3.1.1 Concentrations of monoterpenes

The MTs showed the highest concentrations of the terpenoids, with the mean MT sum being 400, 440 and 430 pptv in summers (July–August) 2011, 2015 and 2016, respectively (Table 1). All the MTs except p-cymene showed clear maxima in summer. p-Cymene is also known to have anthropogenic sources (Hakola et al., 2012). The variations in the MT sum measured with the GC-MS2 and GC-MS3 were similar (Fig. S1 in the Supplement), but their direct comparison was not possible, due to the different sampling times.

Long-term MT concentration measurements were previously conducted at this boreal forest site with PTR-MSs (Lappalainen et al., 2009; Kontkanen et al., 2016). Those PTR-MS measurements were conducted close to the forest canopy at heights of 14 m (2006–2009) or 16.8 m (2010–2013). The median July MT concentration measured between 2006 and 2013 was 382 pptv. In previous studies in March 2003, 2005 and 2006, the MT concentrations measured in short campaigns with PTR-MSs have been higher than in our measurements in April 2016 (Table S3). Yassaa et al. (2012) measured concentrations lower than ours in a campaign in July–August 2010. Large spatial differences in concentrations especially for terpenoids were expected depending on the sampling point at the site (Liebmann et al., 2018). In our study the sampling site was at the edge of the forest whereas in previous studies by Lappalainen et al. (2009) and Kontkanen et al. (2016) it was in the upper canopy level and in Yassaa et al. (2012) above at the canopy at a height of 24 m, where it is expected, due to transport and chemistry, that concentrations are lower compared to our measurements.

In our measurements, the MT concentrations showed high peaks in May 2016 (12 May at 03:04 and 14 May at 04:10–06:10), 1 June at 01:10 and 9 September at 23:35, which clearly deviated from the other data. Based on the wind directions, these peaks may have been due to high concentrations coming from the site of operations of a sawmill, a wood mill and a pellet factory in Koreakoski, 5 km southeast of Hyytiälä. The influence of this factory on the MT concentrations has also been observed previously (Eerdekens et al., 2009; Liao et al., 2011; Williams et al., 2011; Hakola et al., 2012). These samples were not used in further analysis.

α-Pinene showed the highest concentration of the measured MTs (50 % of the MT sum) followed by Δ³-carene, β-pinene and limonene (Table 1). The MT distribution was very similar for nighttime (photosynthetically active radiation, PAR, <50 µmol s⁻¹ m⁻²) and daytime (PAR >50 µmol s⁻¹ m⁻²); only 1,8-cineol and linalool showed slightly higher fractions during the day. Similar MT distributions have also observed at the site (Hakola et al., 2012; Yassaa et al., 2012) and resemble one of the emissions of local trees (Bäck et al., 2012; Hakola et al., 2006, 2017).

The diurnal variability in the MT concentrations at the site was driven by vertical mixing; low values were measured during the day when mixing was highest and the highest values during nights with the lowest mixing (Figs. 2 and 3). This has been observed also in earlier studies of MTs at this boreal forest site (Hakola et al., 2012; Kontkanen et al., 2016). Similar diurnal variation was found by Bouvier-Brown et al. (2009) in a ponderosa pine forest in the Sierra Nevada Mountains of California. However, this observation of MTs is in contrast to the diurnal variation in MT concentrations measured in the Amazon tropical rainforest by Yáñez-Serrano et al. (2017). Light-dependent emissions found in the Amazon rainforest (Jardine et al., 2015) could explain this. In boreal forests emissions are strongly temperature dependent and may also continue during nights with only lower rates if temperature is sufficiently high (e.g. Hakola et al., 2006).

The diurnal variation in concentrations was highest in June, concomitant with the highest variation of the MLH (Figs. 2 and 3). The MLHs during the day (at 12:00–16:00) in June and July were 1605 and 819 m, respectively, while during the night (at 00:00–04:00) in June and July they were 87 and <60 m, respectively. The monthly mean MLH, which roughly describes the mean dilution volume of the emissions, was also twice as high during our measurements in June than in July (Table 1). Since the lidar used for measurements of the MLHs was unable to detect MLHs <60 m, we also used temperature difference between heights 125 and 4.2 m to roughly describe the vertical mixing. The correlation of monthly mean diurnal variation in MT sum concentrations with temperature differences at the site was high ($R_{\mathrm{MT}}^{\mathrm{2}}=\mathrm{0.85}$ in July). Individually measured values also showed relatively favourable correlation with temperature difference ($R_{\mathrm{MT}}^{\mathrm{2}}=\mathrm{0.46}$ in July, Fig. S2b). 1,8-Cineol and linalool did not follow this general diurnal pattern of MTs indicating different sources. 1,8-Cineol is the only MT known to also show clearly light-dependent emissions from Scots pines growing at the site (Hakola et al., 2006).

3.1.2 Concentrations of sesquiterpenes

Few data are available on atmospheric SQT concentrations and emission data are also much sparser than for MTs. In our measurements SQTs showed seasonal variation similar to that of the MTs, but their concentrations were much lower (Table 1). SQTs are very reactive and therefore their contribution to the local chemistry can still be significant. The highest 30 min mean for the SQT sum (103 pptv) was detected on 25 July at 03:15, coinciding with high temperature and a shallow mixing layer. The concentrations of SQTs did not increase during the sawmill episode, in contrast to that observed for MTs. They were more reactive and, if emitted, were probably depleted during transport from the sawmill to the site.

