Determination of Aerosol Characteristics and Direct Radiative Forcing at Pune

Simultaneous measurements of aerosol optical depth and incoming global solar flux were carried out with a MICROTOPS-II Sunphotometer and Eppley Precision Spectral Pyranometer over Nowrosjee Wadia College (NWC, Pune) as the nodal centre, and at Pune University (PU) and IUCAA Girwali Observatory (IGO), in a field campaign mode during December 2010–May 2011. Data was analyzed to determine the surface short-wave (SW) aerosol direct radiative forcing (ADRF, ∆F per unit 0.1 AOD) and to study the characteristics of the aerosols. The results indicate that ADRF shows significant day-to-day variability and co-varies with AOD. The cause of the day-to-day variation in ADRF is an anticorrelation between ADRF/AOD, and differences in the daily maximum minus minimum RH and temperature. At NWC, ADRF ranges between –37.7 W/m (highest) and –5.9 W/m (lowest). For 500 nm, ADRF takes values in the range –17.3 ± 7.1 W/m to –54.2 ± 5.5 W/m at PU, whereas the corresponding values at IGO are –15.1 ± 2.1 W/m and –36.6 ± 6.4 W/m. Monthly ADRF is at the minimum level in winter and maximum during the pre-monsoon period. The magnitude of AOD shows significant diurnal variability. In winter, the mean AOD diurnal percentage departure at 500 nm is positive in the morning and negative during the afternoon, and this is reversed in the pre-monsoon period. The diurnal cycle of AOD is related to the prevalent meteorological conditions, surface-based nocturnal temperature inversion in the atmospheric boundary layer (ABL), and influx of aerosols from different source regions.


INTRODUCTION
It is well known that in spite of a very small contribution to the mass of the atmosphere, it is found that aerosols significantly influence the Earth's radiation budget and climate (Shaw, 1983;Crutzen and Andrae, 1990;Charlson et al., 1992;Andreae, 1995;Subbaraya et al., 2000;Moorthy et al., 2001;Devara et al., 2005).Aerosols produce direct radiative effects in the atmosphere by increasing backscattered solar radiation and by absorbing solar and long wave radiation (Satheesh and Ramnathan, 2000;Houghton et al., 2001).They indirectly affect climate by changing the microphysical properties of clouds and their life span resulting into the modification of planetary albedo and precipitation regime thereby affecting hydrological cycle (Ramnathan et al., 2001).Based on their properties, aerosols can either have positive or negative contribution to radiative forcing in the atmosphere.When absorbing aerosols are present in the atmosphere, a positive forcing producing warming is found while if a negative radiative forcing is found it gives rise to a cooling effect.However, these effects exhibit large regional variations owing to the short residence times of aerosols which contribute to significant spatiotemporal variations in the aerosol concentrations, their chemical and optical properties.The spatial and temporal variations in aerosols highlight the need to quantify aerosol radiative properties and forcing on a regional scale.However, the limited information on aerosol properties and dynamics, particularly in the troposphere, is a major uncertainty.In fact, the confidence in current climate change predictions is still very low (Santos et al., 2008).
One major goal in making radiation measurements is to quantify the relationship between aerosols and the surface fluxes in this region.The changes in the surface fluxes, or radiative forcings, due to the varying aerosol quantity and composition are critical factors that drive climatic processes on planetary and local scales (Graham, 1995).At the surface, the radiative fluxes are dependent on the entire atmospheric column that interacts with the solar radiation.Knowledge of the radiative forcings and forcing efficiencies is essential in better understanding the major sources of uncertainty in modeling climate change in global circulation model (Hansen et al., 1998).

EXPERIMENTAL METHODS AND OBSERVATIONS
In order to assess the short-wave (SW) aerosol direct radiative forcing over Pune and adjoing regions, an intensive field campaign was conducted at Nowrosjee Wadia College (NWC) as nodal centre, and in short-term campaigns at Pune University (PU) and IUCAA Girwali Observatory (IGO) situated at about 80 km to the north-west of Pune from December 2010-May 2011 on completely clear sky, cloudless conditions.Measurements of short-wave (280-2800 nm) global solar flux, aerosol optical depth and surface meteorological parameters were carried out by employing ground-based Eppley Precision Spectral Pyranometer (PSP), MICROTOPS-II Sunphotometer and WXT-510 Weather Transmitter.

