Seasonal Characteristics, Sources and Pollution Pathways of PM 10 at High Altitudes Himalayas of India

The present study represents the annual and seasonal concentration of PM 10 over different sites (Darjeeling, Nainital, Mohal-Kullu) across the Himalayan region of India from July 2018 to December 2019. The collected PM 10 samples were analyzed for carbonaceous aerosols [organic carbon (OC), elemental carbon (EC), water-soluble organic carbon (WSOC), primary organic carbon (POC), secondary organic carbon (SOC)] and major trace elements to inspect their possible sources. The annual average concentrations of PM 10 over Mohal-Kullu, Nainital, and Darjeeling were recorded as 57 ± 32 µ g m –3 , 65 ± 41 µ g m –3 , and 54 ± 17 µ g m –3 , respectively. The high OC/EC ratio and significant correlation of OC with EC and WSOC with OC indicated a significant effect of biomass burning aerosols over the study sites. Principal component analysis/absolute principal component score (PCA/APCS) resolved four major sources: crustal/soil dust (26.6%), biomass burning/fossil fuel combustion (28%), vehicular emissions (28%), and industrial emissions/coal combustion (17%). Identification of the source region using the potential source contribution function (PSCF) and concentration weighted trajectories (CWT) showed that PM 10 was mainly transported from the northwestern part of India (Haryana, Punjab), the northeastern region of Pakistan, the Thar Desert, and Indo-Gangetic Plain (IGP), which contributed to dust-related aerosols over the Himalayan region of India.


INTRODUCTION
In recent decades, rising urbanization and booming industrialization have led to the release of various pollutants, predominately particulate matter (PM) (Cheng et al., 2007;Lim and Turpin, 2002). The surge in particle load in the atmosphere has given rise to several environmental (climate change) and human health problems, especially across South Asia (Cheng et al., 2007;Fuzzi et al., 2015;Panwar et al., 2020;Tsai et al., 2012). The concentration and components of aerosols depend on their originating sources, atmospheric processes, and prevailing climatic conditions for the distribution of aerosols (Andreae and Crutzen, 1997;Rastogi and Sarin, 2006). Carbonaceous fractions, the primary constituent of ambient PM, are a source of great concern globally due to their impact on regional and global climate along with the atmospheric system of Earth (Garza et al., 2008;Hansell, 2005;Lim and Turpin, 2002;Phairuang et al., 2020;Ram and Sarin, 2011;Ramanathan and Carmichael, 2008). On the basis of physical and chemical characteristics, carbonaceous aerosols (CAs) are categorized as organic carbon (OC) and elemental carbon (EC). OC is released into the atmosphere from combustion, known as primary OC, or formed by gasto-particle conversion of volatile organic compounds (VOCs) (referred to as secondary organic aerosols [SOA]) (Gupta et al., 2007;Haque et al., 2016;Jimenez et al., 2009) whereas, EC is formed due to the partial combustion (low temperature and low oxygen availability) of carbonaceous fuels. CAs comprise a significant proportion of water-soluble organic carbon (WSOC), function as cloud condensation nuclei (CCN), are hygroscopic, and play a pivotal role in haze formation, posing a health risk (Hegde et al., 2016;Izhar et al., 2019). Besides carbonaceous fractions, PM constitutes various trace elements such as iron, zinc, manganese, and heavy metals such as cadmium, arsenic. Temperature, movement and flow of surface waters, atmospheric circulation, and wind speed all influence the transport of heavy metals and other pollutants (Briffa et al., 2020). Trace metals can adversely impact human health due to their physical presence within the lungs, which hinders gas exchange and allows adsorbed metals to leach into the cells as well as the bloodstream (Baboolal et al., 2020). Heavy metals have high density, atomic weights, or specific gravity (Duffus, 2001;Popoola et al., 2018), and they come into the environment majorly through the deposition of particulates, vehicular emissions, road dust re-suspension, and construction activities (Shrivastav, 2001) causing severe physiological health effects (Bollati et al., 2010;Khillare et al., 2004).
While aerosol chemistry, temporal variation, and source identification have mainly focused on several urban agglomerations across India, therefore, the study of Himalayan aerosols has also piqued the interest of experts Gajananda et al., 2005;Kaushal et al., 2018;Kumar and Attri, 2016;Ram et al., 2008;Sarkar et al., 2014Sarkar et al., , 2015Sheoran et al., 2021;. The Indian Himalayan Region (IHR) is considered a pristine, ecologically fragile, rich in biodiversity, and highly vulnerable region on Earth Yang et al., 2021;Yuan et al., 2020). In addition, urban development raised energy consumption, causing perturbation to the Himalaya's temperature and deteriorating its air quality (Ram and Sarin, 2010;Sharma et al., 2021). Being far from intensive anthropogenic activities, the loading of CAs increases with long-range transport into the IHR along the valleys, accelerating the melting of glaciers by depositing on the snow and ice surface (Yuan et al., 2020(Yuan et al., , 2021. Previous studies (Cong et al., 2015;Streets et al., 2003;Xu et al., 2018;Yang et al., 2021;Zheng et al., 2017) show that pollutants are transported over the Himalayas during the pre-monsoon season, which coincides with the yearly extreme fire season in South Asia. Therefore, an in-depth study of particle chemistry gives a detailed description of their sources which helps in the development of appropriate mitigation initiatives that may enhance the air quality (Bond et al., 2013;Cao et al., 2006;Pope et al., 2009;Rai et al., 2020;Ramgolam et al., 2009;Sharma et al., 2017Sharma et al., , 2018Ramana et al., 2010). However, only a few studies using receptor models for chemical characterization and source identification have been conducted in the Himalayas (Adak et al., 2014;Panwar et al., 2020;Rai et al., 2020;Sharma et al., 2020a;Sharma et al., 2021;Sheoran et al., 2021;Soni et al., 2020). This study highlights the annual and seasonal variation of PM10 (aerodynamic diameter < 10 µm) and its components (carbonaceous and elemental) at Mohal-Kullu, Nainital, and Darjeeling and identifies the possible sources of carbonaceous species and elements of PM10 using principal component analysis/absolute principal component scores (PCA/APCS). To identify the specific source regions, we performed the backward trajectory, potential source contribution function (PSCF), and concentration weighted trajectory (CWT). Therefore, the present study on the year-long aerosol chemistry across the Indian Himalayas would undoubtedly be a valuable addition to the existing knowledge and database on atmospheric aerosol pollution.

