Variations in the aerosol optical properties and types over the tropical urban site of Hyderabad, India

Authors


Abstract

[1] Aerosol measurements over the tropical urban site of Hyderabad, India, provide a way of determining the variability of the aerosol characteristics over a duration of 1 year (October 2007 to September 2008). The meteorological pattern over India, mainly driven by the regional monsoons, has a great effect on the amount and size distribution of the aerosols. Higher aerosol optical depth (AOD) values are observed in premonsoon, while the variability of the Ångström exponent (α) seems to be more pronounced, with higher values in winter and premonsoon and lower values in the monsoon periods. The AOD at 500 nm (AOD500) is very large over Hyderabad, varying from 0.46 ± 0.17 in postmonsoon to 0.65 ± 0.22 in premonsoon periods. A discrimination of the different aerosol types over Hyderabad is also attempted using values of AOD500 and α380–870. Such discrimination is rather difficult to interpret since a single aerosol type can partly be identified only under specific conditions (e.g., anthropogenic emissions, biomass burning or dust outbreaks), while the presence of mixed aerosols, without dominance of the coarse or accumulation mode, is the usual situation. According to the analysis the three individual components of differing origin, composition and optical characteristics are (1) an urban/industrial aerosol type composed of aerosols produced locally and all year round by combustion activities in the city or long-range transported (mainly in spring) biomass burning, (2) an aerosol type of mineral origin raised by the wind in the deserts (mainly in premonsoon) or constituting coarse-mode aerosols under high relative humidity conditions mainly in the monsoon period, and (3) an aerosol type with a marine influence under background conditions occurring in monsoon and postmonsoon periods. Nevertheless, the mixed or undetermined aerosol type dominates with percentages varying from 44.3% (premonsoon) to 72.9% (postmonsoon). Spectral AOD and α data are analyzed to obtain information about the adequacy of the simple use of the Ångström exponent for characterizing the aerosols. This is achieved by taking advantage of the spectral variation of lnAOD versus lnλ, the so-called curvature. The results show that the spectral curvature can be effectively used as a tool for aerosol types discrimination, since the fine-mode aerosols exhibit negative curvature, while the coarse-mode particles are positive.

1. Introduction

[2] Aerosols play an important role in the Earth-atmosphere system and can affect general atmospheric circulation patterns [Kristjànsson et al., 2005; Lau et al., 2006] and biochemical cycling [Xin et al., 2005]. However, there is large uncertainty regarding the overall aerosol climatic effect [Morgan et al., 2006; Remer and Kaufman, 2006; Yu et al., 2006]. Because of the variety of aerosol sources, their short atmospheric lifetimes and the dynamic processes that may alter them after generation, their physical and chemical characteristics are highly inhomogeneous in space and time. Therefore, the accurate assessment of the aerosol impact on radiative transfer is a complex task, since various aerosol types cause different effects on the solar radiation spectral distribution [Kaskaoutis and Kambezidis, 2008a]. Indeed, a full description of all the characteristics that control aerosol interaction with solar and terrestrial radiation is needed. Better estimates of the aerosol radiative effects at a planetary scale are required to reduce the uncertainties. To achieve this, it is necessary to obtain a better characterization of the aerosol physical and optical properties [McComiskey et al., 2008; Tafuro et al., 2007]. To understand the effect of aerosols on the Earth-atmosphere system, it is essential to study and understand their physical, chemical, and optical properties [Moorthy et al., 1999; Satheesh et al., 2002]. Therefore, numerous studies have been undertaken to characterize the aerosol properties at several worldwide locations [Dubovik et al., 2002; Pace et al., 2006; Kaskaoutis et al., 2007a; El-Metwally et al., 2008; Ogunjobi et al., 2008].

[3] The atmospheric aerosols may be of natural origin such as the windblown mineral dust or of anthropogenic origin such as those from industry, automobiles or other human activities in urban areas [Kaskaoutis et al., 2007a]. They may originate from gas-phase reactions of low volatile vapors in the atmosphere [Anttila et al., 2007]. The natural aerosols, due to their dominant share (80%), evidently play a vital role in global-scale climate, whereas the anthropogenic aerosols play crucial role in regional-scale climatic features [Ramachandran, 2004]. So, it is required to delineate the effects of natural and anthropogenic aerosols as far as regional features are concerned. El-Metwally et al. [2008] found that Cairo's aerosols are a mixture of three individual components produced by different mechanisms: local urban activities as “background pollution,” biomass burning in the Nile delta as “pollution-like” and wind erosion in Sahara as “dust-like.” Furthermore, Pace et al. [2006] found three individual aerosol types in Lampedusa, mainly consisting of European pollution, Atlantic clean marine and African dust particles.

[4] Aerosol mass burden and the extinction efficiency of aerosols are the most comprehensive variables to determine the AOD. From the spectral AOD, the Ångström formula [Ångström, 1964] is used in order to obtain the turbidity coefficient (β) and the wavelength exponent (α), which give information about the columnar burden and aerosol size, respectively. Recent research has forwarded to other related variables, such as the curvature of the spectral AOD (in the form of a lnAOD versus lnλ plot), which can be used to have some insight in the aerosol size distribution [Eck et al., 1999; Kaskaoutis et al., 2007b]. The curvature can be utilized in conjunction with the AOD for the discrimination of the different aerosol types, and can also be used as an indicator of the relative influence of fine- versus coarse-mode particles in the aerosol size distribution [Schuster et al., 2006].

[5] India is densely populated, industrialized and in the recent years has witnessed an impressive economic development. Aerosols over and around India not only affect the Indian monsoon but also the global climate [Satheesh et al., 2006]. The growing population coupled with revolution in industry has resulted in higher demands for energy and transport. With more and more urbanization the usage pattern of fossil and biofuels are leading to changes in aerosol properties, which may cause changes in precipitation and can decelerate the hydrological cycle [Ramanathan et al., 2001]. Over urban areas of India aerosol emissions from fossil fuels such as coal, petrol and diesel oil dominate [Habib et al., 2006]. Furthermore, the Indian subcontinent exhibits different land characteristics ranging from vegetated areas and forests to semiarid and arid environments and tall mountains. India experiences large seasonal climatic variations, which result in extreme temperatures, rainfall and relative humidity. These meteorological and climatic features introduce large variabilities in aerosol optical and physicochemical characteristics at spatial and temporal scales.

[6] The scope of this work is to analyze the results of Sun photometer measurements performed between 3 October 2007 and 30 September 2008 over the tropical urban site of Hyderabad, India. The AOD measurements performed by Microtops-II several times a day provide a unique opportunity for determining the temporal variability of aerosol characteristics. The 1-year measurements allow to characterize the seasonal pattern of aerosol properties, whose influence on the environment is thought to have an impact at least at regional scale. This seasonal pattern is expected to be important because (1) Hyderabad is a location where aerosols from different origins are mixed and in proportions that depend on the activity of their sources some of which are highly seasonal, (2) of the seasonal variation of the forest fires affecting the area [Badarinath et al., 2007a, 2007c], and (3) of the specific seasonal meteorological pattern driven by the monsoons and the location of the Inter-Tropical Convergence Zone (ITCZ). Previous studies over Hyderabad were mainly focused on examining the variations in black carbon (BC) concentrations, AOD, UV index and solar radiation during short time intervals or under specific events, such as biomass burning or dust transport, by comparing the observations of “normal” and “polluted” days [Latha and Badarinath, 2005; Badarinath et al., 2007b, 2007c, 2008, 2009]. The present study is among the first over Hyderabad focusing on the seasonal pattern of aerosol properties and types and aiming at associating them with local emissions, regional climatology and long-range transport.