Solar Radiation Measurements
The surface short-wave global solar irradiance in the spectral range 280-2800 nm was measured by using recently calibrated and ventilated Eppley Precision Spectral Pyranometer (PSP).It has sensitivity of about 9 milli-volts per 1 W/m 2 and has a linear response of ± 5% up to 2800 W/m 2 .PSP was operated in conjunction with MICROTOPS-II Sunphotometer from morning till evening at 1-minute interval on cloud-free days during December 2010-May 2011 in field campaign mode from NWC, PU and IGO.Data at an interval of 1-minute was recorded on PC based data logging system and was used to get half-hourly solar flux values for each day.It is well known that the broad-band Pyranometer using thermopile detectors suffer measurement errors and biases due to their cosine-law response (incidentangle dependence), temperature dependence, and the socalled zero (thermal) offsets (Dutton et al., 2001).The cosine response error for the present pyranometer was 1% for incident solar zenith angles (Z) less than 70° and the temperature dependence errors were at most 1% for daytime air temperature of ~10°-40° that were encountered during the study period.Zero offset depends on the difference between sensor temperature and effective sky temperature and may be different during the day.This effect is usually smaller for ventilated pyranometer (used in the present study) in comparison to unventilated pyranometer (Markowicz et al., 2008).Drummond and Roche (1965) have shown that ventilation reduces zero offset by using the air to maintain a more uniform temperature over the surface of the instrument.By considering aforementioned uncertainties, we estimated the overall relative accuracy of solar irradiance measurements by using the Pyranometer to be within 2% (at most ± 15 W/m 2 for the total band global irradiance) for Z < 70°.

Aerosol Optical Depth Measurements
The aerosol optical depth (τ pλ ), AOD is a key parameter which determines the extent of aerosol's direct influence on Earth-atmosphere radiation balance at the surface and at the top of the atmosphere.AOD was measured at 10 minutes interval from morning till evening yielding about 50 observations daily during cloud-free conditions by operating a handheld MICROTOPS-II Sunphotometer (manufactured by the Solar Light Company, Inc.) at spectral channels centered at 440, 500, 675, 870 and 1020 nm wavelengths.AOD at any wavelength was calculated from the total optical depth (TOD, τ λ ) of the atmosphere by using the extraterrestrial solar flux corrected for the Sun-Earth distance, ground level measurements of the direct solar flux at that wavelength, and the optical depth due to Rayleigh scattering of gas molecules.For each individual observation, the optical depth at specific wavelength was determined using Lambert-Beer-Bouguer law as: where m is the relative air mass defined as the secant of the solar zenith angle (Z), I λ is the intensity of ground reaching solar radiation, and I 0,λ is the intensity of solar radiation at the top of the atmosphere, determined for each wavelength using the Langley plot technique based on the measurements at Mauna Loa observatory initially and subsequently at Mt. Sinhgad (amsl = 1450 m) situated to the south-west of Pune (Shaw, 1983).It is found that the error in calibration of the sunphotometer (I 0,λ ) is < ± 0.5% for measurements at different wavelengths for air mass one (for overhead sun), which yields ~0.005-0.03error in optical depth.At larger air mass, the errors in AOD decrease.Overall error in AOD measurements is ± 0.03 (Devara et al., 2001;Porter et al., 2001;Ichoku et al., 2002).Measurements were only made when there were no observable clouds in the vicinity of the Sun.Data were further screened for the influence of clouds.For this, the derived AOD values differing from daily mean AOD by 3 times the standard deviation are treated as cloud contaminated (Pandithurai et al., 2007).A total of 5900 measurements made over 118 days for the period of field campaign is considered for analysis.

