Novel Technique for Profiling of Aerosol , Ozone and Water Vapor during Winter Using Mobile Radiometers over a Hilltop Station Panuganti China

Following a novel approach, the vertical distributions of columnar aerosol optical depth (AOD), precipitable water content (PWC) and ozone (TCO) have been determined using compact, multi-filter, solar radiometers during winter period of 2002–2003. These profiles were obtained by making measurements at different altitudes while ascending to/descending from a rural site, elevated up to an altitude of about 1450 m above mean sea level (AMSL). Besides the wavelength dependency, the profiles depict significant layer structures which are explained on the basis of concurrent atmospheric stability parameters. The aerosol size distributions which were obtained from the inversion of spectral dependence of AOD at different altitudes show a mixture of power-law, monoand bi-model distributions indicating the influence of aerosols originating from both anthropogenic and natural sources. The present results, representing a rural high-altitude station, are compared with those reported over a few selected similar high-altitude stations in North India, and also observed at a nearby urban station (Pune) in Central India to infer modulation of terrain-induced meteorological parameters on aerosol source strength. The importance of the experimental approach and profiles of AOD and pre-cursor gases over areas with scarce alternative measurements, networks with sporadic presence of ground sites and limited number of satellite retrievals is highlighted.


INTRODUCTION
There has been mounting evidence of the importance of aerosols in global climate change through their direct and indirect effects (IPCC, 2007).The magnitude of these effects is poorly constrained because of our limited knowledge of the processes that control the distributions as well as physical, chemical and optical properties of aerosols.Atmospheric aerosols and precursor gases exhibit large temporal and spatial variations due to variety of production, transport and removal processes.Aerosols in the boundary layer are directly produced from the natural and anthropogenic processes while those in the troposphere and aloft are largely due to gas-to-particle conversion processes.Hence, vertical (altitude-resolved) distributions of columnar aerosol and precursor gas parameters in the height region, encompassing boundary-layer and free troposphere, play an important role in the transport and transformation of these constituents from source regions due to processes such as entrainment, mixing etc.Moreover, such measurements yield light extinction for a certain layer.Airborne measurements in this direction utilizing tracking sun-photometers (AATS) and coincident satellites (MODIS, TOMS, GOES) have been reported in the literature (Russell et al., 1999;Livingston et al., 2003, Shinozuka et al., 2011).Combined lidar and sun-photometric measurements of columnar aerosol optical parameters over Taipei, Taiwan have been reported by Chen et al. (2009).Their results showed sensitivity of columnar water vapor on the association between aerosol extinction, size distribution and phase.Similar kind of measurements made over a tropical urban station (Pune, India) revealed that the trends in columnar AOD are not confined to boundary-layer but also extend up to stratospheric altitudes.In addition, the results indicated that the boundary-layer AOD contributes more than 20 per cent to the total column AOD (Devara et al., 2012).
Measurements over high-altitude stations yield background levels of aerosol concentration (Arya and Jain, 1997;Reiter et al., 1984;Mahadevan et al., 1989;Jain, 2001;Dani et al., 2003;Jain and Arya, 2004;Sagar et al., 2004;Ghude et al., 2005;Jain et al., 2007;Gautam et al., 2011;Srivastava et al., 2012).So, it would be possible to examine and assess the extent to which the 'clean' remote areas have been affected by growing urbanization/industrialization (Jaenicke, 1979).Moreover, background sites with presumably cleaner environments and clear-sky conditions offer an excellent opportunity to calibrate the performance of the optical monitoring sensors/equipment (Jayaraman, 1999;Sumit et al., 2012).Added, these stations lie in the boundary-layer during daytime and in the free troposphere during the nighttime, thus provide good opportunity to investigate the transport/mixing of aerosols and gases from/between the boundary-layer to/and the free troposphere.Under seasonally varying wind patterns, these stations being in the free troposphere during night, becomes very important for regional study of transport of pollutants, particularly during early morning/night transition period.
Due to large heterogeneity in geography, climate, urbanization and population, the Indian subcontinent provides an interesting scenario where regional scale features would dominate distribution of any atmospheric constituent including aerosols.Systematic studies on atmospheric aerosols were initiated in India under the Indian Middle Atmosphere Program (IMAP) and Aerosol Climatology and Effects (ACE) projects (Krishnamoorthy et al., 1999) of Indian Space Research Organization -Geosphere Biosphere Program (ISRO-GBP).Although, aerosol properties have been measured at many sites in India using direct and remote sensing techniques from ground, aircraft and ship in several campaign modes in the last more than two decades, only a few of them have fairly long-term data (Krishnamoorthy et al., 1999;Devara et al., 2002;Gautam et al., 2009;Dey and Girolamo, 2010;Dani et al., 2012 and others).In spite of unique importance, as explained above, aerosol measurements over high-altitude stations are very sparse (for example, hilly and forest regions in particular) in India (for example, Guleria et al., 2012;Srivastava et al., 2012;Vijayakumar and Devara, 2013).The present paper aims at (i) to obtain vertical distributions of aerosols and pre-cursor gases (ozone and precipitable water content in the present study) using a mobile dual multi-filter solar radiometer, (ii) to describe the experiments conducted at Sinhgad, a high-altitude, hill-top, remote (rural) station, and (iii) to present principle of operation of radiometers, method of data acquisition and analysis, and discuss the results obtained and summarize.