2. Regional Characteristics, Meteorology, and Data Set

2.1. Site Description

[7] With its ∼5.5 million inhabitants, Hyderabad (17.47° N, 78.43° E) is the fifth largest city in India, and is also considered as one of the most polluted [Naseema Beegum et al., 2009]. This is a direct result of the growth in population and associated activities that have been observed during the last decades. In addition to this, particles produced by seasonal sources can be transported to Hyderabad at certain times of the year. This is the case of mineral dust produced mainly in premonsoon, though not exclusively, and the mass burning of the crop residues as well as forest fires in the dry period of the year. Because of its potential effects on people's health, the persistence of high levels of particulate concentrations and BC over Hyderabad [Naseema Beegum et al., 2009] is a matter of great concern for its inhabitants. Accurate determination of the aerosol properties is necessary in order to quantify these effects.

2.2. Meteorology Patterns

[8] The well established linkage of dynamic and thermodynamic features, such as the setup of a low-level jet, warm pool and land-ocean thermal gradients influences forthcoming monsoonal features over the Indian subcontinent [Lau et al., 2006]. Hence, besides the dynamic and thermodynamic linkage, the role of aerosols over India is also expected to play a critical role on monsoon features over the region [e.g., Devara et al., 2003] and partly on global climate through the radiation budget.

[9] The mean composite vector wind at 925-mbar level over the Indian region is shown in Figure 1 for the four seasons considered. These maps were obtained from NCEP/NCAR reanalysis and are the mean composites in each examined period, while the wind speed is shown in colored scale. In the winter, the winds over the Indian subcontinent are generally low, exhibiting larger speeds over the oceanic areas. The wind flows mainly from eastern/northeastern directions, carrying significant amounts of polluted air masses over the Arabian Sea (AS) in certain cases. Fair weather conditions with clear skies exist during winter with continental air masses passing over the region [Pandithurai et al., 2007]. Low-level inversions in the morning and evening hours, and haze in the morning occur during this period along with the incursion of dry polar continental air in the wake of low-pressure systems. In general, during premonsoon season (March–May), the weather over and around India is very hot with a daily maximum temperature around 40°C, while the surface winds are mostly gusty, especially over northern AS, northern Indian Ocean (NIO) and Bay of Bengal (BoB). The dust content over northwestern India (Thar Desert) is at its maximum and often dust exposures affect Hyderabad, while cumulonimbus clouds develop around late afternoon to evening hours. Development of low-pressure systems due to increased heating over land start in premonsoon, when all India has the same pressure distribution, with only slightly higher pressure over the AS and the BoB [Pandithurai et al., 2007]. The Indian summer monsoon (June–September) is a part of a large-scale circulation pattern, which develops in response to the thermal gradients between the warm Asian continent in the north and cooler NIO in the south. A strong southwesterly flow in the lower troposphere brings a substantial supply of moisture into India, which is released as precipitation almost across the entire country. In this period, the most strong and gusty winds occur all over the region. Monsoon and aerosol loading in the atmosphere are very intricately related to each other because of the amount and type of the aerosols, which act as cloud condensation nuclei (CCN), together with the available moisture in the atmosphere that determines the amount of rainfall occurring over the region. In the postmonsoon period (October–November 2007) a low-pressure system was developed over South India resulting in moderate winds from northeastern directions over AS and continental India, and in a persistent western flow over NIO.

Figure 1.

NCEP/NCAR reanalysis mean vector wind composites at 925 mbar level for winter (December–February), premonsoon (March–May), monsoon (June–September), and postmonsoon (October–November) seasons during the measurement period (3 October 2007 to 30 September 2008).

[10] Table 1 shows the monthly mean values of ambient air temperature, wind speed and direction, atmospheric pressure, relative humidity (RH) and accumulated precipitation in the period October 2007 to September 2008 over Hyderabad. The air temperature is high varying from 22.0°C ± 4.4°C in January to 31.2°C ± 4.2°C in May. The surface-level wind follows variable directions, from southeasterly in the period October to March to southwesterly in the period April to September. The large standard deviations in each month, and especially in the period October–December, indicate the strong variability in wind direction within these months. Furthermore, the winds are generally low, enhanced in the monsoon period only. During this period the prevailing wind is southwesterly and the aerosols over Hyderabad have a possible influence of marine air masses reaching the region from AS and the Indian Ocean. In the premonsoon, the winds are predominantly westerlies and northwesterlies. The atmospheric pressure shows lower values during the monsoon period. The RH exhibits high values, mainly in the monsoon period with monthly mean values above 80% (July to September). In March–May, also called dry season, the RH values are the lowest, and the dry and semiarid landscapes become the sources of mineral or desert dust. On the other hand, the dry weather and the absence of precipitation cause the favorable conditions for the occurrence and spread of forest fires. Precipitation is absent in winter and premonsoon, except April, while it is high in the monsoon months, with large rainfall amounts in August and September.

Table 1. Monthly Mean Values, as Well as Standard Deviations, of Several Meteorological Parameters at Hyderabad During the Period of Measurements, October 2007 to September 2008
MonthsTemperature (°C)Wind Speed (m s−1)Wind Direction (degrees)Pressure (mbar)Relative HumidityRainfall (mm)
January22.0 ± 4.41.2 ± 0.9126.8 ± 74.9948.4 ± 3.558.2 ± 23.80
February24.1 ± 3.91.3 ± 0.9117.8 ± 76.9945.4 ± 2.560.0 ± 22.70
March26.0 ± 4.01.2 ± 1.0149.3 ± 75.8946.7 ± 2.654.0 ± 23.30
April30.0 ± 5.31.3 ± 0.9191.1 ± 93.8947.4 ± 2.945.7 ± 21.812.56
May31.2 ± 4.22.3 ± 1.7249.6 ± 60.8945.1 ± 2.033.1 ± 13.70
June27.7 ± 3.22.9 ± 1.2233.9 ± 21.1944.1 ± 1.759.2 ± 16.6130.48
July27.4 ± 2.82.3 ± 1.3221.2 ± 56.8943.4 ± 2.378.6 ± 16.597.94
August25.0 ± 2.51.4 ± 1.3255.7 ± 73.8944.3 ± 2.785.3 ± 12.8212.7
September25.3 ± 2.91.8 ± 1.1211.9 ± 65.7945.2 ± 2.884.0 ± 14.3264.1
October23.7 ± 4.21.5 ± 1.1136.4 ± 117.8947.5 ± 3.167.5 ± 19.320.9
November21.6 ± 5.31.5 ± 1.3117.2 ± 117.1950.4 ± 1.958.1 ± 23.020.9
December21.8 ± 4.41.1 ± 0.9134.4 ± 106.5950.0 ± 3.161.0 ± 19.80

2.3. Sun Photometer Measurements

[11] The spectral AOD over the urban region of Hyderabad was measured at six wavelengths centered on 380, 440, 500, 675, 870 and 1020 nm using a handheld multichannel Sun photometer (Microtops-II). Previous studies using the Microtops-II showed a typical error in AOD measurements of ±0.03 [Morys et al., 2001; Porter et al., 2001; Ichoku et al., 2002]. The full width at half maximum (FWHM) bandwidth at each of these wavelength channels is 2.4 ± 0.4 nm, and the accuracy of the Sun-targeting angle, i.e., field of view (FOV), is ∼2.5°. Great care has also been taken to avoid any error in sun targeting the Microtops-II by mounting the instrument on a tripod stand. Measurements of AODs with the same Microtops-II have been used in several studies by the same authors, while the overall procedure of retrieving spectral AOD from solar irradiance measurements is described by Badarinath et al. [2007c].

