the effects of nucleation. Though the exact process of the nucleation cannot be simulated with only this ion nucleation process, the rough tendency of the active nucleation and its effect on the aerosol layer can be calculated through ion nucleation. The results of case (bj in the last section, except the profile of [SO,], were used as the initial conditions. The initial profile of [SO,] is that obtained in case (b) plus 40 ppbv between 22 km and 27 km altitude. This was deduced in Section 2.2.2. as the increased concentration of SO2 after the eruption of El Chichon. The sulfuric acid solution droplet aerosols grow from the increased

(e) The small eddy diffusion coefficient was used in the first six months and then the two diffusion coefficient profiles in Fig. 10 were used alternatively every six months without ion nucleation. Then to examine the effect of ion nucleation calculated

we

(f) case (d) plus ion nucleation, and (g) case (e) plus ion nucleation. The resulting first six months scattering ratio profiles of case (d) are shown in Fig. 22. It is clear that the layer broadening is too fast in this case, if compared with the approximate observed profile of the upper layer during this period. The first six months scattering ratio profiles of case (e) in Fig. 23 show better coincidence with the approximate observed profile than in case(d). Figure 23 also shows the scattering ratio profiles of the seventh to twelfth months. As in Fig. 23, case(e) also reconstructs the layer broadening from about the sixth month after volcanic injection, i.e. since October. Figure 24 shows

40

t

106um

-I

Fig. 22. The scattering ratio profiles of case(d). The numbers attached to the profiles are the month after injection.

The El Chichon volcanic cloud in the stratosphere

Scattering

1137

ratio

Fig. 23. The scattering ratio profiles of case (e). The numbers attached to the profiles are the month after injection.

the t? variations in the model simulation. From April to September the slope of the line in case(d) is too steep. Comparing the calculated layer width with the observed width (Fig. S), the variation of the layer profiles in case (e) coincides better with the observed profiles. This better coincidence with those observed in case (e) indicates that the assumed small eddy diffusion coefficient in summer is more plausible. The tendency is the same in case (g) (Fig. 25). The deduced D values from Figs. 24 and 25 are 2.0 x lo* cm* s- ’ and 5.8 x lo* cm* s-’ in summer for cases (e) and (g), respectively, but are 2.0 x 10” cm* s-’ and 7.4 x 103 cm* s-1 for cases (d) and (f), respectively. The D values in winter are 3.3 x lo3 cm* s-’ and

Case Cd)

c

1

“““‘I”” 2 3

4

5

6

7

6

9

10

II

I2

Month

Fig. 24. The a2 variation in case(d) and case (e).

Fig. 25. The a2 variation in case (f) and case (g).

5.0x I03cm2s- ’ from Figs 24 and 25 for cases (e) and (g), respectively. Though a smaller diffusion coefficient than the value deduced in Section 2.2.1 is assumed above 20 km, the derived D values are rather larger than 3.1 x lo* cm* s- I in case (g). Figure 26 and 27 show the time variations in the mixing ratio profile of SO2 and the concentration profile of OH in case (e). SO, e-folding decay time is shorter than 3 months (7 x lo6 s) at 22 km. At the other altitudes, where SO2 increased, the decay time is shorter than the time at 22 km. Though the OH concentration is smaller between 22 km and 27 km where SO2 was increased, as determined in Section 2.2.2, the decrease of OH is by a factor 4 in the first month after SO2 injection, but OH concentration recovers to its steady state value within about 9 months. In spite of the decrease of OH concentration, SO, e-folding time is shorter than 10’ s at 24 km with the reactions and reaction rate constants used in this model. Figure 28 shows the time variation of [H,SO,+ HSO,] for case(e). For three months after the injection of SO2, [H,SO, +HSO,] is more than one or two orders ofmagnitude larger than the steady state value at !he altitudes where the injection occurred. This makes the growth rate of the aerosol droplets larger by the same orders of magnitude, i.e. the particles show rapid growth at these altitudes. Figure 29 shows the calculated [H2SOa] time variation in case (e). Figure 30

1138

Mixing

ratlo

(ppbv)

ii -

30

9 c 5

a

20

t

_I

IO0

IO Mixing

ratlo

(ppbv

1

Fig. 26. The mixing ratio profiles of SO, in case(e). The numbers attached to the profiles arc the month after injection.

