Journal of Geophysical Research: Atmospheres

Airborne observations of dry particle absorption and scattering properties over the northern Indian Ocean

Authors


Abstract

[1] Airborne measurements of the absorption and scattering coefficients of dry aerosols, at 525 and 545 nm, respectively, were performed over the northern Indian Ocean during the Indian Ocean Experiment (INDOEX). In the marine boundary layer the dry absorption coefficient decreased from about 25 Mm−1 close the Indian subcontinent to 2 Mm−1 close to the Intertropical Convergence Zone (ITCZ), which reflects the removal and dilution of anthropogenically produced aerosols during their transport over the Indian Ocean. The dry scattering coefficient initially decreased with distance from the Indian subcontinent, from 70 Mm−1 at 8°N to 10 Mm−1 at 2°N, resulting in a dry single scattering albedo between 0.6 and 0.8 (0.7–0.9 at ambient relative humidity). At further distance from the Indian subcontinent the scattering coefficient and single scattering albedo increased, indicating the increasing importance of aerosols of natural origin. Using Mie theory, the refractive index of the anthropogenically produced aerosols has been estimated to be 1.50–0.06i.

1. Introduction

[2] Sulfate aerosols are generally thought to cause a negative radiative forcing at the top of the atmosphere (TOA), directly due to scattering of solar radiation, and indirectly due to changes in the prevalence, lifetime, extend and albedo of clouds, which are a result of enhanced cloud droplet concentrations and reduced precipitation [Intergovernmental Panel on Climate Change (IPCC), 2001]. This negative radiative forcing cools Earth's surface and the lower atmosphere. Sulfate aerosols may thereby partly counteract the positive radiative forcing (warming) by greenhouse gases.

[3] Absorbing aerosols (e.g., soot), however, exert a positive radiative forcing at the top of the atmosphere, due to the absorption of solar radiation. This causes a cooling of the surface, since less radiation reaches Earth's surface, and a warming of the lower atmosphere. Indirectly, absorbing aerosols can also lead to a surface warming, because the solar heating of the boundary layer can evaporate clouds and allow more solar radiation to reach the surface. The net radiative effect of absorbing aerosols at the surface strongly depends on the cloud fraction, surface albedo, and the vertical aerosol and cloud distribution [Ramanathan et al., 2001a].

[4] Both the direct and indirect effects of absorbing aerosols are important over the northern Indian Ocean, since the ocean is partly covered with a layer of trade wind cumulus clouds embedded in a pollution/haze layer [Satheesh and Ramanathan, 2000; Ackerman et al., 2000]. During wintertime, pollution from south and southeast Asia is transported over the northern Indian Ocean with the northeast trade winds toward the ITCZ. This causes the build-up of an extensive aerosol layer [Lelieveld et al., 2001]. A unique characteristic of this pollution layer is its light-absorbing nature, which reduced the diurnal average surface solar heating of the ocean by as much as 20 Wm−2 during the Indian Ocean Experiment in February–March 1999 [Ramanathan et al., 2001b].

[5] This paper describes the spatial distribution of the absorption and scattering coefficient in the marine boundary layer over the Indian Ocean during the INDOEX experiment. This experiment took place in February/March 1999 from Hulhule airport, the Maldives (4°11′N, 73°31′E). Measurements of the absorption and scattering coefficient on board of the Cessna Citation research aircraft are presented and compared to Mie calculations.

2. Instrumentation

[6] The absorption coefficient measurements were performed using a custom-built, filter-based light absorption photometer similar in operation to a Particle Soot Absorption Photometer (PSAP). The instrument operates at a wavelength of 525 nm and has a sample spot size of 2.85 mm diameter. It uses a Pallflex membrane filter similar to the filter used in the commercially available PSAP instruments. Light which is transmitted through the filter is detected using Siemens DPW21 detectors. The flow is controlled by a critical orifice which generates a flow rate of 1.8 L/min. The instrument is much smaller in size as the commercially available PSAP, and hence has a different geometric setup.

[7] The instrument produces a continuous measurement of absorption by monitoring the change in transmittance across a filter. The change in optical transmission of the filter caused by particle deposition is related to the optical absorption coefficient using Beer's law.

equation image

where A is the area of the sample spot and V the volume of air drawn through the filter during a given time period. I and I0 are the transmissions through the exposed filter and a nonexposed filter, respectively.