Study Area
The sites were chosen to symbolize the vast stretch of the IHR in the Indian subcontinent, which incorporates the western, central, and eastern regions of the Indian Himalayas, respectively dust-laden activity over the Thar Desert as well as southwest Asia. During the winter months (January-February), domestic heating increases CAs emissions from wood-burning. A detailed description of the sampling location is discussed in previous studies (Hegde et al., 2016;Sagar et al., 2015;Sharma et al., 2020b;Sheoran et al., 2021).
Darjeeling: The study site, Bose Institute, Darjeeling (27.01°N, 88.15°E, 2200 m a.s.l.), is located in the eastern Himalayas and surrounded by Nepal, Sikkim, and Bhutan from the west, north, and east sides, respectively. Geographically, Darjeeling is adjacent to the IGP region (a global hotspot for heavy aerosol loadings)  and is majorly under the influence of tourist activities and biomass burning (Adak et al., 2014). During the monsoon season (June-September), the site has wet summers due to heavy rainfall. A detailed description of weather, topography, and influence of sources on particulate concentrations at the study site are discussed in previous studies Rai et al., 2020;Sharma et al., 2021).
Four distinct seasons have been experienced in India [winter (January-February), summer (March-May), monsoon (June-September), and post-monsoon (October-December)] during the year . These seasons are classified according to India Meteorological Department (IMD).

Aerosol Sampling
A respirable dust sampler (flow rate: 1.2 m 3 min -1 ) was used to collect aerosols (24 h) (PM10) on a pre-baked (550°C for 5 h) Pallflex Tissuquartz filters (20 × 25 cm 2 ) at three urban locations (Bose Institute, Darjeeling; ARIES, Nainital; G. B. Pant NIHE, Mohal-Kullu) from July 2018-December 2019. The pre-(before sampling) and post-(after sampling) weights of the filters were recorded using a weighing balance (resolution: ± 10 µg). The gravimetric mass of PM10 was calculated as the difference of post-and pre-filters mass (in µg) divided by the sampled air volume (m 3 ) passing through the filters. Before the chemical analysis, the filters were kept in a deep freezer at -20°C.

