Contents lists available at ScienceDirect

Atmospheric Research

journal homepage: www.elsevier.com/locate/atmosres

Influence of anthropogenic emissions on wet deposition of pollutants andrainwater acidity in Guwahati, a UNESCO heritage city in Northeast India

Rajyalakshmi Garagaa, Supriyo Chakrabortyb, Hongliang Zhangc, Sharad Gokhalea, Qiao Xued,Sri Harsha Kotae,⁎

a Department of Civil Engineering, Indian Institute of Technology Guwahati, Indiab Center for Climate Change Research, Indian Institute of Tropical Meteorology, Pune, Indiac Department of Environmental Science and Engineering, Fudan University, Shanghai, ChinadDepartment of Environment, College of Architecture and Environment, Sichuan University, Chengdu, Chinae Department of Civil Engineering, Indian Institute of Technology Delhi, India

A R T I C L E I N F O

Keywords:Northeast IndiaUNESCO siteWet depositionAcid neutralizing capacityIsotope analysisHYSPLITPMF

A B S T R A C T

Guwahati, the largest urban corridor of Northeast India, is one of the United Nations Educational, Scientific andCultural Organization (UNESCO) world heritage sites and one of the 200 eco-regions in the world. The presentstudy investigates the characterization of chemical components and sources of precipitation samples collected inGuwahati during June 2016–June 2017. Acidic rain events occurred throughout the year, with a frequency of64% and 87% during monsoon and non-monsoon seasons, respectively. Higher contributions of the acidicspecies (SO4

2− and NO3−) coinciding with poor neutralizing capacity of crustal species (Ca2+, K+ and Mg2+)

led to acid rain in this region. Isotope analysis (δ18O and δD) indicated that monsoonal and non-monsoonal rainswere of marine and non-marine origins, respectively. This is further supported by the back trajectory analysis asmajority of the individual rain events during d-excess< 10‰ i.e. monsoon, indicated their maximum con-tribution from ocean while, during d-excess> 10‰ i.e. non-monsoon, the trajectories originated from water-inland. The enrichment factors (EF) for Pb, Zn, Co and Cu were>5, indicating the dominance of anthropogenicsources in this region. The Positive Matrix Factorization (PMF) along with isotope analysis identified marine(40%) as the major source in monsoon and industrial emissions (28%) in non-monsoon, indicating rainwaterevaporation is more of ocean and continental origin during monsoon and non-monsoons, respectively. This studysuggests the need of further studies and implementation of stringent anthropogenic regulations not only in localbut also at regional and global scale, in this acid rain prone region.

1. Introduction

Acid rain, a global ecological problem is defined as the precipitationhaving pH<5.6, the value at which the cloud water will be in equi-librium with atmospheric CO2, causing adverse effects on ecosystems(Charlson and Rodhe, 1982). India, being 2nd most populous country inthe world, identified as 11th mega biodiversity center in the world and3rd in Asia sharing approximately 11% of its total plant resources(Ramakrishnan, 2007). Northeastern India, which is a gateway of muchof India's variety of species, spotted as part of Indo-Burma biodiversityhotspot, which is one of the 35 Global biodiversity hotspots (Chatterjeeet al., 2006). However, the increasing anthropogenic activities likeexpansion of industrial activities, acid rains and huge consumption offossil fuels have been causing the detrimental effects to the variety of

ecosystems (Mehr et al., 2019; Singh and Agrawal, 2007). Modeling ofglobal patterns of deposition fluxes of sulfur and nitrogen during lastdecade, indicated the high depositions in southeastern China, north-eastern India, Bangladesh and central Europe (Vet et al., 2014). Thesestudies observed that the wet and dry deposition of sulfur and nitrogenis high in northeastern India, having good agreement between modelestimates and measurements. Also, due to acidification of sulfur andNitrogen, 7–17% of the global area of natural ecosystems is at greatperil (Bouwman et al., 2002).

Rapid urbanization, expansion of industrial activities, huge con-sumption of fossil fuels and population explosion led to air pollution,which is rising at an alarming rate in world's cities. According to worldhealth organization (WHO) report (WHO, 2016), out of world's 15 mosthighly polluted cities, 14 cities belongs to India posing a major

https://doi.org/10.1016/j.atmosres.2019.104683Received 9 May 2019; Received in revised form 11 September 2019; Accepted 13 September 2019

⁎ Corresponding author.E-mail address: harshakota@gmail.com (S.H. Kota).

Atmospheric Research 232 (2020) 104683

Available online 16 September 20190169-8095/ © 2019 Elsevier B.V. All rights reserved.

T

environmental risk to human health. In Indian scenario, where highconcentrations of particulate matter (PM) range from thoracic to al-veolar size fractions, wet scavenging plays a critical role (Gobre et al.,2010; Kulshrestha et al., 1999). Emissions of gaseous and particulatepollutants released into atmosphere are washed out by rains to someamount due to solubilizaton, which can cause ecological imbalance andaffect the biogeochemical cycle (Cao et al., 2009; Kulshrestha et al.,2014). This process of wet scavenging takes place in two differentpathways namely below-cloud scavenging (washout) and in-cloudscavenging (rainout). Coarse particles are removed more effectively bywashout process while the rainout process drains the fine particles andgases surrounding the cloud droplets present in the atmosphere (Kajinoand Aikawa, 2015).

Time trend analysis in precipitation (rainwater) chemistry hasgained importance due to its potential in environmental monitoringsuch as, acid deposition (acid rain), trace metal deposition, eu-trophication as well as analysis of the global climate change (Xiao,2016). Though this practice has been extensively used in developedcountries, however, only a few studies have explored its potential indeveloping countries like India (Kulshrestha et al., 2014; Tiwari et al.,2012).

Innumerable studies on wet scavenging of atmospheric pollutantshave been reported in India targeting the metropolitan cities, includingnorthern India especially in the nation's capital, Delhi, in western India(Budhavant et al., 2014; Momin et al., 2005), in eastern India (Daset al., 2005; Kulshrestha et al., 2014) and southern India (Kulshresthaet al., 2003). Also, there exists continuous monitoring programs such asBackground Air Pollution Monitoring Network (BAPMoN), Global At-mospheric Watch (GAW) and Precipitation Chemistry Monitoring Pro-gram of the Indian Institute of Tropical Meteorology (IITM) in Pune.However, main focus of these studies was the rainwater characteriza-tion instead of source identification. The rainwater composition is animportant attribute that helps us to quantify the relative contributionsof different sources of atmospheric pollutants. The rainwater compo-sition changes from region to region due to the difference in thesources. Therefore, findings from one location cannot be inferred toother locations, without having the monitoring data, due to the varia-tion of transboundary characteristics of the atmospheric pollution(Akpan et al., 2018). Previous studies reported that rainwater consistsof mixture of chemical species originated from natural (sea salt (Na+,Cl−) and soil dust (Ca2+, K+, Mg2+)) or anthropogenic sources (in-dustrial and vehicular emissions) emits acidic species such as SO4