[12] The spectral AOD values were measured at 30-min intervals, from 0900 to 1700 local time (LT) on cloud-free skies. The AOD data was also subjected to cloud screening using the columnar water vapor values obtained from Sun photometer in addition to sudden departure in AOD values from the mean AOD values for the day. The changes of more than 3 standard deviations from the mean AOD were scrutinized for possible cloud contamination [Smirnov et al., 2000]. The overall error in AOD [Russell et al., 1993] can be due to the diffuse radiation entering the optical channel, the computational error in relative air mass, the deviation of the calibration coefficient with time and the error associated with the uncertainty in the optical depths owing to Rayleigh scattering and absorption by O3, NO2 and water vapor. The contribution of the latter errors in such high AOD values is very low though [Kaskaoutis et al., 2007c]. The absolute uncertainty in the AOD is about 0.03 with the larger values to be depicted at the shorter wavelengths and under low turbid conditions. Perturbed irradiance measurements from possible cloud contamination caused by undetected with naked eye cirrus clouds have been excluded if they conclude to AODs and Ångström exponent values up to two standard deviations from the mean daily value. After these procedures 3336 measurements remained for further analysis. Thus, depending on the sky conditions the number of available data are 1176 in the winter (December to February), 1080 in the premonsoon (March to May), 423 in the monsoon (June to September) and 657 in the postmonsoon (October–November) periods. It should be noted that during the monsoon months the intense cloud cover restricted the number of days for AOD retrievals, since the Microtops-II was not operated during cloud passage. The cloud contamination may also have an influence on the lower Ångström's exponent values in this period.

3. Data Analysis and Methodology

[13] The spectral dependence of AOD is used in this work to compute the Ångström's exponent α. A spectrally averaged value of this exponent, which contains information about the size of the particles or the volume fraction of the fine- versus coarse-mode particles [Schuster et al., 2006], can be obtained by fitting the Ångström's formula [Ångström, 1964]:

equation image

where AODλ is the estimated AOD at the wavelength λ, β is the Ångström's turbidity coefficient, which equals AOD at λ = 1 μm, and α is the Ångström exponent. Although in this definition α is assumed independent to the wavelength, it is well known that depends on it [Kaskaoutis and Kambezidis, 2008b, and references therein]. The Ångström formula is a special case of a more complicated equation valid for a limited range of particle diameters and a limited interval of wavelengths. The validity of this theory presupposes that the Junge power law is valid for the particle radius range, where significant extinction takes place and that the spectral variation of the refractive index does not impose significant variations on the Mie extinction factor [Kaskaoutis et al., 2006]. Taking the logarithms at both sides of equation (1) one obtains

equation image

A more precise empirical relationship between aerosol extinction and wavelength is obtained with a second-order polynomial approximation [King and Byrne, 1976; Eck et al., 1999; Pedrós et al., 2003; Kaskaoutis and Kambezidis, 2006],

equation image

where the coefficient a2 accounts for a curvature often observed in Sun photometry measurements. This curvature can be an indicator of the aerosol particle size, with negative curvature indicating aerosol size distributions dominated by fine-mode and positive curvature indicating size distributions with significant contribution by the coarse-mode aerosols [Eck et al., 1999; Schuster et al., 2006]. In this study, the values of α were computed in the wavelength interval 380–870 nm, applying the least squares method to equation (2). The linear fit to the logarithmic function of equation (2) is the most precise method, although the results may also depend on the spectral interval considered [Pedrós et al., 2003]. The second-order polynomial fit (equation (3)) was also applied to the AOD values at five wavelengths (380, 440, 500, 675 and 870 nm). Although the polynomial fit to equation (3) is more precise than the linear fit to equation (2), large errors can appear especially under low-turbidity conditions [Kaskaoutis et al., 2006]. For limiting these errors, only the cases where the second-order polynomial fit was associated with R2 > 0.95 were considered. It should be noted that the AOD1020 values were omitted from the fits, since they contain larger uncertainties due to the water-vapor effect and detector temperature sensitivity.

[14] As atmospheric aerosols are highly heterogeneous, no single technique or group of techniques is absolutely adequate for an entire characterization of the atmospheric aerosol properties over the extremely wide range of particle sizes, shapes and chemical compositions, or to classify them into groups or types [El-Metwally et al., 2008]. The selection of a particular method depends primarily on the type of application. In the case of atmospheric aerosols monitored by Sun photometers, the columnar mass concentration and the size distribution are vital to understand their source region and optical properties; the first is parameterized by the spectral AOD, while the latter via the Ångström's exponent. Possible correlations linking AOD to α can be sought by examining the scatterplot of α versus AOD. As detailed by several studies [e.g., Pace et al., 2006; Kaskaoutis et al., 2007a; El-Metwally et al., 2008], this visual representation often allows one to define physically interpretable cluster regions for different types of aerosols with different optical characteristics. The importance of these properties is also high for atmospheric and remote sensing studies, apart from other research fields. All these emphasize the need for extensive measurements and analysis of the aerosol optical properties at as many locations on Earth as possible.

[15] In such studies, the selection of the threshold values for the distinguish of the aerosol types can be very important. Therefore, in the present work, the threshold values must be different from those utilized by previous studies, for example, Pace et al. [2006] for Lampedusa, Kaskaoutis et al. [2007a] for four AERONET sites, Kaskaoutis et al. [2007d] for Athens and El-Metwally et al. [2008] for Cairo, directly modulated by the specific characteristics of each location. Since Hyderabad is a densely populated and highly industrialized area, the AOD500, even on days without pollution is rather high, ∼0.35–0.45 [Latha and Badarinath, 2005; Badarinath et al., 2009]. Furthermore, the AOD500 can be above 0.5 during biomass burning conditions [Badarinath et al., 2007c] or >0.6 under the influence of dust plumes [Badarinath et al., 2007a]. Therefore, (1) values of AOD500 < 0.3 with α380–870 < 0.9 represent background conditions with a marine influence, thus corresponding to a Marine Influenced (MI) aerosol type; (2) AOD500 > 0.5 and α380–870 > 1.0 can be used to characterize urban/industrial aerosols under High AOD (HUI), also including the biomass burning episodes, and (3) AOD500 > 0.6 associated with α380–870 < 0.7 are indicative of desert dust particles under High AOD (HDD). Finally, the remaining cases, not belonging in any of the above groups, are characterized as mixed type (MT) or undetermined aerosols [Pace et al., 2006]. In the present work a higher AOD500 threshold value is used from the aforementioned studies for the characterization of the MI conditions, since the AOD500 is relatively large over Hyderabad even on “normal” days (i.e., days with background air pollution levels, e.g., traffic and domestic pollution). A lower threshold α380–870 value for the characterization of the HUI aerosols was selected since the available data set was taken over a location far away from areas directly affected by fresh smoke of biomass burning and the Ångström exponent of pollution may be lower due to hygroscopic growth of fine anthropogenic water-soluble particles into larger accumulation mode size [Eck et al., 2005]. The α380–870 threshold value of 0.7 for characterizing desert dust aerosols was adopted here slightly higher than that (0.5) of Kaskaoutis et al. [2007a] since the desert particles transported over Hyderabad can be mixed with other aerosols (mainly anthropogenic) or to be deposited near the source region. However, because the exact value of α depends significantly on the spectral range used in its determination, the information contained in the α380–870 versus AOD500 scatterplots may be difficult to interpret.

