The QDB, situated in the northeastern part of the QXP, is bordered by the Altyn-Tagh, Kunlun, and Qilian mountains, making it one of China's largest intermontane basins (Zhang, 1987). With an average elevation of 3000 m, the basin features an extremely arid climate characterized by less than 20 mm of annual precipitation in the northwestern region, while evaporation rates exceed 2000 mm (Feng et al., 2022). The QDB is rich in salt lakes, non-ferrous metals, and hydrocarbon resources, with significant coal deposits. As a major salt lake resource area, it hosts 33 lakes, including Qarhan, Dachaidan, and Caka Salt Lake, and faces notable conflicts between resource extraction and ecological preservation. Agriculturally, it cultivates crops like goji berries, quinoa, and forage grass, and hosts China's largest resource-rich circular economy pilot zone. The permanent population of the basin is approximately 400 000, primarily using coal, yak dung, and firewood for domestic heating (Jiang et al., 2020). Additionally, annual tourism peaks from May to September, attracting around 17 million visitors (Qinghai Provincial Bureau of Statistics, 2024a), which likely amplifies atmospheric pollutant emissions.

From January 2020 to March 2023, monthly dust deposition samples were collected from six monitoring stations in the southern QDB. The stations included Xiao Zaohuo (XZH), Golmud (GEM), Da Gele (LTC), Nuo Muhong (NMH), Balong (BLX), and Dulan station (DLX) (Fig. 1). Dry deposition collection employed the glass ball method using Marble dust collector (MDCO) designed dust collection cylinders (Sow et al., 2006). The MDCO sampler operates on the principle of gravitational settling. It uses a collection vessel with a known opening to capture particulate matter that settles naturally from the air. In field measurements, surrogate surfaces or deposition traps are commonly employed to better quantify atmospheric deposition. These surfaces mimic the original ground surface, are easy to deploy, and can be integrated without significantly disturbing airflow. Examples include glass beads (Ganor, 1975; Offer et al., 1992), moist filter paper (Goossens and Offer, 1993), plastic surfaces (Gregory, 1961), and water or antifreeze solutions (Smith and Twiss, 1965). The MDCO deposition vessel is based on the original concept of Ganor (1975), using glass beads as a surrogate collection surface. Due to their very low microscopic roughness, the beads help prevent the resuspension of particles once they have been captured (Goossens and Offer, 1993).

The stainless steel collection device (50 × 30 × 30 cm) contained a plastic sieve container of identical dimensions, with the sieve base positioned 10 cm above the opening and perforated with 0.5 cm diameter holes (Fig. S1 in the Supplement). To minimize dust resuspension during high wind events (Qian and Dong, 2004), two layers of 16 mm glass balls were placed within the sieve container. A high-density polyethylene bag was attached to the base for sample collection. According to Sow et al. (2006), the collection efficiency of the MDCO decreases with increasing wind speed, dropping below 20 % when wind speed exceeds 3 m s⁻¹, and it preferentially collects the finer fraction of atmospheric dust particles ranging from 10 to 31 µm in size (Chow, 1995). Furthermore, collection efficiency decreases significantly as the size of dust-fall increases. Based on our findings, more than 50 % of the collected dust particles are below 40 µm (Fig. S2), whereas the mean particle size of atmospheric dust deposition reported in other studies is approximately 80 µm (Lin et al., 2022a, b). This confirms that the MDCO sampler preferentially collects relatively finer particles. Consequently, this study may place greater emphasis on analyzing the chemical activity of the finer fraction of atmospheric dust, thereby enhancing the detection sensitivity for components such as secondary ions and carbonaceous species in deposited dust. However, this particle-size selectivity introduces a sampling bias that can amplify inter-site differences in deposition flux between sites, complicate local source apportionment, strengthen signals from distant sources (e.g., combustion and vehicle emissions), and weaken the contribution from local, coarse-particle-dominated sources (e.g., resuspended dust and construction dust). These factors may affect the accuracy of both deposition flux estimates and source apportionment. Future research should enhance data comparability and the reliability of ecological assessments through efficiency correction and the combined application of multiple methods. The collection efficiency at each site was calculated based on local dust grain-size distribution and wind speed; details are provided in Sect. S1 in the Supplement. This type of collector has been widely used in many studies to evaluate local dust conditions (Abdollahi et al., 2021; Barjoee et al., 2021; Alzahrani et al., 2024).

In this study, dust samples were collected monthly, with each sampling period lasting 30 or 31 d. The installation height and environment of the samplers are provided in Table S1 in the Supplement. To ensure only dry dust was collected, collection devices were manually-operated covered during rain or snowfall. A total of 37, 39, 23, 30, 16, and 29 samples were obtained from XZH, GEM, LTC, NMH, BLX, and DLX stations, respectively, and no duplicate samples were collected during the sampling period. Laboratory protocols incorporated biannual analyses with negative controls and appropriate control samples. As continuous dust monitoring commenced in 2020, site blanks were evaluated during initial sampling. Stations were classified as Urban (GEM, NMH, DLX) and Rural (XZH, LTC, BLX) based on location characteristics. Consistent with the cold-arid climate in QDB, the HP was defined as October–April, while the NHP spanned May–September.

Materials such as plant remnants, microfauna, and bird droppings were removed from the sample bags with tweezers. The samples were then measured on a balance (0.0001 g) to determine the total dust deposition flux (DF) (Eq. 1) (Yu et al., 2016): 1M=m×30S×K, where M is dust deposition [g (m² 30 d)⁻¹]; m is the sample mass (g); S is the area of the dust collection device (m²); and K is the actual number of sampling days per month (d).

A 100 mg sample was weighed and transferred into a 250 mL bottle. The mixture underwent ultrasonic extraction for 20 min to ensure complete solubilization. The resulting supernatant was then filtered through a 0.45 µm filter for analysis. Based on preliminary experimental results, the concentrations of major ions (K⁺, Na⁺, Mg²⁺, and Ca²⁺) were measured using Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES, NexIon 2000). Anions (Cl⁻, SO42-, and NO3-) were quantified using ion chromatography (IC). To ensure measurement accuracy, samples were organized in sets of twenty, with one randomly selected sample from each group serving as a replicate, achieving an error margin of less than 10 %. The detection limits for the various components were as follows: K⁺ (0.0560 mg L⁻¹), Na⁺ (0.0100 mg L⁻¹), Ca²⁺ (0.0037 mg L⁻¹), Mg²⁺ (0.0390 mg L⁻¹), SO42- (0.0090 mg L⁻¹), NO3- (0.0125 mg L⁻¹), Cl⁻ (0.0100 mg L⁻¹). All standard solutions employed in the analysis were sourced from the National Standard Material Center.