[8] The empirical calibration for the commercially available radiance research PSAP instrument has been applied to the raw signal [Bond et al., 1999], which corrects for the absorption of the filter itself and nonlinearities in the response of the instrument as the filter is loaded. Further corrections are made for scattering, which is partly interpreted as absorption by the PSAP instrument, and for the instrument calibration in the laboratory against a reference absorbing aerosol, performed by Bond et al. [1999]. For the scattering correction the observed scattering coefficients on board the Citation aircraft have been used, which were measured using a M903 Radiance Research nephelometer (see below). These scattering coefficients were not corrected for truncation errors.

[9] Bond et al. [1999] determined the uncertainty in the observed absorption coefficient due to the response of the commercially available PSAP to scattering aerosols to be 2% and the uncertainty in the laboratory calibration 20%. Bond et al. [1999] used the uncorrected scattering coefficient measured with a TSI 3563 nephelometer, for correcting the PSAP measurements. This instrument has a truncation angle of 7–170°. In this study we use uncorrected scattering coefficients measured with a M903 Radiance Research nephelometer, which has a truncation angle of 9–170°. These 2° difference in truncation angle results in a 1.8% higher scattering coefficient measured by the TSI compared to the Radiance Research nephelometer, for the average aerosol size distribution observed in the marine boundary layer during INDOEX. This has been determined using Mie calculations, initialized with the observed average aerosol size distribution in the marine boundary layer during INDOEX and the in this study retrieved aerosol refractive index (see section 5). Adding this 1.8% difference to the 2% uncertainty, mentioned above, the uncertainty of the observed absorption coefficient due to the response of the instrument to scattering aerosols increases to 3%.

[10] The accuracy of the absorption instrument is determined by the uncertainty in the determination of the flow rate and the sample spot size. The sample flow has been measured using a BIOS dry cal flowmeter which has, according to the manufacturer, a maximum uncertainty of 2%. The accuracy of the measurement of the sample spot size has been estimated to be 5%.

[11] The detection limit has been determined from the noise level of the instrument, which was observed during several flight legs in the middle free troposphere, where the absorption coefficient was close to zero. The 2-sigma noise level, or detection limit, during these periods was 0.8 Mm−1 (1 Mm = 106 m), which gives an uncertainty or precision of 6.7% for the average absorption coefficient observed in the marine boundary layer during INDOEX. Hence the total uncertainty of the absorption coefficient measurements in the INDOEX marine boundary layer, which has been determined by the root of the sum of the squares of the individual uncertainties, was 22%. All values are within 95% confidence limits.

[12] The absorption instrument was connected to a backward facing 1/4″ stainless steel inlet, with a total flow rate of 5.2 L/min, allowing only aerosols smaller than about 1 μm diameter to be sampled [Schröder and Ström, 1997]. Using Mie calculations, we determined that this cut-off diameter caused an underestimation of the absorption coefficient in the marine boundary layer of 5–10%, compared to the absorption coefficient when the whole size distribution was considered. The Mie calculations were initialized with the observed particle size distributions, which include particles with a size between 0.010 and 3 μm diameter, and the in this study retrieved aerosol refractive index. The Mie calculations are described in section 5. All marine boundary layer data shown in this paper has been corrected for this underestimation.

[13] Measurements of the aerosol scattering coefficient were performed using a single-wavelength nephelometer (M903, Radiance Research, Seattle, Washington), which operates at a wavelength of 545 nm [Formenti et al., 2001]. The uncertainty in the scattering coefficient measurements due to calibration drifts and lambertian nonidealities of the internal lamp is estimated to be within 10% [Anderson and Ogren, 1998].

[14] The nephelometer was connected to a forward facing, near-isokinetic inlet, which has an upper cut-off diameter of about 3 μm and a total flow rate of 10.5 L/min. The nephelometer operated at a flow rate of 2–3 L/min. Although the optical particle counter, which was connected to the same inlet, detected particles with a diameter up to 3 μm, the smooth 90° bend in the 1/4″ tubing leading toward the nephelometer may have caused super micron particle losses in the sampling line.

[15] Because of geometrical limitations, the sampling volume of the nephelometer is reduced to scattering angles between 9 and 170°. Using Mie calculations, we determined that this truncation caused an underestimation of the scattering coefficient of about 5% in the marine boundary layer. For these calculations, the model was initialized with the observed particle size distributions and the in this study retrieved aerosol refractive index (see section 5). All marine boundary layer data shown in this paper have been corrected for truncation errors.