2−

and NO3−, which are carried away by wind from distant sources

(Hamilto-Taylor and Willis, 1990; Jonnalagadda et al., 1994).The variation in the chemical composition of rainwater depends on

sources, long range transport of air masses, meteorological conditions(Herrera et al., 2009; Kulshrestha et al., 1999; Zunckel et al., 2003).Multivariate receptor techniques are used to identify the sources andtheir contributions of rainwater (Qiao et al., 2018; Qiao et al., 2015a;Qiao et al., 2015b). In Varanasi, dominant sources namely industrialemissions, crustal sources and biomass burning were extracted bymeans of principal component analysis (PCA) (Pandey and Singh,2012). Using PCA, three significant sources, soil dust, sea salt and fossilfuel combustion were reported as major sources, affecting rainwaterchemistry, in Nainital, in the central Himalayas, India (Bisht et al.,2017). This was widely used model in many of the South Asian urbancities (Chakraborty et al., 2016a; Rao et al., 2016; Roy et al., 2016;Tiwari et al., 2016). Rao et al. (2016) used the US EPA's PMF model andreported that the rainwater composition was greatly influenced by soildust (with Ca2+ as dominant cation) and fossil fuel consumption(SO4

2− as predominant anion). In contrary to only few previous studies(For example, see (Balachandran and Khillare, 2001; Kulshrestha et al.,2014)) witnessing acid rain events, most of the studies (For example see(Al-Momani et al., 1995)) reported the alkaline nature (pH > 5.6) ofrainwater due to its neutralization by airborne dust and ammonia re-leased from natural and anthropogenic sources. However, most of these

studies were conducted in monsoon periods (Budhavant et al., 2011;Khemani et al., 1994; Momin et al., 2005) but little attention was paidto non-monsoon rains (Rao et al., 2016). Missing the analysis on non-monsoon rains could result in erroneous conclusions on rainwaterchemistry (Kumar et al., 2014).

The present study was conducted in Guwahati city, Assam, anortheastern state of India, which is one of the UNESCO world heritagesites, known for its wildlife and natural heritage (Garaga et al., 2019).The state has been allied between the great Himalayas and vast floodplains of mighty Brahmaputra river. This area is rich in its mineralsdeposit and is well-known for its fertile land, contributing $46 billion tothe nation's total gross domestic product (GDP), and engaging 69% ofthe state's population in agriculture (https://en.wikipedia.org/wiki/Economy_of_Assam). However, the soil of the state is acidic which re-strict the productivity potential (Kulshrestha et al., 2014) necessitatinginvestigation to understand the cause of acidic nature.

Therefore, the present study attempts a multi-dimensional approachto focus on the following objectives: (1) to analyze the chemical char-acteristics of rainwater in this region; (2) to locate the source of rain-water using isotopic fingerprinting for the monsoon and non-monsoonperiods along with back trajectory analysis; and (3) to characterize theemission sources using EF and PMF.

2. Materials and methods

2.1. Study area

The present study was conducted in Guwahati (Fig. 1) (26°11′14″Nand 91°41′30″E, with an elevation of 50–680m), one of the quick de-veloping cities in northeast India. Sampling site is an industrial hub,clustered by specific types of businesses such as an export promotionindustrial park, a biotechnology park, a skin and health care industry, apharmaceutical company and a machine industry. Presence of Indianliquid petroleum gas (LPG) bottling plant in the city premises has fur-ther aggravated the worsening of air quality. Besides these, construc-tion activities are carried out round the year. The study area is located20 km from the heart of the city and is surrounded by major highways.The temperature averages 23.5 °C, which peaks in July/August (40 °C)and declines in January/December (11 °C). The city receives a meanannual precipitation of 1720mm, most of which is restricted to theMay–October period (http://worldweather.wmo.int/en/city.html?cityId=529).

2.2. Sample collection and pH measurement

Wet deposition samples were collected on the roof of a buildingabout 10m from ground level and 50 cm from the floor of the roof.Sampling was done manually using 2 L borosilicate volumetric flask andpolyethylene funnel of 20 cm diameter (Fig. S1) (Khemani et al., 1989;Roy et al., 2016). The collectors (bottles and funnels) were used forsample collection only after proper washing with detergent and HNO3

followed by deionized water. Collectors were placed as soon as the rainbegan and were retrieved immediately after the rain stopped. Im-mediately after retrieval, pH was measured using digital pH meter in analiquot of the unfiltered sample. Remaining sample amount wastransferred into two clean 100mL polyethylene vials for chemicalcharacterization.

2.3. Chemical analysis techniques and quality control

2.3.1. Trace elements and total organic carbon (TOC)One part (100mL) was acidified with 1mL HNO3 maintaining pH ~

2 and used for metal analysis (Co, Cu, Cr, Fe, Mn, Ni, Pb, Sr and Zn)using Atomic Absorption Spectroscopy (AAS) (Varian Spectra AA-55)and TOC using Shimadzu TOC 5000 total organic carbon analyzerequipped with an Automatic Sample Injector (ASI) (Shimadzu, Kyoto,

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

2

Japan). Prior to analysis, each sample was filtrated through membranefilter (pore size: 0.45 μm). Quality assurance was ensured by using thecertified Standard Reference Material (SRM) 1640a (natural water)from National Institute of Standards & Technology (NIST), USA. TheSRM was treated and analyzed in the same way as the insoluble fractionof rainwater sample. Recovery was within±10% of the certified valuesfor all the elements.

2.3.2. IonsWhile the other aliquot (50mL) was sectioned into two portions in

which one (25mL) was used for ion analysis (Ca2+, Mg2+, NH4+, Na+,

K+, Cl−, NO3−, and SO4

2−) using Ion Chromatograph (Metrohm 792basic IC). Before injection, samples were filtered through a 0.22 μmfilter paper to remove the insoluble matter. Anion column (Metrosep ASupp 5–250/4.0) with suppressor was used for analysis of anions. Asolution mixture of 3.2mM Na2CO3 and 1mM NaHCO3 was used as theeluent and the flow rate was maintained at 0.7 mL/min. 50mM H2SO4

was used for regeneration.Cations were analyzed by a cation column (Metrosep C 4150/4.0).

The eluent was prepared in ul-tra-pure water with 1.7 mmol/L nitricacid and 0.7 mmol/L dipicolinic acid. The eluent flow rate was main-tained at 0.9mL/min. 20 μL of sample was measured using inbuilt loopand injected in to the IC system. Before injecting, all the samples werefiltered through Millipore 0.22 μm PTFE filters. Calibration and quan-tification of components were performed using MERCK referencestandards (CertiPUR) (https://www.merckmillipore.com/IN/en/product/IC-multi-element-standard-VI,MDA_CHEM-109,036#anchor_Product%20Information) (Chaturvedi et al., 2017) of1, 2, 5 ppm for anions and 2, 5, 10 ppm for cations. All the samples werestored in refrigerator at 4 °C.

