1 Introduction

Air pollution has been growing increasingly in many areas of the globe over the past few decades, with far-reaching and harmful effects on the atmosphere and human health (WHO 2016; Saikawa et al. 2019). Particular matter (PM) is one of the most significant contaminants with damaging effects on-air chemistry (air quality and visibility), the health of human beings, snow-based albedo and precipitation, Earth’s global climate-influential radiation budget (Ghosh et al. 2014; Bikkina and Sarin 2019; Pant and Harrison, 2012). The effect of airborne particulates is closely related to their size (Tiwari et al. 2017). Therefore, size-segregated PM studies have shown that they deposited mainly within the tracheobronchial airways (PM10), alveolar region (PM2.5) (Chen et al. 2019), and bloodstream (PM1) (Falcon-Rodriguez et al. 2016), respectively. Several epidemiological studies concluded that the causal risk factors for many harmful human health effects (Pope and Dockery 2006; WHO 2016), including respiratory illnesses (Hanigan et al. 2021; Valavanidis et al. 2008; Kim et al. 2017), cardiopulmonary (Hanigan et al. 2021; Du et al. 2016; Peter et al. 2005; Silva et al. 2016) and carcinogenic effects (Callén et al. 2011; Kim et al. 2017; Masiol et al. 2017; Loomis et al. 2013) are due to particulate pollution.

Among the airborne particulates, the importance of investigating carbonaceous aerosols has recently gained widespread recognition (Sharma et al. 2015, 2017; Pintér et al. 2017; Hedge et al. 2020), as it contributes ~ 30–70 per cent of PM2.5 mass over urban and rural atmospheres (Ram et al. 2020; Dumka et al. 2019). Carbonaceous aerosols are classified into two types, based on their chemical and physical properties, i.e., organic carbon (OC) and elemental carbon (EC) (Kontul et al. 2020; Xu et al. 2015). EC is a light-absorbing component composed of graphite-like organic compounds and emitted directly from incomplete combustion of fossil fuels, biomass, coal, and bio-fuels derived from anthropogenic sources (Xu et al. 2015; Dinoi et al. 2017; Zhu et al. 2020; Zhang et al. 2020). When compared to greenhouse gases (GHGs), EC has a shorter atmospheric lifespan, ranging from a few days to weeks or hundreds of years. OC is the primary carbonaceous fraction (primary organic carbon, POC), emitted from the incomplete combustion of biomass and organic matter (Jafar and Harrison, 2020). OC is composed of various organic compounds, such as polycyclic aromatic hydrocarbons (PAHs), organic acids and alkanes, biphenyls that form secondary process through in situ reactions within the atmosphere (Arhami et al. 2018; Panicker et al. 2021; Srivastava and Naja 2020; Xu et al. 2015). EC has a positive radiative forcing effect in Earth-atmosphere system warming, while different compositions of OC can show different radiative forcing effect (i.e., negative as well as positive) that contributes to cooling and heating of the climate (Zhu et al. 2020; Babu et al. 2021; Aamaas et al. 2017). POC released from anthropogenic (cooking, traffic emissions, domestic heating, biomass and fossil fuel combustion) or biogenic sources (forest fires, vegetation, emissions from terpenes) (Dinoi et al. 2017; Kontul et al. 2020; Xu et al. 2015). Water-soluble organic carbon (WSOC) plays a substantial role in direct and indirect aerosol radiative forcing, hygroscopic properties, and serves as cloud condensation nuclei (CCN) (Panicker et al. 2021; Wei et al. 2019; Jin et al. 2020). WSOC can originate from primary emissions or formed via a series of atmospheric oxidation of volatile organic compounds (VOCs) during transport and biogenic emissions (Jin et al. 2020; Wei et al. 2019). Previous studies (Jin et al. 2020; Rajput et al. 2013; Kirillova et al. 2013; Yu et al. 2014) have shown that biomass burning and secondary organic aerosol (SOA) are the two most significant sources of WSOC. Therefore, it is imperative to understand the chemical properties, temporal of carbonaceous aerosols, which may travel far from their origin because of their high atmospheric lifetime (several days to weeks) (Ram et al. 2020; Joshi 2020; Zhang 2020).

In the era of rapid climate change, aerosol pollution is a significant cause for the deterioration of extremely sensitive Himalayas and the surrounding areas (Hedge and Kawamura 2012; Sen et al. 2018; Ramachandran et al. 2020). Due to their complex topography and numerous glaciers, the Indian Himalayan Region (IHR) affects the Asian Monsoon and is influenced by global climate change (Yuan et al. 2020a). The eastern Himalayas stretches from eastern Nepal to Yunnan (China), across an area of 524,190 km2 (Tse-ring et al. 2010). Anthropogenic activities such as cooking, open burning of agricultural residue and garbage, brick kilns, forest fires, motor vehicles, thermal power generation and other industries generate huge waste (Saikawa et al. 2019; Yuan et al. 2020a; Rai et al. 2020; Apollo et al. 2017). With the emission of carbonaceous particles, an essential trigger of glacial retreat in this fragile region may occur (Li et al. 2020a) and affect the health of a large populace of people, habitats, cryosphere, monsoon patterns, and agriculture (Saikawa et al. 2019; Ramachandran and Rupakheti 2021). A fourteen-month-long study of the carbonaceous aerosols fraction of ambient PM (PM10 and PM2.5) collected in Kangra (north-western Himalayan region) by Kumar and Attri (2015) concluded that the combustion of fuelwood which is a dominant source of carbonaceous aerosols (Sharma et al. 2021). During the land campaign in 2013, Sharma et al. (2014) reported the average concentrations of OC and EC of different locations of the north-western Himalayan region (Palampur, Kullu, Shimla, Solan, and Nahan) as 5.7 ± 1.3 μg m−3 and 2.2 ± 0.7 μg m−3 in PM2.5. Trace gases mixing ratios and particulates concentrations vary spatially and reveal the potential impacts of local activities (Sharma et al. 2014).

Meteorological parameters including rainfall, wind speed, relative humidity (RH), wind direction, and temperature determine the dispersion and transport of airborne pollutants to local and distant regions and establish their temporal and spatial variability (Hernández-Ceballos et al. 2019). Due to their high atmospheric lifespan, carbonaceous aerosols can be transported from long distant source regions (7–10 days in the absence of precipitation) (Chatterjee et al. 2020). Air mass trajectory analysis is, therefore, a crucial scientific tool to determine the origin (local or regional) and transport of pollutants (Chen et al. 2018; Ghosh et al. 2015; Hernández-Ceballos et al. 2019).