4. Results and Discussion

4.1. Seasonal and Monthly Variations

[16] Figure 2 shows the fluctuation of all AOD500 and α380–870 observations during the study period. The results reveal significant day-to-day variability in both AOD500 and α380–870, underlying the influence of varying aerosol types. In the postmonsoon and winter seasons (October to February) the AOD500 peaks are depicted on certain days, while in the premonsoon and monsoon periods (March to September) they are more often. Similarly, the α380–870 values are, in general, higher than 1.0 in the winter and premonsoon periods, while they significantly drop in monsoon. In Figure 2, two AOD500 peaks are depicted caused by very intense dust events that occurred on 10 March and 25 May 2008. Note that on these days the α380–870 present very low values. On the other hand, the increase in fine-mode particles loading due to long-range transport of aerosols from biomass burning is reflected in an increase of AOD and α. Such high values in AOD and α during forest fires have also reported over Hyderabad [Kharol and Badarinath, 2006; Badarinath et al., 2009]. Aerosol properties over Hyderabad can be significantly modified by the advection of aerosols from the adjoining landmasses under favorable wind conditions [Badarinath et al., 2007b, 2009]. Another possibility is that the increased convective turbulence during premonsoon months would be pumping in more surface-level aerosols to higher altitudes, which would decrease in the nighttime because of their tripped in the boundary layer height. In addition to this, the desert dust aerosols, advected by the westerly winds, would also contribute to increased coarse particle abundance.

Figure 2.

Variation of AOD500 and α380–870 over Hyderabad during the period of measurements (3 October 2007 to 30 September 2008).

[17] Figure 3 shows the monthly mean variation of AOD500 and α380–870 in the period October 2007 to September 2008. The vertical bars express 1 standard deviation, while the total number of data in each month is also given. Figure 3 reveals a rather insignificant monthly variation in AOD500, from ∼0.5 to ∼0.7, with lower values in the October–January period and higher in the March–July one. In sharp contrast, the α380–870 values show a pronounced decrease in monsoon period (June to September). This strong decrease in α380–870 values may partly be attributed to cloud contamination in the retrievals caused by thin undetected cirrus clouds. The largest monthly mean AOD500 in March and April are associated with large α380–870 values indicative of fine-mode particles. A possible explanation for this can lie on the fact that this is the time with peaks in forest fires in central and northern India [Badarinath et al., 2007a] and the transport of biomass burning plumes toward Hyderabad by northerly winds (see Figure 9). Moreover, biomass burning aerosols may be produced by the burning of the vegetal residues in premonsoon, before the rains of the monsoon season. The contribution of biomass burning to the overall atmospheric load in BC over India can then increase from 30% [Novakov et al., 2000] to 70% [Alfaro et al., 2003] between the beginning and the end of March.

Figure 3.

Monthly mean AOD500 and α380–870 over Hyderabad. The vertical bars denote 1 standard deviation, while the number of the measurements in each month is given within the bars of the top graph.

[18] Monthly variations in the aerosol size and its influence on AOD are clearly seen over Hyderabad. Coarse-mode and/or growth of fine-mode water-soluble (e.g., sulfate) aerosols due to higher ambient RH result in high AOD500 in the summer monsoon, while during the winter the AOD500 is slightly lower, which is mostly associated with fine-mode particles. Thus, it is observed that in the urban area of Hyderabad anthropogenic sources dominate the aerosol distribution, which could get modulated by the presence of natural aerosols. During monsoon the α380–870 values are being mediated by the higher ambient RH, possible cloud contamination and transport of dust and sea salt in favorable wind conditions. The high values of BC during the winter months have been attributed to the decreased boundary layer height causing low ventilation for mixing the emissions from vehicular traffic [Latha et al., 2004; Badarinath et al., 2007c]. However, Badarinath et al. [2007a] found large daily, monthly and annual variabilities in BC concentrations and aerosol size distribution (nucleation, accumulation and coarse particles), mainly driven by the fire events in the upwind regions adjoining to Hyderabad. Nevertheless, a slight reduction in AOD500 is observed during the monsoon (August–September) due to scavenging effects of rainfall. On the other hand, high AODs during the premonsoon period could be attributed to the increased concentration of continental aerosols due to strong surface winds, which play an important role in lifting the loose soil, as well as other sources such as long-range transport of dust from northwestern India [Badarinath et al., 2007b] or biomass burning since India experiences severe forest fires during January to May each year [Badarinath et al., 2007a; Kharol and Badarinath, 2006]. The AOD values and their seasonal pattern over Hyderabad are similar to those presented over four (Chennai, New Delhi, Mumbai and Kolkata) Indian sites [Ramachandran, 2007].

[19] Figure 4 shows the frequency of occurrence of AOD500 for the four individual seasons. Large changes in the AOD500 values are observed in each season. In the winter a higher frequency is observed for AOD500 between 0.3 and 0.5. A rather similar distribution is shown in the monsoon period, which shows a higher frequency for larger AOD500 values. An almost normal distribution occurs, though, in the premonsoon. The AOD500 ranges between 0.1 and 1.2 in postmonsoon, where the highest frequency is observed for AOD500 values in the range 0.3–0.5. In each season the mean AOD500 values are also given for a direct comparison with the results from other Indian or worldwide urban areas. The cumulative frequency of observing an AOD500 > 0.7 is high (22.7% in winter, 30.4% in premonsoon, 30.1% in monsoon and 8.5% in postmonsoon) while values larger than 1.0 have also been recorded. This important frequency of occurrence for large AOD500 values confirms that the atmospheric load over Hyderabad is usually particularly high. These large AODs are in the same order of magnitude with those presented over Cairo [El-Metwally et al., 2008] and Kanpur [Tripathi et al., 2005] and significantly higher that those at several urban AERONET locations [Holben et al., 2001]. As expected, the aerosol load over Hyderabad is found to be much higher than that observed over the oceanic regions surrounding the Indian subcontinent, i.e., BoB (AOD500 = 0.36 ± 0.12), AS (AOD500 = 0.23 ± 0.09) and NIO (AOD500 = 0.26 ± 0.10), mean and standard deviations during the ICARB campaign [Kalapureddy and Devara, 2008]. The 5-year mean AOD550 is 0.4 in Chennai and Mumbai, while in Kolkata and New Delhi it is >0.55 [Ramachandran, 2007]. The mean AOD500 and Ångström exponent at Visakhapatnam, a costal urban area in BoB, were found to be 0.72 ± 0.39 and 0.88 ± 0.39, respectively in the period May–August 2005 [Madhavan et al., 2008]. The mean AOD500 in the premonsoon season in Pune is 0.42, while in the winter ∼0.38 [Pandithurai et al., 2007]. These values are significantly lower than those over Hyderabad in the same periods, since Pune is not so industrialized and with much less population than Hyderabad.