ABOUT EXPERIMENTAL SITE
The experimental station, Sinhgad (18°21′N, 73°43′E) is a rural (remote) hill station, about 40 km south-west of the city of Pune, India.It is situated on a hill rising 1450 m above mean sea level (AMSL).This fort is a twelfth century old structure and is one of the historical places in Maharashtra State.Fig. 1 illustrates the geographical location of Sinhgad and the associated enlarged picture below shows the view of a mountain-top area with experimental site (closed circle shown by an arrow), where the observations were carried out is a flat terrain with an area of about 0.5 square kilometers, and surface is composed of rocks and rock-dust.Other mountain peaks of comparable height also located in its neighborhood.This place is favored outing site for people in and around Pune during weekends.A few people live at the summit and some tourists visit the area (by foot).Human activities are generally more during the day time (1000-1800 hrs) whereas during night, very few people stay at the fort.As this site is situated in the complex of a Micro-wave tower building owned by Bharat Sanchar Nigam Limited (BSNL), Government of India, it is a protected area and trespassers or tourists are not allowed to enter.The only local source of pollution is wood-burning mainly for cooking and an intermittent small-scale vehicular traffic during daytime.In winter, the villagers burn the dry grass along the slope of the hill to make the way for their transit.Hence, the anabatic (warm upslope) flow during daytime and katabatic (cold down-slope) flow during nighttime are prevalent over the site.These flows help in formation and dissipation of haze within/from the valley floor during winter.During winter through pre-monsoon, the surrounding area is generally dry and production of major pollutants is from local sources such as domestic cooking and also some grass-burning or charcoal-making.Due to these typical conditions and far distance from urban activity, Sinhgad is considered as a rural background location.The weather at Sinhgad during November to February months comprises a synoptic northwesterly to northeasterly circulation, dry ambient atmosphere with relative humidity ~30 to 70% and scanty rainfall.The temperature varies from 4°C (nighttime) to 35°C (daytime).The December month is the coolest of all months with lowest occurring nighttime temperature.The winds are highly variable during daytime.The experimental station experiences aerosol pollution in the boundary-layer during winter season due to its valleylike configuration and associated meteorology.