β-caryophyllene showed the highest concentrations among the SQTs measured followed by longicyclene, β-farnesene and four unidentified SQTs detected only in July and August (Table 1). Nighttime (PAR <50 µmol s⁻¹ m⁻²) and daytime (PAR >50 µmol s⁻¹ m⁻²) distributions of SQT concentrations were very similar; only β-farnesene showed a slightly higher fraction during the day. β-caryophyllene is emitted by local pines and spruces (Hakola et al., 2006, 2017) as well as from the forest floor (Hellén et al., 2006; Mäki et al., 2017; Bourtsoukidis et al., 2018). Aaltonen et al. (2011) and Mäki et al. (2017) also detected longicyclene in forest floor emissions. Laboratory studies have shown stress-related emissions of β-farnesene (Petterson, 2007; Blande et al., 2009), and β-farnesene was also detected in local spruce emissions by Hakola et al. (2017).

The diurnal variation in most SQTs was similar to the variation in MTs, and the concentrations were largely driven by vertical mixing (Figs. 2 and 3). As for the MTs, correlation of the monthly mean diurnal variation in the SQT concentrations with temperature difference between the heights of 125 and 4.2 m was high ($R_{\mathrm{SQT}}^{\mathrm{2}}=\mathrm{0.90}$ in July) while individual measured values also showed relatively favourable correlation with the temperature difference ($R_{\mathrm{SQT}}^{\mathrm{2}}=\mathrm{0.48}$ in July, Fig. S2b). In addition to mixing, a higher chemical sink compared to MTs may affect local SQT concentrations during the day (Zhou et al. 2017). This was supported here by the higher relative diurnal variation in SQTs compared to MTs. The only exception was β-farnesene, which showed almost as high concentrations during the day as at night, indicating different sources than for different SQTs and MTs. A contrasting diurnal variation was found by Bouvier-Brown et al. (2009) in a ponderosa pine forest in California, suggesting that the sources of β-farnesene are mainly light dependent.

3.1.3 Isoprene and 2-methyl-3-buten-2-ol concentrations

The isoprene and 2-methyl-3-buten-2-ol (MBO) concentrations were low (Table 1). Low concentrations of isoprene have also been observed in previous studies (Table S3). In our study, the monthly means in 2016 were 0.3–18 pptv for isoprene and 0.1–30 pptv for MBO. Low levels were expected since the main local trees (Scots pine and Norway spruce) are MT emitters and show only minor emissions of isoprene and MBO (Tarvainen et al., 2005; Hakola et al., 2006, 2017). The highest daily means were measured in July and August together with MTs and SQTs. The emissions of isoprene are light dependent (Ghirardo et al., 2010) while MBO emissions from local trees are mainly temperature dependent (Hakola et al., 2006).

The diurnal variation in MBO coincided with the variation in MTs, with high values during the night and low values during the day. This was expected, due to the temperature-dependent emission of MBO. For isoprene, clear changes in diurnal variation were observed between early summer (April–June) and late summer (July–September) (Fig. 2). In May and June, when emissions are still low due to the early growing season, lower daytime values were detected, but in July and August the daytime concentrations were clearly higher due to high light-dependent emissions. Previously, the gradually increasing isoprene emissions have been associated with the foliage growth period and start when the effective temperature sum (ETS) reaches a threshold value. For example, in tea-leaved willow (Salix phylicifolia L.), lower emissions of isoprene were found when the ETS <400^∘ days (Hakola et al., 1998). During our measurements in 2016, the ETS reached a value 400 on 23 June.

3.1.4 Concentrations of reaction products of terpenes

Methacrolein (MACR) is a reaction product of isoprene, but the monthly average MACR concentrations did not follow the concentrations of its precursor isoprene (Table 1). MACR may also have anthropogenic sources (Biesenthal and Shepson, 1997) and in spring and autumn, when lifetimes are longer and biogenic emissions lower than in summer, anthropogenic influence was expected to be higher. In their studies Biesenthal and Shepson (1997) found that the MACR concentrations near Vancouver were not explained by the photochemical source, while in Toronto MACR correlated with carbon monoxide (CO), suggesting they originate in traffic emissions. In our study, the monthly mean concentrations of MACR (4.8 and 3.3 ppt, respectively) were ca. 30 % of the isoprene concentration in July and August when the isoprene concentrations and biogenic emissions are highest. This is similar to the yields of 25 % and 24 % measured in chamber experiments by Paulson et al. (1992) and Atkinson (1994), respectively.

Nopinone is a reaction product of β-pinene and its monthly mean concentration followed the variation in MTs (Table 1). During the summer months its concentration was 7–13 % of the concentration of its precursor β-pinene. In reaction chamber studies, the yields of nopinone in O₃ reactions of β-pinene have been 19–23 % (Grosjean et al., 1993; Hakola et al., 1994; Winterhalter et al., 2000) and in OH radical reactions 25–37 % (Calvert et al., 2011; Kaminski et al., 2017).

Figure 4Monthly mean diurnal variation in (a) β-pinene and nopinone concentrations; (b) production rates of nopinone from OH (OH +β-pinene) and O₃ (O₃+β-pinene) reactions, destruction rate of nopinone by OH reaction (OH + nopinone) and net formation rates (NFR) of nopinone; (c) limonene and 4-acetyl-1-methylcyclohexene (4-AMCH) concentrations; and (d) production rates of 4-AMCH from OH (OH + limonene) and O₃ (O₃ + limonene) reactions, destruction rates of 4-AMCH by OH (OH + nopinone) and O₃ (O₃ + 4-AMCH) reactions, as well as NFR and NFR with increased yields (NFR increased) of 4-AMCH.