Air quality in the Himalayas, due to its high altitude, has a profound effect in terms of climate change and biogeochemical cycles, as it transports different types of pollutants over longer distances, resulting in alter air quality of the region (Joshi et al. 2020; Sharma et al. 2021; Saikawa et al. 2019). Because of its proximity to the altitudinal gradient, glaciers, topography, and rich bio-diversity, the carbonaceous aerosols are extremely important in the eastern Himalayan region (Lamsal et al. 2017; Sharma et al. 2020, 2021). It is, therefore, essential to investigate the ground-based aerosol measurements and understand the seasonal changes of both fine and coarse particulates over the eastern Himalayan region (Zhang et al. 2018; Sharma et al. 2021; Joshi et al. 2020). With that in mind, this paper aims to identify and estimate the probable pollutant areas over the eastern Himalayan region. The main goal of this study is to monitor the annual and seasonal variations of carbonaceous aerosols in PM10 and PM2.5 at a high-altitude location (Darjeeling). To determine their possible source region of emissions, we used CWT analysis coupled with long-term data on ambient air quality, which is a useful tool for identifying local and regional sources.

2 Methodology

2.1 Study Site Description

The aerosol study was investigated in Indian Himalayan city, i.e., Darjeeling (27.041ºN, 88.266ºE, ~ 2200 m  amsl) (Fig. 1). Geographically, Darjeeling is adjacent to the Indo-Gangetic Plain (IGP), which is a global hotspot for high aerosol loads and is among the few areas of the planet that have endured enhancements (Ojha et al. 2020). Darjeeling is internationally renowned as a tourist destination for its spectacular view of Mt. Kanchenjunga (world’s third highest mountain), along with its tea industry and the Darjeeling Himalayan Railway (“Toy Train”), which a UNESCO World Heritage site. Nevertheless, Darjeeling is most vulnerable to Sikkim, East Nepal, and West Bhutan among the administrative units (Bhattacharaya et al. 2020). The monitoring site is situated within the Bose Institute campus (27.01ºN, 88.15ºE, ~ 2200 m AMSL) located 200 m away from the main town. Mixed emission sources such as local motor vehicle emissions, wood and biomass burning, waste burning, ammonia fertilizer farming practices, coal combustion (from the ‘Toy Train’), and transport pollutants influenced the sampling site (Chatterjee et al. 2010; Sharma et al. 2020, 2021; Bhattacharya et al. 2020).

Fig. 1
figure 1

Source: Google map)

Sampling site in Darjeeling (

Full size image

The Indian Meteorological Department (IMD) reports that the year should be classified as four distinct seasons, i.e., winter (January–February), pre-monsoon (March, April, May), monsoon (June, July, August, September), and post-monsoon season (October, November, December). With heavy monsoon precipitation and cold in winter, the complex labyrinthine mountains with bold spurs, ridges, and valleys greatly vary their weather. The two-tourist season highlights of Darjeeling are pre-monsoon and post-monsoon (Chatterjee et al. 2010; 2020). Meteorological parameters were also recorded continuously during the sampling period using an automated weather station (AWS) installed at the sampling site. Darjeeling is considered a subtropical and temperate climate with wet summers because of heavy rain in the monsoon season. The annual minimum and maximum temperatures are 3.4 °C and 12.2 °C with the monthly average temperature ranging from 5.8 to 17.2 °C. The average rainfall in the monsoon season was 3092 mm (on average, 126 rainy days per year). The monthly average temperature, RH, wind directions, and wind speed during the study period are summarized in Fig. 2. The detailed information about the study site is available in our previous publications and reference therein (Sharma et al. 2020; Rai et al. 2020; Ghosh et al. 2020; Chatterjee et al. 2020).

Fig. 2
figure 2

Monthly average (a) temperature, (b) relative humidity (RH), (c) wind direction, and (d) wind speed during 2018–2019 at Darjeeling

Full size image

2.2 Sample Collection

The (PM2.5 and PM10) sampling system is installed at the rooftop of Bose Institute’s building at an altitude of 15 m above the ground level (AGL). PM2.5 (n = 94) and PM10 (n = 99) samples were collected from August 2018 to July 2019 (PM2.5 sampling was carried out up to June 2019 only). A fine particulate sampler (Envirotech Instrument Pvt. Ltd, India) and respirable particle sampler (Envirotech Instrument Pvt Ltd, India) were used to collect the PM2.5 and PM10 samples, respectively. Through PALLFLEX quartz microfiber filters (M/s. PallFlex Products Co, USA), ambient air was passed at an average flow rate of 1 m3 h−1 for PM2.5 and ~ 1.2 m3 min−1 for PM10 with an accuracy of ± 2% for 24 h. Filters having a size of 47 mm for PM2.5 and 20 × 25 cm2 for PM10 were pre-baked at 550 ºC for no less than 5 h, before their use to remove contaminants present in the filter material. Filters were accurately weighed using a microbalance (M/s. Sartorius, resolution: ± 10 µg) before and after sampling to obtain the mass difference between the initial and final weight. Aerosol mass concentrations (µg m−3) were estimated by PM mass difference (in µg) to the total volume of air (m3) passed during the sampling period. The loaded filters were carefully placed in sealed bags (for PM10) and cassettes (for PM2.5) and stored in a deep freezer (− 20 °C) until or chemical analysis.

2.3 Chemical Analysis

2.3.1 OC and EC Analysis

An OC/EC carbon analyzer (Model: DRI 2001A, Atmoslytic Inc., Calabasas, CA, USA) was used to determine the concentration of OC and EC in PM that follows USEPA Method (IMPROVE carbon protocol). It works on thermal-optical method based on the preferential oxidation of both OC (OC = OC1 + OC2 + OC3 + OC4 + OP) in pure helium and EC (EC = EC1 + EC2 + EC3—OP) in 98% helium and 2% oxygen at elevated temperatures of 140 °C, 280 °C, 480 °C and 580 °C for OC fractions, and 580 °C, 740 °C and 840 °C for EC fractions along with pyrolyzed carbon fraction (OP) (Chow et al. 2004). Its main goal is to check the pyrolysis and carbonization of OC compounds into EC, i.e., laser reflection, and the analyzer’s transmission. Along with field blank filters, a known punch area of 0.536 cm2 of the filter was cut and analyzed in triplicate with 3–5% analytical error (repeatability). The calculated OC and EC values for blank filters were subtracted from the obtained value of sample filters to estimate the correct OC and EC concentrations. There is a detailed overview of the analytical methods and calibration process and reference therein (Sharma et al. 2014; 2020; Rai et al. 2020).

2.3.2 WSOC Analysis

The known punch sizes of PM2.5 (~ 3.46 cm2) and PM10 (~ 7.07 cm2) were cut into four halves and soaked in 20 ml of deionized water (resistivity 18.2 MΩ-cm) and ultrasonicated three times for 10 min each. After filtration, the purified extract was diluted and analyzed for WSOC using a TOC analyzer (Model: Shimadzu TOC-L CPH/CPN, Japan). It works on the principle of catalytically aided combustion oxidation with a temperature of 680 ºC that transforms the carbon components into CO2 which is detected by a non-dispersive infrared (NDIR) gas analyzer. The measured total carbon (TC) and inorganic carbon (IC) values for blank filters were subtracted from the sample filters to estimate the correct WSOC concentrations, which is calculated by the difference between TC and IC measurements. Based on each filter’s triplicate analysis, the WSOC analytical error (repeatability) was calculated to be 3–10%. Details of the analytical methods and calibration process are given in Rai et al. (2020).

