Raw Data

Results and discussion of the old LIDAR observations

5.4 High Altitude Aerosols

5.4.1 Raw Data

200

JOO

400

500

";i' a..£:;

..

⁶⁰⁰

..

~

a: 700

BOO

900

1000

• JOOhPa layer

•500h"a layer

- 700hPalayer

·850 hPa layer

PI PR 3E BL UP SP CT PE DB

Figure 5.7: Mean annual spatial variation of absolutely stable layers at 850,700, 500 and 300 hPa over South Africa. PI denotes Pietersburg, PR Pretoria, BE Bethlehem, BL Bloemfontein, UP Upington, SP Spingbok, eT Cape Town, PE Port Elizabeth and DB Durban (Cosijn and Tyson(1996)).

October, although mixing is taking place, cont inuous loa ding of the ^atrno- sphere has result ed in a higher concentration of aerosols. Fig.(5.5) shows a strong return bet ween 8km and 10 km on October 20th 1997at 18:30. This may be due to the stable discont inuit y at 300hPa ("" 9 km) (see fig.(5.7)).

phot omultiplier and the x-axi srepresent s the alt itude in km.

Ther e are three interesting pointsto note in these plots.

First, cons ider the aerosol layer at 14 km. There is evide nce to support thatthis increased aero sols conce nt ration originates from sugar caneburning.

The upward tro pospheric circul ation has transported this layer to higher altit ude. The estimated thickness ofth is layer is 1 km.

The second feature of int erest is the par tial saturation and recovery of the phot omultiplier between 20 km and 30 km. The mechanical shutter was not in operation at the time which could have solvedthe saturation problem.

Thirdly, the erratic behaviour of the LIDAR signal after 30 km. This is due to int ernal noise from the equipment.

In order to see whether this aeroso l layer persist s for the whole of July, profiles weretaken three weekslat er ,morespecifically on July 23rd 1997 and theresultsareshown in fig. (5.9). Thetimewas takenin intervals of 15 mins as before. The 14 km aerosol layer which appeared on July 3r d 1997 seems to reinforce and appears at greater height. The est imated thickness of the layer is 1 km.

The peaks appearing after 20 km are background noise mixed wit h shot noise when the laser is fired . We can there fore safely conclude that the 14 km aerosol layer persist s for the whole of July and seems to increa se in conce ntrat ion towar ds the end of July. This is obvious because of the continuous loading ofthe atmosphere wit hsugar cane burning whichstarted in the beginning of June. At the same time mixing is also taking place but

JulyJ rd199⁷ 18:)0225shot

200

I

! :~d

10 15 20 25 30 35

Allndelbn

July3rd199~

19:00225she

! ~~ :J

^{o ~&}

10 15 20 25 30 35

PIlitudelkn

JulyJrd I9 1s0 2!UlSts

! :,lb\:

_o P"""

:d

10 15 20 25 30 35

JulyJrd1997 18:45225sho

l : E :J

ID 15 20 25 30 35

Allitudelbn

July3ld 1%

19:15225 she

'

^~

:

^10050o

k£::J

10 15 20 25 30 35

Allitude-br

Figure 5.8: High altitude L1DAR profiles taken on July 3 1997 from 18:30 to 19:30.

probably at a slower rate.

In order to underst and the behaviour of the aerosol layer,five other pro- files were taken in Augu st. Unfortunately, unfavourable weather conditions in the beginning of August has prompted us to take profiles on the third week of August. Fig.(5. 10) shows theresults obtained on August 21st .

From the profiles of August 21st 1997 wecan conclude that the aerosol layer at 14km has disapp eared. At this point we can say that mixin g of the atmosphere has contributed to this disappearence.