Download

The mean diurnal variation in nopinone followed the variations in its precursor (β-pinene) in April–June, but in July–September high values were also observed during the day (Fig. 4). Nopinone is known to be produced both in OH radical and O₃ reactions of β-pinene (Hakola et al., 1994), but it is destroyed only in the OH radical reactions. Since the yields from the NO₃ radical reactions are not available in the literature, they cannot be considered. The NO₃ reactions would increase production especially at night. Deposition may also have an effect (Zhou et al., 2017), but it was not taken into account here. The PR, DR and NFR of nopinone were calculated using the Eqs. (1)–(3). The nopinone yields used for OH radical and O₃ reactions obtained from Hakola et al. (1994) were 0.27 and 0.23, respectively. Change in the nopinone diurnal variation is explained by the balance between its sources and sinks. The concentrations closely followed the NFR variation (Fig. 4). Nopinone is a rather stable molecule and has 5 times lower OH radical reactivity than β-pinene and, in contrast to β-pinene, it does not react with O₃. The results indicate that in May and June, when there was already high light intensity and high OH radical concentrations, but the emissions of β-pinene were still low due to lower temperatures and an early growing season, the nopinone produced may have reacted away during the day, while higher values were measured during the night, when there are no OH radicals, but nopinone is still produced from O₃ reactions of β-pinene. In July and August higher emissions and more rapid reactions of β-pinene with OH radicals resulted in higher daytime concentrations of nopinone. In September the emissions were already lower, but the OH radical concentrations and MLHs were also lower, while higher nopinone concentrations were still detected during the day.

The reaction product of limonene, 4-acetyl-1-methylcyclohexene (4-AMCH), showed very low concentrations and was detected only in June and July (Table 1). The NFR for 4-AMCH was calculated by the same methods as for nopinone using Eqs. (1)–(3). The reaction rates of 4-AMCH with the OH radical and O₃ were only 25 % and 20 % lower than for its precursor. The 4-AMCH yields used here for the OH radical and O₃ reactions obtained from the Hakola et al. (1994) were 0.20 and 0.04, respectively. In a study by Grosjean et al. (1993) the yield from the O₃ reaction was 0.02. The concentrations measured in the present study did not follow the diurnal variation in the NFR, especially in July, when the highest concentrations were measured (Fig. 4). However, PR showed a diurnal pattern similar to that of the concentrations. Studies on limonene reactions by Grosjean et al. (1993) and Hakola et al. (1994) did not take into account that 4-AMCH reacts almost as rapidly with the oxidants as limonene and that the real yields could have been higher. When we increased the yields in our calculations, better agreement was achieved. In Fig. 4d the yields for the OH radical and O₃ reactions were increased by factors of 2 and 3, respectively. In August, the sensitivity of the instrument was less than 50 % of the sensitivity in June and July, due to faulty tuning of the MS, and 4-AMCH was not detected even though calculations indicated higher concentrations than in June.

3.1.5 Concentrations of volatile organic acids

The VOAs showed higher concentrations than terpenoids (Tables 1 and 2). Their atmospheric lifetimes (Calvert et al., 2011; Hellén et al., 2017) are much longer and therefore they can accumulate in the atmosphere and be transported for longer distances. They were expected to have both biogenic and anthropogenic sources and they are also produced in the atmosphere by the reactions of other VOCs (Ciccioli and Mannozzi, 2007). In the present study, the highest concentrations of VOAs in 2016 were already measured in June (Table 2). This was at least partly due to the different measurement periods for the GC-MS2 and GC-MS3 in June and July. The days when the VOAs were measured with the GC-MS2 were in late June when the temperature (18 ^∘C) and PAR were higher than VOC measurements with GC-MS3 in June (temperature 12 ^∘C). The MT sum measured together with the VOAs also showed the highest monthly mean in June.

The daily means of C₃–C₇ VOAs were more highly correlated with the MT sum (R=0.6–0.85) than with anthropogenic compounds such as toluene (R=0.0–0.31), indicating the biogenic origin of these compounds, either directly or through secondary production in the atmosphere. Only acetic acid showed some correlation with aromatic hydrocarbons (R=0.2–0.48). Since the lifetime of acetic acid is longest (1–2 weeks; Calvert et al., 2011), it was expected to be more influenced by the long-range-transported anthropogenic emissions.

The daily means of hexanoic acid were very highly correlated not only with 1-hexanol (R=0.97), but also with other C₆ compounds often referred to as green leaf volatiles (GLVs), e.g. hexanal (R=0.82) and cis-3-hexenol (R=0.83). This indicates that hexanoic acid could also be a GLV or that it is produced from GLVs in the atmosphere. Correlation of the daily means of hexanoic acid with pentanoic and propanoic acids was also high (R=0.89 and 0.80, respectively) and with the MT sum relatively high (R=0.77).