According to the Chinese State Standard “Ambient air and waste gas from stationary sources emission-determination of metal elements in ambient particles” (HJ 777-2015), the concentrations of elements such as iron (Fe), aluminum (Al), silicon (Si), titanium (Ti), copper (Cu), cadmium (Cd), chromium (Cr), manganese (Mn), nickel (Ni), zinc (Zn), lead (Pb), and vanadium (V) were quantified using Inductively Coupled Plasma Mass Spectrometry (ICP-MS) and ICP-OES. A dust sample weighing 0.100 g was placed in a Teflon cup, to which 20.0 mL of a nitric acid-hydrochloric acid digestion solution was added. The sample was heated to reflux at 100 ± 5 °C for 2 h under a watch glass, then cooled. Following this, the inner walls of the cup were rinsed with water, and approximately 10 mL of water was added, allowing the mixture to stand for 30 min for extraction. The extract was then filtered into a 100 mL volumetric flask and diluted to volume with distilled water for analysis. In cases where organic matter content was high, an appropriate amount of hydrogen peroxide was introduced during digestion to decompose the organic materials. Prior to sample analysis, the system was flushed with a rinse solution until the blank intensity value reached a minimum, and samples were analyzed only after the signal stabilized. If the concentration of any element exceeded the calibration range, the sample was diluted and reanalyzed.

This study utilized a combination of wet chemical treatment and thermal/optical reflection (TOR) to analyze carbon elements in dust deposition (Han et al., 2007a, b, 2016). Dust samples were digested stepwise to remove inorganic materials: first samples were treated with 10 mL of 2 N hydrochloric acid (HCl) for 24 h at room temperature to dissolve carbonates and partial metals, followed by centrifugation (4500 rpm, 12 min) to separate the residue; then with 15 mL of a 1 : 2 (v/v) mixture of 6 N HCl and 48 % hydrofluoric acid (HF) for 24 h at room temperature to dissolve silicates and residual metals; and finally with 15 mL of 4 N HCl for 24 h at 60 °C to remove minerals such as fluorite formed during demineralization. After each step, the mixture was centrifuged, and the supernatant was collected. The solid residue was diluted with 200 mL of deionized water and filtered through a pre-combusted quartz fiber filter (Whatman, 450 °C for 4 h, diameter 47 mm) (Han et al., 2007a). This method has been widely applied to measure OC and EC contents in lake sediments and urban soils (Han et al., 2009b; Khan et al., 2009; Han et al., 2011). Studies have shown that the EC collection efficiency of this method is approximately 99.6 % (Zhan et al., 2013). However, because OC includes polar, basic, and mineral-bound compounds that may dissolve in HCl/HF and be lost during filtration, the supernatant from all three acid-digestion steps was collected and analyzed for water-soluble OC using a carbon-nitrogen analyzer (Shimadzu, TOCN-4200, Japan). Total OC was obtained by summing the dissolved OC and the particulate OC measured by TOR.

The filtered samples were air-dried and analyzed for carbon content using a DRI 2001 thermal/optical carbon analyzer (Atmoslytic Inc., Calabasas, CA) at the Institute of Earth Environment, Chinese Academy of Sciences, adhering to the Interagency Monitoring of Protected Visual Environments (IMPROVE) protocol. A 0.544 cm² disc was extracted from the filter and placed in a quartz boat for analysis. During the carbon analysis, the samples were initially heated in a 100 % helium atmosphere, resulting in the production of four organic carbon (OC) fractions (OC1, OC2, OC3, and OC4) at four different temperature levels (140, 280, 480, and 580 °C). The atmosphere was subsequently changed to 2 % O₂ / 98 % He, generating three elemental carbon (EC) fractions (EC1, EC2, and EC3) at three temperatures (580, 740, and 840 °C). Volatile carbon underwent carbonization in an anaerobic environment, indicated by a decrease in laser reflectance, and is referred to as pyrolyzed organic carbon (OPC). In the oxidative atmosphere, OPC was emitted along with the original EC from the filter. The amount of OPC is defined as the carbon evolved until the laser reflectance returned to its baseline value (Han et al., 2007b). According to the IMPROVE protocol, EC is calculated as the total of the three EC subfractions minus OPC (i.e., EC is defined as the sum of EC1, EC2, and EC3, with OPC subtracted). The method enables differentiation between soot and char, as determined by the gradual oxidation of these black carbon subtypes in standard reference materials during the EC1 and the EC2 plus EC3 steps, where char is defined as EC1 minus OPC and soot as the sum of EC2 and EC3 (Han et al., 2007a, 2016).

Please note that in this manuscript, we interchangeably use the terms “EC” and “BC”. While these terms do not strictly refer to the same component, they serve as an adequate approximation within the scope of this study (Seinfeld et al., 1998; Bond et al., 2004). We use “EC” when discussing emissions and modeling components, reserving “BC” for climate-related discussions. Throughout this manuscript, the term “OC” refers to the total organic carbon (sum of the dissolved OC and the particulate OC).

The grain size of dust-fall samples was measured using a laser particle analyzer (Malvern Mastersizer 2000, UK). The particle size distribution was calculated for 100 grain size classes within a range of 0.02–2000 µm. Sample preparation for grain size analysis included wet oxidation of organic matter by adding 10 mL of 30 % hydrogen peroxide (H₂O₂) per 1.5 g dry sample. Carbonates were dissolved by boiling with 10 mL (10 % HCl) over 10 min. The glass beakers were filled with 150 mL distilled water and suspended particles were left to deposit. After siphoning the supernatant water, 10 mL of 0.05 N sodium hexametaphosphat [(NaPO₃)₆] were added, and the residue was dispersed for 5 min in an ultrasonic bath before measurement (Lu and An, 1997), the results are expressed as volume percentages.

Meteorological data used in this study were obtained from the Qinghai Meteorological Bureau during the sampling period. Monthly meteorological parameters, including wind speed (WS), wind direction (WD), temperature (Temp), relative humidity (RH), Precipitation (RF), sunshine duration (SUN), and the frequency of visibility ≤ 10 km (VIS), were collected from four national-level meteorological stations: XZH, GEM, DLX, and NMH. All of these stations are standard meteorological observatories established by the state, equipped with standardized monitoring capabilities, and provided continuous and reliable data for this study (https://data.cma.cn, last access: 13 January 2026). In addition, two atmospheric dust deposition sampling sites (LTC and BLX) were established specifically for collecting dust samples. Due to field-monitoring constraints and research-budget limitations, no meteorological instruments were deployed at these two sites.