[16] Furthermore, the total aerosol mass is determined from the observed aerosol size distribution, which is measured using a combination of two condensation particle counters, a differential mobility analyzer and an optical particle counter. A description of these instruments and the observed aerosol size distributions during INDOEX is given in the work by de Reus et al. [2001]. Uncertainties in the measurement of the particle diameter contribute most to the overall uncertainty of the aerosol volume and mass derivation, which are estimated to be 50% [Curtius et al., 2001].

[17] Due to the high temperatures inside the aircraft, the sample air is heated when it is drawn into the aircraft, which reduces the relative humidity. The temperature increase of the sample air in the marine boundary layer was between 10 and 15 °C, which caused the relative humidity to decrease from the ambient value of about 80% to below 50%. Hence the measured properties in this paper represent (near) dry conditions.

3. Observations

[18] In this section, measurements of the dry volume absorption and scattering coefficient on board of the Cessna Citation research aircraft are presented. The observed parameters are averaged or integrated over the time that the aircraft flew on an isobaric flight level, lasting for at least 10 min, which corresponds to a spatial distance of about 100 km. Due to the changing humidity of the filter in the absorption instrument through rapid changes in ambient relative humidity, when entering or leaving the marine boundary layer, the absorption coefficient measurement was not stable during the first five minutes of an isobaric flight level. Therefore the presented absorption coefficients are integrated over the last 5 min of each isobaric flight level. During the INDOEX experiment 160 isobaric flight levels were flown. On about 80% of them absorption measurements were performed.

[19] Figure 1 shows the altitude distribution of the absorption and scattering coefficients for the entire INDOEX period. A deep aerosol layer was observed, extending from the surface to about 4 km altitude. This deep aerosol layer was observed during most of the individual flights. During 3 (out of 21) flights an elevated aerosol layer with high scattering coefficients, exceeding the boundary layer values, at altitudes between 1 and 3.5 km was observed, which was a common observation on the C130 aircraft over the northern Indian Ocean during INDOEX [Ramanathan et al., 2001b]. This so-called residual continental boundary layer, results from the advection of continental air masses, originating from the Indian subcontinent, over the relatively cold ocean [Mayol-Bracero et al., 2002]. The less frequent observation of this elevated aerosol layer in our data set, might be due to the relatively smaller fraction of time that our aircraft flew near the Indian subcontinent.

Figure 1.

Altitude distribution of the dry absorption (σabs) and scattering coefficient (σscat). Each data point represents the average over an isobaric flight level.

[20] The highest absorption and scattering coefficients were observed in the marine boundary layer, where the absorption coefficient ranged between 2 and 25 Mm−1 and the scattering coefficient between 5 and 80 Mm−1. In the residual boundary layer, between 1 and 3.5 km altitude, the absorption and scattering coefficients were on average 5.0 and 9.4 Mm−1, respectively. Above 4 km altitude, the scattering coefficient gradually decreased from 0.5 to 0.1 Mm−1 while the absorption coefficient was mainly below the detection limit of 0.8 Mm−1.

[21] Since a significant part of the extinction takes place in the lowest part of the atmosphere, we will focus on the spatial distribution of the absorption and scattering coefficient in the marine boundary layer (below 1 km altitude). The north east trade winds transport pollution from the Indian subcontinent and southeast Asia over the Indian Ocean toward the ITCZ. Therefore we will study the data as a function of latitude to determine the change in aerosol properties with distance from the Indian subcontinent. Note that the ITCZ was located south of 5°S during the entire measurement period, hence only measurements north of the ITCZ, in the meteorological northern hemisphere, are reported in this study.

[22] Figure 2 shows the observed absorption and scattering coefficients for all isobaric flight legs in the marine boundary layer, which were generally performed at 150 m above sea level. The data are divided into two different time periods, February and March, since the air mass origin in the marine boundary layer changed from the Bay of Bengal to the Arabian Sea in the beginning of March [Verver et al., 2001; de Gouw et al., 2001].

Figure 2.

Latitude gradient of the dry absorption (σabs) and scattering coefficient (σscat) and dry single scattering albedo (ω0). The filled symbols mark the beginning of the measurement campaign (February 1999), the open symbols the end (March 1999).