For smaller acids (acetic, propanoic and butanoic) daytime maxima were observed, especially in July and August, but for pentanoic and hexanoic acids higher concentrations were observed at night (Fig. 2). The mean diurnal variation in VOAs was not as strong as for the MTs and SQTs. Both direct biogenic emissions and production in the atmosphere were expected to be higher during the day, but since the mixing layer was also higher the VOAs were more diluted. However, due to the longer lifetimes of these acids, the mixing effect during the day was not as strong as for fast-reacting terpenes since the background air may also have contained comparable amounts of these acids. These acids may be transported from distant sources or produced in the upper parts of the mixing layer from the reactions of other VOCs. The losses due to OH reactions during the day and dry/wet deposition during nights were also expected to affect the concentrations (Calvert et al., 2011). In canopy-scale flux measurements by proton-transfer-reaction time-of-flight (PTR-TOF) devices, downward fluxes of acetic acid have been detected, especially at night (Schallhart et al., 2018).

3.1.6 Concentrations of C₅–C₁₀ aldehydes

C₅–C₁₀ aldehydes can be directly emitted or they can be produced in the atmosphere through oxidation of other compounds. Generally, the emissions are much lower than those of smaller aldehydes. Low emissions have been measured, e.g. from grasslands, but emissions of trans-2-hexenal, 2-hexenyl acetate and 2-hexenol from damaged and stressed plants can be significant (Fall, 1999; Hakola et al., 2001). Possanzini et al. (2000) found that larger aldehydes (heptanal, octanal) were emitted from citrus plants when exposed to O₃. There is also some evidence that nonanal can be produced when O₃ attacks the fatty acids on leaf or needle surfaces (Bowman et al., 2003). Hakola et al. (2017) also measured C₄–C₁₀ aldehyde emissions from Norway spruce and found that their magnitudes were similar to that of MT emissions during late summer.

The concentrations of C₅–C₁₀ aldehydes were low; their monthly means remained <10 pptv (Table 1). In the measurements of Hellén et al. (2004) at the same site in March and April 2003, the concentrations were slightly higher (12–16 pptv) but similar. The diurnal variations in C₅–C₁₀ aldehydes followed the variation in isoprene, with low daytime values in June and high values in July and August (Fig. 2).

The daily means of hexanal were highly correlated with MTs and SQTs (R=0.90) in summer (June–August). The daily means of nonanal and decanal showed the highest correlation with β-farnesene in summer. Since β-farnesene emissions are related to stress, this could also have indicated stress-related sources for the emissions.

In July, 24 h samples for analysis of carbonyls with a liquid chromatograph (LC, Praplan et al., 2018) were collected concomitant with the GC-MS3 measurements, similar to Praplan et al. (2017). The concentrations of <C₅ carbonyls were also obtained from those samples. The July means for formaldehyde, acetaldehyde, acetone and butanal were 430, 270, 1820 and 50 pptv, respectively. The concentrations of these smaller carbonyls were much higher than for the C₅–C₁₀ aldehydes. This was expected due to their longer lifetimes and larger emissions (Hellén et al., 2004).

The formation of aldehydes in GC and PTR-MS instruments from organic peroxides has been discussed (Rivera-Rios et al., 2014). However, we measured MACR and hexanal with both LC and the GC-MS3 in July 2016 and the results were comparable (MACR 4.7 and 4.8 pptv and hexanal 8.4 and 12 pptv, respectively). For pentanal even higher concentrations were obtained by LC (July mean 45 ppt). In 2015 when the GC-MS1 was used, the aldehyde concentrations were clearly higher than in 2016 (Table 1) and production from organic peroxides in the GC-MS1 may explain the difference. This indicates that the hypothesis by Rivera-Rios et al. (2014) may hold for some GC instruments, but it is still unclear under which circumstances.

3.1.7 Concentrations of alcohols and acetates

C₄–C₈ alcohols and acetates (including GLVs) generally show very low concentrations; the monthly means were mostly below the detection limits (Table 2). The only exceptions were 1-butanol and isopropanol, both of which were used in instrumentation at the site and therefore showed higher concentrations from leaks and exhaust lines. As for the other BVOCs the highest concentrations of most alcohols and acetates were measured in summer. Most of the alcohols and acetates measured were GLVs, which are emitted due to herbivory or pathogen infection by almost every green plant (Scala et al., 2013) or due to physical damage of plants (Hakola et al., 2001).

3.2.1 Correlation of monoterpene concentrations with temperature

The monthly mean MT concentrations showed very strong exponential correlation with temperature (R²=0.92, Fig. 5a). The site is dominated by Scots pines, which have temperature- and light-dependent emissions of MTs (Tarvainen et al., 2005). The PAR was highly correlated with monthly mean MT concentrations ($R_{\text{Apr–Nov 2016}}^{\mathrm{2}}=\mathrm{0.73}$), but correlation was clearly lower than with temperature.

Figure 5Exponential correlation of temperature with (a) monthly mean MT concentrations in 2011, 2015 and 2016; (b) daily mean MT concentrations in 2016; (c) monthly and (d) daily mean SQT concentrations in April–November 2016 measured at SMEAR II. Error bars show the uncertainty of the mean concentrations calculated from the uncertainty of the measurements. Note that the y axes are in a logarithmic scale.

