Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Recent advances in atmospheric science have shown that the chemical composition of the entire atmosphere of the earth has changed due to human activity, and these changes in gases and airborne particles result in changes in the heat balance of the planet (Ramanathan et al., 2001; IPCC, 2007). The meteorological processes, which constitute the earth's climate, are dependent on and driven by local variations in this energy balance. Thus, there are fundamental reasons to believe that changes in climate may result from these human-induced changes in atmospheric composition. The magnitudes of these changes in climate, however, are poorly known. But it is being increasingly recognised that industrialisation and human activities during the last few decades have already caused irreversible damage to the earth's climate by releasing excessive amounts of trace gases and aerosol particles. Thus, a great deal is at stake in correctly and accurately quantifying all climate forcings. Without this, it will be impossible to develop any reliable predictive capacity for climate responses.
In recent years, scientists have begun to consider quantitatively how anthropogenic aerosols affect global climate (Charlson et al., 1992; Kaufman et al., 2002; Penner et al., 2004; Ramanathan et al., 2007; Sang-Woo Kim et al., 2010). One of the major uncertainties in the magnitude of climate change is related to the radiative forcing effects of aerosols (IPCC, 1996). On global scale, the direct radiative forcing of anthropogenic aerosols is estimated to be in the range from − 0.3 to − 1.0 Wm−2. Yet these estimates are highly uncertain because of large inhomogeneity in optical properties and spatial distribution of aerosols. Although there is much evidence to suggest that radiative forcing due to aerosol is comparable to that of greenhouse gases, but opposite in sign, leading to conclusion that aerosols cause cooling (IPCC, 2001), it is not confidently known just how large such an effect might be. Thus, aerosols are one of the most important but poorly understood factors which influence global climate. The aerosol–cloud–climate interactions have special significance particularly over the tropics. In view of these aspects, there is a need to regularly monitor aerosol distributions over different geographical regions to study their inter-annual, seasonal, and long-term variations. Also, it is important to see if any trend in their changing patterns exists and how they may ultimately affect the regional/global climate. Such long-term datasets of aerosols and results may help better quantification of aerosol radiative effects, aerosol–climate interactions and impacts of aerosol loading on the earth's biosphere.
The physical and chemical characteristics of aerosol can be monitored and studied using several experimental techniques viz., in situ measurements like sampling of air using different instruments such as high-volume sampler, aethalometer, multi-channel spectrometer, or Aitken nuclei counter etc. Although these instruments provide vital information on aerosol and its composition either physical or chemical, it is not true representation of the atmosphere over the station since majority of these observations pertain to surface levels. Polar orbiting satellites provide good spatial coverage but have limited temporal variability and utilisation of geostationary satellites for aerosols is still not well developed. Ground-based remote sensing techniques (both active and passive) are being widely used to monitor spatio-temporal variations in atmospheric aerosols (Krishna Moorthy et al., 1993; Parameswaran et al., 1997; Holben et al., 2001; Sang-Woo Kim et al., 2008). In this study, an attempt is made to examine long-term variations in aerosol optical depth from near-continuous10-year-long measurements made at Pune (18°32′N, 73°51′E, 559 m above mean sea level (AMSL)), India, a tropical fast evolving urban location.
2. Experimental site and meteorology
Pune is an urban city fast changing in the last one decade. It is located at 559 m AMSL in the Deccan Plateau region of peninsular India and is about 100 km east from the Arabian Sea coast. There is hilly area on the western side of Pune (Western Ghats) and also on the southern side. The total geographical area of Pune city is about 450 sq. km. Of this, approximately 38.6% is residential area, 1.8% is commercial area, and 11% is industrial area. An equally large township (Pimpri-Chinchwad) is located on the west/northwest of Pune city almost like a twin city and constitutes various industries including major automotive manufacturers. The Pune–Mumbai National highway with one of the heaviest vehicular traffic runs close to the urban area. The population of Pune is around 3 157 000 as per 2001 census.