Chemical Analysis 2.3.1 OC and EC analysis
The OC/EC carbon analyzer (DRI 2001A, Atmoslytic Inc., Calabasas, CA, USA) was used to quantify the OC and EC using the Interagency Monitoring of PROtected Visual Environments (IMPROVE-A) procedure (Chow et al., 2004). The OC/EC analyzer uses a thermal-optical approach that involves volatilization of OC fractions using helium (at 140°C, 280°C, 480°C, and 580°C), followed by oxidation of EC fractions in 98% helium and 2% oxygen (at 580°C, 740°C, and 840°C). A filter punch (~0.536 cm 2 area) was taken from the filter for the analysis. Detailed information on the method used was referenced therein (DRI 2001A user manual;Sharma et al., 2021). In addition, average blank concentrations of OC and EC were determined from the blank filters to correct sample results.

WSOC analysis
A TOC analyzer was used to determine the WSOC in PM10 samples (Model: Shimadzu TOC-L CPH/CPN, Japan). The analyzer works on catalytic-oxidation combustion at high temperatures (680°C) to convert OC components into CO2. The CO2 was detected using a non-dispersive infrared (NDIR) gas analyzer. A punch size of 30 mm (7.065 cm 2 ) diameter from PM10 was employed for the analysis. Detailed information on the method used to estimate WSOC was discussed in reference therein (Rai et al., 2020;Sharma et al., 2021).

Elemental analysis
We used Wavelength Dispersive X-ray Fluorescence (WD-XRF) Spectrometer (Rigaku ZSX Primus) to determine different elements (Mg, Na, Al, Ca, Fe, Ti, Ni, Cl, S, P, K, Cr, Pb, Cu, Mn, B, Zr, Mo, and Zn) in PM10 samples. In addition, the blank filters were analyzed to accurate the intensities for the analyzed sample with repeatability error in the range of 5-10%. The detailed information on the analytical procedure is presented in Sharma et al. (2020a).

Estimation of POC and SOC
The most common method for estimating POC and SOC is the EC tracer method (Mishra and Kulshrestha, 2021;Srinivas and Sarin, 2014). Both OC and EC are produced by the combustion of fuels, where EC is the primary tracer for POC. (OC/EC)min is the ratio of OC and EC of fresh aerosols emitted during combustion. The following formulas were used to calculate POC and SOC for each season separately:

Enrichment Factors (EFs)
To delineate the origin of trace metal abundances in aerosols, whether crustal or anthropogenic, EFs have been estimated. The EFs of the elements are computed (Taylor and McLennan, 1995) (Taylor and McLennan, 1995). We selected Al (as a reference) element for this study due to its stable and spatially homogeneous distribution in the study area.

Principal Component Analysis (PCA)/Absolute Principal Component Scores (APCS)
PCA uses the Varimax rotation and orthogonal transformation method to recognize the data patterns. Eigenvalues were restricted to more than 1 for the extracted components. In PCA, the data is normalized using the formula: where i denotes the sample number; j denotes the element number; Cij implies the measured elemental concentration i in j th observation; CJ depicts the arithmetic mean (AM) concentration, and signifies the standard deviation (SD) for element i. The formulation of the PCA model is as follows: ( ) 1 1, 2, ..., ; 1, 2, ..., ; 1, 2, ..., where k denotes the number of sources, hkj and gik are the factor scores and loadings, respectively (Song et al., 2006). On the basis of factor loading components, each source contribution is estimated using the APCS method (Thurston and Spengler, 1985). For APCS, normalization of PCA results was done using the formula: After rescaling the scores (known as APCS), linear regression was performed on APCS and the measured concentrations to obtain the accurate source contributions (Song et al., 2006). More information about PCA/APCS model is found in reference therein (Duan et al., 2006;Guo et al., 2004;Larsen and Baker, 2003;Miller et al., 2002;Thurston and Spengler, 1985).