shows 26 km. where about caused

42

the time variation of particle size distribution at The initial size distribution has its mode radius, the size distributions take maximum value, at 0.08 pm. Through the condensation growth by the increased [H,SO, + HSOJ, this particle

-

OH

z t a

3i;

-

‘: +Y

a

20 -

Fig. 27. The concentration profiles of OH in case (e). The numbers attached to the protiles are the month after injection.

radius rapidly increased such that the mode radius of the size distribution became as large as 0.7 pm within 6 months. Figure 31 shows the calculated time vanation of the IBC for case (e). Though the calculated IBC reached its maximum value about 6 months after the injection of SO, the time variation is very different from the observed re-increase from August to December 1982 (Fig. 4). In Fig. 31 the IBC variations below 19.5km and above 19.5 km altitude are also shown. Comparing this figure with Fig. 7, the calculated IBC variations for each height interval are also very diflerent to the observed variations. Though the observed IBC bet\vesn 13.5 and 19.5 km reachesamaximum valueabout 1year after the injection, the calculated IBC below 19.5 km reaches a maximum value about 7 months after injection. Figure 32 shows the size distribution variation in case(g) at 26 km. Figures 33,34 and 35 sho\t the profiles of concentrations of H,SO, and H,SO, + HSO, and the scattering ratio at 1.06 pm for case (g). Comparing Figs. 30 with 32,29 with 33,28 with 34 and 23 with 35, the effects ofadditional ion nucleation on the model can

1139

Concentrotlon

Concentration

Fig. 28.The concentration

profiles of

km-3)

(cme3)

H,SO, + HSO,

in case (e). The numbers attached to the profiles are the month after injection.

be summerized as follows. By the oxidation of injected SO2 the concentration of H2S04 increases at the same altitudes where SO2 increases (Figs. 33 and 29). This increased H#O, causes rapid ion nucleation and forms a large number of fine particles (Fig. 32). Due to the existence of these fine particles the total surface area in case (g) is larger than that in case (e). H,SO, and HSO, are rapidly consumed by attachment to the increased surface area. The concentrations of these components are smaller in case(g) than in case(e) (Figs. 33,34,29 and 28). Subsequently the growth rate of the particles also decreases. Because of this slower growth rate the mode radius of the particle size distribution in case (g) (Fig. 32) is equal to or smaller than half of that displayed in case(e) for the same period after injection. Since the sedimentation velocity is roughly proportional to the particle radius (KASTEN, 1968) the layer descendsmoreslowly in case(g) thanin case(e)(Figs. 35 and 23). Figure 36 shows the variation in the calculated scattering ratio peak altitude for cases (e) and (g).

Because of the slower descending velocity in case(g) the peak altitudes in case(g) are about 2 km above the peak in case (e). As shown in the Figs. 24 and 25, the variations of the half width of the layer are not so affected by ion nucleation. However, in case(e) the layer peak descends 5 km to an altitude of 20 km about 6 months after injection. Then the observed clean level at 2 1km is filled by the layer. In case (g) the layer peak is significantly higher than this clean level for the same period. Figure 37 shows the IBC variation in case (g). The variation in the total value is again very different from the observed variation, as in case(e). However, the IBC variation below 19.5 km is nearer to the observed one than that ofcase (e). The calculated IBC below 19.5 km reaches a maximum value about 1 year after injection. This is about the same length of time as seen under observation. T,his delay in the variation below 19.5 km in case (g) is caused by the slower sedimentation. Figure 38 shows the calculated profile of the channel

TAKASHI SHIBATA,

1140

Mo~owo

FUJIWARAand MOTOKAZLJHIROXO

Concentration

(crr1-~1

Concentration

(cm_31

40 ‘;i t 4 30 3 Y 5 a 20

Fig. 29. The concentration pro&a of H,SO, in case (c). The numbers attached to the profiles are the month after injection.

+A- I Month *

3

Month

-

105-295km

Y

O-

lS5-295km

-

+

10.5-

.._._ 1 IS 5 km

Ob=r*d 135-

Radius (pm)

Fig. 30. The variation of particle size distributions in case (e) at an altitude of 26 km. The numbers attached to the profiles are the month after injection.