Meteorological Measurements
Aerosol optical, physico-chemical and radiative properties are influenced by the surface meteorological parameters like air temperature (θ, expressed in deg K), relative humidity (RH), pressure, wind speed and its direction at the observing site (Maheskumar et al., 2001).In the present work, these parameters are concurrently measured at each observing site by employing Finland).The data obtained is used for correlation studies of AOD and weather parameters.

METHOD OF ANALYSIS
Aerosol direct radiative forcing is estimated by modeling and experimental techniques (Xu et al., 2003;Ramana et al., 2004).In the present work, direct radiative forcing due to aerosols is computed by using experimental method (Jayaraman et al., 1998;Conant, 2000;Latha and Badarinath, 2005).In this method, AODs measured by MICROTOPS-II Sunphotometer were correlated with independent groundreaching short-wave global solar flux measurements in the spectral range 280-2800 nm made by employing co-located Pyranometer.For correlation studies, in order to avoid influence of air mass, both instantaneous AODs and the corresponding short-wave global solar flux were normalized for air mass factor (m = sec Z, Z ≤ 70°) since the slant air column length is found to increase with increasing solar zenith angle.
A plot of normalized short-wave global solar flux against normalized AODs (at λ = 500 nm wavelength) was constructed to yield surface ADRF.The slope of the linear regression fit to each constructed plot yields a change in ground-reaching global solar flux at each wavelength.From this, ΔF per unit 0.1 AOD to be considered as the "surface short-wave aerosol direct radiative forcing" is determined.This is an important parameter in Earth-atmosphere radiation budget studies, since a change in surface solar flux can change the surface temperature and can cause evaporation of water vapour, and can therefore have significant influence on the climate system.The zero AOD intercept gives surface reaching short-wave global solar flux for no aerosol.Fig. 1 illustrates examples of this analysis for 500 nm wavelength on two typical days each in winter and pre-monsoon seasons.It is seen that the ADRF ranges from -27.8 ± 1.4 to -16.18 ± 2.8 W/m 2 .In spite of good correlation coefficient (varies between 0.74-0.93),from figure it is seen that there is some scatter of data points which may be ascribed to the variations in Rayleigh optical depth, resulting from its zenith angle dependence.The procedure is repeated for all the available days of observations at Pune and for the adjoining regions to derive daily, monthly and seasonal variation of ADRF.

Diurnal and Seasonal Variation of Short-wave Solar Radiation
Fig. 2 shows diurnal variation of short-wave global solar flux during winter and pre-monsoon seasons at the observing sites at NWC, PU and IGO.It is seen that the solar flux is symmetric about local noon and the noontime solar flux is less in winter than that in pre-monsoon.This is so because during winter the optical state of the atmosphere over the observing sites is more turbid than that in pre-monsoon. (2) It is found that the Gaussian fit is of good quality as depicted by the correlation coefficient, which takes values 0.99 and 0.98 for winter and pre-monsoon seasons respectively for data at NWC. Figure also shows Gaussian fits for PU and IGO, each having correlation coefficient equal to 0.98.

Diurnal Variation of AOD
Diurnal variability of aerosol optical properties has importance in atmospheric correction and validation of remote sensing data.It can also be used in the determination of aerosol radiative forcing and studying the interaction of aerosols with clouds and humidity (Smirnov et al., 2002).Fig. 3 shows the diurnal variation of AOD at five wavelengths (viz., 440, 500, 675, 870 and 1020 nm) on two typical days each in winter and pre-monsoon during 2010-11 at the observing site NWC.More or less similar pattern prevails at observing sites PU and IGO.The data for NWC is presented here.In general, it is observed that at Pune, the aerosol optical depth is higher in the morning (i.e., during 9:30-11:30 hrs).It decreases to a low value during afternoon superimposed with small peaks around 13:00 to 14:00 hrs.Analysis of data indicates that the diurnal variation on all observing days is somewhat different on different days.However, AOD has strong wavelength dependence, i.e., its value is higher at 440 nm and lower at 1020 nm.Sometimes, 500 nm curve coincides with that of 440 nm due to the proximity of wavelengths used in the observation.This is found to be the common feature on all the observing days.
The varying nature of the diurnal variability of AODs on different days in the current observing seasons at Pune is attributed to the presence of a mixture of aerosols from multiple sources.In order to investigate the influence of air masses originating from various source regions due to longrange transport and ending at the observing site on observed AODs at NWC, 5-day back trajectories are computed using the National Oceanic and Atmospheric Administration Hybrid Single Particle Langrangian Integrated Trajectory (NOAA HYSPLIT) Model (Draxler and Hess, 1998) for these days.5-day period for back-trajectory analysis is due to typical residence time of aerosols in the lower troposphere which is about one week.Since AOD is a measure of the are northwesterly and southwesterly, and originate from North Africa, Arabian Sea, etc. before converging on the observing site.This clearly indicates that the different types of air masses arriving from various source regions, carrying continental aerosols at the observing site since the backtrajectories essentially back-trace the course of aerosol parcels in space (latitude, longitude and altitude) and time (days), starting from the source of investigation at a particular height from the ground (Gogoi et al., 2009).Thus convection, lifting of aerosols from land originating from various source regions along with local meteorological processes plays a significant role in producing varying nature of AOD diurnal variation on different days at Pune.