To find the acidic neutralizing capacity of the alkaline species(Ca2+, Mg2+ and K+) and check the level of acidification from acidrain pollution in the present study region the neutralization factor (NF)for alkaline species was calculated using Eqs. (1)–(3) (Kulshrestha et al.,1995). Therefore, the acid neutralization by Ca2+, Mg2+ and K+ wascalculated using their equivalent and non-sea salt (nss) concentrationsas follows:

= − + −+ + − −NF nss Ca /(NO nss SO )Ca2 2

3 42 (1)

= − + −+ + − −NF nss Mg /(NO nss SO )Mg2 2

3 42 (2)

= − + −+ + − −NF nss K /(NO nss SO )K 3 42 (3)

The nss values of any particular component (nss-C) was evaluatedby using the following equation (Eq. (4)):

− = + +[nss C] C –[Na ] [C/Na ]x x x sea salt (4)

where, x is sample, Cx is measured concentration of an ion in sample ‘x’,and [C/Na+]sea salt is the seawater ratio for particular ion with Na(Keene et al., 1986).

2.3.3. Isotope analysis10mL of the remaining aliquot, was preserved without filtration to

avoid any evaporative losses, for isotope analysis. The isotopic (d18Oand d2H) ratios were determined using an LGR Liquid Water and WaterVapor Isotope Analyzer (Model: TIWA-45-EP). The isotopic ratiosmeasured, agreed within the±1σ variability. Further analytical detailsusing the laser isotope technique were presented elsewhere(Chakraborty et al., 2016b). The proportion of stable isotopes is ex-pressed similar to Eq. (5) known as global meteoric water line (GMWL)(Craig, 1961):

= + >δD (‰) 8 δ O 10 (R 0.95)18 2 (5)

The slope and intercept of any local meteoric water line (LMWL),which is derived from precipitation collected in any specific area understudy, will likely to be different from GMWL reflecting the existence ofunique moisture source in that region (Craig, 1961). The deuteriumexcess (d-excess) defined in (Dansgaard, 1964) and expressed as in Eq.(6) characterizes the atmospheric precipitation and is mainly useful forsource identification.

− =d excess (‰) δD–8 δ O18 (6)

2.4. Source identification

2.4.1. EF analysisThe EF of trace metals with respect to crustal elements to identify

whether certain element is influenced by anthropogenic sources otherthan its natural sources (Acciai et al., 2017; Rahn, 1971). This tech-nique is applied by taking into account of commonly used referenceelements such as Al, Si, Na, Ca, Mg, Mn, K, Fe etc. (Dubey et al., 2012;Yongming et al., 2006). The value of EF is generally calculated as fol-lows (Eq. (7)):

=EF [X/Mn] /[X/Mn]x Sample Crust (7)

where, [X/Mn]Sample and [X/Mn]Crust are the mean concentration of

Fig. 1. Location of the sampling site, Guwahati, the largest city of northeast India.

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

3

target element to Mn in the respective sample and in continental crust,respectively. The concentrations of reference element i.e. Mn is basedon earth crust mean abundance of the metals taken from CRC handbook(Lide, 2007). The enrichment categories based on EF values are: EF≈ 1suggests natural origin, EF= 2–5 is moderately enriched andEF= 5–20 is significant enrichment (Yongming et al., 2006).

2.4.2. PMF analysisThe US EPA's PMF model was employed to identify the sources

contributing to the pollutant concentrations. Its non-requirement ofprior knowledge of contributing source characteristics (i.e. sourceprofiles) improves its feasibility. Based on the given concentration anduncertainty data, PMF provides source profiles and source contributionsto each sample and the detailed description is given in the EPA PMF 5.0guide (Norris et al., 2014). PMF uses the speciated concentration (Xij) ofsample ‘i’ and species ‘j’ and estimates the number of sources ‘l’, thesource profile ‘f’ and the amount of mass ‘g’ contributed by each sourceto each individual sample as expressed in Eq. (1). The residual eij, in Eq.8 is minimized using a least square approach (Paatero, 1997; Paateroand Tapper, 1994).

∑= +=

X g f eijp

l

ip pj ij1 (8)

The robustness of the model lies in the ability to weigh each in-dividual data point using measured uncertainties and respectivemethod detection limit (MDL) (Hopke, 2016). Missing values were re-placed with associated geometric mean values. Results are constrainedto have non-negative source contributions and profiles. The PMF solu-tion minimizes the object function ‘Q', based upon the uncertainties (u)associated with each sample using Eq. (9).

∑ ∑=

⎡

⎣

⎢⎢⎢⎢

− ∑ ⎤

⎦

⎥⎥⎥⎥= =

=QX g f

ui

m

j

k ijp

l

ip pj

ij1 1

1

2

(9)

where, ‘m’ and ‘k’ are total number of samples and species, respectively.In this study, 40 rainwater samples and 18 species were used as

input data for PMF model. Categorization of quality of data was basedon the signal to noise ratio (S/N) and the percentage of samples belowMDL. The species S/N ratio > 1 were categorized as ‘strong’, < 0.2 as‘bad’ and rest as ‘weak’. Thereafter, uncertainties ‘UCij’ were calculatedusing Eq. (10), unless Cij is less than MDLij, where UCij is taken as 5/6thof MDLij.

= × + ×C MDLUC (0.1 ) (0.5 )ij ij ij2 2 (10)

To estimate the uncertainty of the estimated profiles, 100 bootstrapruns with a minimum co-relation of 0.6 were performed.

2.4.3. HYSPLIT analysisSources of moisture are important and their locations play a sig-

nificant role to understand their influence on the study locations.Therefore, air parcel back trajectory analysis was used, which help us tofind the source regions that contributes to the atmospheric pollutantconcentrations (Huang et al., 2008). Data log of Global Data Assim-ilation System (GDAS) model using US National Oceanic and Air Ad-ministration (NOAA) Air Resource Laboratory's (ARL) HYSPLIT model(Draxler, 2011) used to plot the 5-day back trajectories with a startingheight of 2000m above the actual ground level, which is a typical cloudheight during rains (Budhavant et al., 2016; Rao et al., 2016).

