Characteristics and Sources of Atmospheric Formaldehyde in a Coastal City in Southeast China

Abstract

Atmospheric formaldehyde (HCHO) is a major component of oxygenated volatile organic compounds (OVOCs) and plays an important role in O₃ formation and atmospheric oxidation capacity. In this study, seasonal observations of gaseous pollutants (HCHO, O₃, peroxyacetyl nitrate (PAN), CO, NOx, and VOCs) and ambient conditions (J_HCHO, J_NO2, solar radiation, temperature, relative humidity, wind speed, and wind direction) were conducted in a coastal city in southeast China. The average HCHO concentrations were 2.54 ppbv, 3.38 ppbv, 2.53 ppbv, and 1.98 ppbv in spring, summer, autumn, and winter, respectively. Diurnal variations were high in the daytime and low in the nighttime, and the peak times varied in different seasons. The correlation between HCHO and O₃ was not significant in spring and winter, which is likely related to the effects of photochemical reactions and diffusion conditions. The contributions of background (23.0%), primary (47.6%), and secondary (29.4%) sources to HCHO were quantified using multiple linear regression (MLR) models, revealing that secondary formation was the most significant contributor in summer, whereas primary emissions were predominant in spring. These findings help to improve the understanding of the influence of atmospheric formaldehyde on photochemical pollution control in coastal cities.

1. Introduction

Atmospheric formaldehyde (HCHO), which is an important photochemical active volatile organic compound (VOC), plays a crucial role in ozone (O₃) formation [1]. The photolysis of HCHO produces HOx radicals (OH + HO₂), which significantly influence the atmospheric oxidation capacity, thereby regulating overall atmospheric oxidation processes [2,3]. Recent studies further reveal HCHO’s important role in wintertime particulate matter and O₃ pollution, as well as the global sulfur cycle [3,4]. Hence, understanding HCHO sources is essential for controlling ground-level O₃ and improving air quality.

The sources of HCHO are diverse and complex, classified as primary emissions and secondary formation. The primary sources include direct emissions from plants [5], wildfires [6], vehicle exhaust [7], biomass burning [8], oilfield [9], indoor materials [10], ship emissions [11], industrial processes, and solvent usage [12]. The secondary formation of HCHO is generated by the photochemical oxidation of VOCs from anthropogenic and biogenic emissions. Source analysis methods include the multiple linear regression (MLR) model [13,14,15], photochemical age-based parameterization (PCAP) [14,16], the source tracer ratio [17], the positive matrix factorization (PMF) model [14,18,19], and the minimum R squared (MRS) method [20]. Based on the characteristics of similar sources in the atmosphere, the MLR model assumes that primary and secondary sources exhibit linear correlations with selected tracers [13,14,15]. CO [21], SO₂ [22], benzene [21], acetylene [2], and toluene [21], as tracers strongly associated with industrial activities, are representative of primary HCHO emissions. In contrast, O₃ [2] and peroxyacetyl nitrate (PAN) [22], which are characteristic secondary photochemical products, are employed as the indicators of the secondary formation of HCHO. The PCAP method, developed by de Gouw et al., has been applied to evaluate HCHO source contributions in diverse regions, including the southeastern United States [23], Beijing [24], Shenzhen [24], Guangdong [16], and Taiyuan [14]. The source tracer ratio method quantifies primary and secondary sources of HCHO based on formation and removal rates at different time points [17]. The MRS method determines primary emission ratios between HCHO and combustion tracers [20]. Using both MRS and PCAP methods, Liu et al. identified primary sources as the dominant contributor to HCHO during summer in Shanghai [20]. Using the PMF model, Huang et al. attributed the major source of HCHO to secondary formation processes, which accounted for 66.2% of ozone pollution days [19]. Cui et al. compared three distinct methodologies for identifying HCHO sources, with results indicating strong correlations in PCAP and the MLR model [14]. Generally, primary emissions, particularly from vehicular sources, dominate HCHO sources in urban areas during winter, whereas secondary formation processes prevail as the principal source of HCHO in summer [13,25,26]. Notably, they have also been observed to maintain their predominance in summer even under stringent emission control policies [15]. Therefore, the distribution pattern and source apportionment of HCHO should be comprehensively verified in different regions and seasons.