Percentage Diurnal Departure of AOD
In order to delineate the more evident systematic diurnal trend in AOD, sampling procedure used by many researchers (e.g., Smirnov et al., 2002;Singh et al., 2004;Wang et al., 2004;Pandithurai et al., 2007)   departure in the morning (up to 12:30 hrs).In the afternoon, however, the departure gets reversed and is seen to be up to 35%.
For the pre-monsoon season, initially in the morning, AOD departure is negative and has magnitude of about -4 to -10%.As the day progresses, the departure becomes positive in the afternoon (i.e., in the time interval from 13:00 to 17:00 hrs) and varies between 2% and 30%.Thus, on an average, at NWC, for the winter season, AOD depicts positive departure during forenoon (FN) and negative departure during afternoon (AN) in relation to mean AOD.Percentage departure trend in AOD reverses in pre-monsoon season.This indicates that aerosol loading is high during FN and low during AN for winter and the opposite variation pattern prevails during pre-monsoon.

Effect of Meteorological Parameters on AOD Variability
The diurnal variation observed at Pune has its origin in the meteorological conditions prevalent over Pune in addition to the influx of aerosols from different source regions as depicted by the wind trajectory analysis in Fig. 4. Meteorological parameters like air temperature, relative humidity (RH), wind speed and wind direction are analyzed for both winter and pre-monsoon seasons to study their influence on the seasonal diurnal variability of AOD.The seasonal mean diurnal variation of RH and air temperature (Fig. 6(a)) for winter and pre-monsoon seasons during the observing season 2010-11, indicate higher temperature (~35°C) and lower RH (20-70%) during pre-monsoon.In winter, low temperature (up to 30°C) and higher humidity about 75-90% in the early morning hours and lower during The large difference in morning relative humidity (RH m ) and afternoon relative humidity (RH e ) values and low surface temperature during night with calm winds in winter are conducive to the formation of low-level capped inversion in the atmospheric boundary layer (ABL).In the ABL over Pune, a surface-based nocturnal temperature inversion (having intensity of about 8-10°K per 100 meter) is a common observation on most of the days during February and also to some extent in March (Vernekar et al., 1993).
Similar observations at Pune have been reported by Aher and Agashe (1996).The precipitable moisture present in the atmosphere during early morning hours condenses to produce haze particles via gas-to-particle conversion by photochemical reactions (Meszaros, 1981) thereby raising columnar aerosol loading leading to higher AOD in the morning.The inversion normally starts eroding about 2 hours after sunrise taking 4 to 5 hours for complete erosion.Around noontime, convective turbulence provides a thorough vertical mixing of aerosols and other species.This might result in a better vertical homogeneity of aerosols.Convective eddies provide a mean for efficient vertical transport of aerosols and effective exchange between the mixed layer and free atmosphere.Smaller aerosols with less inertia and large surface area (for given mass) are influenced by these buoyant forces and become distributed in the column.After dispersal/evaporation of haze particles around noontime, the columnar aerosol concentration reduces, producing smaller AODs in the afternoon in winter season.