2.4 Estimation of TCA, POC and SOC

To account for the unmeasured H, O, N, and S in organic compounds, a conversion factor (or multiplier) is used to transform OC to OM (OM = f × OC). The f multipliers of 1.4 and 1.8 are depending on the extent of OM oxidation and secondary organic aerosols (SOA) formation, values for f vary from 1.2 for fresh aerosol in urban areas (Chow et al. 2002) to 2.6 for aged aerosol (Robinson et al. 2010). In the present case, the total carbonaceous aerosols (TCAs) were calculated by the proposed method of Srinivas and Sarin (2014), that is the sum of EC and organic matter (OM = 1.6 × OC) of PM10 (the 1.6 factor is used for PM10; whereas, for PM2.5, 1.4 was used (Chow et al., 2004; Robinson et al. 2010)). According to Zhang et al. (2005), a conversion factor of OC > 1.4 for sub-urban aerosols was used. We assumed 1.6 as a factor to convert OC to OM because the sampling location is a desirable sub-urban location (Sharma et al. 2018; Rai et al. 2020).

According to Castro et al. (1999), primary organic carbon (POC) is calculated using the EC tracer approach by considering a minimum OC/EC ratio for each season [(winter 1.40 for PM2.5 and 1.31 for PM10), summer 1.22 for PM2.5 and 1.27 for PM10), monsoon 1.30 for PM2.5 and 1.62 for PM10) and post-monsoon 1.05 for PM2.5 and 1.50 for PM10)]. EC and POC are known to derive from combustion sources, and EC is considered as a robust tracer for POC. The following relation is used to determine POC or OCprimary:

$${\text{OC}}_{{{\text{primary}}}} = \, \left[ {{\text{OC}}/{\text{EC}}} \right]_{{{\text{min}}}} \times \, \left[ {{\text{EC}}} \right] \, + {\text{ c,}}$$
(1)

where (OC/EC)min is the minimum ratio of OC/EC (minimum OC/EC for each season). c denotes the contribution from non-combustion sources, which in this case is negligible. Furthermore, for the estimation of secondary organic carbon (SOC or OCsecondary), the difference between the measured OC and estimated POC concentrations was calculated (Eq. 2).

$${\text{OC}}_{{{\text{secondary}}}} = {\text{ OC}}{-}{\text{OC}}_{{{\text{primary}}}} .$$
(2)

2.5 Backward Trajectory Analysis

Trajectory simulation techniques provide an effective and empirical way to classify areas associated with transporting pollutants to specific geographic regions (NESCAUM 2002). In this type of analysis, the basic methodological approach is to produce enormous amounts of air mass trajectories to consider a large number of transport and dispersion scenarios, such that statistical analysis derives representative information about flow patterns (residence time, range, distance, speed, etc.) (Hernández-Ceballos et al. 2019).

We used the HYbrid-Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (https://www.ready.noaa.gov/HYSPLIT.php) developed by the US National Oceanic and Atmospheric Administration’s (NOAA) to explore the transport and circulation patterns of air pollutants that influence local and regional wind regimes. The Web-based HYSPLIT uses the 1° × 1° GDAS (Global Data Assimilation System) meteorological data at the hour of 0500 UTC to simulate 3 days (72 h) backward trajectories (for each sampling day) using isentropic vertical velocity that is less susceptible to the uncertainties in the raw meteorological data due to the adiabatic vertical movements of air parcels enroute (Adak et al. 2013). The relative importance of long-distance versus regional emission sources was discerned by studying the different heights (100 m, 500 m, and 1000 m) analyzed (Ghosh et al. 2015). Trajectories are subject to interference from surface features (terrain and buildings) at lower altitudes, whereas higher-altitude trajectories may often be above the mixed layer and not reflect the air mass in the mixed layer (NESCAUM 2002).

Meteoinfo map is an application for a geographical information system that examines and visualizes various meteorological data formats (Li et al. 2020b). One of its plug-ins is TrajStat software that uses the HYSPLIT’s trajectory to identify better potential sources and transport of pollutants over the specific region. Apart from trajectory analysis (backwards or forward), it also offers cluster, potential source contribution function (PSCF), and concentrated-weighted trajectory (CWT) analysis.

2.5.1 Cluster Analysis

Cluster analysis is a multivariate statistical technique that uses the Ward hierarchical method to outline the various airflows in the area (Xin et al. 2016; Markou and Kassomenos 2010; Zhang et al., 2018). The Ward’s method (Eq. 3) based on the Euclidean distance is used to describe the latitude and longitude as distance variables between two trajectories, as shown in the following equation (Xin et al. 2016; Chen et al. 2018; Wang et al. 2009):

$${d}_{ij}= \sqrt{\sum_{k=1}^{p}\left({({X}_{i}\left(k\right)- {X}_{j}\left(k\right))}^{2}+ {(({Y}_{i}\left(k\right)- {Y}_{j}\left(k\right))}^{2}\right),}$$
(3)

where a point in p-dimensional space defines p, if the corresponding p-dimensional points are near, two observations are close to each other.

Xi and Yi referred to the position of backward trajectory 1, and Xj and Yj to the position of backward trajectory 2, respectively.

The ‘eyeball’ approach was used to calculate the required cluster number by comparing the mean trajectory figures (Chen et al. 2018). The number of clusters can be reduced by merging groups whose average trajectories are closest and retaining only those clusters that are noticeably different in wind speeds and wind direction. This approach’s main disadvantage is that if two back-trajectories have the same motion path but different speeds, they would be divided into two distinct categories (Xin et al. 2016).

2.5.2 Concentrated-Weighted Trajectory (CWT) Analysis

With simple trajectory analysis, aerosol particle pollution levels being transported to the receptor site cannot be extracted (Mahapatra et al. 2018; Zhu et al. 2018). In this regard, for source and receptor identification, CWT is an efficient trajectory-based receptor model that combines meteorological trajectory nodes (residence time) and pollutant concentrations to measure their pollution contribution to a receptor site (Suzuki et al. 2020; Li et al. 2020b). In the CWT algorithm (Eq. 4), the whole geographic region enclosed by trajectories divided into a grid area with longitude–latitude of (20.00, 70.00) in the northwest, (20.00, 10.00) in the southwest, (100.00, 70.00) in the north-east, and (100.00, 10.00) in the south-east; thus 19,200 grid cells were created. A weighted concentration is allocated to each grid cell (0.5° × 0.5° grid resolution) with the mean pollutant concentrations that cross the grid cell through corresponding trajectories (Zhao et al. 2018; Suzuki et al. 2020), as follows:

$${C}_{ij} = \frac{1}{\sum_{k = 1}^{N}{\tau }_{ijk}}\sum_{k = 1}^{N}{c}_{k}{\tau }_{\mathrm{ijk},}$$
(4)

where Cij is the average weighted concentration in the grid cell, (i, j)th. k: index of the trajectory, N: number of trajectories, Ck: pollutant concentrations measured at sampling location on import of trajectory k, and. τijk: residence time (time spent) of the trajectory k in the grid cell (i, j)th.