The DF in the southern QDB is 5.41 ± 5.33 g m⁻² 30 d, slightly lower than that of the Lake Aibi Basin (10.77 g m⁻² 30 d) (Li et al., 2022), but higher than the surrounding areas of Akatama Salt Lake (2.93 g m⁻² 30 d) (Wang et al., 2014). Specifically, DF was 4.67 ± 4.96 g m⁻² 30 d in rural and 5.97 ± 5.73 g m⁻² 30 d in urban areas. During the HP, DF in rural and urban areas were 4.62 ± 4.15 and 4.95 ± 5.25 g m⁻² 30 d, respectively. In contrast, NHP showed increased DF values of 4.77 ± 4.42 g m⁻² 30 d (rural) and 7.66 ± 6.09 g m⁻² 30 d (urban) (Fig. 2). Notably, Urban DF during NHP demonstrated a 54.6 % increase relative to HP, while rural DF rose by 7.1 % (not statistically significant), contrasting previous findings that associated elevated DF with HP coal combustion (Cheng and Lin, 2009; Gao et al., 2013; Guo et al., 2010; Qi et al., 2018). We hypothesize that the increase in DF during the NHP is attributed to seasonal meteorological variations such as wind speed (Yang et al., 2024), and heightened anthropogenic emissions associated with tourism (Zhang et al., 2011). The NHP coincides with the peak tourist season (May–September) in the QDB, which receives about 17 million visitors annually, a number that continues to rise (Qinghai Provincial Bureau of Statistics, 2024a). This elevated human activity likely enhances local emissions, contributing to increased dust deposition during this period. The influence of meteorological factors on dust deposition will be further examined in Sect. 3.5.

Water-soluble ion concentrations differed between rural (115.31 mg g⁻¹) and urban (72.81 mg g⁻¹) areas (not statistically significant). In rural, water-soluble ion content was greater during the HP than in the NHP, while the opposite trend was observed in urban areas (not statistically significant) (Figs. 3 and S5). We categorized the water-soluble ions in dust deposits into ps and nps based on their sources, following the model of marine aerosols (Zhu et al., 2025). Playa salt content consistently surpassed nps in rural areas across both periods, while urban areas showed the opposite trend (not statistically significant). Notably, during the NHP, ps content in urban and rural increased by 54.46 % and 36.86 % respectively (not statistically significant). Backward trajectory analysis indicated that airflow in both rural and urban areas primarily originated from the northwest QDB and the eastern Tarim Basin during the HP, while during the NHP, it was influenced more broadly by the salt lake of the QDB (Figs. S6 and S7), aligning with the observed variations in ps content. A similar increase in summer sea salt and non-sea salt ions has been reported in Rajkot, India, attributed to ocean wind direction (Gupta and Dhir, 2022).

The ratio of nps-SO42-/NO3- in soluble ions is used to differentiate between coal combustion (fixed sources) and vehicular emissions (mobile sources) (Arimoto et al., 1996; Shen et al., 2008). The higher nps-SO42-/NO3- ratio in urban areas compared to rural areas (Fig. S8) indicates that during the study period, stationary sources (e.g., industrial emissions, coal, and biomass combustion) contributed more significantly to the ionic composition of urban dust-fall (Pipal et al., 2019). In contrast, the ratio was considerably lower during the NHP than during HP, suggesting that mobile sources likely played a more important role in the dust ionic composition during the NHP. Generally, emissions from coal combustion and biomass burning was intensified in northern China during the HP (Liu et al., 2016a), while vehicle emissions dominate during the NHP (Xu et al., 2012), supporting the reliability of this analysis. Nevertheless, further investigation integrating atmospheric emission inventories, source tracers, aerosol physicochemical processes, and meteorological conditions is warranted.

The total concentration of major (Fe, Si, Al) and trace elements (Ti, Cr, Cd, Cu, Mn, Ni, Pb, V, Zn) was determined to be 8.74 ± 5.82 mg g⁻¹, while arsenic (As) remained below the detection limit in all analyzed samples. Crustally derived elements – Fe, Al, Si, and Ti – dominated the elemental profile, aligning with dust composition patterns reported in Ira, Singapore, and Beijing-Hebei regions, China (Joshi et al., 2009; Qiao et al., 2013; Eivazzadeh et al., 2021). In comparison to cities such as Beijing, Shanghai, Xi'an, and Lanzhou, and Junggar Basin the levels of heavy metals in dust from the QDB are relatively low (Jiang et al., 2018) (Table S5).

The low heavy metal content in dust deposition within the QDB can be attributed to the following factors. The region has sparse human activity, lacks heavy industrial zones and dense urban clusters, resulting in low total anthropogenic emissions of heavy metals. Furthermore, the surface soil in the QDB itself has a low background level of heavy metals, primarily derived from natural sources with relatively weak influence from human activities (Nuralykyzy et al., 2021; Chen et al., 2021). From the meteorological perspectives, the basin's high altitude, strong winds, and arid conditions with minimal precipitation favor the dispersion of atmospheric pollutants. This makes the formation of prolonged stagnant weather conditions unlikely, thereby preventing the accumulation of pollutants and the occurrence of high concentrations near the ground. A particularly unique aspect of the QDB is its role as a significant source of salt dust. The recent study indicates that salt dust emitted from the playa lakes within the basin contributes substantially to atmospheric dust deposition (Zhu et al., 2025). These salt dust particles, composed mainly of soluble salts like NaCl and gypsum, may dilute the relative concentration of non-salt components, such as heavy metals, when released into the atmosphere in large quantities. The combined effect of these factors leads to the observed low heavy metal content in dust deposition in this region.

Throughout both the HP and NHP, trace elements concentrations in urban areas were consistently higher than in rural areas (not statistically significant), with the exception of Ti (Figs. 6 and S14). During the HP, urban levels of Zn, Pb, and Cu were significantly elevated compared to rural areas (P<0.05), and Pb also demonstrated a significant increase in urban during the NHP (P<0.01). In rural, the differences in metal concentrations between HP and NHP were minimal. In contrast, urban areas exhibited higher concentrations of all elements except for Ti, Cd, and Cr during the HP, with Fe and V showing notably elevated levels compared to other regions (not statistically significant). These variations in average concentrations indicate that coal combustion for domestic heating in urban areas contributes to increased atmospheric heavy metal levels (Duan and Tan, 2013; Meng et al., 2017). In contrast, Cd and Cr exhibited mixed anthropogenic sources with limited coal combustion contributions, while Ti concentrations remained stable across seasons, reflecting minimal anthropogenic influence.