I9

Month

Fig. 31. The IBC variations in the case(e).

5 km

The El Chichnn volcanic cloud in the stratosphere

q

I

Month

Fig. 32. The variation of particle size distributions in case (g) at 26 km altitude. Thenumbers attached to the profiles are the month after injection. ratio of cases(e) and(g). The ratio is about 1 for case (e), but about 2-5 for case(g), 3-6 months after the injection in the height range where SO, was increased. This difference is due to the different particle growths which are shown in Figs. 30 and 32. The observed profile of the ratio in southern Texas, U.S.A. (27-29’N) 6 months after the eruption (HOFMANN and ROSEN,1983) is seen to be between the profiles of cases (e)and (g) 6 months after injection.

4.

DISCUSSION

As shown in the previous sections, the small dilTusion coefficient in summer above 20 km adequately explains the broadening of the layer at this altitude. This diflusion coefficient value is, however, much smaller

1141

than the value which has been used in the conventional one-dimensional model calculation of stratospheric chemical composition, as described in Section 2.2.1. To reconstruct the profile of the chemical components in the one-dimensional model in which one diffusion coefficient is used, as in cases (a) and (b), a vertical eddy ditTusion coefficient value larger than about 10“ cm* s-’ is needed above an altitude of 20 km. Though theeddydifiusioncoefficientsobtained from the distribution of stratospheric tracers have the values mentioned above, from observations of turbulence values in the order of 10’ cm2 s-I were derived. By usingaircraft,Lr~~~ etal.(l974)estimated that theeddy dinitsion coefficient caused by stratospheric turbulence is ofthe order of 10’ cm2 s-l. CADET(1977), by balloon observation of the turbulence, also obtained values of the same order. CADET (1977) also pointed out that a diffusion coefficient of the order of 10’ cm* s-’ is one order of magnitude smaller than the values often used in one- or two-dimensional stratospheric transport models. He explained this difference through a scale of turbulence. In the usual model calculation, the diffusion coefficient represents the combined effects of small scale turbulence and also global scale disturbances such as planetary waves. Cadet’s observation included only small scale turbulence and he concluded that the larger diffusion coefficient is caused by larger scale global air motion. Since planetary waves cannot propagate into the summer easterly region (MATSUNOand SHIMAZAKI, 1981) there exist no large scale disturbances in this region. The vertical transport of materials is caused mainly by small scale turbulence and the diffusion coefficient in summer is smaller than the conventional values. As the results of case (c) in Section 3.2.2 have shown, the transport by winter eddy dillusion is so large that the main portion of the stratospheric materials are maintained by the transport in winter. By such a variation in the vertical eddy diffusion. the distribution of the chemical components can also be simulated. The result ofcase(c)indicate theseasonal variationin size distribution in Figs. 18 and 19 and in those components in Figs. 20 and 21. Because OCS is transported from the troposphere by eddy diffusion, less reaches above 20 km altitude in summer than in winter. OCS concentration displays lower values above 20 km in summer than in winter. The concentrations of SO*, H,SO, and HSO, also indicate seasonal variations between about 20 km and 30 km in altitude. Of course, horizontal transport cannot be explicitly included in a one-dimensional model. There is a possibility that theabove mentioned seasonal variation

1142

TAKASHI SHIBATA, Momwo

FUJWARA and MOTOKAZU HIRONO

“c

*0 _

(WY)

ww+iv

The

El Chichon

1143

volcanic cloud in the stratosphere

1 i

I

;

Scattering

-

105-29.5km

1

O-

195-295km

-

,‘“‘,

IO

+

rat10

105-195km

‘.‘.”

Fig. 35. The scattering ratio profiles of case(g). The numbers attached lo the profiles are the month after injection.