[23] The absorption coefficient gradually decreased with distance from the Indian subcontinent. In February the absorption coefficient decreased from 25 Mm−1 close to the Indian coast to about 5 Mm−1 close to the ITCZ. In March these values were lower, between 15 and 2 Mm−1. This corresponds to the trend in total aerosol loading, which gradually decreased with distance from the Indian subcontinent and which also showed lower concentrations in air masses originating from the Arabian Sea (March), compared to air masses from the Bay of Bengal (February, see Figure 3).

Figure 3.

Latitude gradient of the dry absorbing and total aerosol mass and the black carbon fraction of the aerosols. The filled symbols mark the beginning of the measurement campaign (February 1999), the open symbols the end (March 1999).

[24] The scattering coefficient ranged between 5 and 70 Mm−1. Between 8°N and 2°N, the scattering coefficient decreased strongly, while further south, it remained relatively constant or increased slightly. The scattering coefficients were also generally higher in air masses originating from the Bay of Bengal (February) than in air masses from the Arabian Sea (March).

[25] From the measurements, described above, the aerosol single scattering albedo (ω0) can be calculated. The single scattering albedo is defined as the ratio of the aerosol light scattering coefficient to the sum of the scattering and absorption coefficient (ω0 = σscat/(σscat + σabs)). The single scattering albedo was generally very low, indicating strongly absorbing aerosols. Down to 2°N the single scattering albedo varied between 0.6 and 0.7. Further south, higher values, up to 0.86 at 4°S, were observed. Note that we corrected the absorption coefficient measurements for the cut off diameter of the inlet, so that the absorption and scattering coefficient measurements, and hence also the single scattering albedo is valid for particles with a size up to 3 μm diameter.

[26] Due to the high temperatures inside the aircraft, the sample air is heated when it is drawn into the aircraft, which reduces the relative humidity. The presented absorption and scattering coefficients are, hence, representative for dry conditions. The scattering coefficient of hygroscopic aerosols is strongly dependent on relative humidity, since hygroscopic aerosols will grow to larger sizes at high relative humidity. This causes the scattering coefficient, and consequently the single scattering albedo, to increase. [Öström and Noone, 2000].

[27] The hygroscopic growth of the marine boundary layer aerosols over the northern Indian Ocean during INDOEX has been studied by Sheridan et al. [2002]. They determined a scattering-based growth factor of 1.58, which is defined as the ratio between the scattering coefficient at 85% relative humidity and the scattering coefficient at 40% relative humidity (σscat,wetscat,dry). Since those conditions are close to the instrument and ambient relative humidity for our measurements, the scattering coefficient at ambient relative humidity is a factor 1.58 larger than the values shown in Figure 2. The ambient scattering coefficient in the marine boundary layer, therefore, ranged from 8 to 130 Mm−1, which resulted in an ambient single scattering albedo between 0.7 and 0.9.

[28] The absorption coefficient is not adjusted for ambient relative humidity, since the dependence on aerosol light absorption is not known, and usually considered to be minor [Hartley et al., 2000]. The mean aerosol radiative properties for dry and wet conditions are shown in Table 1.

Table 1. Average Radiative Properties and Standard Deviation for the Northern Indian Ocean Marine Boundary Layer Aerosols (0–1 km)a
 At instrument RHAt 85% RH
  • a

    The values at 85% relative humidity are calculated using a hygroscopic growth factor of 1.58. Note that the measurements were generally performed at 150 m altitude, and are valid for particle sizes between 0.010 and 3 μm.

σscat (Mm−1)30 ± 1847 ± 28
σabs (Mm−1)11 ± 611 ± 6
ωo0.73 ± 0.090.80 ± 0.07

[29] From the observed absorption coefficient, we calculated the absorbing aerosol mass concentration assuming a specific absorption of black carbon of 8.1 m2g−1 [Mayol-Bracero et al., 2002], which has been determined from chemical aerosol analysis during INDOEX. The total aerosol mass concentration is determined from the observed aerosol size distribution, assuming an aerosol density of 1.7 g cm−3. Using the total and absorbing aerosol mass, the black carbon fraction of the aerosol could be determined, which is displayed in Figure 3. Both the absorbing and total aerosol mass decreased with distance from the Indian subcontinent, resulting in a relative constant black carbon fraction of about 13% in February and 10% in March, over the northern Indian Ocean. This is comparable to the elemental carbon content of 14%, determined from elemental and organic carbon analysis on filter samples, collected in the marine boundary layer during INDOEX [Mayol-Bracero et al., 2002].

