Carbonyl sulfide, a way to quantify photosynthesis
Kooijmans, Linda Maria Johanna
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Kooijmans, L. M. J. (2018). Carbonyl sulfide, a way to quantify photosynthesis. University of Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Carbonyl Sulfide,
a Way to Quantify
Photosynthesis
PhD thesis, 2018
University of Groningen The Netherlands
ISBN: 978-94-034-1079-1
ISBN: 978-94-034-1078-4 (electronic version) Printed by: Ridderprint BV, the Netherlands
The work described in this thesis was performed at the Centre for Isotope Research (CIO), which is part of the Energy and Sustainability Research Institute Groningen (ESRIG) of the University of Groningen, the Netherlands.
Carbonyl Sulfide,
a Way to Quantify
Photosynthesis
PhD thesis
to obtain the degree of PhD at the University of Groningen
on the authority of the Rector Magnificus Prof. E. Sterken
and in accordance with
the decision by the College of Deans. This thesis will be defended in public on Friday 30 November 2018 at 16.15 hours
by
Linda Maria Johanna Kooijmans
born on 16 March 1990
Prof. H.A.J. Meijer
Assessment Committee
Prof. W. Peters Prof. D. Yakir Prof. T. Roeckmann
C
ONTENTS
1 Introduction 11
1.1 Climate change . . . 12
1.2 Vegetative CO2uptake through photosynthesis. . . 13
1.3 A tracer for photosynthesis: carbonyl sulfide . . . 16
1.4 The COS budget. . . 17
1.4.1 Global scale . . . 17
1.4.2 Ecosystem scale . . . 18
1.4.3 Leaf scale . . . 18
1.5 COS measurement techniques . . . 19
1.6 Objective and approach of this thesis. . . 20
2 Continuous and high-precision atmospheric concentration measurements of COS, CO2, CO and H2O using a quantum cascade laser spectrometer 23 2.1 Introduction . . . 24 2.2 Experimental setup. . . 26 2.2.1 Instrumentation . . . 26 2.2.2 Calibration strategy . . . 27 2.2.2.1 Instrument response . . . 28 2.2.2.2 Working standards . . . 29
2.2.2.3 Background and reference strategy . . . 30
2.2.3 Water vapor interference correction . . . 31
2.2.4 Flow schematics for measurements at the Lutjewad station . . . 34
2.2.5 Temperature stability . . . 36
2.2.6 Measurements of COS from flasks . . . 36
2.3 Results and discussion . . . 38
2.3.1 Precision and accuracy. . . 38
2.3.2 Measurement comparison. . . 44
2.3.3 Continuous COS, CO2, CO and H2O observations from Lutjewad . . 48
2.4 Conclusions. . . 50
3 Canopy uptake dominates nighttime carbonyl sulfide fluxes in a boreal forest 53 3.1 Introduction . . . 54
3.2 Field measurements and data. . . 55
3.2.1 Measurement site . . . 55
3.2.2 Instrumentation for measurements of COS, CO2, and H2O . . . 55
3.2.2.1 QCLS for vertical profile and soil flux measurements . . . . 55
3.2.2.2 QCLS for eddy covariance measurements. . . 56
3.2.3 Soil chambers . . . 56 3.2.4 Auxiliary data . . . 56 3.2.4.1 222Rn . . . 56 3.2.4.2 Stomatal conductance . . . 57 3.2.4.3 Meteorological data. . . 57 3.3 Flux derivations. . . 57
3.3.1 The EC-based method. . . 57
3.3.1.1 Eddy-covariance fluxes . . . 57
3.3.1.2 Storage fluxes. . . 59
3.3.2 The radon-tracer method . . . 59
3.3.3 Soil fluxes . . . 61
3.4 Results . . . 62
3.4.1 COS and CO2storage fluxes . . . 62
3.4.2 COS and CO2 nighttime fluxes through the radon-tracer and EC-based method . . . 63
3.4.3 FCOScorrelation with gsCOS, VPD, Tairand u§ . . . 64
3.5 Discussion . . . 65
3.5.1 Vertical distribution of sinks and sources of COS and CO2compared to that of222Rn. . . 65
3.5.2 The effect of canopy layer mixing on flux derivations . . . 66
3.5.3 Sensitivity of FCOS-ECto u§. . . 67
3.5.4 Stomatal control of nighttime FCOS . . . 67
3.5.5 Effect of nighttime COS fluxes on GPP derivation . . . 68
3.6 Conclusions. . . 68
A3 Appendices . . . 70
4 Influences of light and humidity on carbonyl sulfide-based estimates of pho-tosynthesis 75 4.1 Introduction . . . 76
4.2 Methods . . . 77
4.2.1 Site description . . . 77
4.2.2 Branch chamber measurements. . . 77
4.2.3 Stomatal conductance. . . 79
4.2.4 GPP estimates . . . 79
4.2.5 Meteorological data . . . 80
4.2.6 Statistical tests. . . 80
4.3 Results and Discussion . . . 81
4.3.1 Responses of FCOSand FCO2 to light and stomatal conductance . . . 81
4.3.2 Internal conductance of COS limits FCOSduring daytime . . . 83
4.3.3 Seasonal variation of LRU influenced by environmental variables . . 83
4.3.4 Light and humidity-dependent LRU required for accurate COS-based GPP estimates . . . 84
4.3.5 The implications in large-scale GPP estimates. . . 86
4.4 Conclusion . . . 87
A4 Appendices . . . 88
A4.1 Seasonal change of FCOSand FCO2 . . . 89
A4.2 Effect of PAR on daytime LRU . . . 89
A4.4 Correlations of LRU with VPD, gs,COS, RH and T . . . 90
A4.5 Internal conductance . . . 91
A4.6 Temperature response of FCOSvaries over the season . . . 92
A4.7 Fit of LRU against PAR . . . 93
A4.8 Time-integrated LRU. . . 93
A4.9 Chamber types. . . 94
A4.10 Blank measurements. . . 94
A4.11 Respiration. . . 96
5 Sources and sinks of carbonyl sulfide inferred from atmospheric observations at the Lutjewad tower 99 5.1 Introduction . . . 100
5.2 Methodology . . . 101
5.2.1 Measurement sites. . . 101
5.2.2 Measurements of COS CO2and CO . . . 101
5.2.3 Measurements of SF6 . . . 102
5.2.4 Seasonal fit. . . 103
5.2.5 Nighttime ecosystem flux in Lutjewad . . . 103
5.3 Results . . . 105
5.3.1 Diurnal cycle of COS and CO2mole fractions . . . 105
5.3.2 Sources and sinks by wind directions . . . 106
5.3.3 Estimate of nighttime COS and CO2fluxes. . . 106
5.3.4 Sources of COS from agricultural field during ploughing. . . 107
5.3.4.1 Spikes in October 2014 . . . 108
5.3.4.2 Spikes in January 2015 . . . 108
5.3.4.3 Spikes in February 2018. . . 110
5.4 Discussion . . . 110
5.4.1 Understanding the diurnal change of atmospheric COS mole frac-tions . . . 110
5.4.2 COS spikes in the atmosphere . . . 111
5.4.2.1 COS production in agricultural soil. . . 111
5.4.2.2 Anthropogenic source of COS. . . 112
5.4.3 Spatial distribution of COS and CO2sources and sinks. . . 113
5.5 Conclusion . . . 114
A5 Appendix . . . 115
6 Discussion and Outlook 117 6.1 Discussion . . . 117
6.1.1 Measurements of COS . . . 117