Recent studies have investigated the distribution of HCHO in metropolitan regions, including Beijing, Shanghai, Hong Kong, Shenzhen, Guangzhou, Wuhan, and Kolkata [13,15,17,27,28,29]. These areas exhibit elevated HCHO concentration levels driven by intensive vehicular emissions and secondary formation processes. HCHO measurements have also been conducted at high-altitude mountain sites and background stations, where concentrations were found to be associated with biomass burning and regional transport [16,30]. Nevertheless, the seasonal pattern and source apportionment of HCHO in coastal regions are not well-understood. Recent research confirms that coastal marine environments have been identified as significant sources of atmospheric HCHO, with elevated concentrations observed in marine air masses [31]. Additionally, the advection of polluted air masses from upwind regions transports substantial HCHO, exacerbating photochemical pollution [16]. With the rising O₃ concentration in China, an upward trend of O₃ pollution also existed in the southeastern coastal cities [32]. In particular, as a result of climate change, the atmospheric O₃ in coastal cities in southeast China gradually increase in winter, and it is necessary to examine the seasonal distribution and identify the sources of O₃ precursors. In this study, based on systematic and comprehensive field observations and the application of statistical models, we aim (1) to reveal the seasonal distribution characteristics of HCHO in coastal areas, (2) to clarify the key drivers of seasonal patterns of HCHO, and (3) to quantify the primary, secondary, and background sources of HCHO. Therefore, the related results will provide significant guidance for future photochemical pollution control in coastal cities in southeast China.

2. Materials and Methods

2.1. Site Description

The observation region is located in Xiamen, Fujian Province (Figure 1). As a coastal city in southeast China, except for the influence of local emissions, air quality in Xiamen might be affected by the Yangtze River Delta and Pearl River Delta regions during the monsoon period in East Asia [32]. The monitoring site at the Atmospheric Environment Observation Superstation of the Institute of Urban Environment, Chinese Academy of Sciences (IUE, CAS) (24.61° N, 118.06° E), is situated about 70 m above ground, surrounded by high-traffic roads and commercial and residential areas. Jimei and Haixiang Boulevard are two roads to the northwest and northeast of the site. Overall, it is representative of pollution characteristics in the urban area of coastal cities [32]. Four seasonal field campaigns were carried out between 2021 and 2022, including autumn (18 October–19 November 2021), winter (10 January–11 February 2022), spring (10 March–11 April 2022), and summer (30 July–31 August 2022).

2.2. Instrumentation

Atmospheric HCHO was monitored by an online analysis system (FMS-100, Focused Photonics Inc., Hangzhou, China). The system is based on the Hantzsch fluorescence reaction, which generates fluorescent substances by reacting HCHO with derivatives. We also calculated the concentration of HCHO by detecting the intensity of the fluorescent signal emitted. The detection limit was ≤50 ppt, and the accuracy was ±10%. Air quality-related trace gases (i.e., O₃, CO, SO₂, NO, and NO₂) were monitored by commercial gas analyzers (i.e., TEI 49i, 48i, 42i, and 42i, respectively (Thermo Fisher Model, Shanghai, China). PAN was monitored using PANs-100 (PANs-100, Focused Photonics Inc., Hangzhou, China). Total VOC concentration was monitored by gas chromatography–mass spectrometry (GC-MS/FID) (TH-PKU 300B, Wuhan Tian Hong Inc., Wuhan, China) with a single-point calibration each day at 23:00 LT and a multi-point calibration each month using a standard mixture of PAMS and TO-15. The photolysis rate constants (i.e., JO¹D, JNO₂, JHCHO, JHONO, JNO₃, and JH₂O₂) were measured using a PFS-100 photolysis spectrometer (PFS-100, Focused Photonics Inc., Hangzhou, China). Ultraviolet radiation (UV) was measured by a UV radiometer (SUV5 Smart UV Radiometer, KIPP & ZONEN, Delft, The Netherlands). Meteorological elements were monitored by an ultrasonic weather meter 150WX (150WX, Airmar, Milford, NH, USA), including temperature (T), relative humidity (RH), atmospheric pressure (P), wind speed (WS), and wind direction (WD). The boundary layer heights were obtained from the reanalysis data (https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5, accessed on 3 September 2022). A time resolution of 1 h was applied to all instruments for comparison purposes. The details of quality assurance and control have been described in our previous studies [32].