Analysis of the carbonaceous components in QDB dust deposition reveals more intensive coal combustion in rural areas, yet the heavy metal concentrations in atmospheric deposition are lower than in urban area (not statistically significant). This observation can be explained by the following factors. First, pollution sources in rural areas are relatively singular, whereas urban areas are influenced by more complex heavy metal sources. During the heating period, heavy metals in the rural atmosphere of the QDB mainly originate from coal and biomass combustion. In contrast, urban areas are affected by a wider range of sources, including industrial activities, traffic emissions, and others (Liu et al., 2021; Huang et al., 2021a). Additionally, the dense building layout in urban areas hinders pollutant dispersion, leading to accumulation, while the open terrain in rural areas facilitates dilution and diffusion. This pattern, where rural heavy metal concentrations (particularly Pb, Cr, Cd, As, and other elements associated with coal combustion) are lower than those in urban areas during the heating season, has also been observed in studies conducted in Northeast China, Shanghai, Taiyuan, the Yangtze River Delta, and Southern Nigeria (Shi et al., 2012; Liu et al., 2021; Huang et al., 2021a; Liu et al., 2023b; Hilary et al., 2025).

We conducted a PMF source apportionment analysis on soluble ions, trace and carbonaceous elements present in dust, specifically focusing on ps-SO42-, ps-Ca²⁺, ps-K⁺, ps-Mg²⁺, ps-Cl⁻, ps-Na⁺, nps-SO42-, nps-Ca²⁺, nps-K⁺, nps-Mg²⁺, nps-Cl⁻, nps-Na⁺, Fe, Si, Al, Ti, Cr, Cd, Cu, Mn, Ni, Pb, V, Zn, SOC, POC and char-EC, soot-EC. Seven source factors were identified based on prior research and an understanding of potential local sources: salt lakes, soil, traffic emissions, secondary inorganics, biomass and coal burning, and industrial activities (Figs. 7 and S15). A plot of the time series is provided in Fig. S16. The generation of Fig. 7a involved extracting factors identified as the same source from the PMF factor profiles of each site (Urban and Rural) and each heating season (HP and NHP) shown in Fig. S15. The arithmetic mean of the contributions from characteristic species (elements and ions) corresponding to each factor was calculated. Species with average contributions exceeding 20 % were defined as characteristic species of that source in atmospheric dust over the QDB.

The factor profiles for each element in these source categories represent the arithmetic mean of profiles from six stations, with detailed operational methods provided in Sect. S2. The uncertainty of the source contributions was calculated directly from the standard error of the multiple regression coefficients between the deposition flux (independent variable) and the source contribution (dependent variable) at different monitoring sites (Belis et al., 2015; Manousakas et al., 2017). The regression method assumes that all factors explaining the mass have been identified; however, if a significant portion of the mass not directly related to the species in the PMF analysis is omitted, the source contributions may be overestimated, which could be an important additional source of uncertainty. The results are shown in Table S6. It must be noted that this method captures only a portion of the total uncertainty, as it does not include errors from profile uncertainty or rotational ambiguity. The low errors calculated by this method indicate a good model fit. In this study, the sample numbers used for PMF source apportionment were 690, 780, 480, and 750 for the Urban-NHP, Urban-HP, Rural-NHP, and Rural-HP groups, respectively. Limited sample number can increase the risk of model overfitting, reduce the representativeness of source profiles, and potentially lead to the merging of multiple sources into a mixed factor (Norris et al., 2014). Future research will extend the observation period and increase the sample number to further enhance the reliability of the source apportionment results.

The ions ss-Na⁺, ss-Cl⁻, ss-SO42-, ss-Ca²⁺, ss-K⁺, and ss-Mg²⁺ are widely acknowledged as indicators of sea salt (Ambade et al., 2022; Aswini et al., 2022; Gluscic et al., 2023). In this study, we identified ps-Cl⁻ (83.96 %), ps-Ca²⁺ (83.71 %), and ps-SO42- (83.71 %), ps-Na⁺ (83.70 %), ps-Mg²⁺ (83.66 %), ps-K⁺ (82.35 %) as key markers of salt lake sources. The contribution of salt lake sources in rural (12.29 %) was higher than in urban (6.87 %). During the HP, the proportion of salt lake sources in rural areas was 4.97 %, compared to 19.61 % in the NHP; urban showed contributions of 2.16 % during the HP and 11.58 % in the NHP, showing opposite seasonal trends. Backward trajectory simulations indicated that during the HP, air mass in both urban and rural areas mainly originated from the northwestern part of the basin and the eastern Tarim Basin, whereas during the NHP, they were broadly influenced by the salt lake regions within the basin (Fig. S6). The minor wind direction differences and the inter-distributed sampling points (Fig. 1), suggests no substantial geographical disparity between urban and rural sites. Additionally, the ion content derived from playa salts in dust deposits increased during the NHP in both areas. Therefore, we propose that the anomalous increase in salt lake contribution during the urban HP may be closely related to human activities. The enhanced Urban Heat Island (UHI) effect and temperature inversion structures during the HP can alter boundary layer height, turbulence, and deposition conditions, thereby increasing the residence time of externally transported particles within the urban boundary layer and elevating their measured contribution (Cichowicz and Bochenek, 2024). Urban heat sources and heating emissions may also modify local transport pathways, leading to more concentrated deposition of dust originating from playa regions over urban areas. Furthermore, dry road surfaces, increased traffic, and construction activities during the HP can promote the resuspension of previously deposited playa dust. The use and subsequent resuspension of road de-icing salts (e.g., NaCl, CaCl₂) may further amplify the contribution of tracer ions indicative of playa salts (Gertler et al., 2006; Casotti Rienda and Alves, 2021).

The second factor pertains to soil dust, characterized by trace element such as Fe (47.03 %) and Si (36.22 %), along with ions such as nps-Na⁺ (54.21 %), nps-Ca²⁺ (44.05 %), nps-Mg²⁺ (32.30 %) (Pervez et al., 2018; Tian et al., 2021). Additionally, the proportions of elements such as SOC (32.58 %) and POC (20.84 %) suggest that the dust is likely mixed with fossil fuel emissions. Furthermore, V (24.62 %) was detected at multiple sampling sites, indicating contributions from both soil dust and traffic emissions. Mg, Ca²⁺, Al, Si, and Fe are typical tracers for soil dust (Liu et al., 2003; Heo et al., 2009). Notably, the contribution of soil dust in rural areas (21.14 %) exceeded that in urban areas (12.97 %), indicating that soil dust is a major source of atmospheric deposition. In urban areas, the contribution during the NHP was relatively low (6.84 %), likely due to higher wind speeds and the effectiveness of frequent summer precipitation (Zhang et al., 2015a).