2.5.3 Active Fire Data

To obtain the active fire spots over the Eastern Himalayan region during the sampling period, a contextual algorithm that targets the heavy emission from fires of mid-infrared radiation was used with a significant input of brightness temperature (4 µm and 11 µm channels). A particular type of MODIS (Moderate Resolution Imaging Spectroradiometer) Collection 6 data (a combined product from Terra and Aqua, MCD14ML) from Fire Information for Resource Management System (FIRMS) was used. The improved algorithm of MODIS Collection 6 offers high efficiency in detecting small fires, enhanced cloud masking, reduction in a false alarm, and improved sunlight rejection with the best available active fire products. Using open-source GIS software MeteoInfo, data with a high confidence level (> 80%) were collected and plotted. For more details, see MODIS Set 6 Active Fire Product User’s Guide Revision B.

3 Results and Discussion

3.1 Variation in Carbonaceous Components in PM2.5 and PM10

The annual average concentrations of PM2.5 and PM10 were  37 ± 12 µg m−3 (range 16–77 µg m−3) and 55 ± 18 µg m−3 (range 21–116 µg m−3), respectively, that was quite close to the National Ambient Air Quality Standards (NAAQS) of India (annual limit 40 µg m−3 for PM2.5 and 60 µg m−3 for PM10) (Table 1). The maximum monthly average concentration of PM2.5 was recorded in October (51 µg m−3) and the minimum monthly average in January and June 2019 (27 µg m−3). Similarly, the highest monthly average concentration of PM10 was recorded in March (73 µg m−3), whereas the minimum monthly average concentration of PM10 was observed in December 2018 (39 µg m−3). The monthly average mass concentrations of PM2.5 and PM10 are depicted in Fig. 3; whereas, the monthly variation in mass concentrations of carbonaceous species of PM2.5 and PM10 are shown in Fig. 4a, b, respectively. Overall, August 2018 depicts the minimum concentration of OC (in PM10: 3.31 µg m−3; in PM2.5: 2.03 µg m−3), EC (in PM10: 0.82 µg m−3; in PM2.5: 0.98 µg m−3), WSOC (in PM10: 2.75 µg m−3; in PM2.5: 1.08 µg m−3), POC (in PM10: 2.73 µg m−3;  in PM2.5: 1.67 µg m−3), SOC (in PM10: 0.58 µg m−3; in PM2.5: 0.36 µg m−3) and TCAs (in PM10: 6.11 µg m−3; in PM2.5: 3.82 µg m−3) for both PM10 and PM2.5. On the other hand, the maximum concentration of carbonaceous species of PM was observed in November 2018 and March 2019 for PM10 and PM2.5, respectively. Except for EC (March 2019 for PM10: 3.59 µg m−3), POC (December 2018 for PM2.5: 3.51 µg m−3), and SOC (October 2018 for PM10: 2.05 µg m−3) which differ from other species. From the observed results, August 2018 was less polluted, whereas November 2018 and March 2019 were the most polluted months during the entire study period. Low carbonaceous concentration in August 2018 viz., the monsoon month is due to heavy rainfalls that wash off the pollutants. And the high carbonaceous concentration was observed in November 2018 (post-monsoon month) and March 2019 (pre-monsoon month), which are the peak tourist seasons in Darjeeling.

Table 1 Annual as well as seasonal average concentrations of PM2.5 and PM10 and their chemical components (OC, EC, WSOC, TCAs, POC, and SOC) (µg m−3) in Darjeeling
Full size table
Fig. 3
figure 3

Monthly average concentrations of PM2.5 and PM10 over Darjeeling

Full size image
Fig. 4
figure 4

Monthly average concentrations of carbonaceous species (OC, EC, WSOC, POC, SOC, TCAs) of PM2.5 and PM10 over Darjeeling

Full size image

The average ratios of PM2.5/PM10 ranged from 0.59 to 0.72, indicating that over 68% of PM10 in total is in the form of PM2.5. In the present case, the highest seasonal average concentrations of both PM2.5 (41 ± 14 µg m−3) and PM10 (63 ± 21 µg m−3) were recorded during the pre-monsoon season, followed by post-monsoon [PM2.5 (40 ± 11 µg m−3) and PM10 (56 ± 16 µg m−3)] that might be because of the high influx of tourists in these seasons. Also, after the post-monsoon season, winter (51 ± 18 µg m−3) and monsoon (52 ± 12 µg m−3) concentrations were equivalent to each other (for PM10); whereas, for PM2.5, monsoon (38 ± 8 µg m−3) concentration was higher than winter (30 ± 10 µg m−3) (Table 1). The average concentration of carbonaceous species during winter, pre-monsoon, monsoon, and post-monsoon is shown in Fig. 5. For PM2.5, the highest carbonaceous (OC, EC, WSOC, POC, TCAs) concentration was observed in pre-monsoon except for SOC concentration in post-monsoon that depicts more influence of tourism activities (Gajananda et al. 2005; Chatterjee et al. 2021) as well as long-range transportation of pollutants at sampling site of the Darjeeling (Rai et al. 2020). Whereas, lowest carbonaceous (OC, EC, WSOC, POC, SOC, TCAs) concentration was observed in monsoon, showing wash-out of pollutant loadings. Similarly, for PM10, the lowest carbonaceous (OC, EC, WSOC, POC, TCAs) concentration was observed in monsoon except for SOC concentration in pre-monsoon; whereas, the highest carbonaceous concentration was observed in pre-monsoon and post-monsoon due to the peak tourist seasons of Darjeeling. Therefore, emissions from vehicles, coal combustion, cooking, etc. lead to elevation of EC and POC concentrations in pre-monsoon; whereas, crop-residue burning, biomass burning and formation of SOA contribute to higher concentrations (OC, WSOC, SOC, TCAs) in post-monsoon.