Air-Mass Back Trajectory (AMBT) and Probability Functions (PSCF and CWT)
Air-mass backward trajectory (AMBT) has been widely used for the movement of air masses and their long-range transport, which depends on the source and pathway of air masses. We simulated the three days AMBTs using Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model with meteorological data (1° × 1°) of the Global Data Assimilation System (GDAS) (Draxler and Hess, 1998) at different altitudes of 100 m (lower), 500 m (around boundary layer) and 1000 m (higher) above ground level (AGL) (Adak et al., 2014). Different heights were selected to study the significance of transboundary movement and local emission sources, which can interfere with the surface features (buildings) at lower altitudes.
Air parcels contain a specific concentration of pollutants estimated using PSCF (Ashbaugh et al., 1985;Hwang and Hopke, 2007;Li et al., 2017). PSCF divides the potential source region into equal grids of i by j using Eq. (7): nij denotes the frequency of the trajectories passing each grid cell (i, j), and mij denotes the frequency of pollutant concentration exceeding the predefined criteria value (90 th percentile of source contribution). A key limitation of the PSCF is that if the sample concentration exceeds the specified percentile, similar PSCF values can be found in multiple grid cells, making it difficult to distinguish between low and high-impact sources. Therefore, CWT addresses this issue by assigning the species concentration and residence time to its corresponding air masses approaching the receptor location Li et al., 2017;Singh et al., 2018).
In the post-monsoon season, we observed the maximum concentration of OC (Darjeeling: 6.29 ± 2.38 µg m -3 ; Mohal-Kullu: 15.43 ± 9.96 µg m -3 ) and WSOC (Darjeeling: 4.48 ± 1.53 µg m -3 ) of PM10 (Fig. 3), which might be due to the biomass burning (wood burning) and lowering of the planetary boundary layer Ram et al., 2008;Yadav et al., 2013). Nainital has recorded a maximum concentration of OC (8.26 ± 1.15 µg m -3 ) and WSOC (5.66 ± 3.47 µg m -3 ) during the summer season, which shows similarity with previous literature reported over high altitudes (Adhikary et al., 2007;Carrico et al., 2003). High concentration of EC (Darjeeling: 3.01 ± 1.15 µg m -3 ; Nainital: 2.28 ± 1.43 µg m -3 ) and POC (Darjeeling: 3.82 ± 1.46 µg m -3 ; Nainital: 3.69 ± 2.32 µg m -3 ) were found in the summer season. Moreover, higher levels of OC and EC were observed in the post-monsoon and summer seasons, demonstrating the considerable influence of agricultural waste burning followed by rice and wheat harvesting in the northwestern region of the IGP and transborder movement of pollutants to the receptor sites Dumka et al., 2015Dumka et al., , 2021Rastogi et al., 2016;Sharma et al., 2021). During the study period, the monsoon season shows the minimum concentration of carbonaceous species in PM10, which might be due to rain-assisted washout (Ram et al., 2008;Sheoran et al., 2021;Yadav et al., 2013). Table 2 highlights the average seasonal ratios of OC/EC and WSOC/OC at each location. In Fig. 2, the monthly average variation in OC/EC and WSOC/OC ratios are reported. Several researchers ( Andreae and Merlet, 2001;Ram et al., 2008;Saarikoski et al., 2008;Sandradewi et al., 2008;Sudheer and Sarin, 2008) used OC/EC ratio for the identification of CAs sources and differentiate the pollution sources (primary and secondary emissions) which demonstrate higher ratio for biomass burning (BB) and lower for vehicular emission (VE). In this study, the OC/EC ratio exhibited a significant variability, ranging from 1.27 to 6.01 (average: 2.61 ± 1.02) and 1.62 to 10.76 (average: 3.36 ± 1.73) over Darjeeling and Nainital, respectively. The obtained ratios agreed with those reported in the literature (Aggarwal et al., 2012;Ram et al., 2008;Rengarajan et al., 2007;Sharma et al., 2014aSharma et al., , 2014b. The wide range of OC/EC ratios indicates a relatively high fraction of organic aerosols from biomass burning and/or enhanced contribution from SOC formation (Ram and Sarin, 2015;Sheoran et al., 2021). During the winter season, a lower OC/EC ratio (Nainital: 2.33 ± 0.53; Mohal-Kullu: 2.60 ± 0.34) was observed, which is attributable to the dominance of primary organic aerosols from BB (Meng et al., 2018;Ram et al., 2008), whereas Darjeeling shows a lower OC/EC ratio in the summer season (1.74 ± 0.30) attributable to the formation of SOC (Lim and Turpin, 2002;Ram et al., 2008). The positive linear relationship of OC and EC suggests their emissions from common sources like BB or VE, where EC arises from primary combustion sources,  3.01 ± 0.85 0.80 ± 0.12 0.54 ± 0.14 3.70 ± 1.93 0.68 ± 0.19 0.47 ± 0.20 3.13 ± 0.80 0.60 ± 0.33 0.33 ± 0.16 which implies that a major proportion of OC also originated from combustion sources (Dewangan et al., 2016;Kumar and Attri, 2016;Ram and Sarin, 2010;Sahu et al., 2018;Salma et al., 2004;Sharma et al., 2020aSharma et al., , 2021. A weak correlation between OC and EC shows that SOA was formed due to the gas-to-particle transformation of VOCs (Begum et al., 2013). Fig. S2 shows the scatter plots of OC and EC for study sites. OC and EC exhibit a significant linear trend during the winter season (Darjeeling: R 2 = 0.765; Nainital: R 2 = 0.854; Mohal-Kullu: R 2 = 0.912). However, in the summer season, the observed samples are widely dispersed (Fig. S2). The well-defined seasonal variability is attributable to the emission potential of sources and the varying contribution from a variety of sources (fossil-fuel combustion (FFC), BB) (Ram et al., 2008;Ram and Sarin, 2015).