The third factor is traffic emissions, which are particularly pronounced in urban areas. Key characteristic elements and ions include nps-Cl⁻ (56.71 %), Zn (50.13 %), Cd (49.41 %), Pb (47.84 %), Ni(44.22 %), Cu(37.61 %), nps-Mg²⁺ (30.54 %), V (29.63 %), nps-K⁺ (27.74 %), NO3- (24.35 %) and char-EC (46.71 %), soot-EC (38.12 %), SOC (32.25 %), POC (23.72 %) (Adeniran et al., 2017). In rural areas, traffic emissions contributed 8.52 % to atmospheric deposition during the NHP, whereas in urban areas, the contribution was significantly higher at 46.83 %, with 32.91 % during the HP and 60.75 % in the NHP. These findings correlate with previous studies on OC/EC and char/soot ratios, suggesting that carbonaceous elements in the NHP primarily derive from traffic emissions. The traffic emission factor in the QDB represents a mixture of vehicle exhaust and non-exhaust sources (e.g., tire and brake wear, and resuspended road dust). Elements and ions including V, NO3-, Ni, and carbonaceous components primarily associated with vehicle exhaust (Cong et al., 2011; Zhang et al., 2012). For instance, Ni can be emitted from fuel combustion and vehicle exhaust (Pacyna and Pacyna, 2001). In contrast, elements such as Cu, Zn, nps-Mg²⁺, and nps-K⁺ originate from non-exhaust vehicle emissions, including brake and tire wear, as well as the resuspension of road dust (Amato et al., 2014). For example, Zn may derive from the wear of rubber tires on roads (Rogge et al., 1993), Pb emissions may be related to wear (tires/brakes) (Smichowski et al., 2007), and Cu is associated with brake wear (Lin et al., 2015). Furthermore, the presence of crustal elements and ions such as Fe, Si, and nps-Mg²⁺ in the traffic emission factor for Urban-NHP, Urban-HP, and Rural-NHP suggests an additional contribution from resuspended road dust (Chen et al., 2019b).

The fourth factor is coal combustion, characterized by high concentrations of nps-Cl⁻ (75.40 %), nps-SO42- (57.87 %), nps-K⁺ (31.75 %), NO3- (19.10 %), and soot-EC (72.54 %), char-EC (40.63 %), POC (34.10 %) (Kundu and Stone, 2014; Contini et al., 2016). Zhang et al. (2023) found that coal combustion emits particles rich in Cl⁻. Coal combustion was more intense at site LTC in rural areas and at site DLX in urban areas. Coal combustion occurs exclusively during the HP, contributing 30.38 % in rural areas and 23.31 % in urban areas. These results align with earlier studies on carbonaceous aerosols, indicating that the carbon content from coal combustion emissions is higher in rural than in urban aeras. Consistent with northern China, air pollution in QDB urban has declined due to the adoption of clean heating technologies (Zhang et al., 2021; Xue et al., 2023). However, rural coal combustion remains a major source of carbonaceous aerosols during the HP.

The fifth factor, biomass burning, is characterized by significant concentrations of non-precipitating species, including nps-K⁺ (56.99 %), NO3- (34.97 %), Ti (23.26 %) and SOC (43.68 %), char-EC (26.78 %), POC (25.60 %) (Simoneit, 2002; Sulong et al., 2019). K⁺ serves as an important tracer for biomass burning (Cachier and Ducret, 1991). Biomass burning contributes 33.55 % to rural atmospheric dust deposition, with a higher proportion during the HP (41.67 %) compared to NHP (25.43 %). In urban, biomass burning is primarily observed during the HP, contributing only 6.41 %. These findings underscore biomass burning as a major source of carbon emissions in rural settings, aligning with the prevalent use of biomass fuels for cooking and domestic heating in Northern China's rural areas (Meng et al., 2019), where 70 % to 80 % of energy demand are fulfilled by materials such as dung cakes, firewood, charcoal, and crop residues (Tao et al., 2018; Shi et al., 2019). Furthermore, increased biomass burning is also associated with the autumn harvest period (Chen et al., 2017; Li et al., 2021).

The sixth factor pertains to industrial emissions, which are characterized by high concentrations of Mn (61.98 %), Cr (59.22 %), Ti (44.90 %), V (37.17 %), Cu (34.95 %), Cd (28.63 %), Fe (27.47 %) Pb (25.75 %) and Zn (24.03 %) (Almeida et al., 2015; Yao et al., 2016). These elements were consistently detected across all sampling sites, alongside Al (71.17 %), nps-SO42- (28.84 %), Si (22.24 %) and POC (20.95 %), soot-EC (19.97 %), which indicate potential contributions from soil dust. Zn, Cu, Fe, and Mn are major chemical components in industrial emission profiles; Cd is a trace element found in metallurgical industries (Xu et al., 2022), while Zn and Mn emitte from oil combustion, metallurgy, and steel manufacturing processes (You et al., 2017). Pb and Cd are associated with metal smelting and processing (Fang et al., 2021). Industrial emissions showed greater contributions during the HP at XZH and at DLX. In rural areas, industrial emissions constitute 13.57 % of carbon output, with contributions of 13.95 % during the HP and 23.20 % during the NHP. In urban areas, industrial emissions account for 16.12 % overall, with 6.69 % during the HP and 25.55 % in the NHP. Due to the abundance of non-ferrous metal resources (e.g., lead-zinc ores), oil, natural gas, and saline minerals (e.g., potassium, lithium) in the QDB, the primary industrial activities are mining and associated chemical industries. Particularly around the GEM and DLX sites, the presence of numerous Pb-Zn, Fe, and Cu mining enterprises leads to significant contributions from industrial emissions to urban dust fallout, making it one of the major sources of air pollution in the basin.

The seventh factor is secondary inorganic aerosols, primarily composed of NO3- (72.46 %), nps- SO42- (45.79 %) (Liu et al., 2015, 2016a). High mass loadings of NO3- and SO42- are characteristic of typical secondary inorganic aerosols (Huang et al., 2021b). Research indicates that NO3- and SO42- primarily result from the conversion of gaseous precursors, such as SO₂ and NO_x, through photochemical reactions, predominantly sourced from local and regional emission (Liu et al., 2015; Tao et al., 2013). Secondary inorganic aerosols are predominantly observed in urban areas during the NHP, where they contribute 4.70 % to total aerosol sources. This increase is likely due to elevated temperatures and enhanced solar radiation during this period, which promote photochemical activity (Pandolfi et al., 2010).

Dust deposition sources exhibit significant seasonal and spatial variations. In the QDB, coal combustion (26.85 %) and biomass burning (24.04 %) dominate during HP, transitioning to traffic emissions (34.64 %), soil dust (20.04 %) and industry emission (19.37 %) as the primary contributor in NHP. Rural areas predominantly contribute to dust through biomass burning and coal combustion, as well as natural sources like windblown dust and salt lake emissions. This pattern aligns with increased coal usage for winter domestic heating and heightened biomass burning for cooking and heating in rural areas. In urban, dust deposition is briefly influenced by anthropogenic activities, including traffic and industrial emissions, with minimal contributions from domestic heating. Such differences can be attributed to varying economic development models, industrial and energy structures, and levels of human activity (Kataki et al., 2016).

