Preliminary solar spectro-photometric measurements of aerosol optical density at Lagos, Nigeria

Atmospheric Enairomnr Vol. 13. pp. 1611-1615. Pergamon Press Lrd. 1979. hnted in Great Britatn.

PRELIMINARY SOLAR SPECTRO-PHOTOMETRIC MEASUREMENTS OF AEROSOL OPTICAL DENSITY AT LAGOS, NIGERIA

c. 0.

OLUWAFEMI

Department of Physics, University of Lagos, Akoka, Yaba, Lagos, Nigeria (First receiwd 13 June 1978 and infinulform

10 April 1979)

Abstract - Solar spectrophotometric measurements of the aerosol optical density during the dry dusty (Harmattan) season at two carefully selected sites around the Lagos metropolis are presented. At both sites,a probable contribution of the “remnant”, smaller-size, Saharan dust, borne by the North East trades, is indicated in the total aerosol loading of the ambient air. This would tend to increase on days with favourabk meteorological conditions such as that dictatedby the position of the Inter-Tropical Discontinuity (ITD). The values of the aerosol optical density at industrial and rural sites were very similar. Data for the daytime pattern of the aerosol optical density are also presented.

INTRODUCTION

Measurements of direct solar intensities in the u.v., visible and i.r. portions of the electromagnetic spect-

rum constitute a well-known technique for studying atmospheric aerosol structure. The technique usually enables the aerosol extinction parameters to be evaluated. Such parameters are estimates of the integrated vertical aerosol loading of the ambient atmosphere. Aerosol extinction parameters, so determined, do not provide complete information concerning such aerosol properties as detailed size distribution, concentration profiles and chemical compositions. They serve however as a useful baseline for intensive aerosol measurement programmes. Besides, the parameters have direct impact on the earth-atmosphere radiation budget and, therefore, climate. Observation stations must therefore be carefully sited to ensure data collection from strong global source regions. In this regard, the inRuence of aerosols in the NE-trade wind region needs to be properly ascertained. Following the Atlantic shipboard (Cape Verde) measurements of his group, Jaenicke (1976) has found that these aerosols behave like transition aerosols whose properties greatly differ from those of pure continental, maritime and background aerosols. Such aerosols would appear to be conveyed by the “Equatorial air” (Hamilton and Archibald, 1945). Over land, and at the Southernmost coasts of West Africa, the exact influence of Saharan-derived dust will be more difficult to isolate. More so as these dust particles must mix much faster (than over the oceans) with mineral dust of local origin as the former get transported along. Jaenicke’s (1976) and Shutx’s (1976) results have indicated that, at source, the Saharan aerosol body extends vertically from the surface to about 6 km and that large particles (0.1-l pm) are advected more than 1OOOkm from

source before they penetrate the surface. Such particles should therefore be detectable at the surface along most coastal regions of West Africa under favourable meteorological conditions. Using a form of gravitational settling chamber, McKeown (1958) detected dust particles (mostly 0.2 m-O.3 m diameter) at Ibadan (lat. 7” 24N) during the Iiormattan (Saharan dust haxe) season. closer to the Sahara (at Ouagadougou lat. 1225”N), Cerf’s (1976) measurements indicated a preponderance of rather very large particles under similar conditions. The data presented here form part of an on-going study at Lagos (long. 3.3”E, lat. 6.3’N) concerning the structure of the atmospheric boundary layer. Solar photometric measurements using the 3-channel Voixtype photometer (Volz, 1974) are reported. EXTINCTION PARAMETERS

Wten parallel beams of radiation, wavelength I, traverse a distance f in a scattering medium the intensities get depleted as (Bouguer-Lambert’s Law) :

-w = e-u(“” L(4

where I,(I), I(I) P flux density of the radiation at source and at the observation point, respectively and a(L) = the extinction coefficient. Considering the extinction in the entire atmosphere E is replaced by the vertical extent of the homogeneous atmosphere and the relative optical airmass m = set z(p/p,), where p and paare the station pressure and sea-level pressure (= 1013b), respectively. Z is the