Secondary Organic Carbon (SOC)
In Table 1, the mean concentrations of estimated SOC at study sites are reported annually and seasonally. The estimated SOC exhibited an average annual concentration of 2.94 µg m -3 , 2.56 µg m -3 , and 3.94 µg m -3 , corresponding to 54%, 48%, and 36% of the total OC mass over Darjeeling, Nainital, and Mohal-Kullu, respectively. Sheoran et al. (2021) reported the average SOC concentration of 1.99 µg m -3 at Nainital, corresponding to 23% of total OC mass. Previous studies (Pachauri et al., 2013;Shivani et al., 2019) reported the percentage contribution of SOC corresponding to total OC mass as 43-49% for PM2.5 in urban Delhi and accounted for 18%, 25%, and 60% of OC at traffic, rural, and remote sites, respectively, in Agra. On a seasonal basis, the contribution of SOC corresponding to OC over Darjeeling was highest in the monsoon season (59%), followed by post-monsoon (42%), winter (36%), and summer (27%) season. Whereas for Nainital, the SOC fraction to OC increases in summer (55%) and post-monsoon (46%) season and is lowest in winter (26%) season. Mohal-Kullu showed the highest percentage contribution of SOC to OC in the post-monsoon season (41%) and the lowest in winter (26%). One more study in Delhi reported a 50% contribution of SOC on OC, with the increased value in the summer and winter seasons due to the fast oxidation process and fog processing that affects the formation of SOC (Rajput et al., 2018;Sheoran et al., 2021). The higher impact of SOC on OC in the summer season than in the winter season can be attributable to intense solar radiation and higher temperature that enhance the photochemical activities and formation of SOC (Cao et al., 2006;Murillo et al., 2013).

Enrichment factor (EF)
The elements found in Earth's crust (Al, Fe, Ti, and Ca) are known as crustal elements exhibiting low EFs (< 5) in all seasons at all three sampling sites (Fig. 5). At Darjeeling, EFs of P, Cr, Zn, Pb, Ni, Mo, and B were recorded > 10, indicating their anthropogenic origin in all seasons (Chatterjee et al., 2010). Over Nainital, elements like Zn, Pb, B, Cu, Cr, Zr, Mo, Mn, and Ni showed EF > 10 for all seasons, which indicates their anthropogenic origin. Moreover, the highest EF of B, Mo, Cu, and Ni are attributable to the increased anthropogenic contribution such as industrial emissions (IE) (Shridhar et al., 2009). At Mohal-Kullu, EFs of Mg, Al, Na, Fe, P, Zr, Ca, K, and Sr in PM10 were recorded as < 10 in all the seasons, indicating their crustal origin. Moreover, EFs of V, Cr, Ba, and Zn were observed > 10 in all the seasons, indicating their anthropogenic origin.

Source Apportionment of PM10 Using PCA/APCS
SPSS software (IBM, version 26.0) has been used to conduct the PCA analysis. In PCA, Kaiser Normalization and Varimax rotation was applied to maximize the variance of species loadings to identify possible sources of PM10. Only principal components (PCs) having greater than 1 eigenvalue were considered in this study.