The region around the city experiences typical tropical weather with four distinguishable seasons, namely, pre-monsoon (March–May), southwest monsoon (June–September), post-monsoon (October–November) and winter (December–February). Winds are predominantly westerly/southwesterly (marine air mass) during monsoon season and mostly northerly/northeasterly (continental air mass) during the rest of the year. The pre-monsoon season is hot, while during June to September most of the days are cloudy with frequent monsoon showers; and the winter is cold and dry with occasionally hazy sky conditions. Figure 1 shows 30-year (1950–1980) climatological monthly mean variation of day's maximum and minimum surface temperatures (lower panel), and relative humidity at 00 GMT and 12 GMT (upper panel) for Pune meteorological station taken from published ‘Climatological Tables’ of India Meteorological Department (IMD). It is seen that month-to-month variation of day's maximum temperature shows two peaks; a significant primary peak in the pre-monsoon season (March–May) and a secondary peak in post-monsoon (October–November) with a minimum during SW monsoon season. The day's minimum temperature shows only a single broad peak from April to October. Another interesting feature in surface temperature at this location is that the daily temperature range (that is, the difference between day's maximum and minimum) itself is on average around 17–20 °C during winter, and the same is as low as 6 °C during the monsoon season. Relative humidity as expected is high during wet monsoon season and low during winter. These meteorological conditions have a role to play in generation, growth, transport/advection and sink/removal of aerosols and in their day-to-day and seasonal variations.
The observational point is located in the premises of the Indian Institute of Tropical Meteorology (IITM), which itself is almost on the western edge of Pune city. Surrounding the station is rapidly growing urban environment in terms of housing and vehicular traffic. Soil dust, due to open land, was the major source of aerosol particulates earlier which has now changed to typical urbanisation effects like rapid growth in vehicular/transportation, changed land use pattern, construction of housing complexes, roads etc. Formation of aerosols in accumulation mode is considered to be due to gas-to-particle conversion process, whereas coarse-mode particles are attributed mainly to wind-blown dust. The site is surrounded by hillocks as high as 760 m AMSL. The transport/dispersion of pollutants, particularly those in the lower levels of the atmosphere are believed to be affected by the circulation processes evolved due to typical terrain. Earlier, there were stone quarries (east side) and the brick kilns (west side), at a distance of about 1–2 km on either side of the site contributing to the atmospheric extinction. However, now although these activities either stopped or moved away from the site, there is a major activity of construction of residential complexes and transport activities, which load the atmosphere with a variety of aerosol particles at lower levels, which in turn get lifted up by surface wind and by convection and thereby affect visibility and air quality over the entire urban region. The possible aerosol type present over the station is a mixture of water-soluble, dust-like, and soot-like aerosols (Ernest Raj et al., 2003), and the accumulation-mode aerosols are mostly due to anthropogenic activities, particularly from vehicular exhaust (gas-to-particle conversion).
3. Data and instrumentation
Continuous observations on aerosols, ozone, and precipitable water are being conducted at Pune since May 1998 using a set of hand-held compact 5-channel Microtops II (Microprocessor-controlled total ozone portable spectrometer; Solar Light Company, USA) Sun photometer and Ozonometer. The sun photometer provides aerosol optical depth (AOD) at wavelengths centred at 380, 440, 500, 675, and 870 nm (FWHM 10 nm). The ozonometer has three UV channels (305.5, 312.5, and 320.0 nm, having a FWHM band pass of 2 ± 0.3 nm) for total column ozone measurements, and two near-IR channels (940 and 1020 nm) for precipitable water measurements, and also AOD at 1020 nm. A Global Positioning System (GPS) receiver attached with the photometers provides the information on time and location. Typical error in AOD measurements is ± 0.03 (Morys et al., 2001; Porter et al., 2001; Ichoku et al., 2002). A quartz window provides access to the collimator tubes for the five photodiodes and a Sun alignment target. The 5 optical collimators are accurately aligned with full field of view of 2.5°, and the baffles fitted inside eliminate internal reflections. Each channel is fitted with a narrow-band interference filter and a photodiode suitable for the particular wavelength range. A spring-loaded door protects the quartz window when the instrument is not in operation. Initially, the date, universal time, and geographical coordinates of the location needs to be entered manually into the instrument's keypad, or automatically by the GPS receiver. The observer opens the window and points the sun photometer/ozonometer toward the sun until a bright spot of light is centred over the crosshair arrangement in a sun target window provided on the front panel of the instrument. A scan button is then pressed to start a programmable number of rapid scans of each of the five channels. The electrical signals from the five photo detectors are amplified, converted to digital form and a self-contained microprocessor automatically calculates AOD at 6 wavelengths, total column amounts of ozone, precipitable water, and the irradiance at each wavelength.