6.1.2 Nighttime fluxes of COS . . . 118
6.1.3 Controls on leaf COS fluxes . . . 119
6.1.4 The relation between COS and CO2fluxes . . . 120
6.1.5 Non-vegetative ecosystem sources and sinks of COS. . . 120
6.1.6 COS-based GPP estimates . . . 122
6.2 Perspectives and recommendations . . . 122
6.2.1 The role of COS in obtaining GPP estimates . . . 122
Summary 127
Samenvatting 133
References 139
Acknowledgements 153
About the Author 157
1
I
NTRODUCTION
1
1.1.
C
LIMATE CHANGE
Since the start of the Industrial Revolution (ª1750) the global average temperature has increased by 0.85±C (IPCC,2013). Sea ice coverage in the Arctic has shrunk by 40 % since
1979 (Windnagel et al.,2017) and the global average sea level rose by 19 cm between 1901 and 2010 (IPCC,2013). Temperature changes have been observed to be larger over land than over the ocean because land surfaces heat up faster than the ocean. In the Netherlands, for example, the temperature rose by 1.8±C between 1901 and 2013 (KNMI,
2015). Also, precipitation amounts are increasing in the Netherlands, e.g. by 26 % in the 20thcentury (KNMI,2015).
The climate has always changed in the past by natural forces, i.e. due to changes in the Earth’s orbit, changes in solar radiation or volcanic eruptions. However, the changes are currently accompanied by greenhouse gas concentrations that exceed the natural variabil-ity observed in historic measurement records, see for example the CO2mole fractions in
Fig 1.1. These greenhouse gases have the capacity to absorb radiative energy (heat) that otherwise would have been lost to space. Thereby the energy remains in the Earth’s system and warms our atmosphere. The CO2 mole fractions and temperature have therefore co-varied during the last 800.000 years (Fig. 1.1).
Temper ature de viation [ °C] 800 600 400 200 0 − 6 − 4 − 2 0 2
Thousands of years before present
T, Snyder (2016), ice cores.
T, Jones and Mann (2004), ice cores. T, NASA/GISS, in situ.
CO2 Snyder (2016), ice cores.
CO2 Ahn et al. (2012), ice cores.
CO2 NOAA/ESRL, in situ. 1000 1500 2000 200 250 300 350 400 CO 2 mole fr action [ppm]
Year AD
Figure 1.1 | Atmospheric CO2mole fractions (blue color tones) and temperature deviation from a reference
period (orange color tones) over the past 850.000 years. Different data sources are combined to construct the figure. Northern Hemispheric temperature anomalies from in situ measurements since 1980 are relative to 1951–1980 average temperatures (NASA/GISS); global average temperature anomalies in the last 1000 years are
retrieved from ice core records and are relative to the 1961–1990 average temperature (Jones and Mann,2004);
older temperature anomalies are taken fromSnyder(2016) and are against the average temperature over the last
5000 years. CO2mole fractions since 1959 are from in situ measurements done by the Scripps research institute
and later by NOAA/ESRL, older CO2mole fractions are retrieved from ice core records as presented inAhn et al.
1
Scientists have reached consensus that the modern climate change is very likely due to anthropogenic greenhouse gas emissions, i.e. with a probability of 90–100 % (Bindoff
et al.,2013;Cook et al.,2013). Researchers and policymakers face the challenge to come
up with measures against the copious amount of greenhouse gas emissions. At the UN Climate Change Conference in December 2015, country leaders agreed to further reduce greenhouse gas emissions, with the aim to keep the global warming in the 21stcentury below 2.0±C, or even 1.5±C. However,Rogelj et al.(2016) stressed less then a year later that
the intended contributions from individual countries to the Paris Agreement would still lead to a warming of 2.6–3.1±C by the end of the 21stcentury and that more actions are required to stay well below the 2±C target.
The future climate depends on the amount of carbon in the atmosphere, which is influ-enced by the removal of carbon by sinks and the emissions from sources. Anthropogenic emissions are a large source of carbon (e.g. in the form of CO2and CH4), and terrestrial vegetation and the ocean are a net sink (Le Quéré et al.,2018). The exchange of carbon that occurs mainly between the atmosphere, biosphere, land and ocean is called the carbon cycle. Future climate predictions depend on the accuracy of global carbon models and depends on our knowledge of the global carbon cycle. An accurate representation of the amount of carbon present in the atmosphere and ocean is needed, which can be measured at present; however, scientists also need to accurately simulate the sources and sinks of carbon and how they change over time as a response to climate change. Although many of the sources and sinks can be roughly estimated, not all underlying mechanisms are fully understood. For example, it is believed that tree growth will increase—and thereby also the uptake of CO2in plants through photosynthesis—when CO2mole fractions in the
at-mosphere are higher. Still, results of long-term monitoring at ecosystem level over the last decades do not provide consistent evidence for this (Lewis et al.,2009;Van der Sleen et al.,
2014). This example highlights that the underlying mechanisms of photosynthesis are not fully understood. Models are therefore not able to accurately simulate photosynthesis
(Friedlingstein et al.,2014). Moreover, besides uptake of CO2through photosynthesis, CO2
is respired by ecosystems at the same time. To fully understand the mechanisms behind photosynthesis and respiration and their response to a changing climate, researchers need measurements of each flux individually. However, the current measurement networks are only able to resolve the net difference between photosynthesis and respiration and they cannot accurately partition the photosynthesis and respiratory flux. Therefore, these processes cannot be studied separately and our understanding of each of these fluxes is limited. This PhD research deals with this problem. The next section further introduces vegetative CO2uptake and describes why CO2uptake through photosynthesis cannot be
measured directly.