Darjeeling
In the present study, PCA has been carried out with PM10 species to identify the sources using higher factor loadings of PCs. PCA determined four PCs based on factor loadings that explain 69.1% of the total variance (Table 3). The first principal component (PC1) accounts for 30% of the variance that showed the highest loading of Al, Ti, Fe, Ca, Mg, and K, indicating crustal or soil dust (SD) as the primary source Jangirh et al., 2022;Shivani et al., 2019). The EFs of the elements showing higher loadings in Factor 1 also confirmed their crustal origin as the significant source of PM10 over Darjeeling. Air mass trajectory shows that the continental air masses mainly originated from the northwestern regions of Darjeeling and indicated the resuspension of local and wind-blown soil dust particles for coarse mode aerosols and the influence of biomass burning from fine mode aerosols (Chatterjee et al., 2010). The second component was industrial emissions (IE) + coal combustion (CC), which showed the highest loading of Mg, Ni, Cr, P, Zn, and       (Table 3) and contributed 18.2% of the total variance might be originating from small scale coal-based industries and coal engines (Gupta et al., 2007;Jain et al., 2017). Similar results were reported by Chatterjee et al. (2010) that the coal engine used by Darjeeling Himalayan railways is major source of coal combustion markers in Darjeeling. The third source showed the abundance of OC, K, EC, WSOC, S, and Cl in PC3, identified as BB/FFC Jangirh et al., 2022;Song et al., 2006), accounting for 13% of the total variance (Table 3), which is also supported by the ratios of OC/EC and WSOC/OC (Sen et al., 2018). At last, the fourth component explains a 7.7% variance with OC, EC, B, and Zr as dominant species, indicating vehicular emissions (VE) as one of the sources primarily due to the massive influx of tourists around the year  3.5.2 Nainital 22 chemical constituents of PM10 (carbonaceous and elements) have been used to identify the possible sources applying PCA. Four PCs that explain 77.7% of the total variance have been presented in Table 3. The first component explained 48.1% of the variance, and the higher load of K, Al, Ti, Mg, Ca, and Fe accredited to crustal + mineral dust source (Balachandran et al., 2000;Jangirh et al., 2022;Khillare et al., 2004;Sheoran et al., 2021). Recent studies Sheoran et al., 2021) reported mineral dust as the major aerosol source over Nainital using PSCF analysis that the major source locations were the Thar desert and desert areas in East Africa (Dumka et al., 2019;Sheoran et al., 2021). As per Fig. 7(b), the southwest monsoon (SWM) that passes the Arabian Sea transports the dust-laden air masses towards the Himalayas. The second factor accredited to the IE with considerable loading of Br, Mn, Ni, Zr, and Zn accounted for 12.6% variance (Chelani et al., 2008;Jain et al., 2019;Pant and Harrison, 2012). The third component, dominated by OC, EC, Ni, and Mn, accounted for 9.9% of the variance attributed to VE (Chakraborty and Gupta, 2010). At last, the fourth component assigned to BB + FFC shows higher loadings of K, OC, S, EC, Cl, and WSOC (Jain et al., 2019;Jangirh et al., 2022;Khare and Baruah, 2010), with 7.2% of the variance. The high OC/EC ratio is also indicative of BB emissions. During the winter season, biomass/wood burning activities enhanced for heating purposes and long-range transport of air masses from BB source areas in Himachal Pradesh, Uttrakhand (forest fires), and Punjab (extensive agricultural BB) is also the major contributor to BB emission markers over Nainital .

Mohal-Kullu
Three components were resolved with a total variance of 89.7%. The first component was dominated by mixed elements attributed to the VE/IE source with a variance of 70.6% (Table 3). Crustal dust sources have been identified in the second component dominated by Mg, Al, P, S, Cl, and Fe, which accounted for 11.7% of the total variance (Table 3) (Jain et al., 2019;Jangirh et al., 2022). At last, the third component was dominated by the marker elements (OC, EC, S, Cl, K), attributed to BB + FFC (Table 3), and accounted for 7.3% of the total variance Jangirh et al., 2022;Song et al., 2006).
After PCA, APCS were calculated using factor scores and zero concentration (Larsen and Baker, 2003). Then, linear regression was performed on APCS and measured concentrations of species to estimate the actual source contribution. The measured and predicted PM10 concentration using PCA/APCS was depicted in Fig. S4. The predicted species concentration in each source with R 2 was listed in Tables S4-S7 for all sites. Fig. S3 depicts the concentration of chemical species from specific sources determined by APCS. Table S8 reported the identified sources of PM using receptor models over the different sites in the IHR. The percentage contribution of sources were estimated as 10%, 24%, 31%, and 35% for crustal/SD, IE + CC, BB + FFC, and VE, respectively, over Darjeeling (Table 3, Fig. 6). The percentage contribution of sources over Nainital was estimated as 55%, 7%, 13%, and 25% for crustal + mineral dust, IE, VE, and BB + FFC, respectively (Table 3). Over Mohal-Kullu, the estimated percentage contribution of sources was 54%, 15%, 29% for VE + IE, crustal dust/SD, and BB + FFC, respectively (Fig. 6). Majorly the common sources of PM in the IHR are crustal/soil dust/mineral dust, BB/FFC, vehicular activities, and long-range transported aerosols (Table S8). Limited studies are available in the literature to intercompare the identified sources of PM using receptor models (Chatterjee et al., 2010;Panwar et al., 2020;Prabhu et al., 2019;Sheoran et al., 2021;Soni et al., 2020).