3. Results and discussion

3.1. Acidity and chemical composition

Total rainwater samples collected were 40, out of which 17 eventswere during monsoon, and remaining 23 samples belonged to the non-monsoonal period. pH of these rain events varied from 4.59–5.99 with amean of 5.20 (Fig. 2). Further analysis indicated that 64% and 87% ofthe precipitation events during monsoon and non-monsoon season wereacidic in nature, i.e. had pH<5.6. Reference pH of 5.6 is used as anindicator all around the globe to identify the acid rain events as, at thispH, cloud water will be in equilibrium with atmospheric CO2 (Charlsonand Rodhe, 1982). Even though, most of the Indian cities reported al-kaline rain events during the last decade, similar results 52% of acidrain events were reported by previous studies in Assam (Bhaskar andRao, 2017; Kulshrestha et al., 2014). Such acid rain events in this regionsuggests the abundance of SO2 and NOx emissions and lack of bufferingpotential of the soil in this region (Kulshrestha et al., 2014). Also, heavyrainfall (mean annual precipitation of 1720mm) and dense vegetationcover in this part of India makes the soil containing components vul-nerable from lifting into the atmosphere thereby leading to acid rainevents (Bhaskar and Rao, 2017).

The data of trace elements measured in the current study and stu-dies from other UNESCO world heritage sites were shown in Table 1. Inthe present study, in most cases, concentrations were in good agree-ment with those in US (Conko et al., 2004) and Madhya Pradesh (Patelet al. 2001) with few exceptions. The trend during the whole studyperiod was: Zn (20%) > Sr (18%) > Mn (16%) > Fe (11%) > Cu(9%) > Pb (8%) > Ni > (6%) > Co (5%) > Cr (3%). The con-centration of Zn was maximum during non-monsoon while Sr washighest in monsoon. Also, concentrations of the few trace elements suchas Cu, Fe and Sr have increased by 65%, 42% and 67% respectively,during monsoon than non-monsoon. TOC concentrations were similarto those of studies in China (Huang et al., 2010), Japan (Schrumpf et al.,2006) and USA (Velinsky et al., 1986). TOC concentrations in monsoonand non-monsoon seasons were similar.

The average ionic concentrations in the rainwater were in the orderof SO4

2− (65 ± 32 μeq/L) > NO3− (60 ± 31 μeq/L) > Cl−

Fig. 2. (a) Frequency distribution of pH and (b) ionic balance in the 40 rainwater samples collected from June 2016 to June 2017.

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

4

(42 ± 21 μeq/L) > Ca2+ (27 ± 21 μeq/L) > K+ (8 ± 26 μeq/L) > F− (7 ± 10 μeq/L) > Na+ (7 ± 5 μeq/L) > Mg2+

(6 ± 4 μeq/L). Table 1 also shows the comparison of ionic compositionof the rainwater samples collected in the present study along withdifferent UNESCO sites of the world. The acidic species (SO4

2− andNO3

−) were approximately similar to the other UNESCO sites. Theconcentrations of chemical constituents in this study were similar tothat in previous study in the same region (Bhaskar and Rao, 2017;Kulshrestha et al., 2014). While, the concentrations of cations werecomparable to that of other UNESCO sites, anion concentrations in thisstudy were lower than other UNESCO sites. For example, the SO4

2− inthis study was in the same range of China while, concentrations of Na+

in China was atleast 10 times higher than this study. This could be dueto different source contributions in these sites. Higher contribution ofacidic species in this study indicated the influence of industrial sources,coal consumption and thermal power plants in and around the location.Earlier studies reported the abundance of SO4

2− in this region whileCa2+ being dominant in other parts of the country. About 78% ofaverage ionic composition is formed by anions whereas remaining 22%is formed by cations. Maximum concentrations of the anions were ob-served during monsoon except for NO3

− which was high in non-mon-soon. Except Mg2+ (which was maximum in monsoon), all the othercations had high concentrations during non-monsoon. During monsoon,SO4

2− was having the highest concentration among all. NF values forCa2+, Mg2+ and K+ were 0.09, 0.01 and 0.03 respectively, duringmonsoon and 0.1, 0.01 and 0.07 respectively, during non-monsoonperiods. Table 2 compares the NF of these ions of this study with thestudies of UNESCO sites. When compared with those studies, except forCa2+, the other ions are having similar values as of those reported inother UNESCO sites, altogether showing poor neutralization. Suchnegligible neutralization capacity of alkaline species could be thereason for 76% of the acid rain events in this region (Das et al., 2005;Kulshrestha et al., 2014).

3.2. Isotope analysis

Using all the event- based samples collected during the presentstudy, the local meteoric water line (LMWL) in Guwahati (Fig. 3(a)) wasestablished as:

= + =δD (‰) 8.40 δ O 12.01 (R 0.98)18 2 (11)

The slope and intercept of the LMWL (Eq. (11)) are more or lesssimilar to that of GMWL (Eq. (5)). Similar slopes indicate the raindropformation in the tropical monsoon rainforest region of northeast India,under equilibrium conditions (Dansgaard, 1964) and unaffected by theevaporation effect (Craig, 1961). Table S1 in the supplementary textshows the LMWL values of the studies conducted in and around thepresent study location. It can be observed that the values of slopes andintercepts are similar and comparable with the present study.

The average values of δ18O and δD were −4.11‰ and −23.41‰during monsoon, which were much less than the average values duringnon-monsoon i.e., −0.80‰ and 5.60‰. This suggests that the occur-rence of isotopically enriched rain events during non-monsoon, whichwas also observed in a study conducted in southern India (Srivastavaet al., 2014). The d-excess values, which serves as an essential tool toidentify the moisture source, and if d-excess> 10‰, it implies that themoisture source is typically from the continental region, i.e. evapo-transpiration and nominally from the ocean (Wirmvem et al., 2017). Inthe present study, the d-excess values for monsoon and non-monsoon

Table 1Comparison of chemical characteristics of rainwater measured from June 2016 to June 2017 at Guwahati, along with other UNESCO world heritage sites (ions in μeq/L, trace elements in μg/L and TOC in mg/L).

Constituent Guwahatia Jorhatb Jordanc USAd Japan⁎ Russia⁎ Indonesia⁎ Chinaf

Monsoon Non-monsoon

F− 0.32–52.79 0.47–12.16 1.80 NA NA NA NA NA 11.5–59.2Cl− 14.39–78.36 5.78–80.93 7.70 80.60 NA 10.4 29.4 22.4 6.8–1003.2SO4