Download

The daily mean MTs also correlated well with temperature ($R_{\text{Apr–Nov 2016}}^{\mathrm{2}}=\mathrm{0.83}$ and $R_{\text{Jun–Aug 2016}}^{\mathrm{2}}=\mathrm{0.88}$, Table 3 and Fig. 5b). The high correlation with temperature observed indicates that temperature has a major effect on the seasonality of the concentrations and emissions and processes controlling them. In a previous study by Lappalainen et al. (2009) lower correlation (R²=0.50) with temperature was found for the PTR-MS data. However, they used only daytime medians. In our study 24 h averages starting at 08:00 (UTC+2) and ending the next day at 08:00 (UTC+2) were used.

Table 3Correlation of VOC concentrations with temperature at SMEAR II in 2016, intercept (α) of temperature dependence curve, temperature sensitivity (β), and temperature correlations (R²) of monthly (April–November) and daily (June–August) mean concentrations and mixing layer height (MLH)-scaled concentration of individual measurement points (C_MLH). The fitted curves were exponent functions y=αe^βx, where y is the concentration or MLH-scaled concentration, x is the temperature and β is the temperature sensitivity.

MBO: 2-methyl-3-buten-2-ol; MACR: methacrolein; MT: monoterpene; SQT: sesquiterpene.

Download Print Version | Download XLSX

Temperature dependence of MT emissions is often described by the Guenther algorithm (Guenther et al., 1993):

where E_S is the standardised emission potential (µg g⁻¹ dry weight (dw) h⁻¹), T the leaf temperature (^∘C), T_S the standard temperature of 30 ^∘C and β the temperature sensitivity (^∘C⁻¹) of the emissions. Often the value 0.09 ^∘C⁻¹ is used for β to describe MT emissions. In our monthly and daily mean concentration data, the temperature sensitivity was clearly higher (β=0.20 ^∘C⁻¹, Fig. 5 and Table 3). The temperature also affects the vertical mixing of air, and a lower mixing after warm sunny days probably increased the temperature sensitivity of the concentrations. Even though the value 0.09 ^∘C⁻¹ is often used for β to model emissions, it is known to vary (Hakola et al., 2006). Here the temperature sensitivity of the daily mean MT concentration for the summer months (β=0.27 ^∘C⁻¹, June–August) was also higher than for the entire growing season (β=0.20 ^∘C⁻¹, April–November).

To determine the temperature sensitivity of the individual MTs, data from the GC-MS3 were used. Exponential correlations of the monthly means with temperature showed that R²>0.91 (Table 3 and Fig. S3) for all monoterpenoids except 1,8-cineol (R²=0.77), p-cymene (R²=0.72), bornyl acetate (R²=0.71) and linalool (R²=0.25). Tarvainen et al. (2005) found that, in Scots pine emissions, 1,8-cineol was the only MT that was both light and temperature dependent while the others were only temperature dependent. p-Cymene has been detected for example in Norway spruce emissions (Hakola et al., 2017), but it also has anthropogenic sources (Hakola et al., 2012). Linalool is emitted by trees as a result of biotic stress (Petterson, 2007; Blande et al., 2009). Bornyl acetate, linalool and 1,8-cineol showed very low concentrations, which also resulted in higher uncertainty. For the MTs with high (R²>0.91) temperature correlation, the β values of the monthly means varied between 0.15 and 0.26 ^∘C⁻¹, with the values being lowest for camphene and highest for β-pinene.

3.2.2 Correlation of sesquiterpene concentrations with temperature

As for the MTs, the monthly and daily means of the SQTs also showed very strong exponential correlation with temperature (Table 3, Fig. S4). The temperature sensitivity of the SQTs was even higher than for MTs. The SQT emissions from Norway spruce (Hakola et al., 2017) and Scots pine (Tarvainen et al., 2005) were closely correlated with temperature, but the SQT emissions may also have been influenced by light (Duhl et al., 2008). The daily mean β-caryophyllene concentrations showed very high exponential correlation with temperature ($R_{\text{Jun–Aug}}^{\mathrm{2}}=\mathrm{0.96}$) supporting only temperature-dependent emissions. The monthly means of the SQT sum (consisting mainly of β-caryophyllene) also showed very high exponential correlation with temperature (R²=0.97), indicating that seasonality is also driven by the temperature. For the other SQTs, the correlations were lower than for β-caryophyllene. Low concentrations with higher measurement uncertainty and light- and stress-related emissions may have significantly affected the correlations. β-Farnesene is emitted due to biotic stress (Kännaste et al., 2009) and increases simultaneously with linalool in the emissions of Norway spruce and Scots pines (Hakola et al., 2006, 2017). However, the linalool and β-farnesene concentrations did not correlate in our data. Bouvier-Brown et al. (2009) suggested that at least in a ponderosa pine forest β-farnesene emissions may be both temperature and light dependent.

3.2.3 Correlation of isoprene and 2-methyl-3-buten-2-ol concentrations with temperature

Isoprene emissions are both light and temperature dependent (Guenther et al., 1993; Ghirardo et al., 2010). Here correlation of the isoprene daily mean concentrations with light and the temperature activity factor (Guenther et al., 1993) was slightly lower (R²=0.74) than for the temperature only (R²=0.84, Fig. S5). However, the difference in R² is small and, since the concentrations were low and close to the detection limits, no clear conclusions can be drawn from this. The MBO was somewhat better correlated with light and the temperature activity factor (R²=0.76) than with temperature only (R²=0.70). This is in contrast to the Scots pine emissions, in which the MBO was only temperature dependent (Tarvainen et al., 2005).