Figure 4.

Frequency distribution of the AOD500 values in each season over Hyderabad during the period 3 October 2007 to 30 September 2008: (a) winter (December–February), (b) premonsoon (March–May), (c) monsoon (June–September), and (d) postmonsoon (October–November). The mean seasonal values are also given.

[20] The frequency distribution of α380–870 (Figure 5) in all seasons reveals a great dispersion of the values thus denoting variability in the aerosol size distribution. Moreover, the frequency distribution is very different depending on season. In winter the α380–870 values are skewed toward larger values, while in monsoon toward lower. During winter, the modal value is around 1.2, indicating the dominance of submicron aerosols originating from biofuel burning and fossil fuel combustion sources mainly. Moreover, the fact that α380–870 becomes less than 0.5 only rarely in winter shows that the influence of the dust aerosols is very limited. The dominating aerosols are in the submicron size range formed in situ by secondary gas-to-particle conversion processes of the precursors [Badarinath et al., 2007a]. Moreover, the surface inversion associated with the trapping of pollutants give rise to hazy and foggy conditions [Madhavan et al., 2008]. During premonsoon, α380–870 exhibits two modal values around 0.7 and 1.3, ranging from very low (<0.2) to very high (>1.5) values. This clearly indicates that the aerosol scenario over Hyderabad is driven by a combination of pollution and dust patterns due to dust generation from strong surface heating and high winds, as was also shown in Bahrein [Smirnov et al., 2002]. Water-soluble aerosol particles (e.g., ammonium, nitrate, chloride and sulphate) mainly coming from industries and anthropogenic activities play a major role in the nucleation and growth of cloud droplets [Roberts et al., 2002] especially in the monsoon season when RH is very high. The water uptake of such aerosols due to their hygroscopic nature results in size enlargement [Day and Malm, 2001] due to the humidification processes. This fact results in the significantly lower α380–870 values observed in this season, which are partly attributed to the long-range transport of desert dust particles. In postmonsoon, the possibility of high (>1.1) and low (<0.6) α380–870 values is similar with higher frequency in the 0.9–1.0 interval. Values of α380–870 > 1.0 are observed at 65.9% in the winter, 57.3% in premonsoon, 29.2% in postmonsoon and only 8.5% in the monsoon period and are typical of particles located mostly in the submicron range. On the other hand, values smaller than 0.5, and typical of aerosols containing an important proportion of particles larger than 1 μm [Ångström, 1964], are 0.9%, 2.3%, 43.4% and 5.5% in winter, premonsoon, monsoon and postmonsoon seasons, respectively. In Figure 5, the seasonal mean α380–870 values are also given. The large variability in aerosol size around India was established by Kalapureddy and Devara [2008] who found that the portion of α340–1020 < 1 constitutes 2.6%, 37.3% and 74.6% of the total observations over BoB, NIO and AS, respectively. Thus, dominance of fine-mode (α = 1.21 ± 0.11) and coarse-mode (α = 0.86 ± 0.20) aerosol particles has been observed over BoB and AS regions, respectively.

Figure 5.

Same as in Figure 4 but for the α380–870 values.

4.2. Seasonal Discrimination of the Aerosol Types

[21] Figure 6 shows the density plot of AOD500 versus α380–870 over Hyderabad for the four individual seasons. These contour maps were constructed using a step of 0.1 for both AOD500 and α380–870 values. The white gaps correspond to lack of data. A striking feature of the scatterplot obtained is that a wide range of α380–870 is associated with values of AOD500 < 0.4. This indicates that under these conditions the aerosols are probably a mixture of several components differing in size and that the proportion of this mixture depends on season. The assumption that the aerosols over Hyderabad are of differing origins and optical properties is further supported by the fact that the large AOD500 are associated with narrow ranges of α380–870. However, the majority of the points in the scatterplots are confined by intermediate AOD500 and α380–870 values corresponding to a rather undetermined aerosol type, making thus the further characterization of the aerosols difficult, because of the complex combination of natural and anthropogenic factors (including RH, fuel types and emission characteristics) that influence aerosol formation and evolution.

Figure 6.

AOD500 versus α380–870 for each season during the period 3 October 2007 to 30 September 2008. The number of observations is shown in parentheses for each individual season.

[22] Viewing the contour plot some areas of larger density are observed, representative of different aerosol types depending on season. In winter, the maximum density area of (AOD500 = 0.4–0.5 and α380–870 = 1.1–1.2) is representative of the urban polluted aerosols under moderate-to-high turbid conditions. The other AOD500/α380–870 pairs are nearly equally distributed around this area with preferable of larger AOD500. In the premonsoon season, the winter maximum seems to shift into two; one with AOD500 = 0.5–0.7 and α380–870 = 1.1–1.3 with anthropogenic or biomass burning origin probably, and a second with AOD500 = 0.5–0.6 and α380–870 = 0.7–0.8 corresponding to aerosols with larger fraction of coarse-mode particles under turbid conditions. In monsoon, the α380–870 values are significantly lower than those in the previous cases, mainly affected by the density maximum area of AOD500 ∼1.0, α380–870 ∼0.1. The different aerosol sources have a direct effect on the wide range of α380–870 values, from ∼0.3 to ∼1.0, especially for low AOD500 (<0.4), suggesting a range of aerosol types. However, for larger AOD500 the α380–870 decreases to values lower than 0.4, assigning a decreasing trend of the AOD500 versus α380–870 relationship. The postmonsoon is characterized by lower AOD500 (0.46 ± 0.18) and moderate α380–870 (0.89 ± 0.19) values. The maximum density area is observed for the pair (AOD500, α380–870) = (0.35–0.45, 0.9–1.0) indicative of moderately turbid conditions with a rather mixed aerosol field in the vertical. The α values of this magnitude are indicative of bimodal aerosol size distributions, with significant contribution from both fine-mode submicron (radius < 1 μm) and coarse-mode supermicron (radius > 1 μm) aerosols [Eck et al., 2005].

