In this chapter, the results of the derived aerosol bulk properties the optical depth, the single scattering albedo, the aerosol absorption, the ratio of absorption to backscatter, the precipitable water, and the vegetetaion index from spectral airborne measurements are presented and analysed with the purpose to show the potential and accuracy achievable with the newly developed spectral radiometer system. For more details on how the aerosol bulk properties are derived see Bannehr (1990).
The optical depth is one of the most often analysed parameters of the extinction processes for the spectral direct beam irradiance. It is a measure for the column extinction or flux divergence and is the vertically integrated volume extinction coefficient.
Figure 6.1 illustrates the profile of the total optical depth tau_tot(lambda) at three wavelengths on all observation days. The total optical depth tau_tot(lambda) as presented in the Figure 6.1 is the optical depth from a pressure level n to the top of the atmosphere. The optical depth for a layer, delta_tau_tot(lambda), is determined by subtracting tau_tot(lambda) at the lower level n from the higher level n+1 and is independent of the calibration.
Figure 6.1 shows that tau_tot(lambda) is governed mainly by molecular scattering. At 500 nm wavelength the molecular scattering component at ground level is about two to three times larger than the extinction by aerosol particles. Ozone absorption at this wavelength is small (tau_0(500)~ 0.009) and does not contribute much to the total optical depth. With increasing wavelengths, the molecular scattering decreases proportional to ~lambda E-4 and features of the aerosol extinction are more pronounced at 780 nm and 860 nm.
As an indicator of the variation of aerosol content in the atmosphere, it is evident from Figure 6.1 that the total optical depth, çtot(þ), is of limited usefulness. A better parameter is the aerosol optical depth, ça, which is calculated by subtracting the molecular and gas optical depths from the total optical depth. Three wavelengths were chosen to calculate this parameter, one of which is 500 nm, (recognised as the default standard wavelength for measuring the atmospheric extinction). At this wavelength molecular scattering, aerosol extinction, and ozone absorption by the broad Chappuis band occurs. The effect of variations in the ozone mass are relatively small throughout the year and the ozone mass can be approximated by an annual cycle. The values are given by Funk and Garnham (1962). The other two wavelengths to calculate the aerosol optical depth ça are centred at 780 nm and 860 nm. At these wavelengths, only molecular scattering and aerosol extinction are present.
Figure 6.2 shows the aerosol optical depth derived for the three wavelenghts of 500, 780 and 860 nm from aircraft data collected on 3 April 1989 and 22 April 1989 over Lake Alexandrina and on the two consecutive days of the Hincks Park experiment. The corresponding standard deviations are illustrated in the same figures by the symbols on either side of the mean values. The aerosol optical depth of a layer is obtained by subtracting the values between two levels of ça. As þçtot(þ) is independent of the calibration, so is the aerosol optical depth for those layers between the top flight level and the ground independent of the calibration factor. For all wavelengths the aerosol optical depth is between ça=0.05 and ça=0.09 at the lowest flight level (approximately 30 m height). This corresponds to a very low aerosol concentration in the atmosphere, but is typical for semi-arid regions of South Australia. The features are similar for all wavelengths and the shape of the profiles are found to be closely related to boundary layer conditions. In Figure 6.2, the highest data points of each data set correspond to the flight level above the inversion. The observations indicate that about 65% of the aerosol loading is trapped below the inversion. This again coincides with the observed precipitable water with 67% of the water vapour being below the inversion height.
Figure 6.3 illustrates the aerosol absorption profiles for the three wavelengths 500, 780 and 860 nm on all observation days. The aerosol absorption is given as a percentage of the total available flux. In order to illustrate the variation during the flights, the standard deviation is included in the diagram. Looking at all three diagrams, a slight increase in the overall absorption and its variability with increasing wavelength is indicated. For the 500 nm wavelength the average aerosol absorption derived varies between 1.3% and 2.8% on average for the highest flight level (always above the inversion). For the lowest flight level, which was always about 30 m above ground, it varies much more and is between 3.4% and 6.7%. Similar percentages were inferred by Pascoe (1986), also on the Eyre Peninsula. Forgan (1979) calculated the aerosol absorption from broadband pyranometer measurements. His numbers lie between 1% to 3% for Adelaide (South Australia). Ball and Robinson (1982) stated a slightly higher absorption (7.5%) for central USA, which compares well with the 780 nm and 860 nm wavelengths.