As a general practice, observations are carried out from sunrise to sunset on all days when the sky is clear, at 10 minute-intervals during the periods following sunrise and preceding sunset and at 30 minute-intervals throughout the rest of the day. Observations are also made on partly cloudy days mostly in the months of May–September whenever the field of view to the sun is clear of clouds. An important advantage of the present sun photometer when many clouds are present is that, an AOD or ozone measurement can be made within a few seconds when the sun is not obstructed. Thus, on each clear sky day about 25 to 30 measurements of AOD at each of the 6 wavelengths spread over the daytime are made. It was not possible to operate these systems throughout the daytime during the southwest monsoon months of July to September due to persistent cloud presence and rain, and even if it was operated, it was only for a small duration/part of the day. The average of such data may not be truly representative of the day's average and, hence, such data has not been presented here resulting in a data gap during the months of July and August. Thus, quality-controlled daily mean data of AOD at 6 wavelengths obtained on 1572 days during the period 18 May 1998–31 December 2007 has been used in this study to examine long-term trends and seasonal variations in aerosol loading over this tropical urban station. The number of available days of AOD measurements during the four seasons of pre-monsoon, SW monsoon, post-monsoon and winter are 542, 334, 78, and 618 respectively. The results are presented and discussed below.
4. Results and discussion
The daily mean data of aerosol optical depth at 6 wavelengths for the entire 10-year period (May 1998–December 2007) is taken, and overall average AOD, standard deviation, and coefficient of variation (COV) (in %, computed as (ratio of standard deviation to mean) × 100) at each wavelength are computed and shown in Table I. As per spectral (wavelength) dependence of AOD, AOD at the shorter wavelength, 380 nm, is highest (0.593 ± 0.11) and it decreases with increasing wavelength (0.220 ± 0.09 at 1020 nm). Smaller-sized aerosols contribute maximum to solar extinction at shorter wavelengths and due to their relative abundance and longer lifetime in the atmosphere, AOD at shorter wavelength is higher. The long-term mean AOD at 500 nm at this tropical continental urban station is 0.447 with a standard deviation of ± 0.08. A multi-year (2001–2008) analysis of aerosol optical properties retrieved from a sun/sky radiometer at Gosan, Korea, showed that monthly mean AOD at 675 nm ranged between 0.12 and 0.36 with high AODs (>0.33) from April to June (Sang-Woo Kim et al., 2010). In the present study also it was found that overall mean AOD at 675 nm was 0.317 ± 0.05 with a high value (0.333 ± 0.04) during the pre-monsoon (March–May) months. Thus, the long-term optical measurements made at two stations in the Asian region seem to be in agreement because of near-similar climatic conditions. It is observed that variability in AOD at this location is quite high in magnitude with the CV ranging between 34 and 49%. Variation is relatively higher at longer wavelengths. This is again because larger-sized aerosols contribute maximum to extinction at longer wavelengths and they have relatively smaller lifetimes in the atmosphere. Any larger-sized aerosols released into the atmosphere due to surface winds are removed faster by sedimentation. Atmospheric aerosols are, by nature, thus highly variable in space and time. At this tropical location, with its varied weather conditions (wet and dry seasons), the high values of CV are mainly due to the large seasonal variability. To examine further the intra-seasonal variation, the long-term mean AOD and corresponding standard deviations are computed separately for the three seasons, namely, pre-monsoon (March–May), post-monsoon (October–November) and winter (December–January) and shown in Table II. Wavelength dependence of AOD during the three seasons is similar to that seen for the entire 10-year period. Interestingly, at 500 nm, the seasonal mean AODs are nearly alike and close to the overall yearly mean. But at the two shorter wavelengths (380 and 440 nm) AODs are higher in winter than in the pre-monsoon, whereas, at the three longer wavelengths (675, 870 and 1020 nm) AODs are much lower in winter than in the pre-monsoon. But from long-term (1997–2005) AOD measurements at Thessalonika, Greece, Kazadzis et al. (2007) have shown that mean AOD at 340 nm varies from 0.33 in winter to 0.53 in summer. Such a seasonal behaviour could be due to origin of different types of air mass in that region. Generally, aerosols of relatively smaller size (mostly of anthropogenic origin) contribute maximum to aerosol extinction at shorter wavelengths. Aerosols of larger size and of natural origin (mostly soil dust) contribute to the extinction at longer wavelengths. Thus, the seasonal mean AOD data shows that more number of larger-size aerosols seem to be present during the pre-monsoon months. This is because at this location during the pre-monsoon, surface winds are strong, hot and dry conditions prevail, and so a large amount of aerosols of soil origin get lifted into the surface layers and later get carried aloft by convective processes. In winter, strong nocturnal inversions are frequent and whatever aerosols due to various human activities (domestic cooking, vehicular emissions, industrial emissions, etc.) are let out into the surface layers get trapped in the lower atmosphere due to less ventilation. Also, due to calm wind conditions, aerosols of soil dust type are less during winter. Therefore, AODs are relatively higher at shorter wavelengths during winter compared to that in the pre-monsoon season. Mean AODs during the transition period of October–November (post-monsoon) are slightly less in magnitude at all wavelengths compared to that in the pre-monsoon and winter periods. Most of the aerosol load that gets into the atmosphere during the pre-monsoon is removed by rain washout and cloud scavenging processes during the summer monsoon (rainy) season. COV on a seasonal mean scale is around 20% or less at all wavelengths except at 870 nm. This is less compared to the variation in annual mean and implies that nearly 15–20% of the total variation is due to seasonal effect.