1611 LE.13 12-A

’

solar zenith an&. Thus, Equation (11, with the sun as source, becomes

where z’(n) = c+)H

(3)

is the optical thickness whose constituents, t:(A), r:(i) and b’(A) arise from molecular scattering, selective absorption and extinction by atmospheric haze (aerosol) respectivefy. It is usual to write Equation (2) with the natural logarithmic basechanged to 10. The primed quantities, now unprimed, become the components of the optical density. Thus, the aerosol optical density is

where J,(A), J(1) = sun photometric indications (proportional to solar spectral &radiance) at wavelength 1 for extraterrestrial irradiana and irradiance at the observation point, respaXivcly, F = correction factor for the mean sunearth distance. Combining Equation (3) with Angstrom’s (1929) expression for the scattering co&icient, the optical thickness is ~‘(2) = constant 1-2,

(5)

where

micro-ammeter. A diopter enables the equipment to be pointed at the sun and also measures optical air mass. The observations from two channels (0.5 and 0.88 pm) are presented in this paper. The equipment was precalibrated at delivery but the extraterrestrial constants were crosschecked before operational use by the Langley method with the sun as source (see Flowers er al.. 1969). The linearity of the equipment could not be assessed as a sufficiently good standard was not available for comparison. The observations were made for some months at two locations about 6 km apart. One of the observation points (the main Campus of Lagos University) is located beside the Lagos Lagoon, well away from busy roads and strong sources of local dust. The other (Ilupeju Industrial estate) is NW of the former and towards the Lagos hinterland. These two locations are hereafter referred to as sites 1 and 2 respectively. Lagos is approximately at sea level so that air mass readings m are practically for sea level pressures. The observations spanned the tail end of the coastal rainy season (October 1977) and the dry season (November 1977-February 1978). A total of 190 sets of observations were anaiysed. Norm&y, observations were made under approximateiy cloud-free exposures at sunrise (-0,900 LMT), at local noon and in the fate afternoon f_ 1500-1700 LMT). On cloudy days, observations were made when possible. Supplementary ground level and radioson& data were supplied by the Nigerian Meteorological department for both Oshodi (located around the industrial estate) and Ikeja Airport (some 3 km NW of site 2). RESULTS

a=v-

2

(61

is the wavelength exponent. Y is the exponent in the Jungc (1963) formula for aerosols described by the power law (Bullrich, 1964):

dN(4 5

Cr_v

dlog r

’

The constant C depends on the number of particles with radii between r and r f dr per unit interval dr. Hence, through Equations (S), (6) and (7) a is coupled to the size distribution. Generally Qiies between 0 and 4, being smaller for bigger particles. v = 3 for a mass d~~ibution with qua1 particle masses within qua.l logarithmic radius interval dlogr - the so calfed homogeneous hazy atmosphere. This corresponds approximately to the natural aerosol size distribution (Bullrich, 1964). MEASUREMENTS

Solar energy is accepted on interference filters centred on Green (0.5 pm) and two i.r. wavelengths (0.88 and 0.94 bm). A photovoltaic Silicon detector is thus actuated and the output current J(n) is read on a

Measurements at the two sites were compared by evaIuating monthly mean aerosol optical density and the wavelength exponent as shown in Table i. There appears to be no significant difference between measured values at the two sites. In a bid to associate prevailing atmospheric aerosol loading with air-mass types, weekly means were plotted for the entire duration of observations as shown in Fig. 1. The error bars are standard deviations. As the observations were made mainly during the dry season, the influence of the dust-laden NE trades was anticipated. Vertical arrows indicate the time of first dust-haze occurence as observed by the Met~ro~o~l Service. At Lagos, this harmattan effect was at peak during the latter half of December 1977 to the middle of January 1978 as judged by the latitude of the Inter-Tropical Discontinuity (ITD] deduced for Lagos and plotted in Fig. 1. The ITD veered North and South of Lagos city during this period and would appear to have taken-on a mean latitudinal position of 6.65”N which is very close to Lagos. This situation is coincident with recorded rises in aerosol optical density. For periods other than the peak Harmattan, a ciear-cut identification coutd not

Preliminary solar spectra-photometric

1613

measurements

Table 1. MonMy and overall means of the aerosol optical density for 2 wavelengths and the wavelength exponent in Lagos at two measurement sites during the dry season Site 1 Months