Time series analysis (Fig. S16) indicates that atmospheric dust deposition in rural areas primarily originates from natural processes and dispersed anthropogenic combustion due to sparse human activities. During the NHP, the contribution of salt-lake emissions increases in May and June. Strong winds in spring and summer in southern QDB transport saline soil particles, making them a key source of dust deposition (Zhu et al., 2025). Soil-derived dust shows variations highly synchronized with the salt-lake source, with elevated contributions also occurring in May and June, reflecting the natural erosion of exposed surfaces under high wind conditions (Zhang, 2010). Traffic contributions rise noticeably in July and August, likely linked to increased summer tourism and transportation. During the HP, contributions from biomass and coal combustion increase, peaking from November to January. This peak is attributed to the widespread use of biomass fuels such as livestock dung for decentralized heating in rural areas, leading to increased particulate matter emissions (Chen et al., 2023). In contrast, contributions from salt lakes and wind-blown dust decrease significantly, likely due to lower wind speeds and reduced natural source emissions in winter (Jiménez et al., 2018; Li et al., 2019a; Yang et al., 2024).

In urban areas, where human activities are more concentrated, pollution sources are dominated by centralized anthropogenic emissions (industry, traffic, and centralized heating), with temporal trends driven by both industrial activity and the heating cycle. During the NHP, industrial contributions increase, especially in towns such as GEM and DLX, where surrounding mining operations constitute a major anthropogenic source (Zhu et al., 2016). Traffic-related contributions follow a trend similar to that in rural NHP, also peaking in July and August, reflecting the common influence of increased vehicle flow during the tourism season. Meanwhile, precursors such as SO₂ and NO_x from industrial and traffic emissions undergo photochemical and gas-to-particle conversion under summer sunlight, enhancing the formation of secondary inorganic aerosols (e.g., SO42-, NO3-) (Zhang et al., 2013; Ma et al., 2017). During the HP, trends in salt-lake emissions, soil dust, and coal/biomass combustion are generally similar to those in rural areas. However, unlike in rural regions, traffic contributions in urban areas continue to increase during the HP, highlighting the higher intensity and less seasonally-variable nature of human activities in towns (Peng et al., 2021).

This study observed that the contribution of biomass burning to atmospheric dust deposition in rural areas of the QDB during the HP was higher than that of soil dust. Given that the collected dust samples had particle sizes >10 µm, while biomass burning typically emits aerosols in the submicron range, we propose several potential explanations. Firstly, during the HP, factors such as increased soil moisture and snow cover significantly suppress soil dust emission, resulting in a lower intensity than in other seasons (An et al., 2018; Yang et al., 2019). Simultaneously, biomass and coal burning for heating increases substantially, leading to intense, short-term emissions of fine particles. Although it was fine initially, these high-concentration ultrafine particles can undergo coagulation or coalescence, aggregating with each other or onto pre-existing coarse particles, thereby increasing their size (Butler and Mulholland, 2004; Kulmala et al., 2004; Li et al., 2020). Furthermore, fine particles from biomass burning (e.g., carbonaceous materials) may mix internally with coarse particles like soil dust or salt dust from the QDB, forming internally mixed particles (Li et al., 2003; Hand et al., 2010). During source apportionment, such coarse particles are more likely to be attributed to the biomass burning source. Additionally, the QDB is a significant source of salt dust (Zhu et al., 2025). Salt dust particles (e.g., halite, gypsum) provide excellent condensation nuclei for soluble substances emitted from biomass burning, greatly promoting hygroscopic growth (Li et al., 2003; Kumar, 2010; Wang, 2013). The basin's topography also favors stable inversion layers, inhibiting pollutant dispersion and allowing particles more time to grow, mix, and age in the atmosphere. Prevailing wintertime winds may also transport pollutants from surrounding regions into the basin.

Moreover, the dust in this study was collected using a passive sampler. Over 90 % of the collected dust particles were smaller than 100 µm, with approximately 25 % less than 10 µm (Fig. S2), indicating the presence of fine particles (<10 µm), albeit in a relatively small proportion. The particle size distribution of atmospheric dust deposition is similar to that of TSP, with both primarily consisting of particles smaller than 100 µm. Using PMF source apportionment, this study identified a notably high contribution from biomass burning in rural areas, particularly during the heating period. Similarly, studies on atmospheric TSP in Iran (Ashrafi et al., 2018), the QXP (Lulang) (Zhao et al., 2013), Northeast China (Jia et al., 2024), and Qingdao (Liu et al., 2022b) have also reported significant contributions from biomass and coal combustion. This suggests that contributions from biomass and coal combustion can indeed be observed in particles larger than 10 µm. Finally, the PMF model may have uncertainties in resolving sources with similar chemical profiles. If the chemical compositions of local soil dust and biomass burning particles overlap after long-range transport and complex atmospheric reactions, the model might not fully separate them (Cesari et al., 2016).

Meteorological parameters critically influence the accumulation and dispersion of airborne pollutants. To examine their impact on the chemical composition of dust deposition in this study, monthly data for temperature, humidity, precipitation, wind speed and direction, sunshine duration, frequency of visibility ≤ 10 km were obtained from four monitoring stations (XZH, GEM, NMH, DLX) during the study period (Table 1 and Fig. S17). Data from the LTC and BLX sites were excluded from this analysis due to a lack of on-site meteorological instrumentation.

Based on ANOVA results, wind speed and temperature were significantly higher during the NHP (P<0.05), promoting dust emission and consequently increasing atmospheric dust flux (Jiménez et al., 2018; Li et al., 2019b; Yang et al., 2024). However, at DLX, wind speed showed no significant seasonal difference, while precipitation and relative humidity were significantly higher in the NHP (P<0.05). Since precipitation is a known suppressor of dust deposition (Li et al., 2019b), meteorological factors were not the primary driver of the elevated dust flux at DLX during the NHP (Fig. S18); other factors, such as increased tourism, likely played a more important role. Additionally, correlation analysis revealed that dust and salt dust flux were positively correlated with the frequency of visibility ≤10 km and wind speed (P<0.05) and negatively correlated with relative humidity (P<0.05), consistent with previous findings (Wei et al., 2023).

Pearson correlation analysis further elucidated the relationships between dust chemical components and meteorological parameters. Ions derived from salt lake (ps ions) and crustal elements (e.g., Al) showed significant correlations with wind direction and speed (P<0.05, Fig. S19), indicating the dominant control of wind on the emission and transport of both playa salt and soil dust. During the HP in both rural and urban areas, relative humidity was negatively correlated with ps ions and crustal elements (P<0.05), suggesting that higher humidity promotes the removal of airborne particles (Deshmukh et al., 2011). Playa salt ions are hygroscopic, so higher relative humidity facilitates their deliquescence (Rörig-Dalgaard, 2021). The mean annual relative humidity in the QDB (about 34 % RH) is notably lower than Wuhan (75.4 % RH) (Zang et al., 2021), Handan (63.35 % RH) (Meng et al., 2016), Pearl River Delta region (67.05 % RH) (Yue et al., 2015), Jorhat, India (80.29 % RH) (Rabha et al., 2021) and other regions, providing favorable conditions for the long-range transport of salt-lake-derived particles.