SOLAR RADIOMETER, PRINCIPLE OF OPERATION, DATA ARCHIVAL AND METHOD OF ANALYSIS
Two compact, on-line, multi-filter solar radiometers (MICROTOPS-II, Sunphotometer Model 540 and Ozonometer Model 521, manufactured by M/s Solar Light Co., USA, factory-calibrated every year) were used in the present study.Fig. 2 displays photograph of the multi-filter solar radiometer of both sun-photometer and ozone versions, mounted on a wooden platform fixed to a tripod for achieving high stability, time synchronization between observations and easy focusing of radiometer to the Sun's disk.Each channel is fitted with a narrow-band filter and a Gallium Phosphide detector.The radiation, captured by the collimator and band-passed by the filters falls onto the photodiode, produces an electrical output proportional to radiant power (irradiance).These outputs measured at each filter are amplified and analog-to-digital converted, and finally stored, together with the time of observation provided by the built-in master clock, in the memory for further analysis.The radiometer is equipped with built-in algorithms for computing ozone, precipitable water content and AOD from the output of the amplifier recorded for each filter.Dual multi-filter solar radiometers equipped with Global Positioning System (GPS) and automatic weather monitor, carried with a vehicle, have been used for the measurement of aerosols, ozone, precipitable water content along with met parameters (pressure, temperature and wind).
Observations were carried out from bottom of the hill to top/peak, thus at different altitudes during the mobile track having atmospheric pressure difference of about 75 mb, covering over 12 height intervals.Each set of observations, covering all these pressure (altitude) levels, took less than an hour.On some experimental days, measurements were made both during onward and return journey.For each measurement, the site location (latitude and longitude) and pressure values have been fed to the sun-photometer/ ozonometer as per the instrument's requirement.The instruments were operated concurrently and obtained instantaneous values of AOD, TCO, PWC and meteorological parameters at each altitude on clear-sky days.The vehicle's engine was put off during the measurements to avoid contamination due to vehicular emissions.The variations in solar irradiance measured for different solar zenith angles (air mass) at each filter have been used for obtaining aerosol optical depth (AOD).The variations in AOD for spectral bands centered at 380, 440, 500, 675, 870 and 1020 nm have been utilized to retrieve columnar aerosol size distribution (ASD).The TCO and PWC were obtained by following the differential optical absorption method, applied to the irradiance observed by the radiometer around the UV and NIR wavelengths.A brief description of the methodology is given below.
Solar radiation traversing through the terrestrial atmosphere undergoes extinction due to three processes, namely, Rayleigh scattering by air molecules, Mie scattering by aerosols, and molecular absorption.Solar radiation is a composite of monochromatic radiations.Molecular and ozone (significant absorbing gas in the wavelength band considered in the present study) optical depths were computed by using the expression of Kneizys et al. (1980) and Teillet (1990), respectively.The sun photometer works on the principle of measuring the solar radiation intensity at some specified wavelengths and converts it to optical depth by knowing the corresponding intensities at the top of the atmosphere (TOA).The TOA irradiance at each wavelength was calculated via the well-known Langley method.For this, the expression given by Kasten and Young (1989) for the air mass computation was used.The instrument measures the irradiance signals at different wavelengths in mV, from which the absolute irradiance in Wm -2 is obtained by multiplying the signal with calibration factor (Wm -2 /mV).The calibration relies on a high-performance voltage reference with the temperature coefficient ≤ 0.001% per degree Celsius and long-term stability of ~0.005% per