Observed 13 5- 19 5 km

. ,

Month

in the model is compensated for by meridional transport. However, since the mean meridional circulation near 30 km altitude is from the summer hemisphere to the winter hemisphere (MATSUSO and SHIMAZAKI,1981) the meridional transport should intensify such variations. The accuracy of the observations Df sulfur gases are not so good as to observe such a systematic time variation in these gas components. However, the seasonal variations described above should be confirmed by more accurate observations in the future. As in Figs. 3 1and 37, the calculated time variations of the IBC cannot simulate there-increase ofthe observed

26 -

25 -

23

.9

I

I

I

2

1

3

I

4

I

5

I

6

I

7

I

8

i

9

Month

Fig. 36. The variation of the calculated scattering ratio peak altitude for cases(e) and (g).

Fig. 37. The IBC variations

in case (g)

IBC from August to December 1982. The characteristics of the time variations of the IBC in the model correspond well to the variation of n in cases(b) and (c) in Fig. 9. This is because, as shown in Fig. 26, SO1 efolding decay time is shorter than about IO’ s at 22 km and because in the first 6 months the diffusion coefficient is so small that this e-folding time can be taken as the oxidation time in the model. OH depletion does occur in the model calculation (Fig. 27). However, the degree of depletion is not so large as to extend the oxidation time much longer than 10’ s. Volcanic eruptions also inject large amounts of H,O into the stratosphere. This H,O becomes the source of OH through reaction R 16. Though this process was not considered in the model, the process shortens the oxidation time of SO,. Soon after the eruption of Mt. St. Helens in May 1980, SEDLACEKet al. (1980) and CHUANer al. (198 1) observed sulfuricacid aerosols. They inferred a very fast oxidation of SOz. TURCO et al. (1983) explained this oxidation mechanism as the heterogeneous chemical reaction of SO2 on the surface of ash particles. This process also shortens the oxidation time of SO1. The results of the simulation and the above discussions imply that explanation (b)in Section2.2.2 is not of essential importance and that the observed reincrease of IBC is not caused by the decrease in oxidation time of SO,. The erects of horizontal transport must be taken into account to explain the reincrease in the IBC after the initial decrease in August, as in explanation (a). Due to the better correspondence of the variation in the IBC below 19.5 km. as in case(g), with the observed

TAKASHI SHIBATA. MOTCWO

1144

FUJIWARA and

N(r~O15~m)/N(r20.25~m)

MO~~KAZUHIRONO

Nlr~0.15pm~/N~rlO25~ml

Fig. 38. The variation ofchannel ratio profiles in case(e). The numbers attached to the profiles are the month after injection. The approximate channel ratio profiles observed by HOFMANN and ROS.EN (1983) at Texas (2729”N) on 23 October 1982 and at Laramie (41”N) on 5 November 1982 are shown by the broken lines.

variation and by the observation that the main portion of the cloud is in the upper easterly region in the stratosphere, the following explanation of the reincrease of IBC can be made. As shown in the observations by REITERet al. (19833, LABITZKEet al. (1983) and D'ALTRIO et al.(1983), in the higher latitudes the El Chichon cloud was detected first in the lower layer below 20 km. As mentioned above. planetary waves can propagate as far as about 20 km from the troposphere because the wind is westerly below this altitude in summer. The existance of planetary waves in the lower region makes horizontal transport of the materials faster than in the upper region above 20 km. If, as in the simulation of case (e), formation of fine particles does not occur in the volcanic cloud. the particles in the cloud grow faster than in case (g), in which nucleation occurs, and the cloud falls into the westerly region below 20 km faster than in case(g) (Figs 31 and 37). The particles falling into this region are rapidly transported to the higher latitudes. The formation of fine particles in the cloud decreases the particle sedimentation velocity and suppresses horizontal transport. As a result, the nucleation helped to trap the main part of the cloud within the upper layer at lower latitudes (about YS-35’N) until November 1982 (Section 2.2.2) and contributed to the IBC variation observed at Fukuoka.

5. CONCLUDINGREMARKS From the observations and numerical calculations the following conclusions can be drawn. (a) The stratospheric vertical eddy diffusion coellicient in the summer easterly region is of the order of lo2 cm2 s- ’ or smaller. (b) It is predicted that the stratospheric materials of tropospheric origin would be in smaller concentrations in the summer easterly wind region. (c) The observed re-increase of the IBC (integrated backscattering coefficient) from September to December 1982 was not caused only by the delayed oxidation ofinjected SO, but also by seasonal variation of the meridional advective and diffusive transport from lower to higher latitudes in these periods. The nucleation in the cloud makes the simulated results more plausibly fit the observed results. (d) As estimated by ARNOLDand BOHRKE(1983), HSO, would also condense on H$O,-Hz0 liquid aerosol particles. Acknowledgements-The

would like to thank H N. KUGUMIYAfor their assistance in these observations. The authors are indebted to many members of the Meteorological Agency. The YAMAMURA,

stratospheric pressure, temperature and wind data were supplied through the courtesy of the Fukuoka Meteorological Observatory.