Fig. 5
figure 5

Seasonal average concentration of carbonaceous species (OC, EC, WSOC, POC, SOC, TCAs) of PM2.5 and PM10 over Darjeeling

Full size image

The relation between atmospheric concentrations of OC and EC offers qualitative knowledge about the sources of carbonaceous species in PM. It is noted that if OC and EC are both released into the atmosphere by the same primary source, the two carbonaceous species should be closely related (Dinoi et al. 2017; Sharma et al. 2020; Rai et al. 2020). The scatter plots of OC versus EC concentrations, in both size fractions, during winter, pre-monsoon, monsoon, and post-monsoon are given in Fig. S3 (in supplementary information) for Darjeeling. The solid lines indicate the linear regressions of data. Good correlations were found in winter (R2 = 0.77, R2 = 0.88 with p < 0.05), pre-monsoon (R2 = 0.85, R2 = 0.87 with p < 0.05), and post-monsoon (R2 = 0.74, R2 = 0.80 with p < 0.05); whereas, weaker correlation is found in monsoon (R2 = 0.09, R2 = 0.63 with p < 0.05) in PM10 and PM2.5, respectively, suggesting their common sources (Sharma et al. 2014; Ram and Sarin 2010).

The OC/EC ratio varies significantly from source to source due to the varying strengths of the various emission sources (Yang et al. 2019). Besides, the presence of a minimum OC/EC ratio indicates that samples contain almost entirely primary carbonaceous compounds which can be influenced by various factors such as meteorology, local sources, and long-range aerosol transport (Dinoi et al. 2017). In India, biomass burning emissions (from wood fuel, agriculture waste, and cow dung) have been reported as a major contributor of carbonaceous aerosols, while coal-based emissions are predominant in eastern India (Ram and Sarin 2010). The OC/EC ratios ranged from 1.05 to 2.87 in PM2.5 and 1.27 to 5.25 in PM10. The average OC/EC ratio in PM2.5 (Table 2) was higher in post-monsoon (1.94 ± 0.45) > pre-monsoon (1.92 ± 0.36) > winter (1.87 ± 0.36) > monsoon (1.85 ± 0.41); whereas for PM10, monsoon (3.37 ± 1) > post-monsoon (3.01 ± 0.9) > winter (2.04 ± 0.37) > pre-monsoon (1.74 ± 0.30). The higher OC/EC highlights the clear prevalence of the organic carbon species over EC, whereas the low ratio indicates the higher emissions from fossil fuel (coal and vehicular exhaust) combustion in Darjeeling. This may be attributed to local heating systems and biomass burning which is an important source of primary-originated OC (Sharma et al. 2021).

Table 2 Average ratio of PM2.5 and PM10, and OC/EC, WSOC/OC, and SOC/OC of PM2.5 and PM10 in Darjeeling
Full size table

The complexity of the OC reaction pathways, as well as the large number of products produced by photochemical and thermal oxidation reactions, may render in the quantification of SOC contributions to carbonaceous aerosol (Dinoi et al. 2017). Atmospheric VOCs, which undergo gas-to-particle conversion in the presence of some oxidizing agents (ozone, hydroxy radicals, free radicals of oxides of nitrogen) and solar radiation, are essential for the formation of SOA in the atmosphere (Alves and Pio 2005). The higher ratio of SOC/OC (Table 2) indicates the high contribution of SOC to OC, and vice versa. In the present study, the higher ratio of SOC/OC was observed in post-monsoon for PM2.5 (0.44 ± 0.12) and monsoon for PM10 (0.48 ± 0.16) can be attributed to increased emissions of volatile organic precursors, combined with a stable atmosphere and longer residence time, which may strengthen atmospheric oxidation of volatile organic compounds. During study, the lowest ratio of SOC/OC was observed in winter (0.23 ± 0.13) for PM2.5 and pre-monsoon (0.25 ± 0.11) for PM10 may be attributed to direct emissions from combustion sources (coal and other fossil fuel combustion).

WSOC is one of the key components of carbonaceous aerosols because it contributes to the density of the cloud condensation nucleus (CCN) which can influence the global climate (Jin et al. 2020). The WSOC/OC (Table 2) ratios ranged from 0.22 to 0.78 and 0.18 to 0.81 with an average of 0.52 ± 0.16 and 0.69 ± 0.21 for PM2.5 and PM10, respectively. In PM10, the average of WSOC/OC was quite high that may come from cement dust, soil dust, and biogenic aerosols such as algae, pollen, etc. A broad range of WSOC/OC ratios indicates that emission sources, their intensity, and the contribution of SOA at the sampling site are all vulnerable to temporal variability (Ram et al. 2012). When compared to emissions from biomass burning, the vehicle emissions ratios of WSOC/OC are typically low because of poor solubility for organic fuels such as diesel, gasoline, etc. Besides, Ho et al. (2007) suggested that WSOC/OC ratios are used as an indicator for the formation of secondary aerosols and the presence of aged aerosols, which can further transport over long distances. The average WSOC/OC was found to be 0.50 and 0.71 in winter, 0.52 and 0.61 in pre-monsoon, 0.53 and 0.65 in monsoon, and 0.54 and 0.79 in post-monsoon for PM2.5 and PM10, respectively. The highest WSOC/OC was observed in the post-monsoon season indicating the enhanced contribution of fresh secondary aerosols, as shown in our previous study (Rai et al. 2020). This may be due to the aging of organic compounds and being absorbed by the atmosphere during the production of secondary organic aerosols (SOA). A correlation between WSOC and OC was observed to evaluate the presence of secondary aerosols, as shown in Fig. S4 (in supplementary information). A very significant correlation was found in pre-monsoon (R2 = 0.91, R2 = 0.73 with p < 0.05), and post-monsoon (R2 = 0.85, R2 = 0.83 with p < 0.05) in PM10 and PM2.5, respectively, suggesting the abundance of emissions of secondary aerosols. Whereas, weaker correlation in winter (R2 = 0.58, R2 = 0.63 with p < 0.05) and monsoon (R2 = 0.17, R2 = 0.62 with p < 0.05) in PM10 and PM2.5 was observed indicating the presence of different emission sources.

3.2 Transport Pathway of PM2.5 and PM10

3.2.1 Backward Trajectory Analysis

The techniques of trajectory modeling provide an analytical and visually compelling way to classify geographical regions associated with the long-range transport of pollutants to particular locations. The annual (Fig. S5 in Supplementary Information) and seasonal (Fig. 6) air mass backward trajectories were plotted at the height of 100 m, 500 m, and 1000 m to understand the pathway of pollutants arriving at the sampling site of Darjeeling.

Fig. 6
figure 6

Seasonal air mass backward trajectories and cluster map of PM10 and PM2.5at a height of 100 m, 500 m, and 1000 m AGL at Darjeeling

Full size image

At the height of 100 m, the trajectories were expected to persist in the boundary layer throughout the year, as shown in Fig. 6 representing well-mixed air masses that lead to visibility conditions encountered at ground level. However, orographic effects and other surface features are more strongly influenced by these trajectories and appear to cluster along defined paths, and many trajectories drift as they are triggered by localized surface flows (NESCAUM 2002). Consequently, the air parcel transporting from the IGP, central highlands of India, Nepal, Bay of Bengal (BoB) contributes to regional pollution. By dividing these trajectories into different seasons, the local pollution sources are observed in pre-monsoon and winter seasons; monsoon brings the air masses from the semi-arid region and BoB; post-monsoon and winter show transport of some transboundary pollutants.