It can be clearly seen that the total aerosol absorption is higher in autumn during the Lake Alexandrina field exercise than during the Hincks Park experiment in summer. Whether the differences are induced by an annual cycle caused by the cyclonic pattern, a local feature, or simply by a particular weather situation cannot be established because of a lack of sufficient data. However, for all observations and wavelengths it is obvious that very little or no absorption occured in the lowest layer close to the ground. For the 780 and 860 nm wavelengths the aerosol absorption is even negative, implying the limits of the accuracy achievable with the current instruments and method. Interesting is the primary aerosol absorption which is always found in the highest layer for all profiles except for 22 April 1989 where the absorption is uniform throughout the boundary layer. Noteworthy are the profiles of 3 April 1989, where the major absorption layer is found between 818 and 878 hPa. In this thin layer alone 4.7% of a total of 6.7% of the incoming energy are absorbed. The high aerosol absorption in this thin layer is connected with the boundary layer inversion. Underneath the inversion light aerosols are concentrated. The shape of the
aerosol absorption profiles is very similar to the profiles of the aerosol optical depth (Figure 6.2). The differences in the shape of the profiles indicate the variation in the absorption characteristics of the aerosols.
The differences in the aerosol absorption profiles of 4 and 5 December 1989 are caused by the change of the air mass. On 4 December 1989 the high pressure cell west of Tasmania brought, after showers in the Northwest of Australia, moist clean air with a northwest airstream to the Eyre Peninsula. This high pressure cell moved fast towards the East of Tasmania on 5 December 1989 resulting in dry air with winds from northeast.
In this study it is particularly interesting that the observed absorbed energy is not strikingly high, but coincides with a very low aerosol optical depth (Figure 6.2). This indicates a strongly absorbing aerosol type which is further manifested by a low single scattering albedo and a high ratio of absorption to backscatter as will be shown in Figures 6.4 and Figure 6.5.
The main feature of the calculated single scattering albedo þ0 profiles (Figure 6.4) are the extremely low values at all heights when compared with values found in literature. Only the single scattering albedo as Pascoe (1986) has derived is <0.5 and coincides with present spectral observations.
The profiles of the single scattering albedo for the three wavelengths show apart from the wavelength dependence similar features and therefore allow a general discussion. In Figure 6.4 þ0 decreases before it increases or remains constant with increasing height, suggesting that the aerosol type is different above the boundary layer than below. The profiles also prove that the single scattering albedo is quite variable and not constant throughout the atmosphere. It is a similarly prominent profile such as the ones found for the aerosol optical depth and aerosol
absorption where the major aerosol loading and absorption is observed for the single scattering albedo on 3 April 1989. þ0 increases in the highest layer between 878 and 818 hPa considerably for 500 nm. For the other two wavelengths an increase is visible, too, but it is less pronounced. This increase in scattering in the highest layer might be caused by concentration of various particle types enclosed below the inversion. As already found for the optical depth and aerosol absorption, the þ0 profile is uniform on 22 April 1989. On 4 December 1989 the higher values in the profiles (except for the highest layer at 500 nm) are explained by the moist air mass reaching the observation site from northwest.
Furthermore, it is a remarkable fact that the wavelength dependence is not as evident as in the other profiles. The values obtained during the other three observations below the inversion are close to each other and decrease steadily, suggesting the prevalence of the same aerosol type on all days.
The values of the single scattering albedo for all wavelengths and heights were less than 0.5, which is very low and suggests the presence of highly absorbing particles such as soot, iron, or iron oxide, possibly released into the atmosphere by mining activities in these regions (Twidale et al., 1985). Another highly absorbing particle is hematite, which could be transported southwards from the dry, sparsely vegetated interior of Australia by northerly winds. These highly absorbing aerosol particles, in conjunction with low humidity conditions, prevent the deposition of water vapour onto the aerosol. Consequently, the single scattering albedo þ0 is low and the imaginary part of the refractive index is high.