Even though the diurnal variation in most MT, SQT and MBO concentrations did not follow the ambient temperature, isoprene showed the highest concentrations during the day, while the 30 min mean concentrations were exponentially correlated with the ambient temperature (Fig. S6, R²=0.64). Due to the close link between isoprene production and light, isoprene is produced and emitted from trees only during the light hours and is therefore detected in the atmosphere only during the day while the MBO, MTs and SQTs are also emitted from storage pools inside the needles or leaves during the night, and due to lower vertical mixing the ambient air concentrations are higher at night (Ghirardo et al., 2010).

3.2.4 Correlation of terpenoid reaction product concentrations with temperature

The nopinone concentrations showed very clear exponential correlation with temperature ($R_{\mathrm{daily}}^{\mathrm{2}}=\mathrm{0.80}$) due to the temperature dependence of its precursor (β-pinene) and more rapid production on warm and sunny summer days. The temperature sensitivity of the nopinone daily means (β=0.25 ^∘C⁻¹) is similar to the sensitivity of its precursor β-pinene (β=0.27 ^∘C⁻¹).

MACR, which is a reaction product of isoprene, was as highly correlated (R²=0.86) with temperature in summer as its precursor isoprene (R²=0.84) but the temperature sensitivity was slightly lower (β_isoprene=0.24 ^∘C⁻¹ and β_MACR=0.17 ^∘C⁻¹). Similar to its precursor, the 30 min mean concentrations of MACR also showed low exponential correlation with temperature (R²=0.32, Fig. S6), but the monthly means of MACR did not correlate with temperature (R²<0.01, Table 3).

3.2.5 Correlation of oxygenated volatile organic compound concentrations with temperature

Since the concentrations of most C₅–C₁₀ aldehydes were very close to the detection limits, the results are more scattered but still clearly show strong correlation with the temperature. The highest correlation of the daily means in summer (June–August) was found for hexanal (R²=0.90) and the lowest for octanal (R²=0.36) and decanal (R²=0.43, Table 3 and Fig. S7). The temperature sensitivities of the aldehydes (β=0.08–0.13 ^∘C⁻¹) were clearly lower than for terpenoids (β=0.18–0.67 ^∘C⁻¹). Aldehydes have direct biogenic emissions (Seco et al., 2007; Hakola et al., 2017), but they are also produced in the atmosphere by the oxidation of other VOCs. The correlation of the daily mean concentrations of trans-2-hexenal with the light and the temperature activity factor (Guenther et al., 1993) was higher (R²=0.71) than only with temperature (R²=0.57), indicating a light-dependent source.

As for the isoprene and its reaction product MACR, the diurnal variations in pentanal and hexanal concentrations were also correlated with temperature and temperature sensitivities for the 30 min mean concentrations (β_pentana=0.07 ^∘C⁻¹ and β_hexanal=0.08 ^∘C⁻¹, Fig. S6b), similar to that of MACR (β_MCAR=0.06 ^∘C⁻¹). This indicates that photochemical reactions could be an important source for these compounds as well.

A weak correlation with temperature was also found for the VOAs, but it was lower than for most other VOCs studied (Table 3). Due to the long lifetime of VOAs, the background concentrations and anthropogenic sources were expected to have more of an effect on the concentrations, and therefore their effect of local temperature-dependent emissions and production in the atmosphere remains unclear. Correlation of the daily means was highest for pentanoic acid (R²=0.65, Table 3, Fig. S7). The temperature sensitivity of the butanoic acid daily means (β=0.06 ^∘C⁻¹) was lower than for the other VOAs. The butanol concentrations at the site were strongly affected by the emissions from the particle counters used and were expected to produce butanoic acid.

3.2.6 Seasonality of temperature correlations

The variation in the daily mean concentrations is best explained by the temperature in summer (Table 4). The temperature sensitivity of the MT, SQT and isoprene concentrations was highest during the summer months and lower in autumn and spring. In summer, the emissions from trees were expected to play a major role, but in spring and autumn the relative impact of other emissions (e.g. sawmill emissions) increased.

Table 4Characterisation of the temperature dependence of the isoprenoid concentrations with intercept (α), temperature sensitivity (β) and correlation (R²) of the daily mean concentrations of MT sum and SQT sum measured at SMEAR II in different months in 2016. N is the number of daily means. The fitted curves were exponent functions y=αe^βx, where y is the concentration, x is the temperature and β is the temperature sensitivity.

Download Print Version | Download XLSX

In May, values lower than expected by the overall temperature dependence were detected (Fig. 5b). This is most probably explained by the beginning of the growing season (mean ETS <200) with lower emission potentials (Hakola et al., 2001, 2012). In autumn (September–November), when values were more scattered (Fig. 5b), the emissions from fresh leaf litter were expected to contribute significantly to the concentrations (Hellén et al., 2006; Aaltonen et al., 2011; Mäki et al., 2017). During the colder months when biogenic emissions are low, emissions near the sawmill were expected to show higher relative influence. This was indicated by higher MT concentrations in November than expected by the general temperature correlation (Fig. 5b). However, these higher concentrations were still within the measurement uncertainty.