Table I. Long-term mean and standard deviation in AOD at 6 wavelengths over Pune during the period 1998–2007
Coefficient of variation (%)
Table II. Mean and standard deviation in AOD at 6 wavelengths over Pune for pre-monsoon, post-monsoon and winter seasons during the period 1998–2007
To examine month-to-month variation in multi-spectral AOD, monthly means have been computed for all the 6 wavelengths for the entire period (1998–2007) and shown plotted as histograms in Figure 2. Month number 1 corresponds to daily mean data during January of all the years and so on. Interesting seasonal wavelength-dependence in AOD is seen here. At shorter wavelengths (380 nm and 440 nm) shown in the first two upper panels, the mean AOD is higher during winter months (December, January, February) and the pre-monsoon months (March–May), and it is lower during the middle of the year. Exactly opposite seasonal variation is seen at longer wavelengths (675 nm, 870 nm, and 1020 nm). Here, AOD is smaller in winter months with a broad maximum during the middle of the year. There seems to be no systematic month-to-month (seasonal) variation at the intermediate wavelength of 500 nm. As mentioned above, aerosols of relatively smaller sizes contribute to aerosol extinction at shorter wavelengths and vice versa. After the southwest monsoon rainy season, there seems to be a constant buildup of aerosols of smaller size (mostly of anthropogenic origin) right from October, increasing during winter months and peaking during the pre-monsoon summer season (March–May). Whereas, aerosols of larger size (mainly of soil dust origin at this location) are small in number during winter months due to weak surface winds and stable conditions. They start increasing in number and contribute to aerosol extinction only from March onwards when strong surface winds and high temperatures prevail.
Daily mean AOD at 6 wavelengths for the entire length of the 10-year period (May 1998–December 2007) is shown plotted in Figure 3 to examine the day-to-day variability and long-term trends, if any. Day number 1 in all the 6 panels corresponds to 28 May 1998, and the data gaps are due to non-availability of data during the SW monsoon months. The solid straight line in all the panels corresponds to the linear best fit to the daily mean data of AOD. As mentioned above (Table I), one can see the large day-to-day variation in AOD (34 to 49%) at this location on a daily scale. The linear change of a variable, such as AOD, in a climatological study is often measured with linear long-term trend in terms of percentage change per decade (Zhao et al., 2008). Here in this study, the AOD long-term trend is the slope of the linear regression line for the time series of daily mean AOD. It is clearly seen that a significant long-term increasing trend exists at shorter wavelengths (380, 440, 500 nm). Linear change in AOD (% per decade) at all the 6 wavelengths over the 10-year period is calculated and shown in Table III (column 2 as that for overall data). Increasing trend is observed at all the wavelengths except at the longer wavelength of 1020 nm, where a small decreasing trend (7.5% per decade) is found. At 380 nm the increasing trend is almost 45% per decade. Investigating long-term trend (for 10.5 years) in aerosol variables at a high alpine site, Coen et al. (2007) reported a positive trend of 45–50% per decade during September in the aerosol backscattering coefficient. For a more realistic comparison, Moderate Resolution Imaging Spectroradiometer (MODIS) satellite-derived data of AOD at 550 nm wavelength over the Pune geographical region for the period February 2000–December 2007 has been collected and monthly means have been computed. Figure 4 shows the time series of monthly mean AOD from MODIS along with the linear best-fit line. There seems to be a smaller increasing trend (∼8.4% per decade) in the MODIS-derived AOD at 550 nm compared to an increasing trend of ∼19.4% per decade in ground-based sun photometer derived AOD at 500 nm (Table III). Percentage change per decade in AOD separately for three seasons, namely, pre-monsoon, post-monsoon and winter are computed for all 6 wavelengths and the values are shown in Table III. At Pune, on a seasonal scale, increasing trends are higher in both pre-monsoon and winter months. Observed long-term increasing trends at almost all the wavelengths together with higher mean AODs during winter could be a possible indicator to increasing anthropogenic activities.