The ratio of absorption to backscatter derived from airborne observations (see Figure 6.5) shows a general increase with wavelength. In contrast to the other parameters, there is no clear profile shape visible for the ensemble of observations and wavelengths. An attempt to compare the numbers with the results of other work is difficult, because this quantity has
usually been derived from broadband ground observations and also depends on the solar altitude. Therefore, a statement made about the absolute values can only be a general one. To compare the broadband measurements with the spectral observations, the 780 nm wavelength has been chosen. This wavelength matches closely with the wavelength where the solar energy spectrum can be divided in two equal parts in terms of integrated energy. The ratio of absorption to backscatter as inferred by Unsworth and Monteith (1972), Robinson (1966), Wesley and Lipschutz (1976), and Forgan (1979) is in all cases smaller than in the present work. Only Pascoe's (1986) values are similarly high and found to be between >10 and >190, also gathered on the Eyre Peninsula in South Australia. The results of the spectral airborne radiation measurements, therefore, confirm the broadband determinations of Pascoe (1986).
Furthermore, these high absorbing aerosols are not only found on the Eyre Peninsula, but also in the region of Lake Alexandrina 250 km apart from the Eyre Peninsula, suggesting that wide parts of South Australia show the phenomena of high absorbing aerosols.
The ratio of absorption to backscatter increases with wavelength showing a maximum value of all observations at 500, 780 and 860 nm of 32, 109 and 177. The corresponding minimum values for the same wavelengths are 7.5, 11 and 15. These limits show the wide range of the derived parameter.
Since the aerosol absorption (Figure 6.3) is not strikingly high, the high values of the ratio of absorption to backscatter indicate that the backscatter is very low. Differences in the shape of the aerosol absorption profiles (Figure 6.3) and in the ratio of absorption to backscatter profiles reveal the variation in the backscatter, which is obviously not constant during the time of observation and depends on the aerosol absorption properties. On 22 April and 4 December 1989, for instance, where the aerosol absorption (Figure 6.3) was uniform with height, an increase was noticed at the same time in the ratio of absorption to backscatter profile. This can be explained by a decreasing backscatter with height and is confirmed by the decrease of the single scattering albedo. On 3 April 1989 and 5 December 1989, the ratio of absorption to backscatter profiles and the aerosol absorption profiles show a major decrease in the highest layer while the single scattering profiles (Figure 6.4) show an increase. This decrease of the ratio of absorption to backscatter and the increase of the single scattering albedo establishes that in the highest layer the backscatter increases.
Furthermore, this parameter is sensitive to calibration errors and instrumental uncertainties especially under conditions of a clean atmosphere as it was found in the investigated regions of Australia. Thus, extreme values in the observations might reflect the problems involved when deriving this parameter.
By selecting quasi-monochromatic wavelengths inside and outside of a water vapour absorption band, the amount of water vapour available in the atmosphere can be calculated.
In the current study, the observations were taken in the þåç water vapour absorption band at 820 nm. Additional readings to the left and right of the water vapour absorption band were also determined in order to calculate the precipitable water. The wavelengths chosen were centered at 780 nm and 860 nm. Assuming that the average aerosol optical depth çaer over the two side bands results in the aerosol optical depth at 820 nm, the precipitable water ppw can be derived.
The values for the precipitable water derived from aircraft and radiosonde ascents for all layers and on all days give correlation coefficients of r=0.9989. The difference in the results between the two independent methods is less than 0.050 cm for one layer. Figure 6.6 shows the relationship between the precipitable water computed from airborne radiometer data and that calculated from the aerological aircraft and radiosonde ascents performed in the same areas in which the field excercises took place. The data were gathered on 22 April 1989 and on the 4 and 5 December 1989.
Figure 6.6 Precipitable water of the spectral radiometer versus the precipitable water derived from aircraft ascents.