3.2.7 Temperature sensitivities vs. vapour pressures

The temperature sensitivities (β values) of the most abundant terpenoids were dependent on their vapour pressures (Fig. 6). Vapour pressures estimated with the AopWin^™ module of the EPI^™ software suite (https://www.epa.gov/tsca-screening-tools/epi-suitetm-estimation-program-interface, last access: 27 September 2018, EPA, USA) were used. The vapour pressures used in the calculations are given in Supplement Table S1. Higher β values were found for the terpenes with lower vapour pressure, higher boiling point and higher carbon number. This indicates that temperature sensitivity is driven by the volatility of the compounds. In addition to the temperature sensitivities of the monthly means shown in Fig. 6, the summertime daily means of the terpenes also showed the same dependence on vapour pressures. However, camphene, p-cymene, 1,8-cineol and linalool did not show this dependence either for the monthly or daily means. For these compounds, the temperature sensitivity was lower than expected, based on the volatility. These differences, as previously mentioned, may have been due to the concentrations of camphene and p-cymene affected by the emissions of the nearby sawmill, while 1,8-cineol also showed light-dependent emissions and linalool is emitted from plants, due to stress.

Figure 6Vapour pressure dependence of temperature sensitivities β (^∘C⁻¹) of monthly mean concentrations measured at SMEAR II in 2016. Values for temperature sensitivities and exponent functions of temperature dependence of concentrations can be found in Table 3. Note that the y axis is in a logarithmic scale.

Download

Figure 7Correlation of temperature with (a) MT, (b) SQT and (c) isoprene concentrations multiplied by the mixing layer height (C_MLH) measured at the SMEAR II station in 2016. Note that the y axes are in a logarithmic scale.

Download

Even though the VOAs were less correlated with temperature than terpenes (Table 3), the dependence of temperature sensitivity on vapour pressures was also found for all other VOAs, except butanoic acid, which was expected to be produced from the 1-butanol used in other instruments at the site. For C₅–C₁₀ aldehydes, only monthly means showed this dependence. The summertime daily means of aldehydes were more highly correlated with temperature, but β values still did not follow the vapour pressures.

These dependencies can be used to estimate the type of compound that could explain the missing reactivity found by total reactivity measurements or to assist in the identification of compounds in direct mass spectrometric methods.

3.2.8 Simple proxies for estimating local biogenic volatile organic compound concentrations

Kontkanen et al. (2016) developed MT proxies that are used for calculating concentrations of the MT sum at SMEAR II. The proxies are based on the temperature-controlled emissions from the forest ecosystem, the dilution caused by mixing within the boundary layer and various oxidation processes. Our data show that the monthly and daily means of both the sum and individual MTs and most other BVOC concentrations can be described relatively well, using only the temperature (Tables 3 and 4, Fig. 5), and a simplified proxy for the daily or monthly mean concentrations would be

where α and β are empirical coefficients found in Table 3 and obtained from the correlation of monthly and daily mean concentrations with temperature (Fig. 5) and T is the ambient temperature.

However, describing the diurnal variation in mixing of air must be taken into account. To roughly describe the dilution caused by the vertical mixing we multiplied the concentrations with the MLHs (C_MLH= [VOC] × MLH) and studied the correlation of these C_MLH values with temperature (Table 3 and Fig. 7). All individual measured data points available from the year 2016 were used, except the cases in which MLH was below the detection limit of the lidar (<60 m). Therefore, the highest values during the most stable nights are missing. The correlation of the C_MLH values with temperature was best for the MTs ($R_{\mathrm{MTsum}}^{\mathrm{2}}=\mathrm{0.67}$). The modelling study of Zhou et al. (2017) showed that the variation in MT concentrations was mainly driven by the emissions and mixing, while for faster-reacting SQTs oxidation also plays a role. For oxygenated compounds production in the atmosphere and deposition also affect local concentrations, and therefore correlation of the C_MLH values with temperature was lower than for the MTs (Table 3). In our case the proxy for the concentration of the MT or SQT sum or an individual compound (BVOC_i), when MLH >60 m, would be

where α and β are empirical coefficients found in Table 3 and obtained from the correlation of concentrations multiplied by the MLH (C_MLH) with temperature (Fig. 7), T is the ambient temperature and MLH is the mixing layer height.

3.3.1 Reactivity of the measured biogenic volatile organic compounds

To describe the effects of various compounds and compound groups on the oxidation capacity of the atmosphere, we calculated the OH, NO₃ and O₃ reactivities for the BVOCs studied using the measured concentrations and reaction rates with different oxidants (Eq. 4). The OH reactivity of the MTs was clearly higher than for any other VOC group at this boreal forest site, showing the importance of the MTs for local OH chemistry (Fig. 8a). The OH reactivity of the MTs was 10 times higher than for SQTs, even in July when the SQT concentrations were highest. The OH reactivity of the other compound groups were minor (approximately 4 % in July). Based on additional measurements in July, the contribution of carbonyls <C₅ (formaldehyde, acetaldehyde, acetone and butanal) was minor. However, even when reactivities of all the BVOCs, anthropogenic VOCs and other OH-reactive compounds measured at the site were added up, the OH reactivity was much lower (<50 %) than the total reactivity measured at the site by Sinha et al. (2010), Nölscher et al. (2012) and Praplan et al. (2018). In a previous modelling study at the site, the OH reactivity of the MTs was also highest, but the other VOCs showed almost as high a contribution (Mogensen et al., 2011). These other VOCs included 415 compounds mainly consisting of second- or higher-order organic reaction products, but not SQTs. Even with these reaction products, approximately 50–70 % of the measured total OH reactivity remained unexplained.