For nps ions, nps-Cl⁻ (a tracer of coal combustion) during the HP was positively correlated with wind direction (P<0.05, Fig. S19), indicating downwind transport. In contrast, nps-K⁺ (a tracer of biomass burning) was negatively correlated with wind speed and solar radiation (P<0.05), consistent with the observed peak in combustion contributions during winter when solar radiation is the lowest. Increased wind speed may suppress biomass-burning intensity (Zhang, 2024), while higher solar radiation can reduce smoldering-phase nps-K⁺ emissions and enhance photochemical removal of aerosols (Kuang et al., 2020; Huang and Gao, 2021).

For carbonaceous components (Fig. S20), in the rural NHP, soot-EC, OC, and the char/soot ratio were positively correlated with precipitation and negatively with the frequency of visibility ≤10 km (P<0.05). In the rural HP, the char/soot ratio was positively correlated with precipitation. In the urban NHP, char-EC, soot-EC, and their ratios showed positive correlations with humidity and precipitation (P<0.01), while POC was positively correlated with the frequency of visibility ≤10 km (P<0.01). Total carbon, OC, and SOC were negatively correlated with wind speed (P<0.05). In the urban HP, EC and char-EC were negatively correlated with the frequency of visibility ≤10 km; POC was positively correlated with wind speed and solar radiation but negatively with the frequency of visibility ≤10 km (P<0.05); and the char/soot ratio was positively correlated with wind speed and negatively with the frequency of visibility ≤10 km (P<0.01).

Overall, wind speed was negatively correlated with TC, OC, SOC, POC, and the char/soot ratio (P<0.05). This pattern aligns with findings from Xi'an (Cao et al., 2009), New Delhi (Tiwari et al., 2015), and Karachi (Bibi et al., 2017), although it was not observed in rural areas (Sharma et al., 2002). Increased wind speed enhances ventilation, promoting aerosol dispersion and reducing carbonaceous aerosol concentrations (Saha and Despiau, 2009). The low mean wind speed (about 2.30 m s⁻¹) may also indicate that carbon emissions are predominantly local rather than distant sources (Shen et al., 2021b). Positive correlations of precipitation and relative humidity with soot-EC, char-EC, OC, and the char/soot ratio likely reflect the hygroscopic nature of carbonaceous aerosols: under humid conditions, particles absorb water, increase in mass, and undergo accelerated dry deposition. For instance, PM_2.5 deposition rates can increase 2–3 fold as relative humidity rises from 30 % to 70 %, thereby shortening their atmospheric residence time (Wu et al., 2016). The negative correlations of EC, char-EC, and POC with the frequency of visibility ≤10 km align with findings in cities such as Suzhou (Zhang et al., 2010) and Beijing (Kong et al., 2021), attributable to the strong light-absorbing (EC) and light-scattering (OC) properties of these components, which jointly enhance atmospheric extinction and reduce visibility (Zhang et al., 2010; Kong et al., 2021). Weaker correlations in rural areas indicate that carbonaceous components are less influenced by local meteorological dilution and removal processes, pointing to more complex source influences.

In summary, seasonal variations in atmospheric dust deposition result from the combined effects of meteorological conditions and anthropogenic emissions. In urban areas, meteorological factors, especially wind speed and relative humidity, play a dominant role in the dispersion and removal of local anthropogenic pollutants. In rural areas, dust composition is more strongly influenced by large-scale wind-driven dust transport and regional combustion activities.

Wind direction analysis provides a valuable supplement to the PMF model (Watson et al., 2008). Studies have shown that emissions from sources such as traffic, biomass burning, coal combustion, and salt lakes exhibit significant directional patterns, which can be effectively visualized using polar plots (Saraga et al., 2021). In this study, bivariate polar plots were generated for the concentrations of POC, SOC, char-EC, soot-EC, and selected ions/elements (Figs. S21–S24), as well as for the source contributions resolved by the PMF model (Figs. S25 and S26), to illustrate their directional provenance. Notably, the directional patterns of the various aerosol components exhibited distinct differences and did not fully align with the local prevailing wind directions.

The polar plots of key tracer ions and elements (Figs. S21–S24) showed patterns consistent with the corresponding PMF-resolved source factors (Figs. S25 and S26), supporting the validity of the source apportionment. Specifically, the following pairs showed similar directional distributions: Total-ps and the salt-lake source; nps-K⁺, POC, and char-EC with biomass burning; nps-Ca²⁺ and Si with soil dust; nps-SO42-, nps-Cl⁻, and SOC with coal combustion; Fe and Cd with industry; Ni and Zn with traffic emissions; Fe and Al with industry; and nps-SO42- together with NO3- with secondary inorganic aerosols. These spatial consistencies reinforce that the selected chemical tracers are associated with their respective PMF-identified sources.

Wind speeds in rural areas are generally higher than those in urban environments, with predominant wind directions between west (W) and north (N), indicating the potential long-distance transport of pollutants. Specifically, emissions from saline lakes primarily originate from the west-northwest (WNW) direction. Back trajectory modeling suggests that HP levels are mainly attributed to salt lake emissions from the northwestern QDB, while in NHP, sources include long-distance transport from salt lakes in the Tarim Basin (Figs. S6, S7). For industrial emissions during the HP, sources primarily originate from medium- to short-distance transport in the W direction, with NHP periods showing contributions from both the W and northwest-northwest (NNW) directions. In contrast, soil dust in HP is predominantly sourced from long-distance transport from the W direction, whereas in NHP, it is largely of local origin. Biomass burning during HP mainly arises from local sources, while its contribution during NHP is minimal and comes from both local and long-distance transport, consistent with PMF results.

In urban areas, lower wind speeds and varying dominant wind directions between the HP and NHP lead to different sources of pollutants. During HP, emissions from saline lakes and coal combustion primarily originate from short-distance transport in the WNW direction, while biomass burning is sourced from the W. In contrast, industrial, traffic emissions, and soil dust are predominantly attributed to local sources. In NHP, pollutants demonstrate a dispersion trend from their sources to urban areas. Industrial emissions primarily come from the WNW to north-northeast (NNE) directions, soil dust from the southwest (SW), and vehicle emissions from the W direction. Notably, both salt lake and secondary inorganic sources exhibit similar directional patterns, indicating a common origin from the SW direction. The presence of processing enterprises near salt lakes suggests that secondary aerosols may form from emitted SO₂ and NO₂ (Hewitt, 2001). These findings align with previous analyses showing that urban industrial and traffic emissions mainly derive from local sources, while coal and biomass burning predominantly originate from surrounding rural areas, corroborating the role of nearby soil sources in salt lake and soil emissions.