2.3. Multiple Linear Regression Model

According to the modeling assumptions, CO was selected as the primary source tracer, and PAN and O₃ were chosen as the secondary source tracers [28], as shown in Equations (1)–(3). Although PAN and O₃ are both the products of photochemical reactions, they have different peak hours and presented different seasonal correlations, due to the thermal decomposition of PAN. In this study, PAN was a secondary tracer in spring, summer, and autumn, but O₃ was a secondary tracer in winter. The model analysis assumed that collinearity between tracers is possible.

[HCHO] = β₀[Background] + β₁[Primary] + β₂[Secondary]

(1)

$$\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{m}\mathrm{a}\mathrm{r}\mathrm{y}={\displaystyle \frac{\beta 1{\left[\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{m}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}}{{\beta }_0+{\beta }_1{\left[\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{m}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}+{\beta }_2{\left[\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}}}$$

(2)

$$\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{y}={\displaystyle \frac{{\beta }_2{\left[\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}}{{\beta }_0+{\beta }_1{\left[\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{m}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}+{\beta }_2{\left[\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}}}$$

(3)

$$\mathrm{B}\mathrm{a}\mathrm{c}\mathrm{k}\mathrm{g}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}={\displaystyle \frac{{\beta }_0}{{\beta }_0+{\beta }_1{\left[\mathrm{P}\mathrm{r}\mathrm{i}\mathrm{m}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}+{\beta }_2{\left[\mathrm{S}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{y}\right]}_{\mathrm{i}}}}$$

(4)

where β represents the coefficient of the multiple linear regression. The primary sources represent vehicle emissions and industrial combustion sources, the secondary sources represent the photooxidation of VOCs, and the background sources represent natural emissions sources. The background HCHO value of 0.57, obtained from [28], for southeastern coastal cities was selected as the intercept in this study. Ten distinct transformations (e.g., linear, ln, sqrt, x², x³, 1/x, 1/ln(x), 1/sqrt(x), 1/x², and 1/x³) were applied to the concentrations of HCHO, primary tracers, and secondary tracers. The linear (untransformed) form was ultimately selected for regression modeling. At the 95% confidence interval of the regression coefficients, the simulated and observed HCHO values of four seasons fit well (R² of 0.41–0.72), with a significant value of 0.00, indicating that the results of the model analysis are statistically significant (

Table 1. Comparisons of the calculated coefficients for various sources in different seasons.

Season	β0	β1	β2	R2	Sig.
Spring	0.57	0.00403	0.822	0.58	0.000
Summer	0.57	0.00412	2.26	0.72	0.000
Autumn	0.57	0.00300	0.0207	0.46	0.000
Winter	0.57	0.00137	0.791	0.41	0.000

).