1.2.
V
EGETATIVE
CO2
UPTAKE THROUGH PHOTOSYNTHESIS
Small pores on the surface of plant leaves, so called stomata, allow air to enter the leaf. When CO2has entered a plant cell it diffuses into mesophyll cells in the middle layer of a leaf (Fig. 1.2). In these cells, photosynthesis occurs in two stages. In the first stage, energy of photons (light) is converted to chemical energy in the form of molecules. Hydrolysis takes place in this process, which means that H2O is used and as a bi-product O2is being
produced, which is the O2in the atmosphere that we breath. In the reaction that follows
1
sugars (glucose). This reaction is catalyzed by the enzyme RuBisCo and when the reaction takes place, plants release water vapor through transpiration. Overall, these processes that occur during photosynthesis can be written in chemical form as:6CO2+ 12H2O°°°! Clight 6H12O6+ 6O2+ 6H2O.
The rate of CO2assimilation in plants is largely dependent on the catalytic capacity of the Rubisco enzyme which depends on temperature (Cen and Sage,2005). Besides of the activity of the RuBisCo enzyme, the availability of light, carbon and water is another requirement for photosynthesis to happen. For that reason the stomatal aperture largely controls photosynthesis: plants adjust the opening and closing of stomata to optimize their internal water balance and the influx of CO2to achieve optimal growth (Fig. 1.2). Therefore, plants typically open their stomata during daytime when there is light for photosynthesis, such that CO2from the ambient air becomes available for the internal cells. At the same
time, plants respond to ambient humidity by closing their stomata when the air gets too dry and they lose too much water to the ambient air due to a large humidity gradient between internal and ambient air. This means that when stomata close due to drought, the availability of CO2in the plant will go down and reduces photosynthesis. The stomatal aperture therefore tightly controls the extent to which photosynthesis can happen.
Figure 1.2 | Schematic cross section of a leaf with the pathways of COS and CO2uptake and transpiration (figure
adapted fromMoene and van Dam(2014)).
Besides taking up CO2during photosynthesis, plants also respire CO2. During respira-tion (Re) plants use the glucose to gain energy, after which CO2is released back into the
1
atmosphere:
C6H12O6+ 6O2°°! 6CO2+ 6H2O + energy.
This process does not require light and happens both during the day and night. Respiration does not only occur in plants, but also in the soil, for example by roots or by organisms that consume plant material (heterotrophs) such as microbes (e.g. bacteria) and macrofauna. On ecosystem scale, the total amount of carbon that is fixed by plants during photosyn-thesis is referred to as Gross Primary Production (GPP). The Net Ecosystem Exchange (NEE) of carbon between the terrestrial biosphere and atmosphere follows from the difference between GPP and Re, where Re is a combination of soil and plant respiration (Fig. 1.3):
NEE = GPP – Re.
A common technique to determine NEE is the eddy covariance (EC) method (Aubinet
et al.,2012). Using this methods, fluxes are determined from fast fluctuations (at 10
Hz) of vertical wind speed and gas mole fractions above an ecosystem. The EC fluxes represent the net exchange of a gas above an ecosystem, which is the NEE in the case of CO2. A large network of EC measurements exists and these data are useful to evaluate and calibrate ecosystem models. The evaluation of ecosystem models would further profit from a partitioning of NEE into GPP and Re. When GPP and Re are separated, these processes can be studied separately and model errors can be better detected (Reichstein
et al., 2005). Unfortunately, there are no methods to directly partition NEE into GPP
and Re. In the last few decades, finding and testing methods to separate NEE into GPP and Re has been a field of research on its own. Flux-partitioning algorithms have been developed, but each of these methods has their flaws (Reichstein et al.,2005). One of the most widely used methods extrapolates nighttime temperature–Re relations to daytime
(Reichstein et al.,2005). However, the disadvantage of this method is that nighttime data
are prone to measurement errors (Papale et al.,2006;Aubinet et al.,2012) and this method assumes that the temperature response of Re is similar between daytime and nighttime, without interference of other environmental parameters. Also isotopic signatures can provide a means to separate GPP and Re because those processes have a different effect on the isotopic composition of CO2that leaves the plant (Fassbinder et al.,2012). For
example,13C and18O isotopes in CO2have been used to separate GPP and Re (e.g.Yakir
and Wang,1996;Ogée et al.,2003;Bowling et al.,2001;Fassbinder et al.,2012), but the
applications are limited by the precision of measurements and by theoretical assumptions
(Wehr and Saleska,2015). Recently, it was shown with improved measurement precision
of13C and18O isotopes that the standard approach to separate NEE into GPP and Re (afterReichstein et al.,2005) overestimated daytime Re by 100 % in a temperate forest, and thereby overestimated GPP by 25 % (Wehr et al.,2016). This overestimation is caused by daytime inhibition of Re by light, which is known as the Kok-effect (Heskel et al.,2013). This effect is not accounted for in the standard flux-partitioning method (e.g. afterReichstein
et al.(2005)), whereas these findings have large consequences for the global carbon budget.