year.As suggested by Devara et al. (2001), Morys et al. (2001), Porter et al. (2001) and many others, determination of zero air mass intercept is one of the important sensitive parameters in the retrieval of AOD, TCO and PWC from sun photometric measurements.Typically, a 0.5% error in the zero air mass measurement gives 0.005 error in optical depth at low air mass.Moreover, the zeros air mass value serves as calibration constant when it is corrected for mean sun-earth distance (typically above 60° solar zenith angle).Possible error sources in the determination of this parameter and solutions have been discussed in the literature (Shaw, 1976;Kremser et al., 1984;Kasten and Young, 1989;Devara et al., 1996).At larger air mass, the errors in AOD decrease.This is the reason why most of the satellites, which are used to derive AOD make measurements near noon solar time.The full width at half maximum bandwidth at each of these wavelength channels is 2.4 ± 0.4 nm, and the accuracy of the sun-targeting angle is better than 0.1°.
Having obtained AOD at different wavelengths, aerosol size distribution (ASD) has been retrieved from the spectral variation of AOD by following the constrained linear inversion scheme (King et al., 1978;King, 1982) with the Fredholm integral as where r is the particle radius, m is the complex refractive index of aerosol particles, Q ext (r,λ,m) is the Mie extinction efficiency parameter and n c (r) is the columnar size distribution.Since n c (r) cannot be written analytically, a numerical approach is followed to separate n c (r) into two parts as n c (r) = h(r)•f(r), where h(r) is rapidly varying function with r and f(r) is slowly varying.Hence the above equation changes to In Eq. ( 2), the quadrature error will be less if f(r) is assumed to be constant.In that case, a system of linear equations results, which may be written as where A = ∫Πr 2 Q ext (r,λ,m) h(r)dr and ε is an error which arises due to deviation between the measured τ a and theoretical τ a (= ∑A ij .fi ).Initially, the Junge exponent (ν) is computed from the wavelength dependence of AOD and used as a zero-order weighting function, h 0 (r).By using h 0 (r) as an initial guess, first order f (1) values are evaluated using the equation where γ is non-negative Lagrangian multiplier and S ε is the measured covariance matrix, H is a mean diagonal matrix and superscript T denotes matrix transposition.This iteration procedure is repeated until the observed τ a comes closer to the re-computed τ a .The total column ozone (TCO, expressed in DU), which is equivalent to thickness of pure ozone layer at standard temperature and pressure, is measured by recording differential absorption of solar light intensity at wavelengths in the UV region (305.5 nm and 320 nm).The measurement at the third wavelength (312.5 nm) is used to correct for particulate scattering and stray light.The columnar precipitable water content (PWC, expressed in cm) is obtained from the radiance measurements made at 940 nm and 1020 nm channels of the radiometer.The estimation of PWC was made by following the differential optical absorption method applied the irradiance data archived at 940 nm (maximum absorption for water vapor) and at 1020 nm (less absorption or almost transmission for water vapor).More description of the instrument and method of deriving AOD, TCO and PWC from the solar irradiance observations including mathematical formulations can be found in the literature (Devara et al., 2001(Devara et al., , 2005)).Albeit the experiments were conducted on many days, useful observations could be obtained only on a limited number of days due to unfavorable sky conditions.Thus the archived clear-sky data on columnintegrated aerosols, ozone and precipitable water content on three typical winter days, namely, 16 December 2002, 03 and 21 January 2003 are presented and discussed in the following sections.Experiment was conducted only during ascent (onward journey) on 16 December 2002 whereas observations were obtained both during ascent (onward) and descent (return journey) on 03 and 21 January 2003.