The source apportionment analysis using the PMF model indicates that in the QDB, rural dust-fall predominantly originates from the combustion of solid fuels, including firewood, yak dung, and coal, accounting for approximately 72.05 % of the total contribution. This proportion significantly exceeds contributions reported for rural areas in Beijing (41 %) (Hua et al., 2018), Agra (54.3 %) (Agarwal et al., 2020), and Beihai, Guangxi Province (66.7 %) (Zhang et al., 2019).

The higher contribution in this study likely reflects the local energy profile, as the sampling site in Haixi Mongol and Xizang Autonomous Prefecture, Qinghai Province, primarily relies on coal, yak dung, and firewood, constituting 58 %, 23.5 %, and 13 % of rural energy consumption, respectively (Jiang et al., 2020; Shen et al., 2021a). In contrast, solid biomass fuels, including wood and yak dung, account for over 70 % of rural household energy consumption in Xizang, with yak dung alone representing 53 % (Liu et al., 2008; Xiao et al., 2015). Similar patterns emerge in South and Central Asia, where biomass fuels dominate residential heating (firewood: 39 %; dung: 29 %) (Amacher et al., 1999; Heltberg et al., 2000; Hoeck et al., 2007; Foysal et al., 2012; Behera et al., 2015; Kerimray et al., 2018). In northern China, rural domestic heating primarily relies on coal (46 %), firewood (23.8 %), and electricity (15.1 %) (Tao et al., 2018), further highlighting the unique energy composition of QDB.

Recent studies have shown that South Asia, Central Asia, and Xizang contribute significantly to high concentrations of atmospheric PM, particularly BC, which accelerates glacier melting in the QXP (Ming et al., 2010; Xia et al., 2011; Chen et al., 2015). The QDB is recognized as a significant dust source affecting the glacier surfaces on the QXP, although it is often overlooked (Dong et al., 2014; Wei et al., 2017; Zheng et al., 2021). Compared to other dust sources, the QDB exhibits higher emissions from coal combustion, giving it a unique influence on the QXP. The organic matter and pollutants, such as PAHs, released from household solid fuel combustion, particularly coal (98 %) and dung (94 %), are substantially higher than those from firewood (Leavey et al., 2017; Secrest et al., 2017; Ye et al., 2020). Consequently, the impact of PM from coal combustion in the QDB on the QXP is significant. Specifically, the presence of BC in PM increases glacier albedo, accelerating the melting of glaciers and snow in the region (Kang et al., 2020) and impacting global freshwater resources (Huss and Hock, 2018). Additionally, BC enhances cloud condensation nuclei, ice number concentration, and cloud cover (Zhou et al., 2025), thereby influencing global climate change. Furthermore, coal combustion releases harmful emissions, including CO₂, NO_x, CO, SO₂ and sulfur trioxide (Munawer, 2018), adversely affecting local human health and exacerbating climate warming on the QXP (Liu et al., 2006; Li et al., 2023), with broader implications for global climate. Therefore, the atmospheric pollutants emission of the QDB deserves considerable attention. However, this study focuses primarily on larger particles, indicating a need for further research on the environmental impacts of carbonaceous aerosols in atmospheric PM within the QDB.

In addition to the distinctive energy consumption structure in the rural QDB, which leads to significant contributions from coal and biomass burning during HP, atmospheric dust deposition in the QDB during the NHP primarily originates from traffic and industrial emissions. The contribution from traffic emissions during the NHP was twice that during the HP. Considering the larger particle size of the dust samples collected in this study, the traffic-related dust is likely derived mainly from vehicle non-exhaust emissions, such as road dust (Gondwal and Mandal, 2021). This indicates that NHP atmospheric conditions are significantly influenced by resource development and tourism. Sampling sites, such as Qarhan Salt Lake, along with GEM and LTC stations within approximately 100 km (Fig. 1), suggest that salt lake resource extraction has a lower impact on regional aerosols than traffic emissions, despite salt lakes being the primary resource. This is likely because salt lake development mainly involves solar evaporation and chemical processes like extraction and adsorption, which emit fewer pollutants compared to other mining methods (Zhen, 2010). Consequently, salt lake resource exploitation exerts a relatively minor effect on local atmospheric carbonaceous aerosol.

Similar salt lakes with comparable environments to QDB, such as Salar de Uyuni in Bolivia, the Atacama Salt Lake in Chile, and Ombre Muerto in Argentina, are rich in lithium resources (Li et al., 2014), making them focal points for resource development. Additionally, Salar de Uyuni, the Atacama Salt Lake, Junggar Basin, and the Great Salt Lake are renowned tourist destinations. This suggests that, in arid basin salt lakes with similar climates and intensive human activity, atmospheric carbonaceous aerosols are likely influenced by resource exploitation and tourism, especially tourism. The study's findings can inform policy decisions regarding unexploited salt lakes in South America, such as Ombre Muerto and Salar de Uyuni. However, while QDB also hosts mineral resources such as copper, iron, and tin, this research focused on larger particles (>100 µm), which are more indicative of local sources. Given that sampling was conducted around the salt lakes, potential impacts from other mineral resource developments may have been underestimated. Further research is necessary to fully assess the environmental effects of carbonaceous aerosols in QDB atmospheric particles.

In conclusion, the localized source profiles developed in this study, such as OC/EC ratios for heating-period emissions and key heavy-metal tracers, provide critical input parameters for regional climate and air quality models. This enables more accurate simulation of the emission, transport, and deposition of carbonaceous aerosols in the QDB and supports quantitative assessment of their contribution to glacier retreat on the QXP. Based on that rural areas in arid and semi-arid basins such as QDB rely heavily on solid fuels for heating, with associated environmental impacts, we recommend that local governments implement targeted clean-energy transition policies. These should prioritize promoting clean heating alternatives (e.g., solar and electric heating) in rural regions, supported by financial subsidies and technical assistance. Simultaneously, the adoption of clean stoves and upgraded biomass fuels should be encouraged to reduce emissions from traditional biomass and coal combustion (Dickinson et al., 2015; Li et al., 2011; Shen et al., 2017). Given the close climatic and environmental linkage between QDB and QXP, we further propose integrating air pollution control in QDB into the broader environmental protection framework of QXP. Establishing a joint regional air pollution prevention and control mechanism for the “QDB–QXP” system would help align emission reduction in the basin with glacier protection goals on the plateau. To strengthen the scientific foundation of such a mechanism, future studies should focus on clarifying the key transport pathways, transformation processes, and quantitative impacts of air pollutants from QDB to QXP glaciers.