Retrieval of Columnar Aerosol Size Distributions from Spectral Attenuation Measurements over Central Himalayas

Extensive measurements of spectral aerosol optical depths (AODs) were made at Manora Peak, Nainital (29.4N, 79.5E, ~1958 m above mean sea level) in the central Himalayas, using a ten channel multi-wavelength solar radiometer during January 2002 to December 2005. Using these spectral AOD values, the columnar size distribution [CSD; nc(r)] function of aerosols have been derived. The CSD, retrieved from spectral AODs are, in general, bimodal (combination of power law and unimodal log normal distribution) with a prominent secondary (or coarse) mode occurring at a fairly large value of radius (r > 0.5 μm), while the primary (or fine) mode either does not appear explicitly or perhaps occurs below the radius ≅ 0.1 μm. The bimodal nature of CSDs indicates the presence of fine as well as coarse mode aerosols over the observational site. The effective radius, total aerosol number content and columnar mass loading computed from deduced CSD shows minimum values during winter (November to February) and maximum during summer (March to June) months. The share of sub micron and super micron aerosols to the total aerosol number concentration (Nt) indicates the dominance of sub micron aerosols to the Nt and it accounts for > 90% during the study period.


INTRODUCTION
Size distribution is one of the most prominent properties of atmospheric aerosols from the stand point of climate impact.Most of the effects of aerosols can be understood only with the knowledge of its size distribution.Many processes in the Earth's atmosphere concerning these aerosols are closely related to the aerosol size distribution therefore it is quite essential to understand the size distribution of aerosols as far as their effect on both climate (Charlson et al., 1987;Russell et al., 1994) and human health (Künzli et al., 2000) are concerned.The effect of aerosols on climate is generally classified into direct and indirect effect.The direct effect comes from the capability of aerosols to scatter and absorb the incoming short wave and out going long wave solar radiation (Charlson et al., 1992), while the indirect effect is a result of the ability of aerosols to act as cloud condensation nuclei; thereby affecting the cloud droplet size distribution, droplet concentration, optical properties, precipitation rate and lifetime of clouds (Twomey, 1977;Ackerman et al., 2000;Rosenfield, 2000).Furthermore, the aerosol size distributions along with particle refractive index and shape are one of the most important parameters, which determine their optical properties such as optical thickness, asymmetry factor, scattering phase function, single scattering albedo and other useful information needed in the aerosol radiative forcing estimation.
In India a number of studies were carried out on the aerosol size distributions using in-situ measurements.However, most of these studies are focused to the low altitude (< 1 km amsl) locations that encompass either urban/semi-urban landmass or oceans adjacent to densely populated and polluted coastal belt (Moorthy et al., 1997;Nair and Moorthy, 1998;Moorthy and Satheesh, 2000;Srivastava et al., 2008) and also the area nearby Indo-Gangetic Plains (Day et al., 2004;Singh et al., 2004).The site specific nature of aerosol properties is clearly reported by Ganesh et al. (2008).Such studies are lacking over the northern part of the Indo-Gangetic Plains, particularly in the Himalayan regions.Investigation from the remote, sparsely inhabited regions have the importance of providing a sort of background against which the urban impacts can be compared (Pant et al., 2006).From this perspective, observations from high altitude stations have prime importance.There is no such study at a high altitude (~2 km amsl) in the northern part of Indian subcontinent except recently started station at Nainital, located in the Central Himalayas (Sagar et al., 2004).In the present investigation we present the columnar size distribution (CSD) function inferred by numerical inversion of spectral aerosol optical depth (AOD) measurements over a period of four years (January 2002 to December 2005).The results and implications of these studies are discussed in the manuscript.

PHYSICAL FEATURE AND GENERAL METEOROLOGICAL CONDITIONS
The observational site Manora Peak, Nainital (29.4 o N, 79.5 o E, ~1958 m amsl) is located in the Central Himalayas at an areal distance of ~20 km from the plains of northern India.It is ~3 km due southwest of the Nainital main city.The observational site is surrounded by the undulated topography from three sides, east, north and west while to the south it has a deep valley region (below the mountain peaks) of the northern Indian plains.Being the hilly terrain the site is free from any major industrial activities and urban impacts.It has thick vegetation of the Central Himalayas, with floating populations of around ~1 million.The observational site has a varied topography as shown in Fig. 1, where the color code represents topographical features of the region surrounding Nainital.The details on the geographical location and meteorological conditions over the site have also been reported in earlier papers (Sagar et al., 2004;Dumka et al., 2008).