RESULTS AND DISCUSSION
From the instantaneous solar flux measurements, columnar AODs at six wavelengths, each centered at 380, 440, 500, 675, 870 and 1020 nm have been evaluated using the internal calibration involving (i) internal barometer/altimeter for monitoring of atmospheric pressure and altitude of the experimental location, and (ii) global positioning system (GPS) receiver that provides the geographical coordinates of the location used for estimating the local air mass.The ozone monitor used in the present experiment utilizes the solar irradiance measured at three UV wavelengths, each centered at 305.5, 320 and 312.5 nm.The columnar PWC was obtained from the radiance measured at two NIR wavelengths centered at 940 and 1020 nm.Thus the measured AODs at the wavelengths of 380, 440, 500, 675, 870 and 1020 nm were utilized to retrieve columnar aerosol size distribution by applying the constrained liner inversion method developed by King et al. (1978).Meteorological parameters greatly influence the aerosol properties (Hänel, 1976;Nilsson, 1979;Devara et al., 1994).Temperature and relative humidity variations lead to gas-to-particle conversion and higher wind speeds which may result in horizontal advection of pollution leading to higher AOD and modified ASD.

Vertical Distributions Features Observed on 16 December 2002
The height profiles of AOD, TCO and PWC along with meteorological parameters (wind, dry-bulb and wet-bulb temperatures) observed on 16 December 2002 are shown plotted in Figs.3(a)-3(c).It is evident from the figure that AOD at all the three characteristic wavelengths decreases with increase in altitude.Moreover, the wavelength dependency (lower AOD at longer wavelength) in these variations can also be seen, particularly in the lower altitudes, which is consistent with the Mie theory.The coarse-mode particles (AODs at 1020 nm) are found to deviate from those at 380 nm and 500 nm, indicating different aerosol chemical and growth processes.The ozone profile started with low value initially, increased to 260 DU around 925 mb and thereafter maintained constant except a layer formation at around 900 mb level, showing almost similar structure as that of AOD.Like AOD, PWC also showed decrease with increase in altitude with a structure around 900 mb.Furthermore, all the three profiles of AOD, PWC and TCO exhibit significant layer structures, similar to those seen in the concurrent wind and temperature profiles.This reveals that the local meteorological parameters play significant role in the altitude structures of AOD, PWC and TCO.
Fig. 4 displays variation of columnar aerosol size distribution (ASD) on 16 December 2002, inverted from the spectral distribution of AOD at different pressure levels (altitudes), following the constrained linear numerical inversion scheme as suggested by King et al. (1978) and King (1982).Each frame of the figure shows a plot of dNc/dlogR, representing the number of particles per unit area per unit log radius interval in a vertical column through the atmosphere versus the radius in microns.Significant changes in ASD due to sudden changes in wind and temperature (both dry-bulb and wet-bulb) can be noted during the study period.One common feature that can be seen in all the size spectra is the dominance of fine-mode particles.Mostly mono-and at some altitudes, bi-modal distributions are seen on this day.This implies that both natural and anthropogenic sources are prevailing over the experimental site, with predominant accumulation-mode particles at higher altitudes resulting from long-range transport phenomenon.

Features Observed on 03 January 2003
On 03 January 2003, the experiment was conducted while ascending to and descending from the experimental site.The vertical profiles of AOD, TCO and PWC together with concurrent meteorological parameters during both onward and return journeys are depicted in Figs.5(a)-5(f).Compared to 16 December 2002, the values of AOD, PWC and TCO are higher on 03 January 2003, which may be due to time of observation.This can also be seen from the dry-bulb temperature recorded on both days.The experiment commenced around 0700 hrs on 16 December 2002 while it started around 1000 hrs on 03 January 2003.So, the aerosol emissions were low on 16 December as compared to 03 January.It can be seen from the figure that there are no much variations in height distribution of multi-spectral AODs, but the wavelength dependency can well be seen.Even though the variations are smaller, two aerosol layers, one around 920 mb and another around 880 mb levels can be seen, which are considered to be due to local meteorology, particularly driven by the wind field.The TCO variations appear to be smaller around a mean value of 290 DU while  the PWC decreased with increase in altitude.The variations in AOD, TCO and PWC during return journey (descending) are seen to be similar to those during the onward journey (ascending) except at the initial portion of the AOD profiles.
The initial decreasing trend during ascending and increasing trend during descending are considered to be due to difference in aerosol emissions at the lower levels.This can be understood from the dry-bulb temperature profiles recorded during both journeys.The aerosol layer formations and the responsible concurrent meteorological parameters seem to be similar for both journeys.
The height variation of aerosol size spectra during onward and return journeys are shown in Figs. 6 and 7, respectively.During both phases, the size spectra at all altitudes appear to follow power-law distribution except at 937 mb level where the spectrum shows mono-modal distribution.However, all the spectra show predominant fine-mode aerosol particles originating from domestic cooking or waste-burning activities associated with convective activity which lifts the fine-mode particles up (thereby dilution in the lower and accumulation in the upper altitudes) in the experimental region, as compared to the earlier experiment conducted in the December month.Thus a sparse contribution from wind-driven fine soil-dust cannot be ruled out.