[23] Figure 7 shows the percent contribution of each of the three different aerosol types over Hyderabad, based on the present AOD and α380–870 threshold values. It is observed that the MT aerosol is the most dominant contributor with varying magnitudes depending on season. The respective influences of the three aerosol types are more balanced in the winter and postmonsoon than during premonsoon and monsoon periods, when the % contribution of the specific aerosol types is larger. The contribution of aerosols of different origin and characteristics to the atmospheric column can be strongly modified in each season. Thus, 47.2% of the HUI aerosols are observed in premonsoon, 30.3% of HDD aerosols in monsoon and 8.7% of the MI conditions in the postmonsoon periods. The MT is predominant in all seasons, except the premonsoon, exhibiting its higher frequency in postmonsoon (72.9%) and winter (62.2%). Quite characteristic is the low percentage of HUI aerosols in monsoon (6.6%), taking into account that the anthropogenic emissions are significant at Hyderabad throughout the year. This can partly be explained by a quick modification of the produced water-soluble local aerosols that can enlarge in size due to humidification especially under the influence of the large RH values in the summer monsoon. Also, significant role may play the dust emissions and transport in this season and the possible adhering of fine-mode pollution particles onto the surface of coarse-mode dust in the mixed aerosol. Since the majority of the aerosols belong to a mixed (or undetermined) aerosol type more attention must be paid to this type, as a result of independent processes. Thus, the mixing could be caused by primary small and large particles. The larger RH values close associated with the lower α380–870 in monsoon may be an evidence of coagulation and hygroscopic growth of the water-soluble urban aerosols, since the higher growth factor includes the hygroscopic and water-soluble particles [Day and Malm, 2001]. However, there is not a detailed analysis in the present study to warrant it and further investigation is needed. Also, the possible cloud contamination in some cases in the monsoon period results in lower α380–870, which associated with the high AODs concludes to the HDD type. On the other hand, the coarse-mode particles transported over Hyderabad can easily be mixed with local pollution or with smoke from forest fires, increasing the AOD wavelength dependence.

Figure 7.

Fraction pies of each aerosol type over Hyderabad contributing to the total in each season: MI (maritime influenced), MT (mixed type), HUI (high AOD urban/industrial), and HDD (high AOD desert dust) for (a) winter, (b) premonsoon, (c) monsoon, and (d) postmonsoon.

[24] Figure 8 shows the percentage frequency of occurrence for each aerosol type as a function of month. In particular, MT was observed in the whole duration of the measurements period. Thus, in certain months, the MT aerosols corresponded to more than 60% of the daily observations. Besides this fact, the dominance of the HUI aerosols is seen during the February–April period. Rice-crop residues are usually burnt in the Indo-Gangetic Plains (IGP) during October/November each year. This large-scale burning constitutes a major source of submicron-sized aerosols, which are known to aggravate lung and respiratory diseases [Wang and Christopher, 2003] and affect Hyderabad on specific days of this period [Badarinath et al., 2009]. The MI conditions show their largest fraction, although very small, from September to December, while HDD makes it's largest contribution in the period May to September.

Figure 8.

Relative (%) frequency of each aerosol type over Hyderabad for each month: MI (maritime influenced), MT (mixed type), HUI (high AOD urban/industrial), and HDD (high AOD desert dust).

4.3. Back Trajectory Analysis

[25] The aerosol load and types at any location are generally influenced by the meteorological conditions (primarily wind and convection) and hence affect the air mass transport over the region for a particular period. Figure 9 shows the 5-day back trajectories of NOAA-HYSPLIT (National Oceanic and Atmospheric Administration Hybrid Single-Particle Lagrangian Integrated Trajectory) version 4 [Draxler and Rolph, 2003] model, where different types of air masses that influence AOD over Hyderabad are revealed in each season. The air masses are computed at three altitudes (500, 1500 and 4000 m) on each day with at least 3 AOD observations at 1000 UTC.

Figure 9.

Five-day back trajectories ending at Hyderabad in each season at 500 m (red), 1500 m (blue), and 4000 m (green). The trajectories were taken from the HYSPLIT model and refer to 1000 UTC.

[26] In winter, there is a well-established difference between the pathways of the air masses at each altitude; thus, as the altitude increases the air mass source regions shift toward western directions. At 500 m the air masses come from northern/northeastern directions originating from Bangladesh and traversing the eastern Indian coast, where densely populated areas and industries are situated. At 1500 m the air masses are of continental origin, northern India and IGP, where often intense fog and pollution-haze conditions occur [Badarinath et al., 2007a]. The lower altitude trajectories favor the transport of pollution aerosols over Hyderabad in winter; thus the relative high AODs compared with those from other urban locations (e.g., Athens, Cairo) in this season [Kaskaoutis et al., 2007d; El-Metwally et al., 2008]. At 4000 m the trajectories are from western directions, originating mainly from Africa and driven by the western synoptic circulation pattern in northern midlatitudes. However, the low occurrence of HDD aerosols in winter (Figure 7) is attributed partly to the mixing processes with the different aerosol types at lower altitudes and partly to the fact that the 4000-m air masses do not interact with the Saharan boundary layer in order for the latter to be capable to uplift and transport large amounts of dust.

[27] In premonsoon, the above situation is somewhat more confused. However, the majority of the lower air masses come from continental India and at 4000 m from northern and northwestern directions. The air masses from the different source regions lead to the formation of different aerosol types as seen in Figures 6 and 7. On several occasions the 4000-m air masses can uplift and transport HDD aerosols over Hyderabad, since they generally interact with the boundary layer over the arid/semiarid locations. Furthermore, in several cases the lower air masses come from the arid locations in northwestern India and Pakistan. In India, fires occur every year from January to May with a predominance during March and April [Badarinath et al., 2007a]. The back trajectories suggest that wind patterns are dominantly northerly with continental air mass conditions prevailing over the study area during the fire season. This leads to enhanced aerosol loading with increased concentrations of BC [Badarinath et al., 2007a] and fine-mode aerosols, thus increasing the α values over Hyderabad. The highly industrialized and urbanized IGP can produce anthropogenic fine particles that may be transported over long distances before they settle down due to gravity. Niranjan et al. [2007] have shown the presence of high-altitude aerosol layers above the boundary layer, at a height between 1.5 to 5 km, at Visakhaptnam located on the east coast of India near to Hyderabad, during the period March–May 2005; they indicated an increase in AOD with no signs of proportional increase in the surface aerosol mass concentration. The air mass trajectories during premonsoon and early monsoon are in close agreement with those observed over New Delhi in the same months, March–June 2006 [Pandithurai et al., 2008].

[28] In the monsoon period, the air circulation pattern is quite characteristic driven by the summer monsoons. Thus, although the boundary layer wind flow (at 500 and 1500 m) is clearly southwesterly, transport of dust aerosols from distinct regions of west Asia (Arabian Peninsula, Iran and Pakistan) occurs at high altitudes (4000 m). The marine and/or desert source regions of air masses in the monsoon period have an evidence of lower α values (Figures 3 and 5) and an occurrence of HDD aerosols (Figure 7). Sagar et al. [2004] found that in northern India the air masses shift from southerly to westerly in the period of winter-spring/summer, when the winds arrive from the vast arid regions of northwest India and others located farther to its west. This change in the air mass type was obviously mostly responsible for the rapid buildup in the AOD after March, as the arid air mass is known to transport large amounts of desert/mineral aerosols from the west Asian and Indian deserts. Finally, in the transition season of postmonsoon, the air mass trajectories exhibit a rather confused pattern, as in premonsoon, where aerosols from different origins can be present in the vertical. The main difference from the premonsoon period is the high-frequency occurrence of upper air mass from marine directions, probably responsible for the more transparent atmospheric conditions. Within the boundary layer, the air masses are mainly originating from IGP, capable to transport anthropogenic pollution and biomass burning aerosols in certain cases, for example, during Diwali festival [Badarinath et al., 2009].