Site 2

W.5 pm)

b(0.88pm)

-01

0.33 0.48 0.57 0.42 0.45

0.19 0.24 0.32 0.25 0.25

0.98 1.23 1.02 0.92 1.04

November 1977 December 1977 January 1978 February 1978 Over all means

be made about the aerosol sources (continental or maritime) due largely to the paucity of supporting

Meteorological data (wind direction of the upper air, for example). This fact, as well as the rather weak circulations typical of tropical situations, ‘prompted the analysis on a weekly basis. The wavelength exponent, a, reaches a maximum early in Harmattan season and decreases during the peak of season. Considering overall means, - a = 1.l for Lagos during this season compared with 0.6 deducible from Cerf s (1976) results in Ouagadougou. This indicates that particle sizes are likely smaller in Lagos than at Ouagadougou. Relative contributions of sea salt and mineral dust to agosol at Lagos, however, still have to be determined. Figures 2(a) and (b) indicate that in most cases the aerosol optical density increases in the forenoon. In a few cases [HO.5 rmr) and b(0.88 run) for October and NO.88 m) for January] a slight decrease is observ& in the forenoon followed by a post-noon enhancement. October curves have bacn dashed to emphasize that only the later half of the month was covered. It is not clear whether the peculiar October results are representative of transition months. Randerson (1973) had reported the tendency for turbidity b (0.5 pm) to in&ease in the afternoon. He associated the phenomenon somewhat with instrumental limitation and the shift in the wavelength of maximum intensity of solar radiation as m gets iarge (m 2 3). Bullrich (in Bullrich and Randerson, 1973) had further indicated a possible

W.SW

0.29 0.48 0.61 0.48 0.47

b(0.88jm)

-K

0.14 0.24 0.33 0.29 0.25

1.29 1.23 1.09 0.89 1.13

contribution of forward scatter to this phenomenon. The diurnal variations observed for b(0.88 pm) in the results presented here appear more complex than for b(0.5 pm) when considered from month to month. In addition, the diurnal variations are much smaller than the month-to-month variations. The effect of a shift of the wavelength of maximum solar intensity with m (Randerson, 1973) may still be hidden in these diurnal variations. The effect of forward scatter is also not obvious. Supplementary data on particle sizes should yield useful information on the latter. Work is in hand to obtain these data. Generally, it will be difficult to compare these diurnal variations with other data based on the measurement of b(0.5 pm) only. Therefore, there is a need for substantial additional multiwavelength data, possibly over a long period, in order to form a reasonable understanding of these variations. Using the observations in the Harmattan pe& when aerosols were most likely to derive from the same source, data were fitted by the least squares to produce the curves in Fig. 3. The observed aerosol optical density and the optical air mass can then be related as b(0.5 pm) = 0.506 - O.OOlm + 0.321 e-“’ and b(0.88 pm) = 0.305 - O.OOlm + 0.215 e-” (1 I m s 3.2). -

1.3

-I

8

0.6

0.6 x^ 2; 0.4

0.0

A 0.7

25 Oct.

30

IO

20

30

Nov. (1977)

IO

I 20 Irmc. I !

30

Hormattan

IO

pmk

I 20 , Jan. I

30

10

20

Feb. (1918)

j

Fig. 1. Variation of the mean aerosol optical density for 0.5 and 0.88pm wavelengths, the wavelength exponent K and the Latitude of the Inter-Tropical Discontinuity during the 1977j78 dry season in Lagos.

1614

oec.

%

Nov. I

---

0.2

t-

--

Oct.

OCL

(b)

(0)

t 0900

I500 "0900

1200

1200

I500

Hn. LMT, GmT + I

Fig. 2. Daytime variations of aerosol optical density at 0.5 and 0.88pm, (a) and (b) respectively.

CONCLUSIONS

0.0

1

/

1

J

Fig. 3. Aerosol optical density b(i) vs optical air mass (m) : (a) for 0.5 pm wavelength, (b) for 0.88 q~ wavelength. Data for the 1977/78 Harmattan peak (16 December 1977-15 January 1978) in Lagos.