The 500 m trajectories are distributed more evenly throughout the south and west with a smaller number of trajectories indicating the shallow surface flows exhibiting a characteristic of chaotic behavior (NESCAUM 2002). Heavy air mass loading at the height of 500 m was observed at the IGP region located below the sampling location that originates from the Thar desert, semi-arid region, central highlands, Nepal, Sikkim as western pollutants, Assam and BoB from eastern and southern contaminants, respectively (Fig. 6). The seasonal transport of pollutants demonstrates the arrival of a significant proportion of aerosols from the IGP region. During pre-monsoon, the Thar desert, semi-arid region, north-east region of Nepal, and Sikkim contribute to pollutant transport; whereas, for the post-monsoon season, Sikkim and the IGP region were the main pollutant emitters. Moreover, monsoon brings pollutants from the semi-arid region, central highlands, north-east region (Assam), and the BoB, whereas winter brings from semi-arid, central highlands, and Sikkim.

The general pattern of high-altitude trajectories (above the boundary layer) is consistent with a strong impact of upper-level winds in the free troposphere. However, the pathway of a parcel that migrates from the boundary layer into the free troposphere may accurately reflect its route that subsides and ultimately resides in the mixed layer to represent better the source region whose emissions contribute to the ambient concentrations observed at the end of the trajectory (NESCAUM 2002). At 1000 m (Fig. 6), the observed transport of air parcels was from transboundary (Afghanistan, Pakistan, Iran, and other south-western countries), marine (the Arabian Sea and BoB), as well as regional (the Thar desert, central highlands, IGP region, north-east region) sources. Winter contributes aerosols from south-western countries as transboundary, Deccan plateau, and local source of pollutants as regional. Monsoon carries aerosols from the IGP, BoB, and north-east regions (Assam and Myanmar). Pre-monsoon loads pollutants from central Asian countries (Afghanistan, Pakistan, Iran), the IGP region, and central highlands, whereas post-monsoon transport pollutants from Pakistan, IGP, and China.

3.2.2 Cluster Analysis

Cluster analysis was conducted at various altitudes (100 m, 500 m, and 1000 m) to compare the prevailing patterns of air mass backward trajectories and each pattern’s frequency. The altitude variation influences the percentage of paths in each cluster, reflecting their mean trajectory (Su et al. 2015; Ghosh et al. 2015). The percentage is much higher in local pathways than in comparatively long-distance clusters compared with a higher altitude (Su et al. 2015; Ghosh et al. 2015). The clusters of annual trajectories were mapped at all heights (100 m, 500 m, 1000 m) for both PM2.5 and PM10, as demonstrated in Fig. 6. The total number of trajectories assigned to each group with their number of polluted trajectories, frequency, and mean PM concentrations are summarized in Table 3. The number of polluted trajectories implies the trajectories cohering with PM equivalent to or greater than NAAQS (India) with a limiting value of 60 µg m−3 and 40 µg m−3 for PM10 and PM2.5. The long-range transport pathways of PM10 depend primarily on the atmospheric advection movement (Chen et al. 2018). Fine PM, on the other hand, has prolonged atmospheric residence time that leads to the contribution in secondary aerosols formation which is produced within the atmosphere by various chemical/physical processes (WHO Joint, 2006).

Table 3 Cluster statistics of each cluster and corresponding mean concentrations of PM2.5 and PM10 (µg m−3) at different heights (100, 500, 1000 m AGL)
Full size table

The cluster map plotted at the height of 100 m gives 3 clusters for PM10 and 4 clusters for PM2.5 (Fig. 6). Cluster 3 is the lowest frequency trajectory (5.3%) in PM2.5 and reflects long-range transport from the Middle East countries. The highest frequency of trajectories (50%) leading to about 30% of polluted trajectories demonstrated by Cluster 1. The highest mean PM2.5 concentration (42 ± 11 µg m−3) in cluster 2 was derived from the most polluted IGP region. The cluster maps of PM10 also depict the highest frequency trajectory (66.7%) and mean PM10 concentration (57 ± 20 µg m−3) originating from Nepal with ~ 37% of polluted trajectories, as illustrated in cluster 1. Cluster 2 is the lowest frequency trajectory (6.9%), demonstrating long-range transport from Central Asian countries (Afghanistan and Pakistan).

Trajectories are mapped at 500 m in height into 2 clusters of PM2.5 and 3 clusters of PM10 (Fig. 6). In PM2.5 maps, cluster 2 of 64.9% comes primarily from the middle IGP region, while cluster 1 is derived from the lower IGP region carrying approximately 52% of polluted trajectories at a mean of 41 ± 12 µg m−3. For PM10, cluster 2 is the main contributor to polluted trajectories (~ 43%) of 45% originated from Nepal. Besides, cluster 1 contributes maximum mean concentration of PM10 (61 ± 15 µg m−3) indicating medium-range transport of pollutant from the Thar desert. The dust was carried to downstream regions and airflow, contributing to dust storm conditions (Chen et al. 2018).

For the upper boundary layer and the low free troposphere that offers the most prominent clusters, a similar analysis was performed. Four clusters were found at higher altitudes of 1000 m with PM2.5 and PM10 (Fig. 6). Cluster 1 of PM2.5 and PM10 denotes the highest frequency of trajectories at 1000 m with 36.2% and 56.9% of pollutants loading, respectively, transported from the Nepal; whereas, cluster 2 represents 30% and 24% of air mass loading in PM2.5 and PM10 that arises from the Nepal–IGP area, respectively. The lowest trajectory frequency contribution of PM2.5 (13.8%) and PM10 (9.8%) was recorded from Bangladesh (through south-western monsoon pattern) and far western countries (through western winds), by cluster 4 and cluster 1, respectively. The mean PM2.5 concentration was almost the same for cluster 2 (39 ± 16 μgm−3); while for PM10, clusters 3 and 4 assessed nearly the same mean concentration with the highest in cluster 1 (67 ± 26 µg m−3).

The results of the cluster analysis concluded that the HYSPLIT trajectory is susceptible to data resolution, when the pollutants transport from short-range (west or south-east), and when the air transports from medium (western) and long-range (north-western), the trajectories are more able to persist unaffected with variations in vertical speed.

