1. Introduction

Long-term observations of aerosol characteristics are important in characterizing their spatiotemporal heterogeneities as well as in evolving climatological models for radiation budget studies. Numerous measurements and impact assessments have been reported from the Indian region in recent years (Satheesh and Ramanathan 2000; Ramanathan et al. 2001; Babu et al. 2002; Tripathi et al. 2005). However, most of these studies have focused on either the urban–semiurban landmasses or the oceans adjacent to the densely populated coastal belt. Investigations from remote, sparsely inhabited regions have the benefit of providing a sort of background against which the urban impacts can be compared. From this perspective, observations from high-altitude stations have significance as the aerosols in this region provide a “far-field picture,” being quite a distance away from potential sources and are more representative of free-troposphere conditions. Recent observations have revealed pockets of high aerosol loading–optical depth in the north Indian regions around the Ganga basin, particularly during the winter season when the prevailing meteorological conditions are favorable (Di Girolamo et al. 2004; Tripathi et al. 2005; Jethva et al. 2005). The observations from high-altitude stations are important, as they are essential in assessing the impacts of human activities on free-tropospheric aerosols and also in the modeling boundary layer to free-troposphere mass exchanges. Furthermore, most of the aerosol sources are located close to the earth’s surface and observations from high-altitude locations provide an opportunity to investigate their features from locations where most of the source fluctuations are smaller. The spectral aerosol optical depth (AOD) is a key parameter in order to build up a comprehensive picture of aerosols and their potential environmental impacts (Moorthy et al. 1999; Satheesh et al. 2002). Extensive measurements of spectral AODs from a high-altitude Indian station (Manora Peak in the Shivalik Hills of the central Himalaya Range) was initiated in January 2002, using a multiwavelength solar radiometer (MWR) under the Indian Space Research Organization’s Geosphere and Biosphere Program (ISRO-GBP) (Sagar et al. 2004). In this paper, we have used an extensive database of aerosol spectral optical depths spread over 3 yr (January 2002–December 2004) to study aerosols at high altitudes and to examine the possible role of synoptic- and mesoscale systems on seasonal and short-period variations in aerosol properties. The results and its implications are discussed.

2. Details of location and measurements

The experimental site, Manora Peak (29°22′N, 79°27′E, ∼1950 m above mean sea level) is located to the south of the hill station, Nainital (see Fig. 1), which has a total population of around 70 000. The site lies in the Manora Peak, Nainital in the central Himalayas. The color code in Fig. 1 indicates the topographical features of the region surrounding Nainital. The foothill stations are located to the southwest of the site at an aerial distance of around 10 km and about 300 m above mean sea level. Farther to the west lies densely populated regions including the Indian capital (New Delhi) and the southern region includes heavily the populated Gangetic plains. To the northeast of the Manora Peak is the hilly terrain of the central Himalayan ranges (>2.0 km MSL), which have essentially no anthropogenic activities and are rather pristine. To the south and southwest are densely populated valley regions (at a mean elevation of ∼300 m).

Regular and continuous measurements of spectral AODs have been carried out since January 2002 from Manora Peak, Nainital, using a multiwavelength solar radiometer described in detail elsewhere (Moorthy et al. 1999; Sagar et al. 2004). The MWR instrument has a field of view of 2° and measures AODs at 10 narrow-wavelength bands centered at 0.38, 0.40, 0.45, 0.50, 0.60, 0.65, 0.75, 0.85, 0.935 and 1.025 μm (with full-width half-maximum bandwidths between 4 and 6 nm). The instrumental details, methods of analysis, and error budget are described in Moorthy et al. (1999) and Sagar et al. (2004). The overall error in the estimation of AOD values is in the range ∼0.02–0.04 at different wavelengths (Shaw 1976). A monthly distribution of the datasets (from January 2002 to December 2004) is presented in Table 1. In all, 605 days of observations were made during the period under investigation. The monthly database set is generally strong for most of the period except during July–September, when the extensively cloudy conditions make it difficult to make observations.