2− 30.60–120.81 14.63–126.31 52.80 53.80 NA 14.7 31.9 24.4 39.4–170.5NO3

− 9.31–120.24 30.07–118.61 39.20 35.70 NA 14.8 21.9 29.8 6.2–34.8Na+ 0.65–15.00 1.96–28.04 10.10 75.60 NA NA 23.9 8.32 7.7–304.3K+ 0.39–7.31 0.77–20.34 6.30 18.40 NA 28 5.84 3.2 0.9–767.6Ca+2 8.98–49.40 8.98–81.09 41.80 71.40 NA 95 16.9 9.25 28.9–812.2Mg+2 1.25–16.25 1.25–15.00 9.80 62.30 NA 26 4.27 3.09 30.6–71.7Co 40–180 40–200 NA NA 0.51 NA NA NA NACr 4–117 27–193 NA NA NA NA NA NA NACu 40–600 20–240 NA NA 0.56 0.62 NA NA NAFe 70–863 26–429 NA NA 15.3 NA NA NA NAMn 21–628 46–834 NA NA 1.24 1.64 NA NA NANi 13–234 24–730 NA NA 1.12 0.26 NA NA NAPb 5–339 41–677 NA NA 1.9 1.24 NA NA NASr 55–876 69–464 NA NA NA 0.13 NA NA NAZn 23–674 153–809 NA NA 5.16 4.77 NA NA NATOC 2.2–9.30 2.7–8.85 NA NA NA NA NA NA NA

NA: not avaiable.aPresent study 2016–17 sampling, India.

b (Kulshrestha et al., 2014): 2005–06 sampling, India.c (Al-Khashman, 2005): 2002–04 sampling, India.d (Verry and Vermette, 1992).f (Qiao et al., 2015b): April 2010 to May 2011 sampling, China.⁎ (EANET, 2012): 2009–2010 sampling, Japan.

Table 2Neutralization factor at different UNESCO world heritage sites.

Area K+ Ca2+ Mg2+ Reference

Guwahati 0.046 0.092 0.007 This workJorhat 0.068 0.45 0.106 (Kulshrestha et al., 2014)Jordan 0.206 0.79 0.69 (Al-Khashman, 2005)Japan 0.96 3.20 0.89 (Takeda et al., 2000)Russia 0.108 0.314 0.079 (EANET, 2012)Indonesia 0.059 0.17 0.057China 0.082 1.90 0.50 (Qiao et al., 2015b)

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

5

were 9.47 and 12.14‰ respectively. To estimate the error of d-excess,an internal laboratory standard was measured for a long time (ca.3 years) (Fig. S2). The standard deviation of d-excess has been calcu-lated (0.75‰) which is reported as error for d-excess. The uncertaintylevel is 0.75‰, is significantly smaller than many previous studies (Forexample see Froehlich et al. (2002)). This could be due to the fact thatmost of the isotopic analysis until about the middle of 2000's was doneby the conventional mass spectrometric technique which typicallyyielded a high error (ca 2‰) in δD analysis (For example see Froehlichet al. (2002)). Since, the present analysis was carried out using an LGRLiquid Vapor and Water Isotope Analyzer, the uncertainty level is muchlower.

Further, in order to understand the seasonal influence on the iso-topic composition of precipitation, the δ18O vs d-excess was plotted asshown in the Fig. 3(b). The distribution shows two distinct differentgroups of data points having two very different trend lines, indicatingsample events belonged to two different groups, monsoon and non-monsoon. A higher slope in the δ18O vs d-excess space represents astrong raindrop evaporation undergoing during the non-monsoonseason and a reduced slope representing low evaporation as a result ofincreased humidity during the monsoon season.

Fig. S3, in the supplementary text shows the HYSPLIT trajectoriesduring monsoon season, having d-excess < 10‰ for most of the in-dividual rain events. For few of the rain events during monsoon, have d-excess > 10‰ which could be due to feeding of secondary water vaporinto the cloud system (Chakraborty et al., 2016a, 2016b). While Fig. S4shows the trajectories during non-monsoon having d-excess > 10‰.Therefore, the trajectories originated from majority of the individualrain events indicated that the maximum contribution from ocean duringd-excess< 10‰ i.e. monsoon and water-inland during d-excess >10‰ i.e. non-monsoon. Furthermore, the d-excess in this study iscomparable to other studies reported in India (Table S2). Therefore, theisotope analysis which is used to identify the geographic origin of themoisture, revealed that the significant amount of rainwater during non-monsoon originated from the water inland i.e. long range transportwhile during monsoon the moisture is transported from the oceanwhich is inline with HYSPLIT analysis.

3.3. Source identification

The source apportionment results carried out using EF analysis fordifferent elements are displayed in Fig. 4. It is evident that the elementsexcept Fe, other elements like Cr, Ni and Sr have the mean EF's > 3. EFof Fe is close to unity meaning it is of geological origin. Whereas, all theother elements i.e. Pb, Zn, Co and Cu have the EF's > 10, indicatingthe influence of anthropogenic sources. Therefore, EF analysis in thepresent study served as an effective tool in identifying the impact ofelemental concentrations by human activities other than natural

sources.EPA PMF v5.0 using the dataset of 40 samples were employed and 4

to 7 factor solutions were tested wherein, mixed and unidentifiablesources were obtained. Hence 5-factor solution with Fpeak of −0.5yielded most reasonable source apportionment results.

Fig. 5 shows the source profiles and its associated percentage ofspecies in a source, which is the median value from bootstrap runs.

3.3.1. Dust emissionsThis factor explained 26% of the total variance with high relative

abundances of Mn (69%), Sr (52%) and Cr (51%). These species aretracers of re-suspended dust and crustal sources and are largely re-ported by many researchers (Crilley et al., 2017; Xue et al., 2010; Zhaoet al., 2011). Since, the study area is close to the highways, it is influ-enced by the abrasive emissions emitted from the highway surfaces.

3.3.2. Marine sourceThis source has maximum (28%) contributions to measured con-

stituents in rain water. High loadings of Na+ (61%), Cl− (40%), Ca2+

(49%) and K+ (38%) characterized this source (Garaga et al., 2018;Kothai et al., 2011; Police et al., 2016; Sharma et al., 2014). The generaluse of K+ as indicator to biomass burning and Cl− as coal combustiontracer in Indian studies, leads to ambiguity however, the combinationof Cl, Na, K, Ca should provide a better signature.

3.3.3. Coal combustionThis factor is mainly characterized by significant levels of Fe (69%),

F− (51%), SO42− (45%), NO3

− (33%) and Cl− (32%) and accounts for9% to the rainwater constituents. Based on the previous studies, thesespecies were reported as tracers for coal combustion and indicating that

Fig. 3. Isotope analysis (δ18O and δD) for the sample events collected during June 2016 – June 2017 (a) Local meteoric water line (b) seasonal influence on isotopiccomposition of precipitation.

0

5

10

15

20

25

30

35

40

E nrichrichme ntFact or

Trace metals

Fig. 4. Distribution of enrichment factors for the trace metals with Mn as re-ference element.

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

6

the surrounding regions have been influenced by the coal consumptionactivities (Machado et al., 2015; Qiao et al., 2015b; Sharma et al., 2016;Tiwari et al., 2016).