July 2Jrd1997 18:3025 shot

"

200~

150

" 100

8 50

o

10 15 20 25 30 35

AltiludeiImI

July2 l "d 199 190 0 25shl!:

20 ~

15

~ l :oO ~

10 15 20 25 30 35

AI:ilu •.tm

Ju~23rd1997 1&45225shot:

~

^g

:~

¹⁰⁰

" 50

o

10 15 20 25 30 35

-.teAm

July 2Jnl 195 19:15ZZ5she

! &;;d

10 15 20 25 30 35

AlliludeAJ1

Figure 5.9: High altitude L1DAR profiles taken on July 23 1997 from 18:30 to 19:15.

In order to obt ain a clear understanding of the aerosol behaviour, we plott ed profiles for the months of September and October. Figs.(5.1l) - (5.13) show the result s obtained on the night s of September 26t h, October 20th and 21st.

From the profiles ofSeptember 26th, it isclear that thepeak s that appear at 12km depict some aero sol concentration at thisaltitude. The appearence of the 12 km aerosols layer can be ex plained by the fact that some tropo- spheric upwelling has transport ed the low alt itude aero sols (from sugar cane burning) to higher altitude.

Perhaps a more int erest ing feature that can be seen from the Oct ob er 20th and October 21st profiles (figs.(5.12) - (5.13)) is the dramati c increase in the lidar return at 28 km. Kuppen (1996) previously detect ed an aerosol layer at 25 km. This might be the same layer with the exception that over

Aug 2lst1997 18:30 50 shotsJ

,:c:::J

^{c~ 100}_'0 ^.

o

10 15 W ~ 30 35

AUudeA<m

AUg21st 199 19:00150shots

' ~:c;:::Id

;; 100

C) , .

o

10 15 20 2S )0 3S

JlJttude/bn

zng21st19, 19:)0 soshot!

" ~:: l::::d

;; 100

C) SO

o

10 is ~ 25 30 35

Ah1ndell:m

Allg21st1997 18:451 50shon

ii~: t:::::d

5100

C) '0

o

10 15 20 25 30 35

J\l1tudelkm

All:2l!t199, 19:15~Osh<1

,:

^U5100

:

^5ij

c::::::d

o

10 15 20 25 30 35

Miudo1<m

Figu re 5.10: High altitude L1DAR profiles taken on August 21 1997 from 18:30 to 19:30.

the sp ace of 1 year some stratospheric upwelling has taken place which has carried the layer to 28 km.

The "28 km layer" (as we sha ll henceforth call it) was detect ed only for the mont h of Octob er 1997 (fig.(5.13)). This maybe due to the fact that weather conditionsmight havebeen ideal in October, that is low water vapour conce nt rations have resulted in strong return from higher altitudes.

This 28km layer was rather intriguing.

Perso na lcorrespondence via e-mailwit h theLIDAR group in Reunion

(H.

Sept26th '9: Sept26lh19'1:

18:45 150shot 19:00 _~!l1ot

l ~~ ~::;;J

_{10 15 20 25 30 35}

^~~~

^o10 15^I 20^I 25 30 35

Ahitudelbn All ndcMIl

Sepr26th1!9?

19:15150shou Sept26th199'

l :10~ _{! :~ ~}

^{19::30150 shot}

10 15 20 25 30 35 10 15 20 25 30 )5

AItirodeibn Alind>/bn

Figure 5.11: High altitude L1DAR profiles taken on September 26 1997 from 18:45to 19:30.

30 O::t:DthI~

18:45150sh

le _{U 1~ , ~} ^I _.dd

10 15 20 25 35

A1titudCl1<r 0et20th 9971

18:30 50<fuots

400

I

! ;~b:",

_{10 15 20}

"

₂₅

, .. ^!":d

_{30 35}

.AlWJde/km

Oct 20th1997 19:00150shot

i~h:

^{o t ,I}

:J

10 15 20 25 )0 35

AIlitudeAJn

o«20lh199' 19:15150sho

l~;b:::J

10 15 20 25 )0 35

AIlitudeibn

Figure5.12: High alt itude L1DAR profiles taken on October 20 1997 from 18:30 to 19:15.

OCt 21s t1997 18:30150shot.

If J _O ~ ~ _~

. . .