EXPERIMENTAL DETAILS
Measurements of spectral aerosol optical depths (AODs) which characterize the integrated extinction of solar radiation suffered during its transit through the earth atmosphere were made by using a multi wavelength solar radiometer [MWR] (Gogoi et al., 2009).The MWR makes continuous measurements of direct solar flux at 10 wavelength bands centered at 380,400,450,500,600,650,750,850,935 and 1025 nm, with FWHM 5 nm.The details of the instrument, method of data analysis and error budget are discussed in earlier papers (Sagar et al., 2004;Gogoi et al., 2009;Rana et al., 2009).Estimates of columnar AODs were made regularly on all clear sky day conditions during four-year period (2002 to 2005) at Nainital.These are averaged to get the monthly mean AOD spectra.A total 38 AOD spectra, thus generated during the above period, formed the basic data set in the present study.
The experimental data considered of spectral AOD, carry information about the aerosol size distribution.Using these spectral AOD (τpλ) values, the CSD function [nc(r)] of aerosols have been determined by numerical inversion of the Mie integral equation, where Qext is the aerosol Mie extinction efficiency parameter, which is a function of the aerosol complex refractive index (m), radius (r) and wavelength of the incident radiation (λ); nc(r) is the columnar number density of aerosols (in a vertical column of unit cross section in an infinitesimal radius range dr centered at r).In defining the nc(r) this way, it is implicitly assumed that the number size distribution is height invariant or averaged over the vertical column.The radii limits r 1 and r 2 to the integral are respectively the lower and upper cutoff radii of the particles, such that only those particles having sizes within the range r 1 to r 2 contribute significantly to Qext.Since the MWR measures only the directly transmitted flux, we used the spectral AODs and Eq.
(1) to retrieve the columnar size distribution.The spectral AODs from the measurements and the corresponding errors formed the inputs.Eq. ( 1) is solved for nc(r), following the iterative inversion procedure described by King et al. (1978); King (1982), as applied by Moorthy et al. (1997).This technique provides fairly accurate information about the aerosol size distributions if the Lagrange multiplier, refractive index, and radius range are carefully selected (King, 1982), particularly when the τ pλ spectra do not show large oscillations.The detailed application of this technique to the MWR data have been discussed by Moorthy et al. (1991); Saha and Moorthy (2004) and Gogoi et al. (2009).In the present study, we set r 1 = 0.05 μm and r 2 = 3.0 μm as optimal after performing a sensitivity analysis of the size range on Qext over the entire wavelength range of measurements used in the MWR to estimate τpλ (King, 1982;Moorthy et al., 1997).The wavelength dependent complex refractive index of aerosols depends on the aerosol chemical composition and as such it is different for different types of aerosols.As the chemical composition was not available to us due to lack of facility, the values of refractive index as a function of wavelengths are taken from the literature for different types of aerosols (Shettle and Fenn, 1979), for estimating the Qext values in Eq. ( 1).The spectral AODs are re-estimated using the direct Mie equation, after each iteration, and are compared with the input AOD spectrum, and the solution are accepted only when the reestimated AOD values agree with those from measurements within the measurement errors.As the measurement errors are used as input, the solutions are weighted by these errors with better accuracy around the size ranges sensitive to more accurate AOD measurements (King, 1982).The details of the retrieved CSDs, from the spectral AOD by means of inversion technique are presented in the following sections.