3. General meteorological conditions

Regular measurements of the surface meteorological parameters such as wind direction (θ, °), wind speed (U, m s⁻¹), relative humidity (RH, %), temperature (T, °C), and total rainfall (RF, cm) are recorded at the site using a set of weather instruments. The monthly mean values of RF, U, and θ during the period under study are shown in Fig. 2. Winds are generally westerly and northwesterly during November–May, shifting to westerly–southwesterly during June–September. Generally, the monthly mean wind speeds are low (<5 m s⁻¹) for most of the year. The rainfall is significant in February (caused by western disturbances) and during June–September (the Indian summer monsoon season), which account for about 68% of the annual rainfall. The western disturbances (WDs) usually originate usually over the Mediterranean Sea–Black Sea areas or the west Atlantic, as extratropical frontal systems, but their frontal properties are lost while moving eastward toward India across Afghanistan–Pakistan. However, even then, an intense WD is capable of producing widespread heavy snowfall–rainfall over the Himalayan region and rains over the northern plains of India. The RH during daytime is generally <50% during November–April and thereafter increases steadily and reaches a peak value of ∼75%–85% during June–September (the monsoon period).

4. Results and discussions

In the following, we examine the temporal and spectral features in AOD and possible linkages of these to the changes in prevailing meteorological conditions.

b. Spectral variation of AODs

The spectral variation of AODs is important, as it is indicative of the changes in the aerosol size characteristics.

In Fig. 6, the variations of the monthly mean aerosol optical depths are shown as a function of wavelength (λ). In general, AODs show a decreasing trend as λ increases. The spectral variations become less steep during the months of April–June due to the relative increases in the AODs in near-infrared (NIR) wavelengths. The spectral variations of the AODs become steeper during winter months.

Spectral variation of aerosol optical depth (τ_p,_λ) can be expressed following the inverse power-law representation of Ångström (1964) as

where α is the wavelength exponent, β is the turbidity parameter, and λ is the wavelength (μm). The value of α depends on the ratio of the concentration of large to small particles and β represents the total aerosol loading in the atmosphere (Shaw et al. 1973). Higher values of α thus indicate an aerosol size spectrum with a relative dominance of smaller aerosols. The values of α and β are obtained by least squares fitting of τ_p,_λ versus λ on a log–log scale. The monthly mean values of α and β are plotted in Figs. 7a and 7b. It was observed that the average value of α decreases systematically from January to March (during 2002 and 2003) and remains at that level during summer, indicating a systematic increase in the relative abundance of large aerosol in the size spectrum. This could be attributed partly to changes in the synoptic conditions, which are conducive for the advection of dust aerosols from the west Asian and western Indian arid regions by favorable winds. The value of α was high during winter months and minimum during summer. On the other hand, β was found to be maximum during summer and minimum during the winter season. These results indicate that there is a relative dominance of submicron particles during the winter season and a dominance of coarse particles during the summer season.

c. Role of boundary layer dynamics on the short-period fluctuations

Examination of instantaneous AODs has revealed significant variations within the daytime. The variations in the AODs at very short-term scales (within a day) are therefore investigated. Out of 605 observation days (during 2002–2004), there are 316 days where both forenoon (FN) and afternoon (AN) AOD observations are available. The difference between FN and AN AODs, Δ = (τ_p,_AN − τ_p,_FN), is estimated on all of these days. Considering the uncertainties in the AOD estimations, a difference in the AODs of less than 0.04 (i.e., |Δ| < 0.04) is considered to be insignificant. The number of data points in the range Δ < −0.04, −0.04 < Δ < +0.04, and Δ ≥ 0.04 are 35, 106, and 175, respectively. Examining from the above perspective, it was found that for nearly 75% of the case covered Δ > +0.04, indicating that on most of the days, the AODs are higher in the AN than in the FN. The deviations, Δ = (τ_p,_AN − τ_p,_FN), of FN and AN AODs are averaged separately for each month and the resulting monthly mean variations are shown in Fig. 8. The figure indicates that generally high AODs persist during the AN relative to the FN, especially during summer. The scatterplot between Δ and the afternoon optical depth (τ_p,_FN) is also shown in Fig. 8 (bottom panel).

The spectral AODs averaged separately for FN and AN, and their difference (Δ), are shown in Fig. 9. Figures 8 and 9 reveals that the FN to AN changes are significant over the entire MWR wavelength range during March–May. However, the difference is larger at short-wavelength range (λ ≤ 0.65 μm). Figure 10 represents the monthly mean variations of α and β for the FN and AN datasets. It is observed that during the AN period α is high compared to the FN value. The higher value of α during the AN period implies the relative dominance of fine–submicrometer aerosol particles.