Since the late ’80s, several studies noticed the link between the plant uptake of carbonyl sulfide (COS or OCS) and CO2(Goldan et al.,1988;Chin and Davis,1995;Kesselmeier and
Merk,1993;Sandoval-Soto et al.,2005). These studies used the CO2to COS uptake ratio
to improve the global sulfur budget with the best estimates of GPP available at that time. This was particularly of interest for stratospheric chemistry, where COS contributes to the
1
formation of the sulfate aerosol layer (Chin and Davis,1995). Later,Montzka et al.(2007) found that the seasonal variation of atmospheric COS shows close resemblance to that of CO2. They concluded that if the atmospheric drawdown of COS is dominated by vegetativeuptake, COS could provide unique insights into carbon uptake by photosynthesis. After that,Campbell et al.(2008) used COS fluxes to constrain GPP, and with that the idea of using COS to partition NEE into GPP and Re was born.
Figure 1.3 | Schematic representation of fluxes in a forest canopy. CO2 is taken up by vegetation (GPP) and
respired by both soil and vegetation (Re), whereas—in an ideal situation—COS is only taken up by vegetation.
1.3.
A
TRACER FOR PHOTOSYNTHESIS
:
CARBONYL SULFIDE
Just like CO2, COS enters the leaf through stomata and diffuses into the mesophyll cells
(Fig. 1.2). In the chloroplast, which is also the place where photosynthesis takes place, the enzyme carbonic anhydrase (CA) catalyzes the hydrolysis of COS to produce CO2and H2S
(Protoschill-Krebs et al.,1992,1996;Stimler et al.,2010a):
COS + H2O°°! HCOOSCA – + H+°°! H2S + CO2.
H2S fulfills several physiological functions in plants (Li,2013). The reverse reaction is
strongly unfavorable and COS is therefore fully taken up by the plant, unlike CO2 that
is re-emitted during respiration (Fig. 1.3). Provided that other sources and sinks in an ecosystem are negligible, the atmospheric drawdown of COS above an ecosystem reflects the uptake of COS by plants. Besides that, COS and CO2share the same pathway from the ambient air to the mesophyll cells (Fig. 1.2) and the uptake of COS and CO2are closely related (Whelan et al.,2018). By using the COS to CO2uptake relationship at the leaf level,
1
COS ecosystem fluxes can therefore be used to estimate GPP following (Campbell et al.,
2008;Asaf et al.,2013):
GPP = FCOS[COS][CO2] LRU1 , (1.1)
where FCOSis the ecosystem flux of COS, [COS] and [CO2] are atmospheric mole fractions of
COS and CO2respectively and LRU is the leaf-scale relative uptake ratio. LRU is calculated
as the ratio of the deposition velocities of COS over that of CO2and deposition velocities
are calculated as the ratio of the leaf flux divided by the mole fraction of the gases.
Following the explanation above, requirements for using COS as a tracer for GPP are that the flux of COS is only a one-way flux, that COS and CO2follow the same pathway into the plant, and that other sources and sinks of COS in an ecosystem are negligible or known, so that they can be corrected for. Besides that, there should be no interaction between COS and CO2(Blonquist et al.,2011). Proper characterization of all sources and sinks of COS is
key in using COS as a tracer for GPP. Therefore, in the last couple of years many studies have focussed on identifying and understanding the sources and sinks of COS.Whelan
et al.(2018) provided an extensive overview of all the knowledge gained on COS in the last
few decades, including some of the work presented in this thesis. The next section will give a brief overview of what is known about the sources and sinks of COS and which questions led to the work in this thesis.
1.4.
T
HE
COS
BUDGET
1.4.1.
G
LOBAL SCALEAnnual mean mole fractions of COS mostly range between 350 and 550 pmol/mol (ppt) around the globe, which makes COS the most abundunt sulfur-containing gas in the atmosphere (Montzka et al.,2007;Brühl et al.,2012). Mole fractions of COS are about 1 million times lower than that of CO2, which has mole fractions around 400 µmol/mol
(ppm). Due to its relatively long lifetime of 1.5–3 years (Montzka et al.,2007), COS can be transported from the troposphere into the stratosphere. COS plays an important role in the global sulfur budget and the formation of stratospheric background sulphur aerosol (Chin
and Davis,1995;Brühl et al.,2012), which have a cooling effect on the Earth’s climate.
On the global scale, vegetative uptake is the largest sink of COS and is estimated to account for 55–64 % of the total COS sink (Launois et al.,2015b). Another sink is the soil, which will be further discussed in the next section. Besides those, small sinks are chemical loss by photolysis in the stratosphere and reaction with the hydroxyl radical (OH) in the atmosphere (Kettle et al.,2002;Berry et al.,2013;Lennartz et al.,2017).
One of the main sources of COS is anthropogenic, which can be either direct or indirect. Indirectly, COS is formed when CS2, which is a gas that is emitted by the rayon industry
(Campbell et al.,2015), is oxidized. Direct emissions of COS are smaller and consist of
biomass burning, coal combustion, aluminum smelting, pigment production, shipping, tire wear, vehicle emissions and coke production (Whelan et al.,2018, and references therein).
Besides anthropogenic emissions, COS is being emitted by the ocean also both di-rectly and indirecly. Direct emissions are through photochemical production and light-independent (dark) processes, which are both poorly understood (Whelan et al.,2018;
1
Lennartz et al.,2017). Indirect emissions are through the oxidation of CS2and dimethylsulfide (DMS) by OH (Chin and Davis,1995). Until recently, ocean emissions were believed to be the largest source of COS (Kettle et al.,2002;Watts,2000;Launois et al.,2015a). How-ever, a study byLennartz et al.(2017) used COS measurements in ocean water to show that the direct oceanic emissions were much lower (130 ± 80 Gg S yr°1) than estimates from
top-down studies (e.g. 813 Gg S yr°1byLaunois et al.(2015a)). Other smaller COS sources
are volcanoes and wetlands.
Large uncertainties remain in the quantification of the sources and sinks of COS and with the current knowledge there is a gap in the COS budget with a source missing. When the sink is larger than the source (which is the current state of the COS budget), this would mean that atmospheric mole fractions of COS are decreasing; however, this is not observed in long-term records of COS mole fractions. Ice core records from the south pole show a trend of increasing COS mole fractions from the mid-1800s to the late 1900s, followed by a decrease of COS since the 1980s (Montzka et al.,2004;Aydin et al.,2008). The extent of the decreasing trend of COS does not match with the size of the gap between the sources and sinks of COS. In fact, the changes that have been observed more recently show strong resemblance with the global rayon production (Campbell et al.,2015). That is, a general decrease in COS mole fractions until the ’90s, followed by an increase since the beginning of the 2000s (Kremser et al.,2015;Lejeune et al.,2017).