Table III. Percentage change in AOD over the period 1998–2007
According to Angstrom (1961), the wavelength variation of AOD (τ) can be expressed as an expression of the form τ(λ) = βλ−α, where β and α are Angstrom coefficients/constants which vary widely depending on environmental conditions and aerosol sources and sinks. Wavelength exponent (α) is a measure of the relative dominance of fine, sub-micrometer-sized aerosol over the coarse aerosols, while Angstrom's turbidity coefficient (β) is a measure of the total aerosol loading. Higher value of α signifies increased relative abundance of fine-size particles. From daily multi-spectral data of AOD, values of α and β are computed for the above 10-year Pune data. Mean values and their standard deviations are given in Table IV. The overall mean value of size exponent α at this urban location is 0.55, and the mean β is 0.76. It is also interesting to note that winter season mean α (0.63) is higher than in the pre-monsoon season (0.48). As a general rule, larger values of α indicate abundance of smaller-size particles, while smaller values of α indicate abundance of larger/coarse size particles. Day-to-day variations in α and β for the period of study (1998–2007) are shown plotted in Figure 5. A linear best-fit line is drawn to the temporal series data to examine the long-term change in both the derived parameters. It is seen that at this continental urban station, both size exponent (α) and turbidity coefficient (β) are increasing steadily at the rate of 25.3 and 8.4% per decade, respectively. Increase in turbidity coefficient (aerosol loading) is in line with the above discussed increasing trends in AODs at almost all the wavelengths. Observation of high increasing trend in Angstrom-size exponent at this location is interesting. A steady increase in α over the 10-year period clearly points out that over the years due to increasing urbanisation and human activities, more and more fine-sized aerosols are being added to the atmosphere over Pune. This should be cause for concern as these fine-mode particles degrade air quality and can have adverse effects on humans and the environment itself. Since a number of motorised vehicles in the urban area is one good indicator of urban/human activity, yearly total number of registered vehicles in the city of Pune for the years 1998–2008 has been shown plotted in Figure 6. The number of vehicles on the road is increasing at a rate of nearly 62 000 vehicles per year (∼7.4% per year). Vehicular emissions contribute directly to primary particles in the atmosphere as well as gases which on nucleation (gas-to-particle conversion) grow into fine-sized suspended particles. Thus, it is very clear that increasing human activities like population growth, vehicular emissions in a fast growing urban centre like Pune is reflected in observations of long-term aerosol optical properties reported here.
Table IV. Long-term mean and standard deviation in aerosol size exponent (α) and turbidity coefficient (β)
Size exponent (α)
Turbidity coefficient (β)
Aerosols are suspended particulate matter which constitute an important atmospheric constituent through aerosol–cloud–climate interactions. Long-term (1998–2007) variations in multi-spectral aerosol optical depth measurements made at Pune, India, a fast evolving urban centre using optical remote sensors revealed the following: (1) long-term mean AOD at 500 nm is 0.447 ± 0.08 with a COV ranging from 34 to 49% in the spectral range 380–1020 nm. Nearly 15–20% of this variation is due to seasonal effect; (2) AODs are higher at shorter wavelengths in winter and at longer wavelengths in the pre-monsoon. This indirectly implies that relatively smaller-sized particles are more in winter and larger-sized particles of soil dust origin are more during the pre-monsoon season; (3) daily mean AOD at all the wavelengths in the range 380–870 nm show significant long-term increasing trends; the increasing trend at 380 nm being nearly 45% per decade. Satellite-derived AODs over the same location also indicate increasing trend; (4) overall mean value of Angstrom-size exponent (α) at this urban location is 0.55 and the mean turbidity coefficient (β) is 0.76. Winter season α value is higher than that in the pre-monsoon season; and (5) long-term increasing trends in α and β (25.3 and 8.4% per decade, respectively) points out that with increasing urbanisation and human activity, more and more smaller-sized aerosols are being added to the atmosphere over Pune.
The authors are thankful to the Director, IITM, for his constant support and encouragement. Several research students who worked in the research group contributed to the sun photometric observational program over the years at different periods of time, and the same is gratefully acknowledged. The authors thank the three anonymous referees for their valuable comments and useful suggestions.