Features Observed on 21 January 2003
Figs. 8(a)-8(c) depicts the height variation of multiwavelength AOD, PWC and TCO during both onward and return journeys on 21 January 2003.It was not possible to record meteorological parameters due to some technical problems.Except TCO, AOD and PWC showed decrease with increase in altitude.In contrast, the AOD profiles showed some fluctuations around respective mean values.Interestingly, all the profiles show a prominent peak around 920 mb pressure level which is considered to be due to sudden wind gradient around that altitude.Aerosol size distributions during onward and return journeys on 21 January 2003 are shown in Figs. 9 and 10, respectively.The size spectra on this day during ascent are found to be different from those during the descent period.The ascent spectra showed mostly bi-modal distribution involving both fine and coarse-mode aerosol particles from natural and anthropogenic sources, and at times the spectra showed power-law distribution with different slopes.The spectra during return journey show power-law or mono-modal distribution almost at all altitudes considered in the study.This implies that the relative contrast in convective activity between the onward and return journeys contributed more number of fine aerosol particles ejected from the lower parts of the atmosphere to higher altitudes.

Comparison with Other Sites
The heterogeneity in space-time variations in aerosol distributions mainly comes from the geographical location and/or experimental terrain, which modifies the wind flow patterns, of the experimental site.In this context, aerosol or pre-cursor gas concentrations over high-altitude sites (in the absence of local pollution sources and long-range transport)  represent background levels, against which the aerosol trends at any site can be evaluated.The complex terrains often complicate the situation.Albeit many high-altitude stations are established over the world, due to unknown degree of variability in natural versus anthropogenic source concentrations, these stations should be used with caution for estimating reliable trends.The columnar AOD, PWC and TCO measured by using ground-based solar radiometer from the present study are compared with those measured at other high-altitude stations of Northern India (Dani et al., 2003;Jain and Arya, 2004;Sagar et al., 2004), and a near-by urban station (Pune) of Central India in Table 1.Two major points are clear from the table -(i) aerosol and pre-cursor gases at remote high altitude locations were different from those observed at highly polluted urban locations, inverse relationship between AOD and station altitude and (ii) short-term changes in AOD, PWC and TCO can be attributed mainly due to changes in the properties and loading of aerosols in the atmospheric boundary-layer over the observing stations.It may be noted here that timesynchronized multi-site measurements (considering elevation of each site as altitude) in a close proximity can also provide height profiles of atmospheric constituents, but in the present study, such profiles at single station by making measurements during the ascent and descent have been demonstrated with better space-time resolution.

SUMMARY AND CONCLUSION
A novel approach wherein portable, on-line, multi-filter, solar radiometers have been used to investigate the vertical distributions of columnar AOD, TCO and PWC over Sinhgad, a high-altitude, rural station.The results obtained on three typical experimental days during the winter of 2002-2003 are presented.The wavelength dependence of AOD is found in accordance with the Mie theory.The aerosol size distributions, inverted from the spectral distribution of AOD revealed, height variation of aerosol size spectra involving mono-and bi-modal distributions, mixing of the local and long-range transported aerosols.The space-time variations in AOD, TCO and PWC and the associated layer structures are noticed to be with local meteorological conditions.The experimental approach presented here would be useful to obtain vertical profiles of AOD, TCO and PWC using portable radiometers carried by vehicle over hilly/mountain regions.Such an approach yields profiles of AOD and precursor gases over areas with scarce alternative measurements, networks with sporadic presence of ground sites and limited number of satellite retrievals.It is being planned to re-visit the site by repeating similar experiments during different seasons over a longer period to understand the influence of latest developments in the form of land-use pattern changes and associated anthropogenic activities such as road construction, irrigation, and plantation etc. on regional climate.

Fig. 1 .
Fig. 1.India map depicting the Maharashtra State and locations of Pune and Sinhgad.Enlarged version of experimental site (hilltop) with an arrow showing the location (filled circle) where the radiometric observations were carried out is shown in a diagram beneath the map.

Table 1 .
Comparison of AOD, TCO and PWC between rural (high-altitude) and urban stations.