3. Results and Discussion

3.1. General Characteristics

The highest O₃ precursor levels (TVOCs: 22.90 ± 10.27 ppbv; NOx: 29.63 ± 17.52 μg·m⁻³) and the lowest wind speed (WS), relative humidity (RH), and boundary layer height (BLH) occurred in spring (1.12 ± 0.59 m·s⁻¹, 61.52 ± 15.61%, and 418.05 ± 359.59 m, respectively). Low WS and reduced BLH were not conducive to the dispersion of HCHO, favoring the accumulation of air pollutants. T, RH, and ultraviolet radiation (UV) were high in summer (30.43 °C, 71.69%, and 24.84 W·m⁻², respectively), while atmospheric pressure (P), WS, and BLH were high in autumn (1013.9 hPa, 1.84 m·s⁻¹, and 581.77 m, respectively). High UV and T in summer provide a favorable condition for the formation of HCHO. Seasonal variations in HCHO concentrations are consistent with JNO₂, indicating the influence of solar radiation intensity on HCHO formation in the region. These are the reasons why HCHO concentrations varied across all seasons. O₃ pollution episodes, defined by the exceedance of the maximum hourly O₃ concentration over the Grade II limit (93 ppbv) of China’s National Ambient Air Quality Standard (NAAQS), predominantly occurred between 13:00 and 15:00 local time. Notably, HCHO concentrations increased significantly 2–3 h prior to these O₃ exceedance events, implying a potential role of HCHO in ozone photochemical production. In autumn, the BLH was higher than that in other seasons, with an average value of 581.77 ± 350.61 m. The elevated BLH creates conditions favorable for pollutant dispersion. Most air pollutants originating from primary sources occurred at high levels in winter, mainly due to weak atmospheric oxidation and their secondary transformation.

Comparisons of atmospheric HCHO levels measured across global monitoring sites are summarized in

Table 2. Comparison of measured atmospheric HCHO (ppbv) around the world.

Location	Mean ± SD	Range	Spring	Summer	Autumn	Winter	References
Xiamen, China	1.59 ± 1.49	0.25–8.34	2.54 ± 1.30	3.38 ± 1.89	2.53 ± 1.18	1.98 ± 1.06	This study
Hong Kong, China 1	2.23 ± 2.87		3.29 ± 0.99	8.80 ± 2.12	3.23 ± 0.77	2.95 ± 1.29	[27]
Shenzhen, China			3.4 ± 1.6	5.0 ± 4.4	5.1 ± 3.1	4.2 ± 2.2	[28]
Taiyuan, China 1	6.27 ± 3.89		5.29 ± 2.98	10.44 ± 4.62	5.42 ± 2.57	4.30 ± 1.97	[33]
Changsha, China 1		1.50–21.67	5.35 ± 1.74	11.47 ± 4.19	7.52 ± 2.62	4.79 ± 1.47	[34]
Orléans, France			2.16 ± 0.59	3.08 ± 2.21	2.28 ± 0.82	1.46 ± 0.4	[25]
Dongying, China				8.59 ± 3.37		3.17 ± 2.12	[9]
Kolkata, India 1				18.84 ± 9.90		14.24 ± 3.79	[17]
Rome, Italy				18 ± 6		10 ± 4	[35]
Zurich, Switzerland				2.35		1.83	[36]
Wuhan, China (2017)	4.90 ± 2.36	1.39–12.00					[13]
Nanjing, China		1.8–12.8	4.1 ± 1.6				[37]
Shanghai, China		NA 2–9.4		2.2 ± 1.8			[2]
Wuhan, China (2023)		0.56–5.92		2.36 ± 1.34			[19]
Qingdao, China				2.4 ± 0.9			[30]
Changzhou, China		4.02–13.89			9.65 ± 2.70		[18]
Shantou, China		2.56–7.31			4.12 ± 1.02		[38]

¹ Unit conversion from μg/m³ to ppb was performed under standard conditions. ² NA: not available.