1.4.2.
E
COSYSTEM SCALEAt the ecosystem scale, the emissions of COS from ocean and anthropogenic sources are typically well separated from the vegetative uptake in natural areas such as forests, grasslands or wetlands. Still, the vegetative uptake of COS must be the dominant COS flux within an ecosystem when COS is to be used as a tracer for GPP; other COS fluxes must be small or well characterized. Besides the presence of the CA enzyme in plants (that catalyzes the hydrolysis of COS), it is also present in soils (Ogée et al.,2016), which makes (oxic) soils typically a sink of COS. The magnitude of soil fluxes were most often found to be linked with temperature (White et al.,2010;Maseyk et al.,2014) and soil water content
(Steinbacher et al.,2004). On the other hand, anoxic soils (like wetlands) are typically
sources of COS (Whelan et al.,2018, and references therein), but COS emissions have also been found in oxic soils, with larger emissions under high temperature and radiation
(Whelan and Rhew,2015;Whelan et al.,2016;Kitz et al.,2017). So far, only few studies
have determined the contribution of soil fluxes relative to the ecosystem fluxes of COS
(Maseyk et al.,2014;Wehr et al.,2017).Maseyk et al.(2014), found that soils were typically
small sinks in an agricultural wheat field, contributing only to 1–6 % of the total ecosystem flux. However, the soils turned into a significant source at the end of the growing season, complicating the use of COS as tracer for GPP. For the application of COS as GPP-tracer it is therefore key that the soil COS fluxes are quantified and well understood.
1.4.3.
L
EAF SCALEThe accuracy of the COS to CO2leaf relative uptake ratio (LRU) is key in translating COS
fluxes to GPP. Several studies have derived LRU for different plant species from cham-ber enclosures (Kesselmeier and Merk,1993;Kesselmeier et al.,1993;Kuhn et al.,1999;
1
et al.,2014). Those LRU values ranged from 0.4 to 9.5 with a median of 1.75 and with 50 %
of the values between 1.48 and 2.46 around the median (see Fig. 3 inWhelan et al.(2018) for the frequency distribution of all published LRU values). LRU used to be reported as con-stant values, which can be valid if COS and CO2fluxes would respond similarly to changing
environmental conditions. However, several studies have shown that LRU depends on light (Stimler et al.,2011;Maseyk et al.,2014;Commane et al.,2015), because the COS and CO2fluxes respond differently to light: the COS flux can continue in the dark because the hydrolysis of COS in plants does not require light (Gries et al.,1994;Protoschill-Krebs et al.,
1995;Stimler et al.,2011), whereas the CO2uptake can only occur under light conditions.
Therefore, the variation of LRU with light needs to be well characterized. Besides the variation with light, changes of LRU with other environmental parameters are not well studied.Stimler et al.(2010a) found from laboratory experiments that LRU changes with temperature. However, no long-term field measurements that monitor the change of LRU under natural conditions over a period of multiple seasons were done.
If the COS uptake is indeed light-independent, this means that vegetative uptake of COS can continue during the night if stomata are not completely closed (Maseyk et al.,
2014). It was long assumed that stomata close during the night. However,Caird et al.(2007) showed that nighttime stomatal conductance exists in a wide variety of plant species. Depending on the extent of nighttime stomatal opening, the nighttime flux of COS can have a substantial contribution to the total COS budget. Therefore, the nighttime COS flux needs to be well quantified, even though COS is not used as a tracer for GPP under dark conditions (because then GPP is zero). Moreover, the fact that COS and CO2 do
not share the same biochemical reactions in the leaf but rather the diffusive pathway between air and the chloroplast was recently put as a motivation to use COS as tracer for diffusive conductance, of which stomatal conductance is the most dominant component
(Commane et al.,2015;Wehr et al.,2017). This would link COS not only to the carbon
cycle, but also to the water cycle, as the loss of water by the plant through transpiration is controlled by the stomatal aperture.
Recently,Gimeno et al.(2017) found emissions of COS by mosses and liverwort that increased with increasing temperature and light. Emissions of COS were also observed from ecosystem fluxes in one summer month in a mixed temperate forest (Commane et al.,
2015). If COS is emitted by plants, it would question the assumption that the COS flux is unidirectional, and challenges the use of COS as a tracer for GPP (Gimeno et al.,2017;
Wohlfahrt,2017).
1.5.
COS
MEASUREMENT TECHNIQUES
Measurements of COS are required to further investigate the potential of using COS as tracer for GPP and to be able to benefit from this method. Preferably, measurements should be available at the surface (within the atmospheric boundary layer), with global coverage and with a fine spatial resolution. Satellite measurements of COS could provide these data, but the satellites that are currently able to measure COS are not sensitive to concentrations near the surface and have limited precision (Kuai et al.,2014). Still, satellite measurements of an Infrared Atmospheric Sounding Interferometer were able to detect large sources of COS over Asia, indicating the benefit of spatial coverage from these tech-niques (Vincent and Dudhia,2017). Total column measurements of COS are made at a few sites using Fourier-transform infrared (FTIR) spectroscopy. Recent developments made it
1
possible to distinguish between COS in the troposphere and the stratosphere from these measurements (Lejeune et al.,2017). Still, these measurements are only available at a few sites worldwide. Currently, studies on COS largely rely on sampling networks at the surface or in situ measurements. The monitoring network as described byMontzka et al.(2007) worked with discrete flask samples that were measured with gas chromatographic mass spectrometry (GC-MS). These measurements have high accuracy (95% of these mea-surements reach an accuracy of < 6.3 ppt), but these samples are taken typically 1 to 5 times per month and can therefore not provide insight into the daily variation of COS, let alone that processes happening within a few hours or minutes can be investigated. Moreover, this instrumentation is typically not suitable for deployment in the field. Devel-opments of Quantum Cascade Laser Spectrometers (QCLS) have pushed the field forward in the ability to do continuous and high-accuracy trace gas measurements, including COS (Stimler et al.,2010b). These instruments have proven to be a valuable tool to do online measurements of COS and CO2(e.g.Stimler et al.,2010b;Commane et al.,2013;
Berkelhammer et al.,2014;Maseyk et al.,2014;Commane et al.,2015;Yang et al.,2018)
and also allows to do high-frequency (10 Hz) measurements of COS (Gerdel et al.,2017). Therefore, these instruments are an improvement to the GC-MS methods that were used before (Stimler et al.,2010b). Nowadays, the QCLS instruments are commercially available from either Los Gatos Research Inc. (Berkelhammer et al.,2014) or Aerodyne Research Inc. (McManus et al.,2010;Nelson et al.,2004). Being able to deploy a QCLS in the field allows field measurements that can help to further enhance our knowledge on the sources and sinks of COS. To take further advantage of this relatively new instrumentation requires elaborate testing of the instrumentation to achieve optimal accuracy and precision. Also, to be able to compare measurements from different sites and with different instruments requires a consistent scale. Therefore, it is key that measurements are accurately calibrated against known calibration standards. Besides that, a high precision is important when the aim is to detect small changes in COS mole fractions.