3.2.3 Seasonal CWT of PM2.5, PM10 and Carbonaceous Species

It is evident from the above results that the sampling site is affected by mixed types of pollutants (influenced by continental, local, and marine inflow). Sarkar et al. (2015) show that the influence of local pollution on Darjeeling was substantially more potent than other high-altitude Himalayan stations, which may be attributed to high anthropogenic emissions from tourism, high population density, unplanned settlements, and also a unique orography and landform with narrow roads led to lack of ventilation and dispersion of aerosols. Adak et al. (2013) reported that the fine mode aerosols were mostly anthropogenic aerosols produced locally. For each of the determined chemical species, the CWT model was applied separately by assigning the measured concentrations to grids concentrated in 'hot spots' that were used as tracers of particular types of PM emissions (Dimitrious et al. 2015, 2020). The grid with a higher value signifies the dangerous effect of particulate and carbonaceous aerosols on the location. To better identify significant pollution regions, only grids with high CWT values were designated (Li et al. 2020b). The well-mixed convective boundary layer (500 m) was studied to reduce the impact of surface turbulence (Yang et al. 2017). Figure 7 depicted a CWT map for Darjeeling with PM10 and PM2.5 concentrations in winter, pre-monsoon, monsoon, and post-monsoon from August 2018 to July 2019. Also, Figs. 8, 9, 10, 11, 12, 13 depicted a CWT map for Darjeeling of OC, EC, TCA, POC, SOC and WSOC concentrations of PM10 and PM2.5 in winter, pre-monsoon, monsoon, and post-monsoon. The CWT maps of PM10 and PM2.5 (Figs. S6–S8) along with their OC (Figs. S9–S11), EC (Figs. S12–S14), POC (Figs. S15––S17), SOC (Figs. S18–S20), WSOC (Figs. S21–S23) and TCA (Figs. S24–S26) at a height of 100 m, and 1000 m are shown in the supplementary information.

Fig. 7
figure 7

Seasonal CWT analysis of PM10 (left panel) and PM2.5 (right panel) (in μg m−3) at 500 m AGL

Full size image
Fig. 8
figure 8

Seasonal CWT analysis of OC of PM10 (left panel) and PM2.5 (right panel) (in μg m−3) at 500 m AGL

Full size image
Fig. 9
figure 9

Seasonal CWT analysis of EC of PM10 (left panel) and PM2.5 (right panel) (in μg m−3) at 500 m AGL

Full size image
Fig. 10
figure 10

Seasonal CWT analysis of TCAs of PM10 (left panel) and PM2.5 (right panel) (in μg m−3) at 500 m AGL

Full size image
Fig. 11
figure 11

Seasonal CWT analysis of POC of PM10 (left panel) and PM2.5 (right panel) (in μg m−3) at 500 m AGL

Full size image
Fig. 12
figure 12

Seasonal CWT analysis of SOC of PM10 (left panel) and PM2.5 (right panel) (in μg m−3) at 500 m AGL

Full size image
Fig. 13
figure 13

Seasonal CWT analysis of WSOC of PM10 (left panel) and PM2.5 (right panel) (in μg m−3) at 500 m AGL

Full size image
3.2.3.1 Winter

Among all seasons, the winter season recorded the lower concentrations for both PM10 and PM2.5 emitted from incomplete combustion of fuels and wood at low temperature, used to combat chills. A persistent thermal inversion with limited dispersion is formed at ground level due to calm wind and fog conditions that entrapped a greater number of fine pollutants i.e., PM2.5 (59%). From the observed results, a higher EC average concentration of 2.7 ± 1.0 μg m−3 in PM10 and 1.8 ± 0.8 μg m−3 in PM2.5 may attributable to combustion activities such as biomass, fuels, and wood (for heating purposes) and coal (from “Toy Train”) over Darjeeling. Li et al. (2020a) have reported the complexity of OC sources which include combustion, fine soil particles, pollen, and secondary organic processes. The average OC concentration of PM10 and PM2.5 in the present study were 12.34 and 5.87 μg m−3, varying from 2.09 to 12.34 μg m−3 and 1.04 to 5.87 μg m−3, respectively. The POC contributes maximum to OC of PM10 (65%) and PM2.5 (76%) emitted predominantly from eastern Nepal and local Darjeeling areas, along with a smaller amount of PM2.5 SOC. In contrast, the IGP, Nepal, and the central highlands were transporting over 2 μg m−3 of SOC of PM10. Due to the incidence of forest fires in Pakistan, the IGP region, the north-eastern and southern parts of India, MODIS fire spots showed good agreement with the observed results (Fig. 14). In the sampling region, the air was polluted by a forest fire and biomass burning fumes as well as continental air mass. Tripathee et al. (2021) indicated that wintertime aerosols in Dhulikhel (Nepal) mainly consist of polluted continental air mass that comprises dust and smoke. Furthermore, WSOC also contributes a fraction of the OC that shows a similar EC trend, that might be emitted from biomass burning, industries, dust, and automobile emissions and form secondary organic aerosol (SOA) (Sharma et al. 2021; Jin et al. 2020). Therefore, the contribution of TCAs is approximately 21% of PM10 and PM2.5 each is primarily from the IGP region, Nepal, and local region. Following the harvest of wheat crops every April–May in the IGP region, large-scale wheat-residual burning is the main source of biomass burning emissions which can emit numerous organic particles (Yuan et al. 2020b).

Fig. 14
figure 14

Seasonal fire count map of India during August 2018–July 2019

Full size image
3.2.3.2 Pre-monsoon

In Darjeeling, the peak tourist season is the pre-monsoon which contributes to the highest concentrations of PM and its carbonaceous species. The majorly polluted regions (western Nepal and IGP) with the highest mean concentration of PM10 and PM2.5 exceeded the NAAQS standard limit as shown in Fig. 7. This could be due to the long-range transport of dust aerosol, nucleating vapors, and precursor gases driven by westerly winds from arid and semi-arid regions of western India and Asia, including the Thar Desert and Arabian deserts, respectively (Rai et al. 2020; Chatterjee et al. 2010; Adak et al. 2013). Moreover, the formation of fine particles in Darjeeling is favored by weather conditions including increased solar radiation and relative humidity (Adak et al. 2013). EC is one of the PM components indicating the presence of pollution from tourist vehicles, dust, smoke, and combustion of coal in the “Toy Train” of DHR. In the current review, the EC concentration of PM10 covers a larger area of Nepal with more than 1.6 μg m−3 compared to PM2.5. Our observations are consistent with the recent research on Dhulikhel (Nepal) by Tripathee et al. (2021). With more than 4 μg m−3 over Nepal, pre-monsoon is the largest contributor of OC of PM10 in comparison to PM2.5. A similar emission trend of POC, SOC, and WSOC of PM10 was observed from Nepal, IGP, and local regions contributing 73%, 26%, and 66% to OC, respectively. In another case of PM2.5, the least SOC concentration and high emissions of POC (65%) and WSOC (53%) were measured among all seasons. This implies that various anthropogenic activities such as combustion of various types of fuels, biomass burning, and forest fire, vehicular emission and pollutants from Darjeeling Himalayan railway (due to influx of tourist) may result in emissions of higher number of organic aerosols. Dust aerosols from the IGP region are vertically advected due to increased convection and pressure gradient, and arrive at the foothills of the Himalayas, combined with carbonaceous species (Adak et al. 2013; Chatterjee et al. 2010). Furthermore, active Terra and Aqua MODIS fire and thermal anomalies (≥ 80%) indicate the prevalence of forest fires across India except in the Thar Desert of Rajasthan and fewer in Bihar and Bangladesh (Fig. 14). Consequently, due to the influx of visitors, the increased anthropogenic activities in Darjeeling and neighboring areas, the contribution of TCA in PM10 and PM2.5 is almost 18% and 20%. Although the contribution appears to be smaller in number, these aerosols may undergo a variety of atmospheric processes leading to the formation of toxic pollutants in the presence of favorable conditions.