. In this study, the average HCHO concentration during the observation period was 1.59 ± 1.49 ppb (ranging from 0.25 to 11.43 ppb), exhibiting a distinct seasonal pattern of summer (3.38 ± 1.89) > spring (2.54 ± 1.30) > autumn (2.53 ± 1.18) > winter (1.98 ± 1.06). The average HCHO concentration was lower than those reported in urban areas of Hong Kong [27], Taiyuan [33], and Wuhan [13]. HCHO concentrations in Xiamen are the highest in summer and lowest in winter, showing a seasonal pattern consistent with other regions, such as Hong Kong [27], Changsha [34], Wuhan [13], Taiyuan [33], and France [25]. A few studies have focused on comparisons of atmospheric HCHO concentrations between different seasons. HCHO concentrations in this study were close to those in Europe but lower than those in South Asia [9,15,17,27,34]. There was a significant increase in HCHO throughout India during all seasons, probably related with the spread of emissions sources, even reached rural regions [16]. Many ports and mining areas exhibited high positive HCHO trends, which also showed new source regions and transport pathways of pollution.

Figure 2 displays the diurnal variation in O₃ precursors and meteorological parameters in four seasons. The diurnal variation in HCHO showed a single-peak distribution, similar to secondary pollutants, such as O₃ and PAN. However, most of them presented different peak hours in different seasons, as reported by previous studies [2,28]. Diurnal variations in CO and NOx exhibited early peaks, suggesting the influence of vehicle emissions on HCHO concentrations in the region. Previous studies have shown that HCHO was predominately affected by primary emissions during the morning peak period [22]. In summer, the diurnal variation in HCHO exhibited the highest peak concentrations, reflecting the impacts of strong photochemical reactions. In contrast, diurnal variation in winter was the smallest. Compared to those during the daytime, the HCHO mixing ratios at nighttime were relatively constant. This was attributed to the replenishment of HCHO primary emissions during the evening rush hour and residual daytime production. Our previous study reported that high HCHO values occurred frequently in the southeast wind direction with a low wind speed, showing the influence of urban plumes with intensive vehicle emissions from downtown Xiamen [32]. In addition, a low wind speed at night was favorable for the accumulation of HCHO due to the decrease in the boundary layer height.

3.2. Correlations with Representative Parameters

Pearson correlation coefficients for HCHO and the related parameters are presented in Figure 3. In spring, HCHO was significantly positively correlated (p < 0.01) with PAN, TVOCs, SO₂, and CO, indicating the influence of secondary formation, vehicular emissions, and industrial sources. Positive correlations between HCHO and Ox, PAN, and TVOCs in summer were also significant (p < 0.01), suggesting that OH-induced secondary photochemical oxidation of VOCs contributed to the formation of HCHO. For autumn, significant positive correlations (p < 0.01) between HCHO and CO, SO₂, O₃, TVOCs, and PAN were found, showing diverse sources of HCHO with both secondary production and primary emissions. Our previous study found that the dominant pathway of daytime HCHO production was the RO + O₂ reaction in spring and autumn [32]. In particular, the CH₃O+ O₂ pathway contributed to total HCHO production rates of above 55%. The sources of CH₃O radicals originated from aromatics, alkenes, and isoprene. HCHO enhanced the production rates of HO₂ +NO and RO₂ + NO, suggesting that HCHO affected O₃ formation by controlling the efficiencies of radical propagation. In addition, HCHO showed significant positive correlations with vehicle emission-related pollutants (i.e., CO and NOx) across all the seasons, especially in winter, similar to other study [26].

For meteorological factors, HCHO showed a negative correlation with atmospheric pressure in spring, summer, and autumn, but not in winter. Meanwhile, HCHO showed positive correlations with T and UV (R > 0.2, p < 0.05), with high temperatures and strong UV being conducive to HCHO production from its precursors. As for the loss pathways of HCHO, the previous study found that the HCHO photolysis loss rates in autumn were significantly higher than those in spring [32]. These results indicated that the favorable meteorological conditions not only made HCHO decomposition more competitive but also limited the high HCHO concentration. In addition, in this study, wet deposition favored the removal of HCHO. Also, HCHO showed a weak negative correlation with wind speed, indicating that the stationary weather was not conducive to the dispersion of HCHO.