Figure 6.7 shows the profile of precipitable water derived from radiometer data from the flights on 22 April 1989 and 4 and 5 December 1989 above Lake Alexandrina and Eyre Peninsula. The numbers are the averages along each flight level, and the corresponding standard deviation is also shown. The precipitable water given in Figure 6.7 corresponds to the integrated amount from a given level to the top of the atmosphere, so that the ppw for a layer can be evaluated by subtracting the ppw of the lower level from the higher one. An interesting feature on all observation days is the variability of the water vapour with height. While the total amount of precipitable water (represented by the numbers close to the ground) on the measured days remains relatively unchanged, the fluctuations within different layers are quite large, and are closely connected to the atmospheric stratification. On all surveyed days, the highest traverses were above the inversion layer, where the amount of water vapour was almost constant for all observations. Calculations reveal that 67% of the total available precipitable water occured below the inversion height. The standard deviations shown in the graph decrease with increasing height for all observations. This feature can be explained by the fact that the variability of ppw with horizontal location is more pronounced near the ground.
Figures 6.7 Precipitable water profile.
In addition to the determination of spectral aerosol properties and the column amount of water vapour in the atmosphere, the spectral radiometer system can also be employed for airborne investigations of the vegetation cover, which is often expressed by the Normalized Difference Vegetation Index (NDVI).
The NDVI is a measure of the physiological activity of plants and is used to characterize the state of the vegetation. Plants show a low reflectance up to a wavelength of about 680 nm. The high absorption of short wavelength energy is used in the plant for photosynthesis. For wavelengths between 680 nm and 750 nm, the reflectance increases rapidly by three to four times, whereas for longer wavelengths the increase is small. In the near infrared the high reflectance avoids an overheating of the plant. Ground that is not covered by any vegetation does not show these characteristics. The increase with the wavelength is almost linear and the slope depends on the soil type and moisture of the surface. By using these spectral characteristics of plants, the state of the vegetation is mostly classified from data of the first two channels of the AVHRR of the NOAA polar orbiting satellite.
The recently designed spectral radiometers were employed to measure the physiological activity of plants from the FIAMS research aircraft. The NDVI calculated from low flight observations is a very useful indicator for investigating the influence of mesoscale changes of vegetation types on mirco meteorological processes.
The radiometer SPECG2 was mounted under the wing of the research vessel so that it is always in nadir position during airborne observations. First tests were carried out using the 500 and 860 nm wavelengths. Only for high flights the signals at these two wavelengths are affected by Rayleigh scattering, Mie scattering, and absorption. By selecting these wavelengths, the influence of the atmospheric water vapour on the NDVI is excluded. The chosen opening angle of the radiometer was 4° so that it coincides with the opening angle of
the radiometrically measured surface temperature. Figure 6.8 shows an example of the surface temperature, the spectral vegetation index and the signal of the two channels computed for a flight from Goolwa to Strathalbyn (South Australia) at 1500 m on 22 April 1989. At this height the ground area "seen" by the instrument is 104 m in diameter. As mentioned earlier, only the highest flight can be analysed, because of malfunction of the radiometer during the lowest flight.
In the Lake Alexandrina region several creeks run into the Goolwa River through dense areas of reed. The high and low values of the surface temperature give some information about the underlying ground and are often used as an indicator for studies of micrometeorological processes. Under clear skies, green vegetation has a lower temperature than non vegetated surfaces. However, Figure 6.8 shows that the surface temperature does not enable any differntiation between water and green vegetation. With the use of the spectral radiometer system the ground cover can be better resolved. Downspikes of the vegetation index in the time series indicate water, and upspikes show green vegetation. Mark 1 shows the Currency Creek region where dense reeds are to be found at the banks of the creek. The central channel of the creek is made clearly visible by a wide region of low surface temperature. Dense areas of reeds are located by a high NDVI at the banks of the creek and also in the centre. Two narrow spikes of low NDVI indicate clear water sections. These reed areas and narrow arms of water are clearly marked on topographical maps of the area. The wide surface temperature depression in the middle (mark 2) of Figure 6.8 indicates the position of the region known as the Black Swamp, which has no open ground cover in the region of the Finnis River where, again, some open water arms are insperesed with regions of dense reeds.
Back to Table of Contents
Next
chapter