4. Comparison With Other INDOEX Platforms

[30] In order to compare the observed scattering and absorption coefficients with those measured on other INDOEX platforms we follow the air mass categories defined by Clarke et al. [2002]. They differentiated between low (σscat < 25 Mm−1), medium (25 < σscat < 55 Mm−1) and high (σscat > 55 Mm−1) aerosol scattering regimes, in order to account for the variability in the data due to the different measurement positions and sample periods of the various platforms. The mean absorption and scattering coefficients and standard deviations for the three air mass categories are given in Table 2. Our measurements agree within one standard deviation of the mean with coefficients determined from the C130 aircraft and the Kaashidhoo climate observatory (KCO), however, the single scattering albedo tends to be lower in our measurements. The Ron Brown research vessel shows lower values for the absorption coefficient in the middle and low scattering regime, which can be explained by the more time that the Ron Brown spent close to the ITCZ in relatively clean air masses [Clarke et al., 2002].

Table 2. Average Dry Radiative Properties and Standard Deviation for N Indian Ocean Marine Boundary Layer Aerosols (0–1 km) for Three Dry Scattering Coefficient Regimes, Defined by Clarke et al. [2002]
 Low, σscat < 25 Mm−1Medium, 25 Mm−1 < σscat < 55 Mm−1High, σscat > 55 Mm−1
σscat (Mm−1)13.5 ± 6.338.4 ± 8.169.6 ± 10.4
σabs (Mm−1)6.4 ± 3.815.5 ± 5.817.2 ± 8.1
ωo0.71 ± 0.120.72 ± 0.070.80 ± 0.10

[31] Using the data from two intercomparison flights, which were performed during INDOEX, the measurements onboard the Citation could directly be compared with the measurements onboard the C130 aircraft. However, differences in altitude (up to 100 m) between the two aircrafts caused a large difference in the observed wind direction on both aircrafts (up to 60°), which makes the comparability of the measurements questionable. During the two intercomparison flights, one isobaric flight level was flown in the marine boundary layer. During this period, the scattering coefficient measured onboard the C130 was a factor 2.1 and the absorption coefficient a factor 1.4 larger than the scattering and absorption coefficients measured onboard the Citation aircraft. One reason for this difference could be the difference in airspeed of the two aircraft, which forced the Citation aircraft to fly relatively slow, which may have caused nonisokinetic sampling, leading to an undersampling of larger particles and subsequently to an underestimation of the scattering and absorption coefficient. Moreover, a comparison between the in-situ absorption and scattering coefficient measurements in the marine boundary layer onboard the C130 aircraft and other INDOEX platforms shows that higher scattering and absorption coefficients were observed onboard the C130 compared to the other platforms [Clarke et al., 2002].

[32] A comparison with data obtained with a lidar instrument, which was situated at Hulhule airport during INDOEX [Müller et al., 2001a, 2001b] was performed on 18 February 1999. Backward trajectory analysis showed the advection of air masses from Southeast Asia across the Bay of Bengal at altitudes above 1 km. Below, the air masses originated from India after which they also crossed the Bay of Bengal. After sunset, the aircraft made a slow descend through the observation path of the lidar. The extinction coefficient at 532 nm, measured by the lidar and the extinction coefficient at 525–545 nm, obtained from the scattering and absorption measurements onboard the aircraft are shown in Figure 4a. The lidar extinction profile represents a 1-hour sequence of observations, only including those individual profiles, which passed cloud screening [Müller et al., 2001b]. The signals for the extinction coefficients were smoothed with a gliding average of 600 m. The uncertainty of the lidar extinction measurements was about 20%. The retrieval procedure for the extinction profiles is described by Müller et al. [2001a]. The uncertainty of the extinction coefficient derived from the aircraft measurements has been assessed by the root of the sum of the squares of the uncertainty in the scattering and absorption coefficient measurements, and was 24%.

Figure 4.

(a) Intercomparison between the lidar and measurements onboard the Citation aircraft. The solid line shows the extinction coefficient at 532 nm measured by the lidar, the markers the scattering (asterisks), absorption (crosses) and extinction coefficients (filled squares), derived from the aircraft measurements. (b) Altitude profile of the relative humidity and (c) accumulation mode number concentration (Dp > 0.12 μm) measured onboard the aircraft.