Overall, in this study, a strong positive correlation between air pollutants (e.g., CO, NO₂, and SO₂) and HCHO indicated the impacts of vehicle emission and industrial sources. The previous study also reported that high concentrations of HCHO were frequently found near industrial areas [22]. On the other hand, HCHO also showed positive correlations with secondary pollutants (O₃ and PAN) and its precursor VOCs, indicating that HCHO was derived from the secondary generation of the photooxidation of VOCs, which is similar to most studies conducted on coastal sites [38].

3.3. Source Apportionment of HCHO

To understand the sources of HCHO, a multiple linear regression model was used to calculate the contributions of various sources in different seasons (Figure 4 and

Table 3. Relative contribution of different sources to HCHO, based on the MLR model.

Location	Season	Background Source (%)	Primary Source (%)	Secondary Source (%)	R2	References
Xiamen, China	Spring/Winter	23.0	47.6	29.4	0.41–0.72	This study
Hong Kong, China (YL)	Spring/Winter	24	39.8	36.2	0.35	[26]
Hong Kong, China (TC)	Spring/Winter	48.3	21.3	30.4	0.21	[26]
Anhui, China	Spring/Winter	29.0	49.2	21.8	NA 1	[39]
Beijing, China	Summer	5	76	18	0.65	[15]
Wuhan, China (ZY)	Summer	15.2	17.5	67.2	0.78	[13]
Wuhan, China (JX)	Summer	20.2	32.4	47.4	0.60	[13]
Wuhan, China (ZY)	Winter	9.5	73.5	17.1	0.74	[13]
Wuhan, China (JX)	Winter	15.2	58.6	26.2	0.71	[13]
Pearl River Delta (PRD)	Winter	6.2	54.4	39.3	0.77–0.79	[40]
Baoding, China	Winter	41.1	50.6	8.4	0.33	[41]
Mount Tai, China	Summer	34	22	44	0.83	[42]

¹ NA: not available.

). During the daytime, the primary source contribution to HCHO followed the order of spring > summer > autumn > winter, while the seasonal pattern of secondary source contribution was summer > autumn > winter > spring. Background sources were the greatest contributors to HCHO in winter (31.1%) and the smallest contributors in summer (14.5%). On average, the atmospheric HCHO sources were primary sources (47.6%), followed by secondary (29.4%) and background (23.0%) sources (Table 3). The contribution of primary sources in Xiamen was higher than that of Hong Kong (36.2%) and comparable to that of Anhui city (49.2%), which was attributed to the large number of vehicle-related primary emission sources in the urban area [26,39]. In other cities, the primary sources of atmospheric HCHO also included vehicle exhaust and combustion emissions.

As shown in Figure 4a, the primary sources (50.9%) in spring were much larger than the secondary (30.3%) and background (18.8%) sources, and the primary sources were higher than 50%, even at midday, probably due to the influence of vehicle emissions from the freeway near the monitoring site. In general, the main contributors to HCHO in summer are secondary sources, which are associated with favorable meteorological factors, such as high temperatures, low humidity, and strong UV radiation [13,42,43]. In this study (

Table 4. Relative contribution of various sources to atmospheric HCHO in different times.

Season	Period	Spring	Summer	Autumn	Winter	Mean
	Daytime	18.8	14.5	20.9	31.1	21.3
Background (%)	Nighttime	22.8	21.1	22.4	33.1	24.9
	Difference	−4.0	−6.6	−1.5	−2.0	−3.6
	Daytime	50.9	45.5	41.9	34.4	43.2
Primary (%)	Nighttime	59.8	67.2	47.6	35.8	52.6
	Difference	−8.9	−21.7	−5.6	−1.4	−9.4
	Daytime	30.3	40.0	37.2	34.5	35.5
Secondary (%)	Nighttime	17.4	11.7	30.0	31.1	22.6
	Difference	13.0	28.3	7.2	3.4	12.9