Influence of Transport Pathway of PM10
3.6.1 Backward trajectory In order to understand the path of particulate air masses reaching the sampling sites, we plotted annual and seasonal AMBTs at all heights (100 m, 500 m, and 1000 m), as shown in Figs. S5 and S6. Annual trajectories at different heights (Fig. S5) for all sites were mainly approaching from the local areas of Uttrakhand, Jammu and Kashmir, the Thar Desert, Himachal Pradesh, Afghanistan, the IGP region, and the Bay of Bengal (BoB). In addition, AMBTs at 500 m AGL for all sampling stations were studied to reduce the surface turbulence impact (Yang et al., 2017). Fig. 7 depicts the AMBTs at 500 m AGL at all receptor sites originating from local and long-distance regions where the IGP region was the major pollutant emitter. During the summer season, the heavy loadings of AMBTs arriving at Darjeeling (Fig. 7(a)) originated from the semi-arid (the Thar desert, Rajasthan, Punjab) and the IGP region, Assam, Nepal, and Sikkim. Significant transport of AMBTs was observed during the monsoon season, mainly originating from BoB, as reported by Rai et al., 2020Rai et al., , 2021. In contrast, local BB activities are a prominent source of PM during the winter season.
During the summer season, the AMBTs arriving at Nainital (Fig. 7(b)) originated from Pakistan, Afghanistan, Iran, the Thar Desert, and the IGP region of India. The IGP region was the major source region during the winter season, where biomass/wood burning activities intensified due to prevailing meteorological conditions (low mixing height and low ambient temperature) . Similar to Darjeeling, AMBTs originating from BoB reflected the influence of marine air masses at the receptor site in the monsoon season Sheoran et al., 2021).
During the summer and winter seasons, the AMBTs arriving at Mohal-Kullu (Fig. 7(c)) originated from the local and upper IGP region, Afghanistan, Pakistan, and a few passing over the Arabian Sea. On the other hand, during the post-monsoon season, the heavy loadings of air masses came from northwestern countries (Iran, Afghanistan, Pakistan) (Gajananda et al., 2005).

Potential source contribution function (PSCF)
In order to determine the potential source areas of PM10 at the receptor sites (Darjeeling, Nainital, and Mohal-Kullu), the PSCF was done. The grids with a probability < 0.1 are marked as transparent, while other grids are shown in colours (lower probability grids are represented in lighter colours, and higher probability are shown in darker colours). Figs. S7-S10 depicts the annual and seasonal PSCF plots of PM10 at all heights (100 m, 500 m, and 1000 m AGL). A detailed description of PSCF results at sampling sites was as follows: Darjeeling: Overall, heavy loadings of westerly air masses were coming from Rajasthan, Uttar Pradesh, Madhya Pradesh, and other surrounding regions. In the summer season, majorly AMBTs originate from IGP and Nepal region Rai et al., 2020Rai et al., , 2021, while in the winter as well as post-monsoon season, PSCF shows the source regions are Punjab, Haryana, Himachal Pradesh, Jammu and Kashmir ( Fig. S9(a)).
Nainital: During the summer season, most of the air masses approaching the receptor site came from Afghanistan, the Thar Desert, Pakistan, and the IGP region. During the monsoon season, southwest monsoon (SWM) airflows across the Arabian Sea transport dust-laden air masses over the receptor site ( Fig. S9(b)). In addition, local biomass/wood burning activities lead to domestic heating in the winter season. In the post-monsoon season, the source regions are Himachal Pradesh, Uttrakhand, Punjab, and nearby areas where extensive agricultural residues burning and forest fires occur, contributing to the long-distance transport of pollutants at the receptor site Sheoran et al., 2021).

Mohal-Kullu:
The IGP and the Thar Desert were the primary source region over the receptor site. PSCF plots showed that the dominance of air masses was mainly from the west, northwest, southwest, south, and southeast directions (Fig. S9(c)). AMBTs and its PSCF analysis conclude that the source regions of pollutants are locally-originated from the continental landmass and transboundary movement of air masses (Fig. S9). Though the PSCF analysis successfully recognizes the source regions in the geographical area, the grid cells having a concentration of pollutant greater than the predetermined value have the same PSCF value, and the grid cells which are devoid of any endpoints and known pollution sources could not be considered as source region in this model. Therefore, annual and seasonal CWT was also done for all sampling locations (Figs. S11-S14).