[33] Since the lidar determines the aerosol radiative properties at ambient relative humidity, the observed dry scattering coefficient onboard the aircraft has been corrected for ambient conditions. This has been done using the observed relative humidity during the descend over the lidar, which is shown in Figure 4b, and the scattering based growth factor at different relative humidities, f(RH), which has been observed at the Kaashidhoo Climate Observatory during INDOEX [Clarke et al., 2002]:

equation image

where a = 0.841 and b = −0.368. Since we can only measure the absorption coefficient on isobaric flight levels, no absorption data could be obtained during the descend over the lidar. Moreover, since the scattering coefficient was much lower during the rest of the flight compared to the descend over the lidar, we did not use the absorption coefficient measured on isobaric flight levels during other parts of the flight. Instead, we calculated the absorption coefficient using the observed dry scattering coefficient and the average dry single scattering albedo observed in the marine boundary layer (0.73, see Table 1). Finally, the extinction coefficient has been determined by adding the calculated absorption and wet scattering coefficient.

[34] The altitude profile of the scattering coefficient, measured onboard the aircraft shows a similar pattern as the extinction coefficient profile measured by the lidar, which is supported by the accumulation mode particle number concentration profile (Figure 4c). Although some assumptions had to be made about the humidity correction and the single scattering albedo, the extinction coefficients derived via these two very different methods compare very well. Only at about 3 km altitude, the extinction coefficient derived from the aircraft measurements is slightly higher compared to the extinction coefficient from the lidar. The humidity correction and single scattering albedo are both derived from boundary layer measurements, which certainly induces additional uncertainties in the calculation of the extinction coefficient at higher altitudes. The lidar measurements, for example, indicate an increase in single scattering albedo with altitude [Müller et al., 2001b], which would cause an overestimation of the absorption coefficient at higher altitudes. However, the good agreement between the two measurement techniques suggests that this assumption can be applied for the lower 4 km of the atmosphere. Moreover, from the good agreement between the two instruments, we can conclude that aerosol extinction, above 1 km altitude, is dominated by aerosols smaller than 3 μm diameter.

[35] Extinction profiles below 1 km altitude could not be determined with the lidar because of an incomplete overlap between outgoing laser beam and the field of view of the receiver telescope [Müller et al., 2001a]. The profile of the particle backscatter coefficient at 532 nm, which can be retrieved almost down to the surface, showed a strong increase between 1 km and the surface, which is comparable with the increase in the scattering coefficient observed onboard the aircraft.

5. Model Calculations

[36] Using a Mie extinction code [Bohren and Huffman, 1983] we calculated the scattering and absorption coefficient based on the observed aerosol size distributions and assuming a certain refractive index.

[37] The refractive index that provided the best agreement between the measured and modeled scattering and absorption coefficients appeared to be 1.50–0.06i (see Figure 4). At very low observed absorption coefficients, however, the model overestimates the absorption coefficient. This suggests that it might not be realistic to use a constant refractive index for the entire measurement region. It is very well possible that sea salt aerosols and other biogenically produced aerosols, of less absorbing nature, play a more important role at further distance from the Indian subcontinent, while the total anthropogenic aerosol number concentration decreases in concentration. This would cause a change in the apparent refractive index with latitude, which is not taken into account in the model calculations.

[38] Due to the calibration of the optical particle counter (OPC) with ammonium sulfate particles and latex spheres, an error is made in determining the size of strongly absorbing aerosols by the OPC [Collins et al., 2000]. For the accumulation mode particle size range, where the anthropogenically produced absorbing aerosols dominate, this error is below 10% (P. Guyon, personal communication, 2002). Hence the uncertainty in the modeled absorption and scattering coefficient is estimated by performing Mie calculations including 10% uncertainty in the observed particle size. This results in a 45% uncertainty for the scattering coefficient and 35% for the absorption coefficient. The error bars on the observed absorption and scattering coefficient in Figure 5 are determined by the instrument uncertainty, of 22% and 10%, respectively.

Figure 5.

Comparison between the measured and calculated dry absorption and scattering coefficient. The solid line shows the 1:1 relation between model and measurements.