), the main sources of HCHO in summer were also primary sources (45.5%), while the difference between primary and secondary sources was only 5.5%. During the daytime, photochemical reaction was strong, and it showed a clear diurnal variation, especially from 11:00 to 14:00, with the maximum concentration occurring at 13:00, accounting for a total of 49.8%. Meanwhile, the average contribution (40%) of secondary sources in summer was comparable to that of Taishan (44%) [42], higher than that of mega-cities, such as Beijing (18–23%) [15] and Hong Kong (30.5%) [26], and lower than that of urban (67.2%) and suburban (47.4%) areas [13], probably due to the levels of their precursors and actual photochemical conditions. Moreover, the concentration of HCHO should consider its formation and destruction rates (i.e., net production), although they were not considered in this model. For autumn, secondary formation was the dominant source from 13:00 to 16:00, and the maximum occurred at 14:00, with an average of 42.4%. Previous studies have shown that primary sources were the main sources of ambient HCHO in winter due to the reduced rate of HCHO formation under adverse photochemical reaction conditions [26,40]. However, as shown in Figure 4, the contribution of all three sources was more stable. In winter, secondary sources (34.5%) in Xiamen were higher than those in Baoding (8.4%) [41], Hong Kong (27.8%) [26], and Wuhan (17.1–26.2%) [13] and close to those in the YRD (39.3%) [40]. This suggests that secondary production in coastal cities in southern China still occurred in winter, which may be related to the transport of aging air masses and atmospheric oxidation capacity caused by other oxidants, such as halogen radical reactions. Our recent study found that the enhancement in the chlorine (Cl) radical chemistry facilitated the elevations of both O₃ and atmospheric oxidation capacity during the winter daytime [44]. We will further study the influence of Cl radicals on the formation of HCHO during the wintertime in the future.

To further investigate the influence of regional transport on the variation in HCHO concentrations in the monitoring site, backward trajectories were conducted in all seasons (Figure 5). Backward trajectory clusters in spring were similar to those in summer, which is mainly correlated with relatively clean oceanic air masses originating from the western Pacific. According to the distribution of potential source areas around the study area, major emissions sources causing high HCHO values mainly originate from the southwest of Fujian Province. Meanwhile, in autumn, air masses with the highest HCHO concentrations predominantly came from the eastern coastal regions of China, while air masses in winter presented the lowest HCHO concentrations because of lower temperatures. It brought polluted air masses via industrial areas, such as Ningbo and Wenzhou in Zhejiang Province and Ningde, Putian, and Quanzhou areas in Fujian Province, to the monitoring site. Air masses carrying anthropogenic VOC emissions and biomass burning plumes might have significantly enhanced HCHO concentrations [16]. Therefore, in this study, the transport of aging air masses from the northeast direction in autumn and winter might have led to high contributions of secondary sources to HCHO. Moreover, the background sources were relatively high in winter due to weak diffusion conditions. In addition, the presence of sources that cannot be recognized by primary or secondary tracers might have resulted in a high contribution of background sources [42].

4. Conclusions

In this work, we conducted a comprehensive field campaign in the eastern coastal region across four seasons and analyzed the seasonal characteristics and sources of atmospheric HCHO by statistical methods and multiple linear regression models. The distribution of atmospheric HCHO concentrations followed the order of summer > spring > autumn > winter. Atmospheric HCHO showed a single-peak diurnal variation similar to that of O₃ and PAN, with the largest diurnal variation occurring in summer. On average, the main contributors to atmospheric HCHO in Xiamen were primary sources (47.6%), followed by secondary formation (29.4%) and background (23.0%) sources. Secondary formation accounts for the largest contribution to HCHO in summer and should not be underestimated, even in winter. Further research must focus on the influence of Cl radicals on the formation of HCHO in coastal areas. This study highlighted that targeting vehicular emissions or secondary VOC precursors in Xiamen city might mitigate both HCHO and O₃.