~

10 15 W 25 ~ JS

A1I1odelbn

o«^21st19 9~

19:00150 sh e

, :~

; 100 U 50

o

10 15 20 25 30 35

.A1titudelkm

0et2 1,t 1997 18:45150shot

'

^~

=

¹⁰⁰

h:LJ

u 50

o

10 15 20 25 30 35

Altitude&rn

Figure5.13: Highaltitude L1DAR profiles taken on October 211997 from 18:30 to 19:00.

Bencherif, 1997,per s. com rn.l) and in New Zeal and (G. Bodeker, 1998, per s.

comrn.") indicat ed that they did not det ect aerosolsat such high altitude. In order to com pare the Durban old LIDAR system with that used in Reunion Island, we describe briefly in the next sect ion the LIDAR system used in Reunion Island.

TheLIDARsystem in Reunion consistsof a Nd- YAG laser as transm itter , operating at a wavelen gth of532 nm (frequency doubled) with a rep etition rat e of 10 Hz, and the energy per pulse is300 mJ.

There are two ch annels for data acquisit ion:

(i) Cha nnelA. This is used for measurement ofverti cal temperaturepro- filesand relative den sity in the altitude range 30 to 80km.

(ii) Channel B. This is an extension ofchannel A towards lower alt it ude

1Dr Hassan Bencherif, Department of Physics, University of Reuni on, 15 av H.

Cassin,BP 7151,97715,St-Denis cedex,Reunion , Fra nce.

2Dr Greg Bodeker, National Institut eof "Vater and AtmosphericResearch, P.O. Box 50061,Omaka u, Cent ral Ot ago, NewZealand.

and allows measurement oftemperatur e and density to bedone inthe range 10 to 45 km. It also corrects for overlap with the alt it ude range of channel A.

The receiver is a Cassegrain telescope type with a system of photomul- tipliers (PMTs) as detector. Each PMT has an int erference filter centred on 532 nm with a bandwidth .6.>' = 1 nm and is equipped with electronic shutt ers. The latter prevent saturationof the PMTs during LIDAR returns from low alt it ude. The PMTs are also cooled by thermo- element s ("Peltier effect") which are themselvescooled by water. Thisreducesconsiderably the dark current in the PMTs.

The laser system and the acquisition channels used in Reunion Island have been installed recently at Durban and form part of the new LIDAR syst em .

The wavelength(589 nm) ofthe old LIDAR syst emof Durban is marginally different from that of Reunion Island (532 nm) but both areequally capable of detecting the 28 km layer . We now make a deeper analysis of our result s by computingthe extinct ion coefficients dueto aerosols.

As will be pointed out lat er (Chapter 6) the Klett inversion method is a moreaccurate method to retrieve aerosolextinct ions from the LIDAR dat a.

Duetothe stronglaserreturnsfrom lowaltit ude (0 - 10 km),sat ur ation ofthe photomultiplier swamped most of the relevant features. We therefore opted toapply Klett inversion methodto analyse features above 10 km. Figs.(5.14)- (5.15) show plots of the extinction coefficient versus altitude in the 10 km to 35 km range for July 23r d 1997 and August 21st 1997 lidar profiles re-

spectively. Figs.(5.16) and (5.17) show plots ofthe extinction coefficients in the altitude range 10-15 km and 20-32 km for the same date October 20th 1997. Plots of extinction coefficients were also made for October 21st 1997 in the altitude range 10to 15 km and 20 to 30 km and these are shown in figs (5.18) and (5.19) respectively. The programme Klet t_ Inv.for was used to calculat e the extinct ion coefficient s using the Klett inversion method. A listing of the source code isgiven in Appendix A.

The programme requires 3inputs:

(i) the raw datafile,

(ii) the lower altitude limit for the integration (Zo), (iii) the upper altitude limit for the integration (zm).

The values Zo and Zm are chosen according to the range the integration is carried out.

Returningto figs.(5.12) - (5.13) there are two features of interest:

(i) the aerosol layer between 14 km and 15km, (ii) the aerosol layer between 28 km and 30 km.