RETRIEVED SIZE DISTRIBUTIONS AND ITS PARAMETERIZATION
By following the procedure discussed in preceding sections, nc(r) is retrieved from a set of spectral AOD measurements.The CSDs deduced by the inversion of spectral AOD are presented in a graphical form by plotting n c (r) against r on a log-log scale.The CSDs retrieved during the study period are examined for the general features.An examination of all the retrieved CSDs, revealed a bimodal [a combination of power law (PL) and unimodal (UM)] in nature.A typical example of such type of distribution is shown in Fig. 2.This type of size distribution is characterized by a secondary mode occurring at a fairly large value of radius (r > 0.5 µm), while the primary mode may not be well developed or occurs at or below the radius 0.1 µm because the lower radii limit used in the inversion processes is much higher than that.The peak number density for secondary mode is generally two to three orders less than the primary mode number density.
In order to understand the observed changes in the deduced CSDs in terms of physical parameters of the aerosols, as well as to present them in the form that can be used for computation, it is essential to parameterize the CDSs using proper analytical functions, which depend on the physical properties of aerosols such as geometric mean/mode radii (rmi) and the geometric mean standard deviations (σmi).For this purpose we have used bimodal (PL + UM) log normal distribution using the equation: where N 01 and N 02 are scaling parameters which depend on the aerosol concentration, r m2 and σ 2 are respectively the mode radii and standard deviation of the secondary mode and ν is power law index.By evolving a fit between the retrieved CSDs and appropriate analytical functions with minimum rms error, the power law index, mode radii and standard deviations are deduced as applicable.In addition to the above, the other parameters representing the physical state of the aerosols were estimated following the expressions: For a given aerosol size distribution, R eff is a measure of the total volume to area of aerosols and gives the radius of an equivalent mono-dispersion that would exhibit the same scattering properties (Moorthy et al., 1997;Satheesh et al., 1999; and reference cited therein) where ρ is the density of aerosols, taken as 2.0 g/cm 3 (Junge, 1963).(5) The columnar content of the accumulation (Na) and coarse mode (Nc) aerosols were also estimated, considering particles smaller than 0.5 μm in radius to represent the accumulation regime.

RESULTS
While dealing with the monthly, seasonal and spectral variation of AODs at Nainital, it has been observed that there is a systematic change in the wavelength dependence of τp from winter (November-February) to summer (March-June) months (Dumka et al., 2008).These changes in spectral variation are indicative of the changes in CSD of aerosols.In the following paragraph, the above aspect are examined using CSDs retrieved from the λ dependence of τp values over Nainital and the results and implications are discussed.

Monthly and Seasonal Variation of CSDs
Representative CSDs for each month of the study period have been obtained by inverting the monthly mean τpλ values.The retrieved CSD generally show a steep fall in columnar number density nc(r) with increase in r from 0.05 μm (at the lower particle size; depicting an inverse power law type dependence), followed by a well defined secondary mode (coarse particle; depicting a unimodal type distribution) as shown in Fig. 2. At still larger size the number density n c (r) decreases again with r (see Fig. 2).The typical CSD obtained for December 2004 (Fig. 2) has two panels, the lower panel (b) representing the retrieved CSDs in a log-log scale, while the upper panel (a) depicts the measured AODs (solid points with error bars) along with the reestimated AODs (continuous line) from the retrieved CSDs as shown in panel (b).In Fig. 2, the n c (r) decreases initially upto r = 0.5 μm, where it falls by six orders in magnitude from its value 10 13 /m 2 /μm at r = 0.07 μm.From r = 0.5 μm onwards the value of nc(r) increases gradually, reaching a secondary peak (nc(r)~10 10 /m 2 /μm) at r = 1.0 μm and then continuously decrease with r.In such type of CSDs, for the smaller size (accumulation/fine mode aerosols) particles, sometimes no mode is explicitly seen and at times its presence at radii below r 1 is indicated by the slanting nature of the CSD towards the smaller values of r.Thus, the slanting nature of the size distribution indicates the occurrence of a primary/fine mode at a value of r lower than the lower limit (r 1 ) as considered in the inversion process.This is also physically justifiable as the number density cannot increase indefinitely because of the processes like as coagulation, which limits the concentration of the sub micron aerosols.This leads to the formation of an accumulation mode aerosols (Hoppel et al., 1990), by rapid transformation of smaller size aerosols to the larger ones.
In all, 38 such size distributions are obtained during the period of January 2002 to December 2005.It has been observed that all these retrieved CSDs can be analytically represented by a combination of power law and unimodal log normal distribution in nature in the optically active size range.The composite plots of monthly mean CSDs retrieved for each year are shown in Fig. 3, following the same convention as Fig. 2. The nature of all CSDs remains more or less same throughout the period under study.
As the aerosol properties over the Manora Peak, Nainital, have shown the well defined seasonal variations (Dumka et al., 2008), it is imperative to examine its signature in the size distribution.For this the daily AOD values at each wavelength were arranged by grouping them in seasonal ensembles mean AOD spectra, representative of the different seasons were obtained.The