1.6.
O
BJECTIVE AND APPROACH OF THIS THESIS
Overall, COS has been presented as a promising new tracer, but the overview in the previous sections showed the challenges that emerge from detailed investigation of COS fluxes and its relation with fluxes of CO2. Several aspects of COS, related to both the understanding
and quantification of the sources and sinks, are unresolved, but are essential for the use of COS as tracer for GPP. The goal of this PhD research is to determine accurate COS-based
GPP estimates. This thesis focusses on a few aspects to improve the understanding of the
sources and sinks of COS that are key to reach that goal. It would not be possible to cover all aspects of the COS budget in a four-year PhD research; therefore, the scope of this study is refined to do the following:
First, a QCLS was tested for its ability to do continuous and high-accuracy measure-ments of COS and CO2. Laboratory tests are performed to find optimal routines for calibration of the measurements and to carefully quantify all uncertainties associated with processing of those measurements. The aim was also to build a robust system to continuously measure ambient mole fractions of COS and CO2. This system should be able
to operate autonomously with little operator attention in the field. This work is presented in chapter 2 of this thesis. All of the aspects related to instrumentation are key for further studies on COS in this thesis, as well as studies done by other research groups.
1
The following chapters focus on characterizing the fluxes of COS in a boreal forest and a coastal agricultural site. Using the setup for instrumentation that was developed in chapter 2, the ongoing COS research in Hyytiälä, Finland (led by the University of Helsinki) was extended with a number of extensive field campaigns in three consecutive years (2015–2017). This work was done in collaboration with researchers from the University of Helsinki, the University of California, Los Angeles, and the Open University, Milton Keynes. As a result of the field campaign in Hyytiälä in 2015 the nighttime fluxes of COS could be characterized, which is presented in chapter 3. With the COS uptake being independent of light, the nighttime fluxes of COS have the potential to have a substantial effect on the COS budget. Two measurement techniques (eddy-covariance and the radon-tracer method) were used to quantify nighttime fluxes of COS in the summer and autumn of 2015. Besides that, measurements of mole fractions, soil fluxes (Sun et al.,2018a) and stomatal conductance were used to determine what drives the nighttime COS fluxes.
To be able to accurately translate the measured COS ecosystem fluxes in Hyytiälä to GPP estimates it was aimed to investigate the size and variability of LRU at the site. The COS and CO2fluxes were measured at the leaf-level using branch chambers in 2017.
In chapter 4, the light-dependence of LRU is characterized and the variation of LRU with other environmental parameters was studied. The knowledge of LRU and COS soil fluxes was used to get COS-based GPP estimates and to show the sensitivity of those GPP estimates to variations in LRU. Also the application of leaf-level LRU to regional scale modelling is discussed in this chapter.
In chapter 5 the variability of atmospheric mole fraction measurements of COS and CO2is investigated from an agricultural site at the coast of the Netherlands. Collocated
measurements of other trace gases such as CO and222Rn helped to understand the sources and sinks of COS. Furthermore, the coastal location of the Lutjewad station allows for separation of air masses from the sea and from inland, where more biogenic, agricultural and anthropogenic signals can be expected from inland activities.
Finally, the findings are synthesized in chapter 6. This chapter discusses the role that COS can have in obtaining GPP estimates and the advancements that need to be made for the wide application of COS for that purpose.
2
C
ONTINUOUS AND
HIGH
-
PRECISION ATMOSPHERIC
CONCENTRATION MEASUREMENTS
OF
COS, CO
2
, CO
AND
H
2
O
USING
A QUANTUM CASCADE LASER
SPECTROMETER
Carbonyl sulfide (COS) has been suggested as a useful tracer for gross primary production as it is taken up by plants in a similar way as CO2. To explore and verify the application
of this novel tracer, it is highly desired to develop the ability to perform continuous and
high-precision in situ atmospheric measurements of COS and CO2. In this study we have
tested a quantum cascade laser spectrometer (QCLS) for its suitability to obtain accurate
and high-precision measurements of COS and CO2. The instrument is capable of
simul-taneously measuring COS, CO2, CO and H2O after including a weak CO absorption line
in the extended wavelength range. An optimal background and calibration strategy was developed based on laboratory tests to ensure accurate field measurements. We have derived water vapor correction factors based on a set of laboratory experiments and found that for COS the interference associated with a water absorption line can dominate over the effect of dilution. This interference can be solved mathematically by fitting the COS spectral line separately from the H2O spectral line. Furthermore, we improved the temperature stability
of the QCLS by isolating it in an enclosed box and actively cooling its electronics with the same thermoelectric chiller used to cool the laser. The QCLS was deployed at the Lutjewad
This chapter is published as: Kooijmans, L. M. J., Uitslag, N. A. M., Zahniser, M. S., Nelson, D. D., Montzka, S. A.,
and Chen, H.: Continuous and high-precision atmospheric concentration measurements of COS, CO2, CO and
H2O using a quantum cascade laser spectrometer (QCLS), Atmos. Meas. Tech., 9, 5293–5314,
doi:10.5194/amt-9-5293-2016, 2016.