Seasonal Spectral Variation of AOD and Angstrom Exponent (α)
The spectral dependence of AOD provides information about physical properties of aerosols.These can be derived from Ångström empirical formula as (Ångström, 1961), where wavelength exponent α is an indicator of fraction of accumulation-mode aerosols (radii< 1 μm) to coarse-mode aerosols (radii > 1 μm).β, a turbidity coefficient (which equals τ pλ at 1000 nm), is a measure of columnar aerosol loading.Values of both α and β are determined by evolving linear least squares fit between τ pλ and λ (in μm) on log-log scale over the spectral range of Microtops.The slope and intercept of the regression line are α and ln β respectively.From this, α and β values are estimated for winter and premonsoon seasons for the spectral dependence of AOD shown in Fig. 7(a).α takes the values 1.21 (± 0.06) and 0.65 (± 0.16) for winter and pre-monsoon respectively while corresponding β values are 0.22 (± 0.04) and 0.32 (± 0.09).
Magnitudes of α and β values depict transformation of winds.This is also seen from the frequency distribution of Angstrom exponent, which is mono-modal during winter (modal value = 1.15) while it is bi-modal in pre-monsoon with primary and secondary modal values equal to 0.25 and 0.95 (Fig. 7(b)), respectively.

Short-Wave Aerosol Direct Radiative Forcing
Day-to-Day Variation Analysis of the data indicates that there is a significant day-to-day variation in surface short-wave aerosol direct radiative forcing viz.(ADRF), ΔF per unit 0.1 AOD (Figs. 8(a), (b), (c) and (d)).Figure also shows day-to-day variation of AOD at 500 nm, daily maximum minus minimum temperature (T max -T min ) and daily max-min relative humidity difference (RH max -RH min ).It is found that ADRF and AOD co-vary while there is an anti-correlation between ADRF/ AOD and daily max-min RH and temperature differences.This shows that the surface meteorological parameters of temperature and RH exert a considerable influence on the day-to-day variation of AOD and ADRF.Daily ADRF varies from highest value of -37.7 W/m 2 to lowest value equal to -5.9 W/m 2 .Winter days generally show low surface forcing as compared to the pre-monsoon days.Table 1 shows day-to-day variation of ADRF on different observing days during March and April at PU and IGO, respectively.
It reveals that the ADRF takes values in the range -17.3 ± 7.1 W/m 2 to -54.2 ± 5.5 W/m 2 at 500 nm at PU. Similarly, the corresponding ADRF values at IGO are -15.1 ± 2.1 W/m 2 and -36.6 ± 6.4 W/m 2 .This indicates that there is relatively more change in ADRF values at these field sites.This is due to large fluctuations in daily AODs.Large variations in AODs are ascribed to two types of aerosol sources viz., local and influx of aerosols due to long-range transport.The local sources include dust and industrial soot like aerosols.Analysis of 5-day back-trajectories at PU and IGO during short-term campaign reveals that the source regions of long-range aerosol influx vary on day-to-day basis.Further, aerosol behavior is also controlled by the prevailing atmospheric conditions (i.e., variations in RH and air temperature) as well as due to diffusion and aging processes such as humidification, coagulation, scavenging by precipitation and gas-to-particle conversions processes.