4.4. Optical Properties of the Different Aerosol Types

[29] In this section, the optical properties of the aerosol types are further investigated. Figure 10 shows the relative (%) frequency of occurrence for positive and negative a2 values as a function of month. The existence of positive (negative) curvature indicates a significant contribution by coarse- (fine-) mode aerosols in the size distribution [Eck et al., 1999; Schuster et al., 2006]. It is shown that the negative a2 values are predominant in the period January to April, and the positive a2 in the rest of the year. Comparing Figure 10 with Figure 8, it is concluded that in months with significant occurrence of HUI aerosols the curvature is negative, while in months with significant fraction of MI or HDD aerosols the curvature is positive. In the first period (January–April) the comparison of Figures 10 and 8 reveals that the majority of the MT aerosols exhibit negative curvature, while in the second period (May–December) positive. This can further discriminate the initial origin of this mixed “undetermined” aerosol type. Gas-to-particle conversion, followed by coagulation and direct emission of aerosols in the accumulation mode must be the main processes explaining production of the HUI aerosols. This consists of the presence of a dense motorized traffic within Hyderabad that releases very fine particles found in the accumulation mode [Badarinath et al., 2007a]. Indeed, the biomass burning is known to release particles in the accumulation mode but also coarser particles resulting from the condensation of organic compounds [Reid et al., 1999]. The negative curvature is much greater for aged fine-mode aerosols due to particle growth by coagulation [Eck et al., 2003]. This highlights why Ångström exponent alone does not give all the information for the determination of the fine/coarse mode relative contributions.

Figure 10.

Frequency of occurrence (%) for the positive and negative a2 values obtained from equation (3) for each season over Hyderabad.

[30] Figure 11 shows the scatterplots of the coefficient a2 (curvature in the polynomial fit) against AOD500 for the four aerosol types depending on season. The correlation between a2 and AOD500 provides information on the atmospheric conditions under which α is independent from wavelength, so the spectral variation of AOD can be accurately described by the simple Ångström formula (equation (1)). The data lying on or near the a2 = 0 line correspond to the monomodal Junge size distribution or to bimodal lognormal aerosol size distribution without curvature. Near to zero a2 values can also be caused by a unimodal coarse-mode size distribution, i.e., dust dominated [Eck et al., 1999]. Negative a2 values (fine-mode aerosols) are depicted, unexpectedly, for the MI type and correspond to observations associated with larger errors in the a2 retrievals caused by uncertainties in the polynomial fit (equation (3)) under low turbid conditions [Kaskaoutis et al., 2006, 2007b]. Furthermore, some positive a2 values are presented for the HUI type, thus showing that the initial fine-mode aerosols have been modified by coagulation, condensation and gas-to-particle conversion resulting in greater size and less negative or even positive a2 values [Schuster et al., 2006]. Quite different graphs are revealed in each season. In winter and postmonsoon, and partly in monsoon, a2 presents high positive values (in the majority of the cases) for low AOD500, which are reduced as AOD500 increases. These high positive values at low AOD are likely due to large uncertainty in computed a2 for these cases, due to AOD uncertainty of ∼0.03. In premonsoon, a2 values are mostly negative for the lower AOD500 values. It is interesting to note the smaller fraction of positive a2 values of MT in winter and premonsoon, highlighting the great fraction of fine-mode particles in its component; in monsoon and postmonsoon the majority of MT cases correspond to positive a2.

Figure 11.

Seasonal correlations between coefficient a2 and AOD500 for each aerosol type over Hyderabad for (a) winter, (b) premonsoon, (c) monsoon, and (d) postmonsoon.

[31] It is concluded that the different aerosol types are rather difficult to be distinguished based on the a2 values, in contrast to the results in the work by Kaskaoutis et al. [2007b]. However, these researchers investigated the aerosol properties in different environments, where the aerosol types were clearly distinguishable, whereas the present study focuses over Hyderabad, where aerosols from different origins are present. So, except MI and some portion of HDD (at higher AODs) the other types are still unable to be discriminated unambiguously. Note also that for low AOD500, there is a wide variability in a2 values (both positive and negative), thus implying large curvature in equation (3). It was found that in such cases the uncertainties in the polynomial fit dramatically increase as well as the errors in the a2 values. Kaskaoutis and Kambezidis [2006] also found that the curvature of the polynomial fit becomes greater as the aerosol loading alters from high to low. On the other hand, coarse-mode HDD aerosols present near to zero a2 values as the turbidity increases. This is in accordance with previous studies [e.g., Eck et al., 1999; Kaskaoutis et al., 2007b].

[32] Figure 12 is drawn for further aid in a clear discrimination of the aerosol types, especially of HDD, which is confined to lower α380–870 values. In general, for a specific value of α380–870, a great spread of a2 occurs, especially for intermediate α380–870 values (0.7–0.9). This is in agreement with the findings of Schuster et al. [2006], who showed that different size distributions with similar α can produce large differences in curvature; they verified that the curvature alone is not enough for describing the aerosol particle size, as stated elsewhere [e.g., Eck et al., 1999]. Yet, it is possible to better discriminate between the different aerosol types by plotting α380–870 versus a2. This is mainly achieved for the HDD aerosols (low α380–870 and positive or near zero a2), but it is rather confused for identifying HUI from MT aerosols in the winter and premonsoon periods. In contrast, the HUI type is more distinguishable in the monsoon and postmonsoon periods. Despite the great differences in the α380–870 and a2 values, there is a similarity between the graphs in each season, mainly conducted by the negative a2 values as α380–870 increases, thus implying a general decreasing trend in their correlation. Furthermore, dominance of negative curvature was found over Visakhapatnam, India, highlighting the abundance of columnar submicron aerosols [Madhavan et al., 2008]. They found that the positive curvatures dominated during the pre-rainy days, while the negative curvatures during the post-rainy days indicating that the coarse-mode aerosols had a wet deposition and the fine-mode ones acted as CCN. This was also observed at Hyderabad in the monsoon period mainly regarding the MI and the MT.

Figure 12.

Same as in Figure 11 but for the correlation between a2 and α380–870.

[33] Figure 13 shows the frequency distribution of AOD500 for the four aerosol types according to the discrimination criteria. Very different distribution is revealed depending on season, with the most pronounced differences to be observed in the monsoon period, when the HDD aerosols are nearly equally distributed in a wide range of AOD500 values (0.6–1.1). Note also the very low frequency of the other types (MT and HUI) for AOD500 > 0.6. The winter and postmonsoon distributions present some similarities, despite the lower frequency of the HUI aerosols in postmonsoon, while in premonsoon there is a clear discrimination between HUI and HDD aerosols, with the latter to dominate for AOD500 > 1.1. It was found (not presented) that all these cases correspond to dust outbreaks from the Thar Desert. During winter and postmonsoon the above situation is somewhat different. AOD500 maxima can then be associated either with urban pollution and mixed aerosols or, to a lesser extent, with dust particles. Large fraction of MT aerosols for AOD500 > 0.5, where also HUI and HDD aerosols can be found, is observed only in winter and premonsoon.

Figure 13.

Number of cases of the different aerosol types based on AOD500 values for each season over Hyderabad. HUI in red, MI in blue, HDD in yellow, and MT in cyan for (a) winter, (b) premonsoon, (c) monsoon, and (d) postmonsoon.