Concentration-weighted trajectory (CWT)
CWT calculates the contribution of the source region in terms of PM concentration. Figs. S11-S14 illustrate the annual and seasonal CWT plots of PM10 at all heights over each study site. However, measured concentrations of particulates were assigned to grids (termed as hot spots) for CWT analysis. The value of the grid signifies the magnitude of the effect of PM pollution over the observational sites.
Darjeeling: The PM10 source regions were mainly local and regional, including the middle and upper IGP region (Fig. S13(a)). The maximum residence duration of AMBTs coming from the IGP region was similar to previous literature (Adak et al., 2014;Sen et al., 2018). During the winter season, Nepal and the IGP region of India contribute maximum to PM concentration, whereas lower CWT originated from the western region and concludes that diverse sources originated from Afghanistan, the Thar Desert, and Pakistan (Rai et al., 2020Sharma et al., 2020a). The summer and post-monsoon seasons were considered peak tourism seasons when massive anthropogenic emissions of tourist vehicles and toy trains were observed over the study area. Sen et al. (2018) concluded a similar observation that western IGP contributed to particulate concentration in the post-monsoon season due to BB activities for cooking, heating, etc. Nainital: Fig. S13(b) shows the CWT plot of PM10 over Nainital at 500 m (AGL), showing the contribution from the probable source zone with high loadings. During the winter as well as postmonsoon season, the PM sources were primarily local, IGP region, and from neighboring areas of Pakistan. Therefore, the PSCF and trajectory results aligned with the high CWT value, are the potential source areas.
Mohal-Kullu: In all seasons, Fig. S13(c) shows that the significant sources of PM were local regions with higher CWT values. In addition, the long-range transport originated from IGP and nearby areas with lower CWT values.

CONCLUSIONS
This study examined the carbonaceous and elemental composition of PM10 collected over Darjeeling, Nainital, and Mohal-Kullu from July 2018 to December 2019. Seasonal variation of carbonaceous components and elements in PM10 at different sites were assessed to identify the prominent sources (PCA/APCS) of PM10 in the high altitude stations of the IHR and their source regions using PSCF and CWT analysis. Major outcomes of the present study are furnished below: 1. The annual mean concentration of PM10 over Nainital, Darjeeling, and Mohal-Kullu was recorded as 65 ± 41 µg m -3 , 54 ± 17 µg m -3, and 57 ± 32 µg m -3 , respectively. The mass concentration of PM10 showed considerable seasonal variability, which might be attributed to long-distance transport of pollutants, local pollution sources, and prevailing climatic conditions. 2. Carbonaceous component concentrations were highest in the winter and post-monsoon seasons (owing to increased burning operations throughout IGP and the Himalayas for household heating) at all three locations. In contrast, the lowest concentration was found during the monsoon season (due to the washout of pollutants). 3. OC/EC ratio and WSOC/OC ratios were used to identify CAs and differentiate the sources as primary and secondary. During the winter season, a low OC/EC ratio is attributable to SOC formation. The high OC/EC ratio and their significant correlation indicate that the BB is one of the major sources in the IHR. WSOC/OC ratio showed that aged aerosols dominated the Himalayan region, and long-range transported aerosols from the IGP region also contributed to PM sources over the Himalayas. 4. The enrichment factor shows the significant contribution of crustal/mineral dust in coarse mode aerosols over the study sites. 5. The PCA revealed four common sources over the IHR: crustal/SD (26.6%), BB + FFC (28%), VE (28%), and IE + CC (17%). 6. PSCF and CWT identified the source areas and showed that PM10 was primarily transported from northwestern parts of India (Haryana, Punjab), a northeastern region of Pakistan, and the Thar Desert. In contrast, the IGP and the Thar Desert contributed to dust-related aerosols over the receptor locations.
The findings of this study will be helpful in identifying the specific sources for emission inventories and effectively assessing the climate impacts across Himalayan regions. Further, it is expected that the present study will establish a long-term air quality monitoring network over the Himalayan region of India.

Author's Contributions
The conceptualization and design of the study were planned by SKS; Data collection and analysis were performed by NC, PS, SM, AR, JCK, RL, AC, NV, MN, TKM, and SKS; the first draft was written by NC. Data interpretation was carried out by NC, SM, JCK, RL, AC, NV, MN, TKM, RKK, and SKS. All the authors read, reviewed, and approved the final manuscript.

Data Availability
The datasets developed during the current study are available from the corresponding author on reasonable request.