Physical Parameters of Aerosol from CSDs
With a view to quantifying the monthly and seasonal changes in the CSDs in terms of the physical parameters of aerosols, these CSDs are parameterized using appropriate analytical functions as discussed above, and the parameters like r m , ν etc have been determined.The seasonal variations of the r m and σ m of the mode are shown in the lower and upper panels of Fig. 5, respectively.No systematic seasonal pattern is seen either in r m or in σ m .However, Fig. 5 indicates that the secondary coarse mode is broader and is generally consistent whereas the primary small/fine mode is explicitly not visible.In summer, the variability is either larger or comparable to those seen in other seasons.
With a view to examining the changes in columnar properties associated with the temporal changes in spectral AOD, the variation of the effective radius (Reff), median radius (R), columnar mass loading (mL), and columnar number concentrations of coarse mode aerosols Nc and total aerosols Nt are examined.Further, we have divided the Nt into two groups, the accumulation mode (small aerosol particles) and the coarse particles mode (larger aerosol particles) by integrating the CSDs, function from r 1 to 0.5 μm and from 0.5 μm to r 2 respectively.As the accumulation mode concentrations mainly pertains to the anthropogenic and transported (long range) species, while the coarse mode would represent mechanically generated particles, mostly due to the natural processes.As such the ratio Nc/Na would represent a measure of the relative abundance of the number concentration of natural (coarse) aerosols Nc with respect to the anthropogenic abundance.
The temporal variations of monthly mean Nt, Na, Nc and the ratio Nc/Na are shown in Fig. 6 from bottom to top panels respectively.The accumulation mode particles Na being larger than the coarse mode particles Nc by an order of magnitude, the ratio Na/Nc is mainly determined by Na.It was observed that during a year Nt, Na, Nc and Nc/Na increases from a minimum value during winter to a maximum value during summer months.An increase in the ratio of Nc/Na clearly indicates the dominance of the coarse mode aerosols in the size spectrum during the summer seasons as compared to the winter season and it is considered to be responsible for flattening of the AOD spectrum  ( Dumka et al., 2008).Moreover, the total columnar abundance Nt also increases approximately by a factor of 10, as the season changes from winter to summer, indicating a large increase in the columnar abundance.This increase is more for Nc than for Na and as a result the ratio N c /N a also increases.These results in a change in the aerosol spectrum which is also reflected in the effective radius (R eff ) and the columnar mass loading (m L ) as shown in the Fig. 7, which depicts the seasonal variation of R eff and m L in bottom and top panels respectively.Both m L and R eff , remain low during winter and high during summer.Though the seasonal variation of Reff is not conspicuous during 2003 and 2004 but m L shows the pattern consistently in all the years.The increase in the m L and R eff is attributed to the increase in the relative abundance of coarse aerosols in the size spectrum during the summer seasons.The value of R eff depends on the relative dominance of large to small particles where as the value of m L depends both on N t as well as the concentration of coarse mode aerosols (Moorthy and Satheesh, 2000).As such, the relative abundance of large aerosol increases the value of R eff while an increase in the columnar abundance of accumulation and coarse mode leads to an increase in m L .As the R eff is defined as the ratio of the total volume to area, the increase in the R eff during the summer seasons is attributed to the increase in the relative abundance of coarse mode aerosol concentration.This partly explains the absence of the constancy in R eff from year to year.Further, during the summer there is a data scarcity due to the onset of monsoon over the experimental site which resulted to unequal weightage to the aerosol parameters in different years.The inter parameter dependencies are examined in Fig. 8, which shows a scatter plot of m L and N c /N a against the R eff in the bottom and top panels respectively.It is clearly seen that, the relative dominance of N c (or N c /N a ) leads to significant increase with the R eff (Fig. 8; top panel) with a correlation coefficient of +0.53 at a significant level of 0.02.Similarly the columnar mass loading m L is also positively correlated with R eff ; with a much lower correlation coefficient +0.38.These similarities with earlier observations indicate that R eff is a batter proxy for monitoring the changes in the coarse mode abundance.