[34] Similarly, Figure 14 shows the frequency distribution of the aerosol types with respect to α380–870. In the scale of α380–870, HDD and MI are expected to fall in the lower end of α380–870 values, since both types are characterized by coarse-mode particles. Nevertheless, it is surprising to note that the MI type is absent for α380–870 values lower than 0.4. Note that the larger possibility of observing lower α380–870 (Figure 14) in premonsoon and monsoon periods than in the rest of the year shows that the importance of the mineral component is generally greater in these seasons. This is further confirmed by examining Figure 13 showing that only the HDD aerosols are responsible for the very large AOD500 > 1.1 in these seasons. MT plays a significant role in the distribution of the α380–870 values in the winter and monsoon periods, since this is the dominant type for a wide range of α380–870 values, 0.7–1.3 for winter and 0.3–1.2 for monsoon, hindering the discrimination of the other aerosol types. On the other hand, in premonsoon, the HUI type clearly dominates for α380–870 > 1.0. Therefore, it is concluded that the AOD500 or α380–870 values alone cannot identify any aerosol type, except of few cases, due to their similar and broad distributions.

Figure 14.

Same as in Figure 13 but based on α380–870 values.

[35] The mean seasonal values with the standard deviations of AOD500, α380–870 and a2 are given in Table 2 for each aerosol type. The AOD500 and α380–870 values for each aerosol type are mainly defined by the threshold values for the type identification, but it is worth noting the higher AOD500 in premonsoon as well as the lower α380–870 values in monsoon independently from the aerosol type. Thus, regardless of the aerosol type the aerosol load over Hyderabad is larger in the premonsoon season directly affected by the dust exposures and/or the biomass burning. On the other hand, the AOD500 is lower in postmonsoon indicative of more transparent atmospheric conditions. As regards the a2 values, the large standard deviations do not allow safe conclusions. In each season, as shown in Figures 11 and 12, they can be negative or positive. However, it is worth noting the negative mean a2 values for the fine HUI aerosol type and the positive for the coarse HDD and MI types. The MT exhibits negative a2 (dominance of fine-mode aerosols) in winter and premonsoon and positive (coarse-mode aerosols) in the other seasons.

Table 2. Mean Values and Standard Deviations of AOD500, á380–870, and a2 for Each Aerosol Type and Season Over Hyderabad in the Period October 2007 to September 2008
Aerosol TypeSeasonAOD500á380–870a2
HUIwinter0.672 ± 0.1301.178 ± 0.107−0.642 ± 0.320
 premonsoon0.701 ± 0.1411.218 ± 0.116−0.501 ± 0.474
 monsoon0.639 ± 0.1291.148 ± 0.099−0.377 ± 0.340
 postmonsoon0.654 ± 0.1271.139 ± 0.161−0.012 ± 0.329
     
HDDwinter0.917 ± 0.2050.605 ± 0.080.061 ± 0.377
 premonsoon0.969 ± 0.4130.552 ± 0.1210.024 ± 0.450
 monsoon0.869 ± 0.1730.324 ± 0.1850.202 ± 0.265
 postmonsoon0.855 ± 0.2120.485 ± 0.1340.217 ± 0.274
     
MIwinter0.249 ± 0.0380.774 ± 0.1401.048 ± 0.957
 premonsoon0.280 ± 0.0100.844 ± 0.0570.084 ± 0.207
 monsoon0.261 ± 0.2480.638 ± 0.1320.025 ± 0.760
 postmonsoon0.248 ± 0.0380.770 ± 0.0740.981 ± 0.707
     
MTwinter0.477 ± 0.1941.026 ± 0.183−0.295 ± 0.546
 premonsoon0.542 ± 0.1760.902 ± 0.224−0.268 ± 0.624
 monsoon0.506 ± 0.1540.617 ± 0.2400.011 ± 0.567
 postmonsoon0.428 ± 0.1180.877 ± 0.1660.467 ± 0.524

5. Conclusions

[36] The Indian continent is unique with regard to aerosol sources, characteristics and their interaction with the Indian monsoon system. Seasonal variations in the aerosol optical depth and its spectral variation analyzed using a Microtops-II Sun photometer during October 2007 to September 2008 over Hyderabad were investigated. The aerosol optical properties exhibited large temporal variation due to the prevailing meteorology, air mass trajectories, and atmospheric conditions. The AOD500 presented a rather insignificant monthly variation with larger values in March–April and June (∼0.7) and lower in November and January (∼0.5). In contrast, the α380–870 exhibited a pronounced annual pattern with low values in the monsoon period and high in winter and premonsoon. The gas-to-particle reaction products of the exhausts from the industrial and urban emissions, combined with dust and sea salt during favorable wind conditions contributed to the aerosol loading in the Hyderabad urban area. The results of the study confirmed the large variability of the aerosol optical properties. They also showed that the use of detailed spectral values of AOD and Ångström exponent makes it possible to obtain the basis for characterizing some aerosol properties. Using the relationship between AOD500 and α380–870 three aerosol types were classified over Hyderabad in order to represent different atmospheric conditions (i.e., urban/industrial, marine influenced and desert dust aerosols). The cases, which did not belong to any of the above types, were characterized as mixed (undetermined) aerosols. The presence of each aerosol type seemed to exhibit strong temporal variability, mainly affected by the season, the long-range transport and the mixing processes in the atmospheric column. Thus, the premonsoon season presented the higher occurrence of the HUI aerosols; the HDD aerosols were mainly found in the monsoon period, while the MI conditions were identified, although rare, in the postmonsoon period. The AOD500/α380–870 contour plots over Hyderabad revealed the existence of different aerosol types on certain days, months and seasons in contrast to the dominant scenario of a mixed aerosol type in the vertical over this area. To examine the effect of long-range transport, 5-days back trajectories were computed using the HYSPLIT model. The air masses in all seasons came from varying directions, mainly northwestern in the winter and premonsoon periods. In the monsoon season the lower altitude air masses originated from AS carrying significant amount of marine aerosols over the region, while the upper atmosphere air masses came from the northwest arid and semiarid areas. This resulted in the dominance of the coarse-mode aerosols in this season. The analysis showed that the inclusion of the curvature of lnAOD with respect to lnλ did not significantly improve the aerosol type discrimination over Hyderabad. This is attributed to the mixed aerosol field over the region, since particles of different origins and sectors can be present in the atmospheric column. The dominance of positive and negative a2 values of the MT in monsoon/postmonsoon and winter/premonsoon, respectively, suggested that the fine-mode aerosols play a much more important role in the mixed-type aerosols in winter and premonsoon and the coarse-mode particles in the rest of the year. On the other hand, the HUI presented negative a2 values while the HDD near zero or positive, which are in agreement with other studies in the literature. The information retrieved in this way can be further used in remote sensing applications, or climatological studies, as the ground-based stations are spread all over the world and are becoming more standardized.

Acknowledgments

[37] The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://www.arl.noaa.gov/ready.html) used in this publication. The NCEP reanalysis scientific team is also gratefully acknowledged for providing the meteorological fields. The authors thank ISRO-GBP for funding support, and Director, NRSC, and Deputy Director (RS&GIS-AA) for all the encouragement.

Ancillary