A variance analysis was performed on the curves using the F-test. The ratio of variance about the regression to that due to the regression was 8.2 for the b(0.5 pm) fit and 1.Ofor the b(0.88 pm) curve for 12 and 1 degrees of freedom. The 10,5 and 1% F-values (from Tables) were 3.18, 4.75 and 9.33 respectively, indicating that the ratio was just insignificant for the b(0.5 pm) fit and more insignificant for the b(0.88 pm) fit. These suggest that the fits are likely to be rather fair representations of the data.

Preliminary measurements of the aerosol optical density, using the solar spectrophotometric technique in dry dusty weather, have shown that the aerosol optical density at O.Yand 0.88 m wavelengths were similar near a Lagos industrial estate and beside the Lagos lagoon, some 6 km SE of the former site. Both the aerosol optical density measurements and the ITD statistics suggest that the values of the spectral aerosol extinctions during this seasons are due, in part, to dust of non-local origin; a Saharan-derived component is very probabk. The average particle size inferred from the results are smaller than the sizes near the Sahara desert (Ouagadougou). This further indicates a removal, en-route, of giant and very large particles from the Saharan aerosol body, possibly by sedimentation and turbulence, leaving only thesmaller subranges to reach the Southernmost parts of the West African Coast under appropriate meteorological conditions, such as that determined by the mean ITD position. In the majority of cases, the spectral aerosol optical density acquired maximum values in the forenoon. As a whole, the day-time pattern of aerosol production is still imperfectly understood. Acknowledgements - This work forms part of an ongoing boundary-layer structure study programme being done under the auspices of the University of Lagos, Lagos, Nigeria. I wish to thank Dr. F. E. Volz of the Air-Force Cambridge Research Laboratories, Bedford and both Dr. Vernon Derr and Mr. Edward Flowers of the Wave Propagation Laboratory, Boulder, Colorado, U.S.A. for useful discussions. Finally, the cooperation of the Meteorological department, Federal Ministry of Civil Aviation, Lagos, is hereby acknowledged. REFERENCES

Angstrom A. (1929) On the atmospheric transmission of sun radiation and dust in air. Geograph. Ann. II, 156-166. Bullrich K. (1964) Scattered radiation in the atmosphere and the natural aerosol. Adv. Geophys. 10, 101-257.

Preliminary solar spectra-photometric Bullrich K. and Randerson D. (1973) Discussion: A summary of hourly turbidity measurements for Las Vegas, Nevada, during fall months. Atmospheric Environment 7,665~666. Ccrf A. (1976) Atmospheric turbidity in West Africa. Proc. Symp. on Radiation in the Atmosphere, pp. 16-17 (Edited by H. J. Bolk). Science Press, Princeton. Flowers E. C., McCormick R. A. and Kurfis K. R. (1969) Atmospheric turbidity over the United States, 1961-1966: J. Appl. Met. g, 955-962. Hamilton R. A. and Arcbibold J. W. (1945) Meteorology of Nigeria and adjacentterritory. Q. JI R. Ner. Sue. 71, 231-265. Jaenicke R. (1976) Continental aerosol 0ve.r the ocean. Proc. Symp. on Ra&ation in the Atmosphere, pp. 29-31 (Edited by H. J. Balk). Sciencz Press, Princeton. Junge C. E. (1963) Air Chemistry and Radiooctiuity. Academic Press. New York.

measurements

1615

McCartney H. A. (1976) Spectral measurements of aerosol extinction of direct solar radiation in the wavcband 400-800 mn. Proc. Symp. on Radiation in the Atmosphere, pp. 94-96 (Edited by H. J. Bolle). Science Press, Prinatoo. McKeown H. D. J. (1958) Dust concentration in the Harmattan. Q. JI R. Met. Sot. 84280-282 Randerson D. (1973) A summary of turbidity measuremctlts for Las Vegas, Nevada, during fall months, Atmospheric Environment 7,271-279. Sbutz L. (1976) Saharan dust transport in the NE trade wind region over the North Atlantic Ocean. Pree. Symp. on Racy in tke Atmospkere, pp. 68-70 (Edited by H. J. Balk). Science Press, Princeton. Volz F. E. (1974) Economical mu&spectral sun pbotometer for measurements of aerosol extinction from O.Mpm to 1.6pm and precipitable water. Appl. Opt. 13, 1732-1733.