FN/AN Columnar Size Distributions
The aerosol measurements at Manora Peak, Nainital have shown that the AODs are found to be much lower as compared to the low altitude location in India (Sagar et al., 2004;Dumka et al., 2008).Further, afternoon (AN) AODs are generally higher than those of forenoon (FN) AODs, which was attributed due  to the evolution of boundary layer, which transports aerosols and pollutants from the nearby polluted and valley (below the mountain peak) regions to the higher altitude.Further, the role of boundary layer dynamics in influencing the aerosol properties at Manora Peak using multi-year measurements of spectral aerosol optical depths during January 2002 to December 2004 are examined by Dumka et al. (2008).With a view to examining the implication of short period (within a daytime) fluctuations on the CSD and its parameters, we have retrieved CSDs separately for FN and AN hour's.The typical CSDs for FN and AN part of the day are shown in Fig. 9(b), for the month of November.Fig. 9(a) shows the monthly mean AOD values measured during FN and AN period at different wavelengths (solid points with error bars) along with the re-estimated AOD values from the corresponding size distributions (continuous line).In Fig. 9(b) n c (r) is plotted against r on a log-log scale and which is clearly shows that during AN hour's (when the boundary layer has fully evolved) the aerosol number density is higher than the FN hour's (when the boundary layer is shallow), though the nature of size distributions remains bimodal log normal distributions during both FN and AN periods, there is a slight shift in the position of occurrence of modes (secondary mode/coarse aerosols).The retrieved size distributions have been parameterized using bimodal log normal function, following the procedure outlines as above.The value of mode radius and standard deviation during the FN and AN periods are 0.78 ± 0.22 and 0.59 ± 0.35 respectively.This shows that the size distribution shifts towards the abundance of smaller/fine particles in AN hour's; probably due to influx of large quantities of fine aerosols induced into the column.The physical parameters for FN and AN periods are estimated from the corresponding size distributions and given in Table 1.The table shows that there is a significant increase in columnar concentration of total and accumulation mode aerosols in AN period (except in October) and consequently the R eff of the size  - distributions.The effect is more spectacular during March to June period, when the solar heating of the land (over the Indo-Gangetic Plain) is quite intense leading to deeper convective mixing.This would lead to the lifting of effluents in the densely populated plains adjoining in the site to the higher levels, even above the peak, thereby inducting substantial amounts of urban aerosols in the column.Additional processes such as long range transport of fine mineral dust by the synoptic winds would also be playing their role during February to June months (e.g.Dey et al., 2004;Hegde et al., 2007;Nair et al., 2007).Notwithstanding these the local boundary layer dynamics causes perceptible changes in aerosol characteristics (Dumka et al., 2008).This is very important both in air quality assessment and aerosol characteristic to industrial urban areas where high altitude station lies close.

DISCUSSIONS
The CSDs retrieved from the measured AOD spectra consistently shows the bimodal (PL + UM) nature throughout the period under study from 2002 to 2004 (see Fig. 3).The similar type of size distribution is also reported by Rana et al. (2009).The wavelength dependency of AOD depicts a positive curvature towards the longer wavelength in the case of the bimodal nature of the size distribution.The secondary or coarse mode is generally occurring at a value of r > 0.5 µm, whereas the primary or fine mode is not conspicuous or occurring below the value of r < 0.1 µm, lower than the lower limit considered in the inversion technique.The occurrence of such mode suggests the presence of large abundance of nucleation (r~0.001 to 0.1 µm) or accumulation (r~0.1 to 1.0 µm) mode aerosols over the site.The occurrence of such mode is to be expected as the location is a remote, high altitude station lying in the central part of lower Himalayas and away from the strong sources of aerosol production.As the size spectrum is indicative of the particular source and sink of aerosols to which each mode can be attributed, therefore the bimodal nature of CSDs suggests two different sources of aerosols.However, the peak height of the secondary mode is almost 2-to-3 orders of magnitude less than that of the primary mode, which is not conspicuous (Fig. 2 & Fig. 3).Jaenicke (1984) and Hoppel et al. (1990) have suggested that the two modes observed in the CSDs are attributed mainly to two different production mechanisms: gas-to-particle conversion (GPC) and bulk-to-particle conversion (BPC) respectively.The primary small mode is attributed to the aged aerosols from the secondary GPC processes or phase reaction products as well as associated with the human activity on the landmass (Ramanathan et al., 2001).The GPC gives rise to particles size of the order of radius < 0.5 μm, while the BPC gives rise to the larger sized particles of the order of radius > 0.5 μm.After the production, the fine particles (r~0.1 μm) of the size distribution are controlled mainly by the coagulation and condensation processes whereas the coarse mode particles (r > 0.5 μm) are mainly controlled by sedimentation (Pruppacher and Klett, 1978) and impaction scavenging processes.The nature of the size distribution between these radii limits reflects the aerosol sources, sinks and the aerosol transport, which have regional importance.As near the aerosol source, the aerosol characteristics are well associated to the generation or production mechanism while, it is quite different on a global or synoptic scale, when present at large distances from the source region.Moreover, the various microphysical processes bring out the continuous size transformations which are quite important in limiting the concentration of fine mode aerosols.The appearance of bimodal nature of CSDs is indicative of additional sources which are responsible for secondary mode aerosols.As the observational site Nainital is away from the major industrial or anthropogenic activities, the observed changes in the CSDs function are brought in terms of either natural sources and sinks or other long range transport mechanisms.The long range transport is capable of bringing large amount of desert and mineral aerosols from the west Asian and the Great Indian Desert to the site and is responsible for the generation of secondary mode during the study period over the site.In addition to the advection by air masses, the increased solar heating of the landmass in nearby plains and the under lying valley region adjacent to the measurement site during the summer season would result in the increased convection activity and the evolution of the atmospheric boundary layer (ABL) during daytime will cause lifting of the aerosols and pollutants to the higher altitudes.Hence the columnar content as well as the mass loading is increased during the summer seasons.Similar nature of size distribution is also reported near Indo-Gangetic basin by many investigators (Dey et al., 2004;Singh et al., 2004;Jethva et al., 2005).They reported that the bimodal nature of the size distribution is due to the mixing of different type of air masses, having different aerosol populations (Hoppel et al., 1985) and the long range transport from the Great Indian Desert.