3.3.4. Vehicular emissionsThis factor accounted for 18% and contributed to high proportions

of Zn (77%), Cu (74%), Co (44%) and TOC (25%). Earlier studies re-ported that Zn, Cu, Co and TOC are the elemental tracers and reliablemarkers for the vehicle exhaust emissions (Chakraborty and Gupta,2010; Dadashazar et al., 2019; Garaga et al., 2018; Sharma et al.,2014). As this area experiences a huge traffic flow round the year,hence this source is expected.

3.3.5. Industrial emissionsThis factor indicated high loadings of Pb (75%), Ni (57%), Sr (34%),

Cr (34%) and Co (32%) and contributed 19% of the rainwater con-stituents. Previous studies suggested that these elements are mainlyemitted from industries like steel, electroplating, metallurgy, foundry(metal casting), machinery and battery, etc. and are used as key mar-kers to identify industrial emissions (Edwards, 1997; Krishnamurthyet al., 2014; You et al., 2017).

Fig. 6 shows the variation of source contributions in monsoon, non-monsoon and all samples (combined). Marine source (28%) and dustemissions (26%) were the main sources during whole study period(panel (c)). In monsoon samples (when d-excess< 10‰), marinesource (40%) and vehicular emissions (36%) were the major con-tributors (panel (a)) indicating the probable moisture source region isfrom ocean, as observed from most of the backward trajectories during

this period. While on the other hand, the contributions from the marinesource decreased by 16% and industrial emissions increased by 20% inthe rain events occurred during non-monsoon periods, as the rain waterdroplet was more of in-land origin.

4. Conclusion

The present study conducted in Guwahati, one of the fastestgrowing cities of India, where uncontrolled development and increasedanthropogenic activities has led to declined air quality in this regionover few decades. Therefore, the wet deposition of PM analyzed re-vealed a year-long acid rain events (pH < 5.6) in this region. Some ofthe major ions, trace metals and TOC measured in the rainwater sam-ples indicated the low concentrations of crustal species with negliblebuffering capacity of acidic species. The isotope analysis in conjunctionwith back trajectory analysis conducted to understand the source originof rain droplet, identified that the significant source of rainwater duringnon-monsoon was originated from water inland, while during monsoonthe moisture transported from the ocean. Also, enriched isotopic com-position in non-monsoon season (average Deuterium excess (d-excess)of 9.47‰), and isotopically depleted rains were during monsoon period(average d-excess of 12.14‰) was observed during the study period.These results bolster the conclusions from the US EPA's PMF analysisthat showed higher contributions from marine sources in monsoon andindustrial sources in non-monsoon seasons, respectively. Also, the EFanalysis identified the enriched contributions of Pb and Zn which aretypically of anthropogenic origin suggesting that this region needsimplementation scientifically construed control strategies to regulate

Fig. 5. Source profiles resolved by EPA PMF v5.0 model for the rainwater samples collected during the study period.

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

7

the anthropogenic activities which lead to severe acid rains.

Declaration of Competing Interest

None.

Acknowledgements

This work is funded by the Department of Science and Technology,India (ECR/2016/000087). Authors would also like to thank theMinistry of Human Resources Development, India.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.atmosres.2019.104683.

References

WHO, 2016. Ambient Air Pollution: a Global Assessment of Exposure and Burden ofDisease. 2016. World Health Organization, pp. 40–42.

Acciai, C., Zhang, Z., Wang, F., Zhong, Z., Lonati, G., 2017. Characteristics and sourceAnalysis of trace elements in PM2. 5 in the Urban Atmosphere of Wuhan in Spring.Aerosol Air Qual. Res. 17, 2224–2234.

Akpan, A.O., Udosen, E.D., Offiong, N.A.O., 2018. Rainwater Chemistry within theVicinity of Qua Iboe Estuary, Nigeria. CLEAN–Soil Air Water 46, 1700114.

Al-Khashman, O.A., 2005. Ionic composition of wet precipitation in the Petra Region,Jordan. Atmos. Res. 78, 1–12.

Al-Momani, I.F., et al., 1995. Chemical composition of precipitation near an industrialarea at Izmir, Turkey. Atmos. Environ. 29, 1131–1143.

Balachandran, S., Khillare, P., 2001. Occurrence of acid rain over Delhi. Environ. Monit.Assess. 71, 165–176.

Bhaskar, V.V., Rao, P., 2017. Annual and decadal variation in chemical composition ofrain water at all the ten GAW stations in India. J. Atmos. Chem. 74, 23–53.

Bisht, D.S., et al., 2017. Chemical characterization of rainwater at a high-altitude site“Nainital” in the Central Himalayas, India. Environ. Sci. Pollut. Res. 24, 3959–3969.

Bouwman, A., Van Vuuren, D., Derwent, R., Posch, M., 2002. A global analysis of acid-ification and eutrophication of terrestrial ecosystems. Water Air Soil Pollut. 141,349–382.

Budhavant, K., Rao, P., Safai, P., Ali, K., 2011. Influence of local sources on rainwaterchemistry over Pune region, India. Atmos. Res. 100, 121–131.

Budhavant, K., Rao, P., Safai, P., Granat, L., Rodhe, H., 2014. Chemical composition of theinorganic fraction of cloud-water at a high altitude station in West India. Atmos.Environ. 88, 59–65.

Budhavant, K.B., Rao, P.S.P., Safai, P.D., Leck, C., Rodhe, H., 2016. Black carbon in cloud-water and rain water during monsoon season at a high altitude station in India.Atmos. Environ. 129, 256–264.

Cao, Y.-Z., Wang, S., Zhang, G., Luo, J., Lu, S., 2009. Chemical characteristics of wetprecipitation at an urban site of Guangzhou, South China. Atmos. Res. 94, 462–469.

Chakraborty, A., Gupta, T., 2010. Chemical characterization and source apportionment ofsubmicron (PM1) aerosol in Kanpur region, India. Aerosol Air Qual. Res. 10,433–445.

Chakraborty, A., Gupta, T., Tripathi, S.N., 2016a. Chemical composition and character-istics of ambient aerosols and rainwater residues during Indian summer monsoon:

Fig. 6. Source contributions obtained using EPA PMF v5.0 model during (a) monsoon (b) non-monsoon (c) and combined.Authors declare that the work described has not been published previously, that it is not under consideration for publication elsewhere, that its publication isapproved by all authors and tacitly or explicitly by the responsible authorities where the work was carried out, and that, if accepted, it will not be publishedelsewhere in the same form, in English or in any other language, including electronically without the written consent of the copyright-holder.

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

8

Insight from aerosol mass spectrometry. Atmos. Environ. 136, 144–155.Chakraborty, S., et al., 2016b. Atmospheric controls on the precipitation isotopes over the

Andaman Islands, Bay of Bengal. Sci. Rep. 6, 19555.Charlson, R., Rodhe, H., 1982. Factors controlling the acidity of natural rainwater. Nature

295, 683.Chatterjee, S., Saikia, A., Dutta, P., Ghosh, D., Worah, S., 2006. Review of biodiversity in

Northeast India. Background Paper. http://siteresources.worldbank.org/INTSAREGTOPWATRES/Resources/Background_Paper_13.pdf (Accessed 25December 2018).