Figure 8(a) Hydroxyl (OH) reactivity, (b) ozone (O₃) reactivity, (c) nitrate (NO₃) reactivity and (d) production rates of oxidation products (OxPRs) of different VOC groups at SMEAR II during different months in 2016. MTs: monoterpenes, SQTs: sesquiterpenes, HCs: hydrocarbons.

Download

Since O₃ reacted only with unsaturated VOCs, only isoprene, most MTs, SQTs and unsaturated alcohols of the measured VOCs contributed to the O₃ reactivity. From May to September, the SQTs contributed greatly to the O₃ reactivity (Fig. 8b). Even though the MT concentrations were approximately 50 times higher than the SQT concentrations, the O₃ reactivity of SQTs was about 3 times higher than that of the MTs. Hakola et al. (2017) also showed the crucial importance of the SQTs for the O₃ reactivity in the spruce emissions. This indicates that the SQTs, especially β-caryophyllene (Fig. S8d), have much higher effects on local O₃ destruction than MTs. Several studies have shown that the O₃ deposition fluxes measured cannot be explained by stomatal and known non-stomatal sinks modelled, such as reactions with the VOCs measured in the gas phase (Wolfe et al., 2011; Rannik et al., 2012; Clifton et al., 2017). The higher-than-expected impact of the SQTs could explain at least part of the discrepancy.

The NO₃ radicals also mainly reacted with the unsaturated VOCs, and the MTs clearly contributed most to the NO₃ reactivity of BVOCs at the site (Fig. 8c). Of the SQTs only β-caryophyllene was considered, since the reaction rate coefficients were not available for the others. However, β-caryophyllene showed the highest concentrations of all the SQTs, but still did not significantly affect the NO₃ reactivity. Liebmann et al. (2018) measured the total NO₃ reactivity at the site in September 2016, and the BVOCs measured at the same time explained 70 % of the reactivity during the night but only 40 % during the day.

α-Pinene contributed most to the OH, O₃ and NO₃ reactivities, similar to the concentrations of the individual MTs (Fig. S8). However, limonene and terpinolene both have relatively fast rate coefficients with the OH radicals and O₃; therefore, despite being present at lower concentrations, they can greatly impact the formation of secondary products. In addition, limonene shows a higher SOA yield than MTs generally (Lee et al., 2006) and, therefore, it was expected to be more important for SOA production than the concentrations indicate. Of the individual SQTs β-caryophyllene played a major role in contributing OH reactivity, while for the O₃ reactivity it contributed the most (>60 % in June–August) of all the VOCs measured (Fig. S8).

3.3.2 Oxidation products and secondary organic aerosols

Oxidation of VOCs, under various environmental conditions, produces a variety of gas- and particle-phase products that are relevant for atmospheric chemistry and SOA production. To describe this we calculated OxPRs from the isoprene, MT, SQT and OVOC reactions as described in Sect. 2.5.

More oxidation products were produced from SQTs than from MTs (Fig. 8d). The contribution of OVOCs, aromatic hydrocarbons and isoprene was very low. SQTs were very important especially during summer nights (Figs. 8 and 9). The daytime contributions of MTs and SQTs were similar during all months except in July when the SQTs predominated even in midday. In addition, photooxidation of SQTs in smog-chamber experiments generally resulted in much greater aerosol yields than MTs (Hoffmann et al., 1996; Griffin et al., 1999; Lee et al., 2006) and, therefore, they were expected to strongly influence SOA production. However, these OxPRs described very local situations and even though the rapid reactions of SQTs showed very strong local effects MTs also reacted relatively rapidly, producing secondary products on a regional scale. The global emissions of SQTs may be about 20 % of the MT emissions (Guenther et al., 2012), but this is probably a low-end estimate, since evidence for additional unaccounted SQTs and their oxidation products clearly exists (Yee et al., 2018).

Figure 9Diurnal variations in production rates for secondary organic compounds (OxPRs) from the reactions of (a) monoterpenes (MTs) and (b) sesquiterpenes (SQTs) with different oxidants (hydroxyl radical, OH; ozone, O₃; and nitrate radical, NO₃) and (c) contribution of individual biogenic volatile organic compounds (BVOCs) to the total OxPR during different months. The NO₃ radical reactions are missing for October, since the data for calculating its proxy were not available.

Download

α-Pinene is often used as a proxy for all BVOCs, but as shown in Fig. 9, the contribution of α-pinene to the total oxidation reactions was relatively low (approximately 20 %). The most important individual reaction generating oxidation products at the site was the reaction of β-caryophyllene with O₃. For SQTs the contributions of the OH and NO₃ reactions were very low, especially during the summer months (<2 %). For MTs, the O₃ reactions were also the most important, while the OH radicals contributed about 30 % during summer days and the NO₃ reactions were important at night. Peräkylä et al. (2014) stated that for MTs, nighttime oxidation is dominated by the NO₃ radicals whereas daytime oxidation is dominated by O₃. However, as in our study, O₃ also predominated during summer nights. If we account for emissions and concentrations also being highest during summer, O₃ becomes the most important oxidant for the production of oxidation products of MTs at night as well. For SQTs, O₃ oxidation clearly dominates the first step of the reactions. However, the reaction products of MTs and SQTs, which have lost all their double bonds, continued to react with OH and NO₃ and their total contribution was expected to be higher. At nighttime reaction products of MTs may build up in the atmosphere and are oxidised after sunrise with OH radicals, promoting particle growth (Peräkylä et al., 2014). Our results suggest that this also applies for SQTs.