3.2.3.3 Monsoon

High precipitation in the foothills of the eastern Himalayas is due to their proximity to the BoB and their direct exposure to the moisture-laden mountain patterns of the southwest monsoon zone (Cajee 2018). The monsoon season depicts the effect of air masses vertically elevated from the BoB and horizontally transported by convective motion through the southerly flow to the Himalayas (Chatterjee et al. 2010). Although, the possible origin of marine and continental aerosols leads to the influence of mixed aerosols over Darjeeling with minimum concentrations of PM and its carbonaceous components. During the monsoon, almost equal concentrations of EC of PM2.5 and PM10 were measured that might be attributed to fossil fuel combustion and vehicular exhaust. The average OC concentration of PM2.5 and PM10 was 3.6 ± 1.6 μg m−3 and 5.1 ± 2.0 μg m−3 contributing to significant biomass burning from the IGP region, central highlands, BoB, Bangladesh, Assam, and Bhutan. In the north-western Himalayas, Eastern Ghats, and a few central highways and the IGP region, MODIS fire spots supported observed findings (Fig. 14). Due to the below-cloud scavenging process, low concentrations of WSOC and SOC were observed with less than 1 μg m−3. The contribution of POC to the OC of PM10 (51%) and PM2.5 (71%) is strongly influenced by local emissions, the IGP region, and Nepal. The TCAs load of the PM is ~ 11–13% over the IGP region, BoB, and north-eastern region of India that may be attributed to dust and marine aerosols. Heavy precipitation and significant fall in finer mode aerosol particles indicate the washing out  of aerosols; whereas, a comparatively higher concentration of coarser aerosols might be due to the influence of sea salt aerosols (Na+, Mg2+) as described in a previous study by Chatterjee et al. (2010). Results observed were close to previous studies (Adak et al. 2013; Sarkar et al. 2015, 2019; Roy et al. 2016) showing that the effect of wet scavenging and restricts the impact of long-distance transport and local pollution. The heavy dust loading significantly diminishes due to aerosol wash-out from the atmosphere.

3.2.3.4 Post-monsoon

The huge influx of tourists visits Darjeeling in the post-monsoon and is known as the second peak tourist season. The most prominent transport regions of PM through strong westerlies were from Nepal, the Nepal–IGP border, and the local area. The highest concentration of PM is observed due to the peak tourist season, with the greatest contribution of carbonaceous aerosols that could be released due to the presence of high population density, contributing to more tourist vehicle pollution, fossil fuel burning, coal combustion (Toy Train), biomass burning, etc. Significant regions of OC and EC pollution were found to be higher in PM10 (5.9 μg m−3) and PM2.5 (2.2 μg m−3), respectively, arriving from the IGP region through Nepal and local regions of Darjeeling. Tripathee et al. (2021) concluded the influence of continental air mass along with the smoke of biomass burning during the post-monsoon. MODIS active fire spots are in good agreement with these findings showing the massive burning over the IGP, primarily in Punjab and Haryana, where farmers burn crop residue (Fig. 14). The POC contributes about 49% and 57% to OC of PM10 and PM2.5, respectively that might be due to local biomass burning and forest fire episode of IGP. In the PM10 map of SOC, SOC contribution is also very high (> 2 μg m−3) compared to PM2.5, which may be due to the formation of secondary aerosols through the gas-to-particle conversion of VOC. Furthermore, a high concentration of WSOC was observed with a contribution of 80% and 54% to OC of PM10 and PM2.5, respectively, leading to the formation of cloud condensation nuclei (CCN) (Jin et al. 2020). Therefore, the TCA load of the PM is ~ 20% that might be attributed to dust and continental aerosols.

4 Conclusion

This study presents a seasonal progression of PM2.5 and PM10 from August 2018 to July 2019, at Darjeeling, a semi-urban and sub-tropical site of the eastern Himalayan region of India. The annual concentrations of PM10 (55 μg m−3) and PM2.5 (37 μg m−3) were observed to exceeded the threshold value of WHO (20 μg m−3 for PM10; and 10 μg m−3 for PM2.5) by ~ 36% and 27%, respectively, suggesting an increase in levels of coarse mode aerosols over Darjeeling.

  • PM2.5 and PM10 showed remarkable seasonal differences due to fluctuations in local weather conditions, local pollution, and long-range transport. The order of seasonal average concentration of both PM10 and PM2.5, was: pre-monsoon > post-monsoon > monsoon > winter. During the study period, the huge disparity in fine and coarse mode aerosol concentrations may be due to the unique orography and landform of Darjeeling and the thermodynamic conditions of the boundary layer, which either favor or hinder pollutants dispersion.

  • The long-distance transport routes of particulate aerosols rely mainly on atmospheric advection movement. The seasonal air mass back trajectory at different altitudes (100, 500, 1000 m) reveal local, IGP, the Thar desert, semi-arid, central highlands, Nepal, Assam, and the BoB as the common major pollutant transporting regions affected by mixed types of pollutants (continental, local and marine).

  • Clustering of backward trajectory at different altitudes (100, 500, 1000 m) for PM2.5 and PM10 illustrates that the air masses may originate from short-, medium- and long-range transport associated with several polluted trajectories and mean concentration of PM.

  • Following Terra and Aqua MODIS active fire data, the results of CWT analysis of PM (PM2.5 and PM10) and its carbonaceous species (OC, EC, WSOC, POC, SOC, TCAs) are in good agreement. As pre-monsoon and post-monsoon are the peak season for tourists to visit, various anthropogenic activities increased in the town that leads to the emission from vehicles, coal combustion in “Toy Train”, transboundary pollutants, etc. Besides, the favorable meteorological conditions augmented the overall increase in total PM load that may consequently lead to the formation of secondary organic aerosols (SOA) in the atmosphere through chemical/physical processes. During the winter season, although biomass burning in Punjab and Haryana and long-range transport of continental air masses were found to be the major contributor of particulates. Comparatively, a higher concentration of PM was observed in the monsoon season which may be due to more accumulation and less wash-out of pollutants.

The findings of this study will be useful in providing yearly quality assured PM2.5 and PM10 concentrations for efficient air quality mitigation policies in the Himalayan region, as well as invalidating model results for future air quality studies. In future research, we will work on source apportionment and factor contribution of potential source regions to enhance the accuracy and scientific quality of our models.