The temporal changes in AODs and their λ dependence appear to have been strongly influenced by the dynamics of the boundary layer and its variation over the year associated with the changes in solar heating. During nighttime, the shallow nocturnal boundary layer acts as a capping inversion (Stull 1989), confining the emissions from these regions well below the mountain peak. After sunrise, the heating of the land surface results in setting up of convective motions, which gradually raise the inversion to higher altitudes (Kovalev and Eichinger 2004), leading to increased vertical mixing that brings up the air with high aerosol concentrations to higher levels. This would result in an increase in the concentration of anthropogenic aerosols at Manora Peak, as is observed in the present study.

We have also examined the role of the evolving boundary layer in transporting aerosols from the polluted valley region to Manora Peak. To estimate the mixed layer height, we used radiosonde data (from the nearest India meteorological department, New Delhi). The measured temperature (T) and pressure (P) are used to compute the potential temperature, T_P (Stull 1989):

where P is in hectopascals and T is in kelvin. The altitude profiles of T_P are used to infer the atmospheric boundary layer (ABL) characteristics. During the afternoon, the potential temperature profiles revealed a well-developed ABL. It consists of a surface layer (SL) close to the ground where T_P decreases sharply with altitude, representing a superadiabatic regime. Above the SL, T_P becomes nearly steady with altitude (due to turbulent mixing), representing an adiabatic lapse rate. This region of nearly constant T_P is the well-mixed layer (ML), the base of which is at the top of the SL. In Fig. 11, we show the percentage of occurrence of the mixed layer height exceeding the altitude of Manora Peak (our observation site) during each month. When the mixed-layer height exceeds the altitude of Manora Peak, then aerosols from the polluted valley below reach the observation site. It was observed that during March–June and October, the percentage of occurrence exceeded more than 50%. Further, AODs during the afternoon showed a positive correlation during summer (with a correlation coefficient r of +0.49), with the maximum temperature (T_max) observed during the daytime (see Fig. 12). Even though this correlation is not very high, it is significant (P < 0.001) for the 46 pairs of observations in Fig. 12. The significant scatter is attributed to the usage of T_max as a proxy for the strength of convection, which would in turn determine the depth of the ABL. The deeper the ABL, the more efficient will be the lifting of the pollutants from the valley to the peak and the larger will be the increase in the AOD in the afternoon. On the other hand, AODs during the forenoon showed negative correlation (with a correlation coefficient r of −0.13) with T_max. The investigations reported above indicate the significant role of boundary layer dynamics in transporting aerosols from the polluted valley to the pristine atmosphere at higher levels.

Similar types of studies have been reported by Bhugwant et al. (2001) in the case of aerosol black carbon. Bhugwant et al. (2000, 2001) showed that the high BC values during the AN period at high-altitude sites can be attributed to the vertical transport of aerosols from the nearby polluted urban and valley regions, which were initially confined to lower heights during the night and early morning hours due to the low-level inversions, but are released to greater heights (∼2 km) as the boundary layer evolves. At night, most of the aerosol concentrations are trapped by the low-level capped inversions, and as the land gets heated up during the day, the increased convective activity leads to a breaking of the capped inversion causing the mixing of aerosols and pollutants vertically and horizontally, thereby leading to increased AODs in the AN period. Aerosols over the Indian subcontinent may have implications for cloud microphysical properties and the radiation budget.

5. Conclusions

Multiyear measurements of spectral aerosol optical depths (AODs) made at Manora Peak in the central Himalayas have revealed the following.

1. The AODs at 0.5 μm were very low (≤0.1) in winter and increased rather steeply to reach high values (∼0.5) in summer.
2. The monthly mean AODs vary significantly (by more than a factor of 6) from January to June.
3. Strong short-period fluctuations (within a daytime) were observed in AODs.
4. The boundary layer dynamics was found to play a key role in transporting aerosols from the polluted valley region to higher altitudes, causing a large contrast in AODs between forenoon and afternoon.
5. The seasonal variation in AODs while examined in conjunction with synoptic-scale wind fields has revealed that the transport of dust aerosols from arid regions to the valley regions adjacent to the observational site and their subsequent transport upward by boundary layer dynamics are responsible for the summer increases.