Monthly/Seasonal Variation of ADRF
For studying monthly/seasonal variation of ADRF, its daily values are grouped season-wise for 500 nm (which is a mid-visible wavelength) producing monthly variation.Results shown in Fig. 9(a) indicate that at NWC, ADRF at 500 nm is minimum (i.e., -16.7 ± 2.6 W/m 2 ) in winter months (December, January and February) and maximum  (i.e., -27.9 ± 2.9 W/m 2 ) during pre-monsoon months (March, April and May).Similarly, monthly mean value of ADRF (given in Table 1) during March at PU is -31.9 ± 14.6 W/m 2 at 500 nm while the corresponding value during April at IGO is -26.1 ± 7.9 W/m 2 .Thus during pre-monsoon, ADRF is nearly same at NWC and IGO whereas at PU, the corresponding values are about 1.2 times higher than those at NWC and IGO.Pandithurai et al. (2004) have reported mean aerosol surface radiative forcing of -33 W/m 2 during dry seasons of year 2001-02 at Pune.In a road campaign experiment conducted during February 2004 in the central Indian region, as a part of Indian Space Research Organization Geosphere-Biosphere Program (ISRO-GBP), Jayaraman et al. (2006) have found surface forcing (i.e., reduction in the surface solar flux) ranging between -27.4 W/m 2 and -30.1 W/m 2 for onward and return journey, respectively.Satheesh et al. (2010) have found that a monthly scale reduction of surface irradiance due to the presence of aerosols varies from -30 W/m 2 to -65 W/m 2 at a highly polluted urban site, Bangalore in Southern India.
Similar observations have been reported by Sarkar et al. (2005) over India and by Dey and Tripathi (2008) over Kanpur region in the Indo-Gangetic basin.Sarkar et al. (2005) have shown the climatology of surface forcing over India and adjoining regions indicating maximum surface forcing over India due to build of aerosols because of dust events in Sahara during summer.During winter months, the surface forcing is lower but still considerable and mainly restricted to north Indian plains.Dey and Tripathi (2008) have found that the anthropogenic components viz., scattering water-soluble and absorbing BC contribute more than 80% to the composite aerosol optical depth at 500 nm in the winter whereas the natural dust contributes more than 55% in summer months.Aerosols induce large negative surface forcing (more than -20 W/m 2 ) with higher values (more than -30 W/m 2 ) during pre-monsoon season, when the transported natural dust add to the anthropogenic aerosol pollution.Thus, the present results corroborate well with the reported values in the literature.
Analysis of past AOD data at Pune University (Aher et al., 2001) and the present AOD data at NWC indicate that AOD is maximum in February (transition month) and summer (March/April) and minimum in winter (December-January).It is found that the observed high value in March/ April is primarily due to increased aerosol inputs in summer due to surface heating, long-range transport of aerosols, wind-blown dust and mechanical production of aerosols.In winter, AOD at Pune is about 4/5 of its high value in March/April.Low AOD in winter implies low aerosol input due to colder ground surface and wet removal processes showing that low winter AOD is mainly produced because of anthropogenic component.The observation that ADRF is minimum in winter and maximum in pre-monsoon, correlate well with the seasonal variation of AOD at Pune.

Frequency Distribution of Aerosol Radiative Forcing
The frequency histogram of ADRF at 500 nm is shown in Fig. 9(b) for all the days of observation at NWC.The frequency distribution for ADRF is found to be bi-modal and is rather narrow.The primary and secondary modal values of probability distribution are -27.5 W/m 2 and -17.5 W/m 2 , respectively.Gaussian regression is found to fit the probability distribution and regression equations have the same form as given in Eq. ( 2) with correlation coefficients of 0.81.

CONCLUSIONS
The main findings of the present study are as follows: (1) The short-wave global solar flux shows systematic seasonal diurnal variation pattern which is symmetric around local noon.The variation is represented by Gaussian distribution law at all the observing sites and is found to be dependent on the optical state of the atmosphere.
(2) Magnitude of AOD shows significant diurnal variability which is different on different days.In winter, mean Fig. 1.Examples of linear regression fit to the normalized short-wave global solar flux and normalized AOD at 500 nm.

Fig. 3 .
Fig. 3. Diurnal variability of AOD at different wavelengths during winter and pre-monsoon seasons at NWC.

Fig. 8 .
Fig. 8. Day-to-day variation of short-wave aerosol direct radiative forcing, AOD, daily maximum minus minimum temperature and relative humidity (RH) difference.

Fig. 9 .
Fig. 9. (a) Monthly variation of short-wave surface forcing and (b) Frequency distribution of short-wave aerosol direct radiative forcing at 500 nm at NWC during 2010-11.

Table 1 .
Short-wave surface aerosol direct radiative forcing over (∆F per unit 0.1 AOD) Pune University and IUCAA Girwali Observatory.