Chaturvedi, M., Tiwari, R., Kulshrestha, U., 2017. Atmospheric reactive nitrogen fluxesand scavenging through wet deposition over Mathura (India). J. Indian Geophys.Union 21, 319–326.

Conko, K.M., Rice, K.C., Kennedy, M.M., 2004. Atmospheric wet deposition of traceelements to a suburban environment, Reston, Virginia, USA. Atmos. Environ. 38,4025–4033.

Craig, H., 1961. Isotopic variations in meteoric waters. Science 133, 1702–1703.Crilley, L.R., et al., 2017. Source apportionment of fine and coarse particles at a roadside

and urban background site in London during the 2012 summer ClearfLo campaign.Environ. Pollut. 220, 766–778.

Dadashazar, H., Ma, L., Sorooshian, A., 2019. Sources of pollution and interrelationshipsbetween aerosol and precipitation chemistry at a Central California site. Sci. TotalEnviron. 651, 1776–1787.

Dansgaard, W., 1964. Stable isotopes in precipitation. Tellus 16, 436e468 Google Scholar.Das, R., Das, S., Misra, V., 2005. Chemical composition of rainwater and dustfall at

Bhubaneswar in the east coast of India. Atmos. Environ. 39, 5908–5916.Draxler, R.R., 2011. Hysplit (Hybrid Single-Particle Lagrangian Integrated Trajectory)

Model Access Via NOAA ARL Ready Website. http://ready.arl.noaa.gov/HYSPLIT.php.Dubey, B., Pal, A.K., Singh, G., 2012. Trace metal composition of airborne particulate

matter in the coal mining and non–mining areas of Dhanbad Region, Jharkhand,India. Atmos. Pollut. Res. 3, 238–246.

EANET, 2012. Data Report on the Acid Deposition in the East Asian Region 2010. AcidDeposition Monitoring Network in East Asia.

Edwards, J., 1997. Coating and Surface Treatment Systems for Metals: A ComprehensiveGuide to Selection.

Froehlich, K., Gibson, J., Aggarwal, P., 2002. Deuterium excess in precipitation and itsclimatological significance. In: Technical Report no. IAEA-CSP-13/P.

Garaga, R., Avinash, C.K.R., Kota, S.H., 2019. Seasonal variation of airborne allergenicfungal spores in ambient PM10—a study in Guwahati, the largest city of north-eastIndia. Air Qual. Atmos. Health 12 (1), 11–20.

Garaga, R., Sahu, S.K., Kota, S.H., 2018. a review of air quality modeling studies in India:local and regional scale. Curr. Pollut. Rep. 4, 59–73.

Gobre, T., et al., 2010. Chemical composition of precipitation in the coastal environmentof India. Bull. Environ. Contam. Toxicol. 85, 48–53.

Hamilto-Taylor, J., Willis, M., 1990. A quantitative assessment of the sources and generaldynamics of trace metals in a soft-water lake. Limnol. Oceanogr. 35, 840–851.

Herrera, J., Rodríguez, S., Baéz, A., 2009. Chemical composition of bulk precipitation inthe metropolitan area of Costa Rica, Central America. Atmos. Res. 94, 151–160.

Hopke, P.K., 2016. Review of receptor modeling methods for source apportionment. J. AirWaste Manage. Assoc. 66, 237–259.

Huang, K., Zhuang, G., Xu, C., Wang, Y., Tang, A., 2008. The chemistry of the severeacidic precipitation in Shanghai, China. Atmos. Res. 89, 149–160.

Huang, X.-F., et al., 2010. 5-year study of rainwater chemistry in a coastal mega-city inSouth China. Atmos. Res. 97, 185–193.

Jonnalagadda, S., Makadho, J., Matinde, N., Karimanzira, R., Makarau, A., 1994.Chemical composition of rainwater and air quality in Zimbabwe, Africa. Sci. TotalEnviron. 144, 261–271.

Kajino, M., Aikawa, M., 2015. A model validation study of the washout/rainout con-tribution of sulfate and nitrate in wet deposition compared with precipitationchemistry data in Japan. Atmos. Environ. 117, 124–134.

Keene, W.C., Pszenny, A.A., Galloway, J.N., Hawley, M.E., 1986. Sea-salt corrections andinterpretation of constituent ratios in marine precipitation. J. Geophys. Res.-Atmos.91, 6647–6658.

Khemani, L., et al., 1989. Spread of acid rain over India. Atmos. Environ. (1967) 23,757–762.

Khemani, L.T., et al., 1994. Atmospheric pollutants and their influence on acidification ofrain water at an industrial location on the West Coast OF India. Atmos. Environ. 28,3145–3154.

Kothai, P., Saradhi, I., Pandit, G., Markwitz, A., Puranik, V., 2011. Chemical character-ization and source identification of particulate matter at an urban site of NaviMumbai, India. Aerosol Air Qual. Res. 11, 560–569.

Krishnamurthy, N., Vallinayagam, P., Madhavan, D., 2014. Engineering Chemistry. PHILearning Pvt. Ltd.

Kulshrestha, U., Sarkar, A., Srivastava, S., Parashar, D., 1995. Wet-only and bulk de-position studies at New Delhi (India). Water Air Soil Pollut. 85, 2137–2142.

Kulshrestha, U., et al., 1999. Measurements of acid rain over Indian Ocean and surfacemeasurements of atmospheric aerosols at New Delhi during INDOEX pre-campaigns.Curr. Sci. 968–972.

Kulshrestha, U., Kulshrestha, M.J., Sekar, R., Sastry, G., Vairamani, M., 2003. Chemicalcharacteristics of rainwater at an urban site of south-central India. Atmos. Environ.37, 3019–3026.

Kulshrestha, M.J., Singh, R., Duarah, R., Rao, P., 2014. Influence of crustal aerosols onwet deposition at a rural site of North-East India. Int. J. Environ. Stud. 71, 510–525.

Kumar, P., Yadav, S., Kumar, A., 2014. Sources and processes governing rainwaterchemistry in New Delhi, India. Nat. Hazards 74, 2147–2162.

Lide, D., 2007. CRC Handbook of Chemistry and Physics. Vol, A52 (1913–1995). pp. 523.

Machado, P.M., et al., 2015. Fluorine determination in coal using high-resolution graphitefurnace molecular absorption spectrometry and direct solid sample analysis.Spectrochim. Acta B At. Spectrosc. 105, 18–24.

Mehr, M.R., Keshavarzi, B., Sorooshian, A., 2019. Influence of natural and urban emis-sions on rainwater chemistry at a southwestern Iran coastal site. Sci. Total Environ.668, 1213–1221.