[39] In the model calculations, presented in Figure 4, the cut-off diameter of 1 μm for the absorption instrument and the reduced sampling volume of the nephelometer are taken into account. The difference in absorption coefficient for the reduced (Dp < 1 μm) and the complete size distribution, and the difference in scattering coefficient for the reduced and the total sampling volume of the nephelometer, calculated by the model, are used to correct the observed absorption and scattering coefficients in the marine boundary layer (Figures 2 and 3).

6. Discussion and Conclusions

[40] In this paper airborne measurements of the dry volume absorption and scattering coefficient in the marine boundary layer over the northern Indian Ocean are presented. The measurements indicate the presence of strongly absorbing aerosols. The absorption coefficient, observed close to the Indian subcontinent, is comparable to values measured over central Europe [Keil et al., 2001] and in different urban areas in the USA and western Europe [Horvath, 1993].

[41] The variability of the absorption and scattering coefficient over the Indian Ocean is high. After dividing the data into three aerosol scattering regimes, the observed absorption and scattering coefficient onboard the Cessna Citation aircraft compare well to those measured on the other INDOEX platforms in the marine boundary layer. Although the single scattering albedo agrees within one standard deviation of the mean with the single scattering albedo measured on the C130 and KCO, it tends to be lower in our measurements. The single scattering albedo at ambient relative humidity, is falls within the range of range single scattering albedos derived from lidar measurements performed from Hulhule airport during INDOEX [Müller et al., 2001a, 2001b]. A detailed comparison of the scattering and absorption measurements onboard the Citation aircraft and the extinction coefficient observed by the lidar instrument shows a very good agreement for the measurements on 18 February 1999. The good agreement between the two instruments, also indicates that that aerosol extinction, above 1 km altitude, is dominated by aerosols smaller than 3 μm diameter.

[42] Based on the optical properties of the aerosols, the northern Indian Ocean region can be divided into two regions. Between 8 and 2°N, the absorption and scattering coefficients gradually decreased with distance from the Indian subcontinent. A similar decrease was observed for the total aerosol number concentration, and attributed to entrainment of free tropospheric air and the removal of particles by precipitation [de Reus et al., 2001]. Anthropogenic aerosols produced in south and southeast Asia dominate the aerosol optical properties in this region.

[43] The concurrent decrease in the absorption and scattering coefficients, between 8°N and 2°N, and the relatively constant single scattering albedo and black carbon content of the aerosols suggest that the aerosols may be internally mixed. This is supported by mass spectrometric particle analysis performed at the Kaashidhoo Climate Observatory on the Maldives during INDOEX. This analysis shows that black carbon particles were always mixed with organics and sulfate [Guazzotti et al., 2001]. A consequence of the presence of internally mixed aerosols compared to externally mixed aerosols is that the removal of absorbing material by wet deposition is facilitated when elemental carbon (hydrophobic) is connected to more soluble substances. Moreover, the absorption efficiency is enhanced, because, in general, internal mixtures yield higher absorption coefficients compared to external mixtures [Horvath, 1993].

[44] South of 2°N, the absorption coefficient decreases further, whereas the scattering coefficient increases slightly with decreasing latitude, which causes the single scattering albedo to increase. In this region, aerosols of different origin, probably marine and biogenic, play a relatively more important role. The absorbing nature of the anthropogenic aerosols is, however, still apparent.

[45] The higher aerosol scattering and absorption coefficients, which were observed in air masses originating from the Bay of Bengal compared to air masses from the Arabian Sea can be explained by a higher aerosol loading. No significant difference in the single scattering albedo was observed.

[46] Using Mie calculations we estimated the refractive index of the aerosols to be 1.50–0.06i. This compares well with results from the lidar measurements during INDOEX. From the lidar measurements a refractive index of 1.58–0.06i has been determined in the aerosol pollution layer between 0.7 and 1.1 km altitude, on 25 March 1999 [Müller et al., 2001b; Wagner et al., 2001].

[47] At large distance from the Indian subcontinent, the absorption coefficient is overestimated by the model, indicating that it is not realistic to use a constant refractive index over the entire northern Indian Ocean, and/or to use a constant refractive index for all particle sizes. The refractive index, that we derived in this study, is, however, representative for the anthropogenic aerosols from south and southeast Asia.

Acknowledgments

[48] We thank the aircraft technicians and pilots from the Technical University of Delft for their support during the field campaign. This work was financially supported by the Swedish Natural Science Research Council (NFR) and by the Max Planck Society (MPG).

Ancillary