CONCLUSIONS
The main conclusions of our study are the following: 1. Aerosol columnar aerosol size distributions retrieved from the inversion of spectral AOD, in general, show bimodal (combination of power law and unimodal log normal) distributions, with a prominent secondary (coarse) mode occurring at fairly large value of r (> 0.5 μm), while the primary peak of fine mode aerosols does not appear explicitly.The basic shape of the CSD does not change significantly with the seasons.2. There is an increase in the effective radius and columnar aerosol mass loading from winter to summer seasons, attributed to the increase in the relative dominance of coarse mode aerosols in the size spectrum during the summer season.3. The aerosol physical parameters estimated from the CSDs have minimum values during the winter seasons and maximum during summer months.4. The share of sub micron aerosol to the aerosol number concentration shows the dominant role of sub micron aerosols to the total aerosol number concentration and it accounts for > 90% during the study period, which indicates that the sub micron aerosols contribute significantly to the aerosol loading during the study period.

Fig. 1 .
Fig. 1.The geographical location of Manora Peak, Nainital over the Indian subcontinent.The color code represents the topographical features (m) of the observational site.

Fig. 2 .
Fig. 2. (a) The input τpλ values estimated from the MWR measurements as the point with error.The continuous line represents τpλ values re-estimated using the retrieved CSDs.(b) typical example of CSD retrieved from monthly mean spectral AOD for the month of December 2004 at Nainital.

Fig. 3 .
Fig. 3. Right panel (b) shows the composite plots of CSDs retrieved from the spectral AODs during 2002, 2003, 2004 and 2005.Left panel (a) shows the spectral AODs re-estimated from the retrieved CDSs (continuous line) along with the corresponding measured AOD spectra (points) that formed the inputs to the inversion.

Fig. 4 .
Fig. 4. Left top panel shows the input seasonal mean of τpλ values as estimated during winter seasons of every year (solid points with error bars), while the continuous lines shows the re-estimated τpλ values from the respective CSDs.The bottom left panel shows the seasonal CSD retrieved from the respective seasonal mean spectral AOD.Similar plots of seasonal mean τpλ, re-estimated τpλ and CSD have been shown in the right top and bottom panels for summer season of every year.

Fig. 5 .
Fig. 5.The seasonal variation of mode radii and standard deviation of the mode radius.

Fig. 6 .
Fig. 6.Monthly variations of columnar aerosols content of the total (N t ) sub micron (N a ), super micron (N c ) and ratio (N c /N a ) over Nainital during 2002 to 2005 are shown in the bottom to top panels respectively.

Fig. 7 .
Fig. 7.The seasonal variation of m L (top panel) and R eff (bottom panel).

Fig. 8 .
Fig. 8. Scatter plot of m L and N c /N a against R eff are shown in bottom and top panel respectively.The dotted line represents the linear fit of the points.

Fig. 9 .
Fig. 9. (a) The input τpλ values estimated for FN and AN hour from the MWR measurements as the point with error.The continuous line represents τpλ values re-estimated using the retrieved CSDs.(b) typical example of FN and AN CSDs retrieved from monthly mean spectral AOD for the month of November at Nainital.

Table 1 .
The physical parameter of the FN and AN columnar size distributions during the study period.