2
atmospheric monitoring station (60 m, 6°21’E, 53°24’N, 1 m a.s.l.) in the Netherlands from July 2014 to April 2015. The QCLS measurements of independent working standards while deployed in the field showed a mean difference with the assigned cylinder value within 3.3 ppt COS, 0.05 ppm for CO2and 1.7 ppb for CO over a period of 35 days. The different
contributions to uncertainty in measurements of COS, CO2and CO were summarized and
the overall uncertainty was determined to be 7.5 ppt for COS, 0.23 ppm for CO2and 3.3 ppb for CO for 1-minute data. A comparison of in situ QCLS measurements with those from concurrently filled flasks that were subsequently measured by the QCLS showed a difference of -9.7 ± 4.6 ppt for COS. Comparison of the QCLS with a cavity ring-down spectrometer showed a difference of 0.12 ± 0.77 ppm for CO2and -0.9 ± 3.8 ppb for CO.
2.1.
I
NTRODUCTION
Carbonyl sulfide (COS) has been suggested as a potential tracer for photosynthetic CO2 uptake (Sandoval-Soto et al.,2005;Montzka et al.,2007;Campbell et al.,2008;Berry et al.,
2013;Asaf et al.,2013), as it follows the same uptake pathway into plants through stomata
as CO2 but is not generally re-emitted by plants (Protoschill-Krebs et al.,1992, 1996;
Stimler et al.,2010a). COS therefore provides a means to partition net ecosystem exchange
into gross primary production (GPP) and respiration. As large uncertainties in the COS budget remain, field measurements of COS and CO2concentrations and fluxes from leaf to ecosystem and regional scale are required for the COS tracer method to be tested and validated (Wohlfahrt et al.,2012;Berkelhammer et al.,2014). Therefore, there is a need for high-frequency and high-precision measurements techniques of COS and CO2.
Several past studies on COS have relied on discrete (flask) samples analyzed with gas chromatographic mass spectrometry (GC-MS;Montzka et al.,2007;Stimler et al.,2010a). For example, the global atmospheric flask sampling network described byMontzka et al.
(2007) has allowed a foundation for understanding COS concentrations over annual cycles on global scale. Although the GC-MS technique can be used for in situ measurements
(Miller et al.,2008;Belviso et al.,2013), this technique does not typically allow for
high-frequency measurements of 1 to 10 Hz. Recent developments of quantum cascade laser spectrometers (QCLSs) have enabled in situ trace gas measurements including COS. These instruments have proven to be a valuable tool for continuous high-frequency measure-ments of COS and CO2up to a frequency of 10 Hz (Stimler et al.,2010a,b;Asaf et al.,2013;
Commane et al.,2013;Berkelhammer et al.,2014;Maseyk et al.,2014;Commane et al.,
2015).
The required measurement precision (in this study we define precision as the standard deviation over a 2-minute period) for studies of exchange processes of COS and CO2
between biosphere and atmosphere depend on the concentration change that these gases undergo in any given experiment. On the regional scale, COS shows seasonal variations typically between ª100 and 150 ppt at continental sites in the Northern Hemisphere (NH) and between 40 and 70 ppt in the Southern Hemisphere (SH) and at marine sites (Montzka
et al.,2007). CO2seasonal variations typically reach up to 15 ppm in the NH and are as
low as 2 ppm at the South Pole (Zhao and Zeng,2014). For the leaf scale, COS and CO2
concentration changes can be substantially larger; for example,Berkelhammer et al.(2014) showed that during branch bag measurements COS generally decreased by 180 to 240 ppt during active photosynthesis and CO2concentrations can easily change by 200 ppm,
2
depending on the setup. Besides the difference in requirements for precision between different experimental setups, the type of analyses intended for a dataset also determines the requirements for precision and accuracy of the measurements. If the intention is to compare atmospheric concentrations across sites, then accuracy is important because data from different sites must be on consistent scales. In contrast, short-term precision is more important than accuracy when differences between heights are to be interpreted (e.g., as in estimation of fluxes from profile measurements). Following the K-parameterization formulation of the flux-gradient method (e.g.,Meredith et al.,2014), F = °K ¢C/¢zΩ, the
precision required to capture the concentration differences between heights (¢C) mostly depends on the size of the fluxes F, the height difference ¢z and the turbulence conditions, which is represented by the eddy diffusivity K and to a lesser extent by the molar density of air Ω. To be able to capture COS fluxes of , for example, 10 pmol m°2s°1over a height
difference of 20 meters, the measurement precision of COS should be better than 0.5 ppt under high turbulent conditions (K = 10 m2 s°1) and 4.8 ppt under low turbulent
conditions (K = 1 m2s°1). If we were to infer the gross fluxes from chamber measurements with ¢CO2(the difference between in- and outgoing chamber concentrations) measurable
from 1 ppm, then, given the leaf-scale relative uptake (LRU) ratio of COS/CO2 1.5–4.0
(Stimler et al.,2010a;Seibt et al.,2010;Berkelhammer et al.,2014), our goal would be to
have measurement precisions of COS better than 1.9–5.0 ppt for COS (calculated from LRU and scaling with ¢CO2and the ratio of COS/CO2mole fractions gives, for example, 1.5*1*(500/400) = 1.9 ppt at the ambient level of 500 ppt and 400 ppm CO2).