We plotted the extinction coefficients for the alt it ude range 10 km to 15 km using the raw data from fig.(5.12) for Oct ober 20th and fig.(5.13) for October 21st. These are shown in fig. (5.16) and fig.(5.18) for the two con- secutive days (October 20th and 21st) . In fig. (5.12), the largeLIDAR return (rv 100 count s) between 14 km and 15 km shows an ext inct ion coefficient of

35 30

20 25 A ltitude (km) 15

10

0.05 ---.--- - - = - t T O l I T - - - ,

0.04 0.03 0.02 0.01

O + - - - + - - - - + - - --+---+---j ....Q)

" - '0..

...

=

.g,au

S

_Q)

ou s:::::

.9...

.s_...u

><

~

Figure 5.14: Extinction coefficient versus altitude in the 10 kmto 35 kmrange- July 23 1997.

s

_~

.... 0.05

~

~

...

_c: 0.04

.~_o 0.03

S _0.02

~

0

() 0.01

...

.s ...

_{o 0}

.s 10

~ ^{15 20 25 30 35}

Altitude (km)

Figure 5.15: Extinction coefficient versus altitude in the 10 kmto 35 kmra nge- August 21 1997.

]:

_0.15

...

&

'-"

i:: ^0.1

.~()

S

_0.05

C.l0 o

c0 0

'J:lo

.8^..., 10 11 12 13 14 15

><

~

Altitude (km)

Figure 5.16: Extinction coefficient versus altitude in the 10 kmto 15 km range- October 20 1997.

a

_0.15

~ l-<

.s

(\)

...

s:: 0.1

.:8o

S(\)

0o_s:: 0.05

.9...

os::

.~ 0

Il.l

20 22 24 26 28 30 32

A ltitu de (km)

Figure 5.17: Extinction coefficient versus altitude in the 20 kmto32 km range- October 20 1997.

j

^0.2

l-<

cD 0.15

S

~

.~~_~ 0.1

EcD

0 0.05

o

.g~

0

o .S';;!

10 11 12 13 14 15

'li

Altitude(km)

Figure 5.18: Extinctioncoefficient versus altitude in the 10 km to 15 km range- October 21 1997.

30 28

24

26

22

20

0.15 ^{- y -- --} - - - -- , - - -.,,---,

[>

.e- o

^0.1

.~u ,...-,..

E~

g

0.05

o s::o

'B 0

.s --t-=---'='--j---''-..-.:..- -f--'<-- - +- - --j- - ----i

JS

Altitude(km)

Figure 5.19: Extinctioncoefficient versus altitude in the 20 kmto 30 km range- October 21 1997.

~ 0.15 km-I. Therefore on the basis of this observation (October 20th 1997) we can deduce that there is an aerosol layer between 14 km and 15 km. This layer can also be seen from the raw data of fig.(5.13) when compared with fig.(5.10) (no aerosol signal). There are two possible reasons for this:

(i) The contribution of sugar cane burning towards the loading of the atmosphere has reached its peak in October rather as was initially expected to July and August.

(ii) In addition to sugar-cane burning,biomass burning either in the form of prescribed burns or wild fires reaches its peak towards the end of the winter season (Rutherford and Westfall 1986). Thus the atmosphere will be richer in aerosols in October than it will be in July and August.

For the higher altitude between 27 km and 30 km, fig.(5.19) shows a remarkable increase in the extinction coefficient of ~ 0.15 km " between 28 km and 30 km. From this value, it is evident that there is an aerosol layer between 28 km and 30 km of thickness ~ 2 km.

This layer is rather intriguing as aerosols have rarely been detected at such high altitude. In the past, Hirono et al. (1974) observed aerosol layers above 25 km during 1972 and 1973 which are reported to be rather variable. Iwasaka and Isono (1977) reported a dust layer above 24 km with a peak intensity at 27 km which has been observed in April 1976 with a two-frequency LIDAR.

5.5 Comparison of Durban old lidar results