REFERENCES &WOLDF.

1982

B~JHRKE TH. BIGGE. K., ONO A. and THOUP~~NW. J.

1983 1970

ARNOLD F. and

authors

N. FUJIWARA and

Nature 299, 134. Nature 301,293. Tellus 22, 550.

The El Chichon volcanic cloud in the stratosphere CADETD. CADI& R. D. CASTL~?MAN A. W. JR, DAVID R. E., MUNKEL~~ TANG I. N. and WOOD W. P.

H. R.,

CHUANR. L., Woons D. C. and M~~x~M~cK M. P. ColJtsoN K. L. D’ALTORIOA. and VIXX)NTIG. DAVISD. D., KLEMHR. P. and PILLINGM. DAVLSD. D. DYER A. J. and HICKSB. B. F10cc0 G. and G~nhls G. FUJ~WARAM., ITAB~T. and HERONSM. FUI~WARAM.. SHIBATAT. and HIRONOM. GMITROJ. I. and VERM~VLENT. GRAMSG. and G. F~occu G. HAMILLP., KIANGC. S. and CADLER. D.

HAJ~OUP., Tuaco R. P, KtANGC. S., TOON0. B. and WHITYXNR. C. HARRIESJ. E. HAYESD., SNETSINGER K, FERRYG., OSER~ECKV. and FARLOWN. HIRONOM., FUJIWARAM., UCHINO0. and ITABET. HIRONOM., FUJIWARAM., ITABET.

1145

1971 1980 1975

J. arms. Sci. 34, 137. 1. geopbys. Res. f& 4495.

1981 1983 1983 1972 1974 1968 1964 1975 1982 1964 1967 1977 1982 1982 1976 1980

Science 211,830. Appl. Optics 22, 1036. Geophys. Res. Left IO, 27. Int. J. them. Kinet. 4,367. Can. J. Chem. 52, 1405. Q. II R. mel. Sot. London 94,545. J. afmos. Sri. 21,323. Rep. lonosph. Space Res. Japan 29,74. J. atmos. terr. Phys. 44+ 811. A.1.Ch.E. Jl 10,740. J. geophys. Res. 72, 3523. J. atmos. Sci. 34, 150. J. Aerosol Sri. 13,561.

1972

Rep. lonosph. Space Res. Japan 26,237.

1977

J. Geomagn. Geoelect., Kyoto 29, 541

1981

Geophys. Res. Lplr. 8, 1019.

1983 1984

Geophys. Res. L.ett IO, 152.

1977 1981 1982 1983 1979

J. geophys. Res. 82. 1435. J. atmos. Sci. 38. 168. J. geophys. Res. 87. 11039. Geophys. Res. Lett. 10,313 Geophys. Res. L&t. 6, 191.

1981 1980

Geophys. Res. Len. 8, 5. J. met. Sot. Japan 58, I27

1976 1968 1983 1983 1970 1981

J. geophys. Res. 81,368. J. appt. Met. 7,944.

ht.

J.&em.

Kiner. Symp. I, 529.

Rev. Geophys. Space Phys. 14,565. Geophys. Res. Lett. 7,974.

and NAGASAWA C.

HERONSM., FUJIWARAM., SHIBATAT. and KUGUMIYAN. HIRONOM. and SHIBATA T. HIRONOM., FUJIWARAM., SHIBATAT.

J. atmos. {err. Phys. (in press). (ATP 1441

and KUGUMIYAN. HOFMANND. HOF~MNND. HOFMANND. HOS%ANND. INN E.

J. and J. and J. and J. and

ROSENJ. ROSENJ. RISEN J. R-NJ.

C. Y., VADER

M. M. M.

M.