Momin, G., et al., 2005. Study of chemical composition of rainwater at an urban (Pune)and a rural (Sinhagad) location in India. J. Geophys. Res.-Atmos. 110.

Norris, G., Duvall, R., Brown, S., Bai, S., 2014. EPA Positive Matrix Factorization (PMF)5.0 Fundamentals and User Guide. Prepared for the US Environmental ProtectionAgency Office of Research and Development. Inc., Petaluma, Washington, DC.

Paatero, P., 1997. Least squares formulation of robust non-negative factor analysis.Chemom. Intell. Lab. Syst. 37, 23–35.

Paatero, P., Tapper, U., 1994. Positive matrix factorization: a non-negative factor modelwith optimal utilization of error estimates of data values. Environmetrics 5, 111–126.

Pandey, J., Singh, A., 2012. Chemical characterization of rain water in a seasonally drytropical region (Varanasi), India. J. Environ. Biol. 33, 629.

Police, S., Sahu, S.K., Pandit, G.G., 2016. Chemical characterization of atmosphericparticulate matter and their source apportionment at an emerging industrial coastalcity, Visakhapatnam, India. Atmos. Pollut. Res. 7, 725–733.

Patel, K.S., Shukla, A., Tripathi, A.N., Hoffmann, P., 2001. Heavy Metal Concentrations ofPrecipitation in East Madhya Pradesh of India. Water, Air, and Soil Pollution 130,463–468. https://doi.org/10.1023/a:1013857923368.

Qiao, X., et al., 2015a. Modeling dry and wet deposition of sulfate, nitrate, and ammo-nium ions in Jiuzhaigou National Nature Reserve, China using a source-orientedCMAQ model: part I. Base case model results. Sci. Total Environ. 532, 831–839.

Qiao, X., et al., 2015b. Atmospheric wet deposition of sulfur and nitrogen in Jiuzhaigounational nature reserve, Sichuan province, China. Sci. Total Environ. 511, 28–36.

Qiao, X., et al., 2018. Wet deposition of sulfur and nitrogen in Jiuzhaigou National NatureReserve, Sichuan, China during 2015–2016: possible effects from regional emissionreduction and local tourist activities. Environ. Pollut. 233, 267–277.

Rahn, K.A., 1971. Sources of Trace Elements in Aerosols: An Approach to Clean Air.Michigan Univ., Ann Arbor. Dept. of Meteorology and Oceanography.

Ramakrishnan, P., 2007. Traditional forest knowledge and sustainable forestry: a north-east India perspective. For. Ecol. Manag. 249, 91–99.

Rao, P.S.P., et al., 2016. Sources of chemical species in rainwater during monsoon andnon-monsoonal periods over two mega cities in India and dominant source region ofsecondary aerosols. Atmos. Environ. 146, 90–99.

Roy, A., et al., 2016. Precipitation chemistry over urban, rural and high altitudeHimalayan stations in eastern India. Atmos. Res. 181, 44–53.

Schrumpf, M., Zech, W., Lehmann, J., Lyaruu, H.V., 2006. TOC, TON, TOS and TOP inrainfall, throughfall, litter percolate and soil solution of a montane rainforest suc-cession at Mt. Kilimanjaro, Tanzania. Biogeochemistry 78, 361–387.

Sharma, S.K., et al., 2014. Variation of OC, EC, WSIC and trace metals of PM10 in Delhi,India. J. Atmos. Sol. Terr. Phys. 113, 10–22.

Sharma, S.K., et al., 2016. Source Apportionment of PM2.5 in Delhi, India using PMFModel. Bull. Environ. Contam. Toxicol. 97, 286–293.

Singh, A., Agrawal, M., 2007. Acid rain and its ecological consequences. J. Environ. Biol.29, 15.

Srivastava, R., Ramesh, R., Rao, T.N., 2014. Stable isotopic differences between summerand winter monsoon rains over southern India. J. Atmos. Chem. 71, 321–331.

Takeda, K., Marumoto, K., Minamikawa, T., Sakugawa, H., Fujiwara, K., 2000. Three-yeardetermination of trace metals and the lead isotope ratio in rain and snow depositionscollected in Higashi–Hiroshima, Japan. Atmos. Environ. 34, 4525–4535.

Tiwari, S., Chate, D.M., Bisht, D.S., Srivastava, M.K., Padmanabhamurty, B., 2012.Rainwater chemistry in the North Western Himalayan Region, India. Atmos. Res.104–105, 128–138.

Tiwari, S., et al., 2016. Nature and sources of ionic species in precipitation across theindo-gangetic plains, India. Aerosol Air Qual. Res. 16, 943–957.

Velinsky, D.J., Wade, T.L., Wong, G.T., 1986. Atmospheric deposition of organic carbonto Chesapeake Bay. Atmos. Environ. (1967) 20, 941–947.

Verry, E.S., Vermette, S., 1992. The deposition and fate of trace metals in our environ-ment. In: General Technical Report NC-150. U.S. Dept. of Agriculture, Forest Service,North Central Forest Experiment Station, St. Paul, MN.

Vet, R., et al., 2014. A global assessment of precipitation chemistry and deposition ofsulfur, nitrogen, sea salt, base cations, organic acids, acidity and pH, and phosphorus.Atmos. Environ. 93, 3–100.

Wirmvem, M.J., et al., 2017. Variation in stable isotope ratios of monthly rainfall in theDouala and Yaounde cities, Cameroon: local meteoric lines and relationship to re-gional precipitation cycle. Appl Water Sci 7, 2343–2356.

Xiao, J., 2016. Chemical composition and source identification of rainwater constituentsat an urban site in Xi’an. Environ. Earth Sci. 75, 209.

Xue, Y., et al., 2010. Source characterization and apportionment of PM10 in Panzhihua,China. Aerosol Air Qual. Res. 10, 367–377.

Yongming, H., Peixuan, D., Junji, C., Posmentier, E.S., 2006. Multivariate analysis ofheavy metal contamination in urban dusts of Xi’an, Central China. Sci. Total Environ.355, 176–186.

You, S., Yao, Z., Dai, Y., Wang, C.-H., 2017. A comparison of PM exposure related toemission hotspots in a hot and humid urban environment: Concentrations, compo-sitions, respiratory deposition, and potential health risks. Sci. Total Environ. 599,464–473.

Zhao, J., et al., 2011. Chemical characteristics of particulate matter during a heavy dustepisode in a coastal city, Xiamen, 2010. Aerosol Air Qual. Res. 11, 299–308.

Zunckel, M., Saizar, C., Zarauz, J., 2003. Rainwater composition in Northeast Uruguay.Atmos. Environ. 37, 1601–1611.

R. Garaga, et al. Atmospheric Research 232 (2020) 104683

9