Measurement instruments for long-term atmospheric trace gas concentration monitor-ing need to meet different requirements than, for example, eddy-covariance measurements. The eddy-covariance technique requires high-frequency data (>10 Hz), which typically adversely affect the precision of the measurements compared to 1 Hz data, and requires an averaging period of about 10 to 30 min. In contrast to the high frequency required for eddy-covariance measurements, lower-frequency measurements (1 Hz) provide useful results over extended measurement periods and enhance the precision of any individual measurement. Furthermore, measurements for long-term monitoring do not require fast response, and thus it is not necessary to operate the instrument at high flow rates. As a mat-ter of fact, low flow rates are preferred so that working standards can be used over a long period. This reduces the additional logistics needed for calibration gases, such as filling, calibration and transportation of the standards (Xiang et al.,2014). Besides in situ measure-ments, flask or canister measurements can be a valuable tool for providing information about ambient concentrations of COS as well. For example, flask measurements were used before when constructing an historical record from firn air (Montzka et al.,2004), during field campaigns (White et al.,2010;Blonquist et al.,2011) and for long-term monitoring
(Montzka et al.,2007). In this research we developed a robust setup for high-precision and
long-term monitoring of ambient concentrations of COS, CO2, CO and H2O at different heights from the Lutjewad monitoring station in Groningen, The Netherlands. To this end we employed a “QCL Mini Monitor” from Aerodyne Research Inc. (Billerica, MA, USA) that can operate autonomously and requires little operator attention. We designed an optimal strategy for ‘zero’ air spectral correction and calibration for accurate measurements and we assessed the correction for water vapor interference. In this paper we aim to evaluate and improve the performance of the instrument. We will show the precision and accuracy of the instrument with over half a year of field data and measurements of working standards, and we compare the measurements with other instrumentation. Furthermore, we
eval-2
uate the total uncertainty of the measurements by combining the uncertainties of scale transfer, water vapor corrections and the measurement precision. Based on the precision and accuracy that we derive from these experiments, we discuss the suitability of COS measurements on this instrument for different purposes; that is, for interpreting profile measurements and comparing concentrations across sites. In addition to the experimental setup for continuous in situ measurements we developed a setup to analyze flasks, which we used to make a comparison with GC-MS measurements of flasks and to assess the laboratory-derived correction for water vapor interference.
2.2.
E
XPERIMENTAL SETUP
Before the actual deployment of the instrument in the field we performed laboratory tests to assess the accuracy and traceability of the QCLS measurements and to develop procedures for applying corrections as needed. Here we describe the laboratory tests and we give detailed information about the instrumentation and field setup.
2.2.1.
I
NSTRUMENTATIONThe “QCL Mini Monitor” that we use is a tunable infrared laser direct absorption spectrom-eter (TILDAS) using a single continuous-wave quantum cascade laser (Alpes Lasers), which is cooled with a Peltier element to -19.8 °C, and using a single photodiode infrared detector (Teledyne Judson Technologies;McManus et al.,2010). The waste heat from both the laser and detector is removed with a recirculating mixture of water containing 25 % ethanol, which is temperature controlled with a thermoelectric chiller, ThermoCube 300 (Solid State Cooling Systems, USA). The instrument was initially set to simultaneously measure COS, CO2and H2O at wavenumbers 2050.397, 2050.566 and 2050.638 cm°1, respectively.
We extended the range of the laser current to include measurements of CO at 2050.854 cm°1. Figure 2.1 shows the simulated transmission spectrum of ambient concentrations of COS, CO2, CO and H2O as obtained through the HITRAN 2012 database (Rothman et al., 2013). The precision and accuracy of the measurements will be discussed in Sect. 2.3.1.
The instrument consists of a 0.5 L astigmatic Herriott style multi-pass absorption cell
(McManus et al.,2010) with an effective path length of 76 m. The cell has a temperature
between 20 and 24 °C, depending on the room temperature and the temperature setting of the thermoelectric chiller. The cell is kept at a constant pressure of 53.3 hPa (40 Torr) with an inlet valve that is controlled by the TDLWINTEL program (Aerodyne Research Inc., Billerica, MA, USA) based on the measured cell pressure. The same software manages the data acquisition and spectral analysis (Nelson et al.,2004) and calculates dry air mole fractions in real time (1 Hz) through nonlinear least square spectral fits combined with the measured cell temperature and pressure, a constant path length and the HITRAN 2012 database cross sections as a function of wavelength. The spectral fit for CO is separated from the fit for COS, CO2 and H2O as there is slight interaction of the CO peak with a
second absorption line of COS. The COS fit close to the CO peak is linked to the COS peak at lower wavenumbers to improve the fitting for CO. This is achieved by fitting the spectra in two steps: first the mole fractions are determined for both COS peaks independently, second the CO concentration is recalculated with the fixed COS concentration derived from the separated COS peak in the first step.
2
2050.3 2050.4 2050.5 2050.6 2050.7 2050.8 2050.9 0.994 0.995 0.996 0.997 0.998 0.999 1.000 Wavenumber [cm−1] Tr ansmittance COS CO2 H2O CO COS CO2 H2O CO T = 298 K, P = 40 Torr, L = 76 mFigure 2.1 | Simulated transmission spectrum of ambient concentrations of COS (500 ppt), CO2(500 ppm), H2O
(1.5 %) and CO (200 ppb) with sample cell conditions: temperature 298 K, pressure 53.3 hPa (40 Torr) and the
absorption path length 76 m. A small water band at 2050.5 cm°1can interfere with COS at 2050.4 cm°1and can
affect the COS correction for water vapor without a split fit at 2050.45 cm°1(Sect. 2.2.3).
be used for re-analysis using the so-called “Playback” mode of the software. The spectral parameters (line shape and position) for the fits are taken from the HITRAN database
(Rothman et al.,2013). The sample spectra are normalized with a ‘zero’ air spectrum to
remove background spectral structures and to remove absorbance external to the multi-pass cell (Stimler et al.,2010b;Santoni et al.,2012). The ‘zero’ air spectrum is periodically determined when the cell is flushed with high-purity nitrogen (99.99999 %), which we will now refer to as ‘background’ measurement. The nitrogen is first passed over a gas purifier (Gatekeeper, CE-500K-I-4R) to remove CO that is often found in such nitrogen cylinders. The frequency of the laser is locked based on the spectrum measurement of the high strength CO2line at 2050.566 as shown in Fig. 2.1. For automatic start-up, a gas sealed
in an aluminum reference cell can be flipped into the optical beam. The reference cell was filled with 8 hPa (6 Torr) COS and 27 hPa (20 Torr) CO. Initially, we could use the peak position of COS in the reference cell to determine the frequency of the laser. However, COS did not last longer than a few months in the reference cell so thereafter the laser frequency was locked only based on the peak position of CO, which did not impact the results.
2.2.2.
C
ALIBRATION STRATEGYTo allow comparison of QCLS measurements with other instrumentation and across differ-ent sites requires traceability to a primary scale. Laboratory tests were conducted to char-acterize the response of the instrument against ambient air standards from NOAA/ESRL, which were subsequently used to transfer the calibration scale to working standards. Moreover, we performed tests to understand the frequency required for background and