J. F., Tuso~ B. J.

and O’HARA D. INN E. C. Y., VED~ERJ. F. and O’HARA D. ITABET., FUJIWARAM., H~RONOM.

and IGARASHIT. JOHNSTONH. S.. KATTEIWORND. and WHITTENG. KASTENF. KRUEGERA. J. LABITZKEK., NAUJOKATB. and MCCORMICK M. P. LAMBH. H. LEFRERE J., PEL~NJ., CAHENC.. FLAMANTP. and MEGIEG. LILLY D. K., WACO D. E. and ADELFANGS. I. MASSIES. T. and HUNTEND. M. MA~SUNO1. and SHIMAZAKIT.

1974 1981 1981

Science [email protected], 1377. Geophys. Res. Letr 10, 24. Phil. Trans. R. Ser. 266.425. C. R. hebd. Sdanc. Acad. Sri.. Paris 292. J. appl. Met. 13,488. J. geophys. Res. f&,9859. Lectures on the tltmospheric

University press (in Japanese). MEIXNERF. X.. GEORGIIH.-W., OOLELNANNG.,

J&ER H. and REITERR. MOOR~GATG. R. and JUNGEC. E. Mossop S. C. POLLACKJ. B. REI?ERR.. JRGERH.. CARNUTH W. and FUNK W. ROEDELW. SOIOFIELDK. SHIBATAT., FUJIWARAM. and H~RONOM.

SHINAZAKIT. and WHITIXN R.C. SMITHSONIAN INSTIT~;TION T~ON 0. B.. Tu~co R. P., HAMMILP., KIANCZ C. S. and WH~XYENR. C. TURCOR. P.. HAMILLP.. TOON0. B., WHITTEN R. C. and KIANG C. S.

1981

Geophys. Res. Lrrt.

1977 1964 1981 1983

Payeophilus 115, 759. Nature 203,824.

1979 1973 1980 1976 1982 1979

.f. Aerosol. Sci. IO, 375.

1979

J. afmos. Sci. 36.699.

Science 211,8

8, 163.

t 5.

Geophys. Res. Lett. 10, 1001.

J. Phys. Chem. Re/. Dara 2.25. Jap. J. appl. Phys. 19, 2205. Reo. GeophJa. Space Phps. 14, I. Sean Bull. 7 (No. 1 l), 3. J. almos. Sri. 36,7 18.

13.

Scien,es

3.

Tokyo

1146

TAKASHISHIBATA.MOTOWO FUJIWARAand MOTOKAZ~~HIRONO 19819

J. geophys. Res. 86, 1129.

198 1b

J. geophys. Res. 86,5313.

1982 1983

Reo. Geophys. Space Phys. 20,233. J. geophys. Res. 88.5299.

1966 1983 1969 1983

U.S. Standard Atmosphere Supplements, U.S. Government Printing Office, Washington, DC. J. geophys. Res. 88, 1457. J. them. Phys. 50,707. Mem. Natn. Inst. Polar Res. Spcc. Issue 29,201.

1982

J. geophys. Res. 87, 3128.

HAMPSONR. F. JR and GRARVIND. (Eds)

1975

SEDLACEK W. A., HEIKENG. H., MR~Z E. J., CUDNEY E. S., PERRINP. R., LEI~R R., and HINCLIFFEL.

1980

Technical Note 866, National Bureau of Standards, Washington, DC. NASA Symposium “Mt. St. Helems Eruption”, Washington, DC, 18-19 November 1980.

Ttmco R. P.. WHITTENR. C., T~ON 0. B., INN E. C. Y. and HAMILLP. TURCOR. P., CICERONE R. J., INN E. C. Y. and Cam?;~ L. A. TURIXIR. P., WHITTENR. C. and Toot 0. B. TUR~IIR. P., Toot 0. B., WHITTENR. C., Hamill P. and KEE~ER. G. U.S. GOVERN~IENT PRINTING @FlCE VI~~IAXOA. A. and ARNOLDF. WE~TENBERG A. A. and de Hm N. YAMA~~URA H.. HIRONOM., UCHINO0. and MAEDAM. Ytm G. K. and DEEPAKA. Reference is also made to the following unpublished material: