Greenhouse gas emissions from three large lakes during the autumn 2020

(1)

Department of Thematic Studies Campus Norrköping

Bachelor of Science Thesis, Environmental Science Programme, 2021

Anja Anderö Nordqvist & Veronica Bohlin

Greenhouse gas emissions from

three large lakes during the

autumn 2020

(2)

1 Rapporttyp Report category Licentiatavhandling Examensarbete AB-uppsats C-uppsats D-uppsats Övrig rapport Språk Language Svenska/Swedish Engelska/English Titel

Växthusgasutsläpp från tre stora sjöar under hösten 2020

Title

Greenhouse gas emissions from three large lakes during the autumn 2020

Författare

Author

Anja Anderö Nordqvist & Veronica Bohlin

Sammanfattning

Metan (CH4) och koldioxid (CO2) är två växthusgaser och stora drivkrafter för globala klimatförändringar. Sjöar är kända för att vara en källa för

CH4 och CO2 till atmosfären. Trots att betydelsen av dessa utsläpp är tydlig är deras storlek och reglering fortfarande osäker på grund av brist på

flödesmätdata från sjöar. De flesta tidigare flödesmätningarna har utförts på sjöar <10 km2 _{och det har påvisats att extrapoleringar inte är direkt}

representativa för stora sjöar. Ny forskning har lett till ett mer allmänt erkännande av sjöars stora betydelse som källa till utsläpp. Trots detta väcker förhållandet mellan miljövariabler, sjöegenskaper, säsongsförändringar och variationen mellan och inom sjöar flera frågetecken. Storskaliga studier om växthusgaser behövs för att bestämma den rumsliga och tidsmässiga dynamiken som finns. I denna studie användes en kammarmetod och manuell provtagning för att undersöka spatiotemporal variabilitet och miljövariabler som kan påverka CH4 flöde och koncentration, samt upplöst oorganiskt

kol (DIC) och pCO2aq (partial trycket av CO2 i vattnet). Provtagningen genomfördes under fem veckor i september och oktober 2020 i tre stora

svenska sjöar. Våra resultat visade generellt varierande CH4 värden mellan de tre sjöarna, vilket indikerade att näringsämnen påverkar mängd och

utsläpp av CH4. Ett mönster noterades där CH4 var högre nära stranden och på ett grundare djup. Det fanns ett samband mellan CH4 koncentration

och miljövariablerna vindhastighet och luft- och vattentemperatur. DIC-värdena var höga i två av sjöarna och låga i en, alla sjöarnas DIC skiljde sig signifikant från varandra. pCO2 hade ingen skillnad inom sjöarna, och det fanns ingen skillnad mellan sjöarna utom i ett fall. Både DIC och pCO2

korrelerade med luft- och vattentemperatur. Studien visar den stora spatiotemporala variationen inom och mellan stora sjöar och att representativa värden för stora sjöar kräver fler mätningar under olika förhållanden för att urskilja hur växthusgaser emitterar och flödar mellan sjöar och atmosfär.

Abstract

Methane (CH4) and carbon dioxide (CO2) are two greenhouse gases and main drivers of global climate change. Lakes are known to be a source of

CH4 and CO2 to the atmosphere. While the importance of these emissions is clear, their magnitudes and regulation are still uncertain due to the

scarcity of flux measurement data from lakes. Most previous flux measurements have been carried out on lakes <10 km2_{and the extrapolations are}

not representative of large lakes directly. Recent research has led to a growing recognition of the great importance of lakes as a source of emissions. Still, the relationship between environmental variables, lake properties and seasonal changes and the variability between and within lakes raises several question marks. Larger scale studies of greenhouse gases are needed to determine the spatial and temporal dynamics that exist. In this study, a floating chamber method and manual sampling was used to investigate the spatiotemporal variability and influencing variables of CH4 flux and

concentration, as well as dissolved inorganic carbon (DIC) and pCO2aq (partial pressure of CO2 in the water). The sampling was conducted during

five weeks in September and October 2020 in three large Swedish lakes. Our results generally showed varying CH4 values between the three lakes,

indicating that nutrients affect the amount and emission of CH4. A pattern was found where the CH4 was higher near the shore and at a shallower

depth. There was a correlation between CH4 concentration and the environmental variables wind speed and air- and water temperatures. Our DIC

values were high in two of the lakes and low in one, all lakes’ DIC differed significantly from each other. The pCO2 did not have any difference

within the lakes, and there was no difference between the lakes except in one case. Both DIC and pCO2 correlated with air- and water temperature.

This study displays the large spatiotemporal variability within and between large lakes and that representative values for large lakes require more measurements under different conditions to distinguish how greenhouse gases emit and flux between lakes and atmosphere.

ISBN _____________________________________________________ ISRN LIU-TEMA/MV-C—21/11--SE _________________________________________________________________ ISSN _________________________________________________________________ Serietitel och serienummer

Title of series, numbering

Handledare Tutor

David Bastviken

Nyckelord

metan, koldioxid, upplöst oorganiskt kol, pCO2, kammarmetod, stora sjöar, metanbudget, koldioxidbudget, växthusgaser, sötvatten,

spatiotemporal, ebullition, diffusion, höst

Keywords

methane, carbon dioxide, dissolved inorganic carbon, pCO2, flux chambers, large lakes, methane budget, carbon budget, GHGs, freshwater,

spatiotemporal, ebullition, diffusion, autumn

Datum

Date 2021-06-02

URL för elektronisk version

http://www.ep.liu.se/index.sv.html

Institution, Avdelning Department, Division Tema Miljöförändring, Miljövetarprogrammet

Department of Thematic Studies – Environmental change Environmental Science Programme

(3)

2

Abstract

Methane (CH4) and carbon dioxide (CO2) are two greenhouse gases and main drivers

of global climate change. Lakes are known to be a source of CH4 and CO2 to the atmosphere.

While the importance of these emissions is clear, their magnitudes and regulation are still uncertain due to the scarcity of flux measurement data from lakes. Most previous flux measurements have been carried out on lakes <10 km2_{and the extrapolations are not}

representative of large lakes directly. Recent research has led to a growing recognition of the great importance of lakes as a source of emissions. Still, the relationship between environmental variables, lake properties and seasonal changes and the variability between and within lakes raises several question marks.

Larger scale studies of greenhouse gases are needed to determine the spatial and temporal dynamics that exist. In this study, a floating chamber method and manual sampling was used to investigate the spatiotemporal variability and influencing variables of CH4 flux and

concentration, as well as dissolved inorganic carbon (DIC) and pCO2aq (partial pressure of CO2

in the water). The sampling was conducted during five weeks in September and October 2020 in three large Swedish lakes.

Our results generally showed varying CH4 values between the three lakes, indicating

that nutrients affect the amount and emission of CH4. A pattern was found where the CH4 was

higher near the shore and at a shallower depth. There was a correlation between CH4

concentration and the environmental variables wind speed and air- and water temperatures. Our DIC values were high in two of the lakes and low in one, all lakes’ DIC differed significantly from each other. The pCO2 did not have any difference within the lakes, and there

was no difference between the lakes except in one case. Both DIC and pCO2 correlated with air- and water temperature.

This study displays the large spatiotemporal variability within and between large lakes and that representative values for large lakes require more measurements under different conditions to distinguish how greenhouse gases emit and flux between lakes and atmosphere.

Keywords: methane, carbon dioxide, dissolved inorganic carbon, pCO2, flux chambers, large

lakes, methane budget, carbon budget, GHGs, freshwater, spatiotemporal, ebullition, diffusion, autumn.

(4)

3

Sammanfattning

Metan (CH4) och koldioxid (CO2) är två växthusgaser och stora drivkrafter för globala

klimatförändringar. Sjöar är kända för att vara en källa för CH4 och CO2 till atmosfären. Trots

att betydelsen av dessa utsläpp är tydlig är deras storlek och reglering fortfarande osäker på grund av brist på flödesmätdata från sjöar. De flesta tidigare flödesmätningarna har utförts på sjöar <10 km2 och det har påvisats att extrapoleringar inte är direkt representativa för stora sjöar. Ny forskning har lett till ett mer allmänt erkännande av sjöars stora betydelse som källa till utsläpp. Trots detta väcker förhållandet mellan miljövariabler, sjöegenskaper, säsongsförändringar och variationen mellan och inom sjöar flera frågetecken.

Storskaliga studier om växthusgaser behövs för att bestämma den rumsliga och tidsmässiga dynamiken som finns. I denna studie användes en kammarmetod och manuell provtagning för att undersöka spatiotemporal variabilitet och miljövariabler som kan påverka CH4 flöde och koncentration, samt upplöst oorganiskt kol (DIC) och pCO2aq (partial trycket av

CO2 i vattnet). Provtagningen genomfördes under fem veckor i september och oktober 2020 i

tre stora svenska sjöar.

Våra resultat visade generellt varierande CH4 värden mellan de tre sjöarna, vilket

indikerade att näringsämnen påverkar mängd och utsläpp av CH4. Ett mönster noterades där

CH4 var högre nära stranden och på ett grundare djup. Det fanns ett samband mellan CH4

koncentration och miljövariablerna vindhastighet och luft- och vattentemperatur.

Våra DIC-värden var höga i två av sjöarna och låga i en, alla sjöarnas DIC skiljde sig signifikant från varandra. pCO2 hade ingen skillnad inom sjöarna, och det fanns ingen skillnad

mellan sjöarna utom i ett fall. Både DIC och pCO2 korrelerade med luft- och vattentemperatur.

Denna studie visar den stora spatiotemporala variationen inom och mellan stora sjöar och att representativa värden för stora sjöar kräver fler mätningar under olika förhållanden för att urskilja hur växthusgaser emitterar och flödar mellan sjöar och atmosfär.

Nyckelord: metan, koldioxid, upplöst oorganiskt kol, pCO2, kammarmetod, stora sjöar, metanbudget, koldioxidbudget, växthusgaser, sötvatten, spatiotemporal, ebullition, diffusion, höst.

(5)

4

Preface

First and foremost, we would like to express our gratitude towards our supervisor David Bastviken for your support, guidance and knowledge in this field - both during the field sampling and the essay process. Somehow, despite an overcrowded schedule, you managed to find time and energy to help us with great optimism. Also, a huge thanks to Ingrid Sundgren for analysing our samples in the laboratory and providing us with the GC results, as well as for always assisting us during the field sampling regarding materials, transportation and questions. Last, but not least, thank you to Malin Bohrn Nilsson for your assistance and company during the field sampling - our designated rower.

Anja Anderö Nordqvist & Veronica Bohlin Linköping 2021-05-24

(6)

5

1. Introduction

The greenhouse effect is the warming of the earth’s surface caused by the atmospheric content of greenhouse gases (GHGs). GHGs absorb infra-red radiation, thereby keeping heat longer in the atmosphere. The greenhouse effect and the gases are necessary for life on earth (Bernes, 2017, p. 24), but due to anthropogenic excessive GHG emissions, the amount of gas in the atmosphere escalates to levels that change the climate and the planet. Two GHGs that are key contributors to the greenhouse effect are methane (CH4) and carbon dioxide (CO2) (IPCC,

2013).

To prevent extensive climate change driven by GHGs, comprehensive knowledge is required and a useful tool for this is well-constrained budgets for CH4 and CO2..The global

budgets consist of several sources and sinks that describe the GHG cycles and their interactions and feedback loops in order to quantify fluxes (Reay, Smith, & van Amstel, 2010, p. 1-5). While the CO2 budget has been studied in detail for many decades, more and more attention is now

being paid to the earth's CH4 budget - partly because CH4 is a very powerful GHG despite being

in low concentrations, partly because it has increased faster than other GHGs in the atmosphere, and partly because this increase has been irregular in a way that cannot yet be explained, which indicates unknown sources and sinks (Friedlingstein et al., 2020; Saunois et al., 2020).

Only about a decade ago it became clear that lakes are one of the largest global CH4

sources (Bastviken et al., 2011), as confirmed by later studies (Saunois et al., 2020), this meant that the global CH4 budget must be revised. Currently, there are issues balancing the global CH4

budget because of uncertainties in lake emissions. Adding the estimated lake emissions led to higher total emissions than the increase that has been seen in the atmosphere. This needs to be researched and explained, and better knowledge of lakes is thus important for understanding the entire global CH4 budget (Saunois et al., 2020).

One of the major uncertainties is that there are significantly more studies from smaller lakes, which has constituted a demand for studies on lakes classified as large lakes (>10 km2₎

(Bastviken et al., 2004; Verpoorter et al., 2014). Large lakes have a total global area of over 2*106 km2, which is almost half of the total global lake area (Verpoorter et al., 2014). Another uncertainty in lakes is how the specific lake characteristics, for example amounts of nutrients, affects the fluxes from large lakes (Fernandez et al., 2020). Thereby, this scarcity of large-lake information has potentially led to biased estimates, e.g. of the 74 studied lakes in Bastviken et al. (2004) only 11 were larger than 10 km2. Studies of the larger and massive freshwater system

(9)

8 are needed to be able to generate representative values (Fernandez et al., 2020; Joung et al., 2019).

Being able to understand these fluxes will be crucial in making reasonable assumptions about how the climate will change in the future (Friedlingstein et al., 2020; Saunois et al., 2020). To contribute to the research area this study examines fluxes and concentrations of CH4 and

CO2 within and between three large lakes in Sweden. To broaden the information, the chosen

lakes have different characteristics regarding e.g., size, depth, nutrients and trophic classification. Considering that most previous lake studies were performed during summer, including previous studies on the selected lakes, this thesis field sampling is executed during autumn, to contribute information from additional seasons.

1.1. Aim and research questions

Due to the need to further develop this research area the aim of this thesis is to contribute more knowledge about large lakes emissions of CH4. Measurements in the autumn help to demonstrate the variability during the year. Since lakes contribute a lot of CO2 to the

atmosphere (Drake et al., 2018), it is also interesting to consider surface water CO2

concentrations (expressed as equilibrium partial pressures abbreviated pCO2aq) and overall

water concentrations of DIC, which both influence CO2 emissions. The research questions are:

- How does the CH4 fluxand concentration, DIC and pCO2aq vary within and between the

large lakes during autumn?

- Is there a correlation between depth, temperature or wind and the flux of CH4, pCO2

and the aquatic concentrations of CH4 and DIC?

2. Background

2.1. CH

4

2.1.1. Gas properties

CH4 is a colourless and odourless organic trace gas which is an important part of the

earth's GHG balance (Bastviken, 2009, p. 783). The atmospheric CH4 has more than doubled

since pre-industrial times and in 2018 the atmospheric methane reached 1857 ppb (Saunois et al., 2020). CH4 has a proportionately short lifetime in the atmosphere compared to other gases,

estimated to 8-12 years. This is because CH4 is oxidized by OH (OH + CH4 → CH3 + H2O) in

(10)

9 The concentration of CH4 is much lower than that of CO2 but the greenhouse warming potential

(GWP) for CH4 is about 28-30 times greater than for CO2, over a 100-year time horizon (Myhre

et al., 2013). Hence, CH4 makes a significant contribution of approximately 20% to the

increased greenhouse gas effect (Bastviken, 2009, p. 783).

The CH4 in the atmosphere consists of both anthropogenic and natural sources.

Anthropogenic CH4 emissions are emitted in the combustion and management of fossil fuels

(such as combustion in industries and natural gas leakage), agriculture, landfills and biomass combustion (Saunois et al., 2020). The natural fluxes that have been known for a long time include wetlands, volcanoes and other geological sources, wildfires, termites, wild ruminants, and certain marine environments (Raey, Smith & van Amstel, 2010, p. 4-12). In recent years, freshwater environments including lakes have been discovered as one of the largest natural sources (Saunois et al., 2020).

Most of the CH4 in the atmosphere is biogenic. The biological production of CH4 is the

final step in the biodegradation of organic matter (OM). The concerted action of multiple microorganisms first performs partial degradation of the OM resulting in a range of degradation products including acetate, CO2 and molecular hydrogen (H2), which can subsequently be used

by methanogenic archaea to produce CH4, i.e. perform methanogenesis (Bastviken, 2009, p.

783). There are three well-studied pathways formethanogenesis: acetoclastic, methylotrophic and hydrogenotrophic (Fenchel, Blackburn & King, 2013, p. 19-20).

Acetoclastic methanogenesis is based on dissimilation of acetate (CH3COOH) through

fermentation (Fenchel, Blackburn & King, 2013, p. 19-20; Wik, 2016). This process is common in freshwater sediments and anaerobic digesters. Methylotrophic methanogenesis is important in some marine sediments and other anoxic systems where methylated substrates are available (Fenchel, Blackburn & King, 2013, p. 19-20). The third pathway for CH4 formation,

hydrogenotrophic methanogenesis, is the reduction of CO2 to CH4 with molecular hydrogen

(H2) (Fenchel, Blackburn & King, 2013, p. 19-20; Wik, 2016). Increased temperatures lead to

an increased CH4 production as the optimum temperature for methanogenesis is higher than the

current global temperatures. Increased temperatures by 10 °C have been estimated to increase methanogenesis approximately fourfold (Bastviken, 2009, p. 783-786).

2.1.2. CH

4

flux from lakes to the atmosphere

Globally, freshwater lakes have long been suggested as a natural source of atmospheric CH4 (Bastviken et al., 2004; Ehhalt et al., 1974; Smith & Lewis, 1992) but the

(11)

10 extent of its importance has gone from a smaller source of emissions to a key source in the last decade (Bastviken et al., 2011; Saunois et al., 2016) with annual global emissions estimate of 103 Tg of CH4 year-1 (Bastviken et al., 2011). Despite this, there is great uncertainty in fluxes

from inland water systems and there is a great need for representative data that can quantify the fluxes according to how they are regulated (Saunois et al., 2020).

In freshwater lakes methanogenesis is the dominant degradation, i.e., produce CH4

(Bastviken, 2009, p. 804). Currently, the traditional fact that the biological production of CH4

takes place in a strictly anaerobic environment is questioned. The discovery that CH4 can be a

by-product of photosynthesis in oxygenated water columns by algae or cyanobacteria, have created a scientific debate. If recognized, it will be of great importance (Günthel et al., 2020; Khatun et al., 2019).

A large proportion, 50->95%, of the CH4 produced in lake sediment can be oxidized.

CH4 can be oxidized to CO2 by both aerobic and anaerobic CH4 oxidation. Aerobic oxidation

generally dominates and occurs mainly where CH4 from the oxygen free environment meets O2

in the surface sediments or in the water column. In the presence of O2, CH4 oxidizes and energy

can be obtained by the aerobic CH4 oxidizing bacteria (MOB). MOB uses CH4 as a source for

energy, and carbon and O2 as electron acceptor (Bastviken, 2009, p. 786). Consequently, large

parts of the CH4 fluxes are controlled by the magnitude of the oxidation process (Bastviken,

2009, p. 788).

The release of CH4 into the atmosphere from freshwater ecosystems can occur in at least four scientifically recognized ways, through ebullition, diffusive fluxes, by release of stored CH4 and flux through aquatic plants (Figure 1) (Bastviken, 2009, p. 792-796).

(12)

11 Figure 1. The release of CH4 from a lake into the atmosphere (modified from Bastviken,

2009, p. 796). Ebullition

The first pathway is through ebullition, easily explained as bubbles that are released from the sediment and move quickly up through the water column. The released bubbles pass so rapidly they thereby largely escape CH4 oxidation. Ebullition occurs to a greater extent from

shallower water where the hydrostatic pressure is lower (Bastviken, 2009, p. 792). Since most of the world's lakes are shallow, ebullition is a very common process (Downing et al., 2006). It has been established that ebullition and plant fluxes account for a range in between 50%->90% of the CH4 emissions from all small and shallow water sources (Saunois et al., 2020).

The hydrostatic pressure has been shown to be affected by weather and a higher incidence of ebullition has occurred during storms and frontal passages. Deeper sediments are not exposed to the same disturbances or turbulence, which reduces the probability of bubbles being released (Bastviken, 2009, p. 792). There can however be high ebullition from deep sediments, where steep sediment slopes lead to transport of large amounts of sediment to small deep areas, e.g., where the large input of new substrates fuel extensive CH4 production and

ebullition (Sobek et al., 2012).

Ebullition as a release pathway is technically challenging to measure since it has a high spatiotemporal variability. This may have led to significant underestimations of CH4

(13)

12 section) has long been a major constraint in estimating the total fluxes from freshwater lakes (Bastviken et al., 2011), and the need for measurements of other emission pathways is desirable (Saunois et al., 2020).

Diffusion

A much slower transport of dissolved CH4 takes place through diffusive fluxes. This

process is regulated by the differences in concentrations of CH4 in surface water and air.

Therefore, it is continuously ongoing since the CH4 concentration in the water is usually higher

than the concentration in the air. The gas is transported from the sediment through the water column through advective processes and/or turbulence enhanced diffusion and then from the water to the atmosphere by diffusion across the water-atmosphere boundary layer. Since the diffusion process is slower, much of the dissolved CH4 is oxidized before it reaches the surface

(Bastviken, 2009, p. 792-796).

The diffusive CH4 flux is regulated by turbulence. It passes through a boundary layer

consisting of a thin surface of water molecules where the thickness determines how fast the gas can pass (Gålfalk et al., 2013). At high turbulence in the surface water the process increases because more water is meeting air (Bastviken, 2009, p. 795). Studies have shown a link between increased wind speeds and higher CH4 emissions, as a cause of turbulence (Jansen et al., 2019).

The boundary layer can be thinned during turbulence such as when waves are created by wind and thus increase gas exchange (Gålfalk et al., 2013).

The velocity of the gas exchange is called k, often termed piston velocity. The piston velocity is a way to represent the physical rate of gas exchange with the atmosphere (Gålfalk et al., 2013). Turbulent diffusion transports mass in the direction of decreasing chemical concentration and is a driving factor in the diffusive fluxes (Hemond & Fechner-Levy, 2000 p. 15-16). When the gas reaches the surface water by turbulent diffusion, the exchange with the atmosphere is driven by the difference in concentration and k. By Fick's law of diffusion, the exchange can be calculated by this equation:

F= k (Cw – Cair), (Eq 1)

where F is the flux, k is the piston velocity (Cole et al., 2010), Cw is the measured gas

concentration in the water, Cair stands for the chemical concentration in the water given

equilibrium with the atmospheric partial pressure (often calculated from Henry's law) (Gålfalk et al., 2013). The k represents at which speed the gas is transported across the boundary layer and depends mainly on the water turbulence (wind). Models therefore often calculate k from

(14)

13 wind speed. The equation part (Cw – Cair) represents the concentration gradient, and when Cw

is equal to Cair, there are no fluxes. If Cw is greater than Cair, there is a flux towards the

atmosphere. Conversely the water can take up gas if Cair > Cw (Cole et al., 2010).

Release of stored CH4 and through aquatic plant flux

In stratified lakes, emissions of CH4 can occur when the water is mixed between

different layers. This storage flux release occurs when the deep anoxic and CH4 rich water

comes up to the surface and results in a rapid diffusive flux release, see Eq 1. Aquatic plants can transport air to the roots to supply them with O2. At the same time, they also can transport

gas in the opposite direction and can act as a shortcut for CH4 from the roots and directly to the

atmosphere without passing through the areas of oxidation (Bastviken, 2009, p. 795-796).

2.2. CO

2

2.2.1. Gas properties

CO2 is an invisible and odourless GHG that is essential to life on earth. The gas is

produced by oxidation of carbon compounds by e.g. cellular energy metabolism and combustion of OM including fossil fuels (Volk, 2008, p. 1-3, 28). It is the most common GHG and is very soluble, compared to e.g. CH4 (which saturation is about 1.6 mol m−3 at 20 ◦C).

CO2has a molar solubility in freshwater of 39 mol m-3 at 20 °C (Casper et al., 2000).

CO2 and its role in the planet's carbon budget have been studied intensively in recent

decades and this has been extensively summarized in the IPCC's reports (Ciais et al., 2013). It is a part of the carbon cycle and when released the gas circulates through the biosphere. Dominant processes for the carbon cycle include photosynthesis and respiration and the solubility equilibrium between the atmosphere and the oceans. Big amounts are absorbed by the ocean or stored in the biomass and soil (Volk, 2008, p. 1-3).

The concentration of CO2 in the atmosphere has increased from approximately 277 ppm

in 1750, to estimates of 410 ppm in 2019. Since the industrial revolution, human activities have greatly modified the exchange of carbon. The emission of fossil fuels is the main reason for the disturbed balance of the natural carbon cycle and the main sources to the net atmospheric increase (Friedlingstein et al., 2020). Although a large part of the increase is due to human activity, emissions from natural sources such as the inland waters and land are also increasing, but there are many uncertainties (Battin et al., 2009).

(15)

14

2.2.2. CO

2

flux from lakes to the atmosphere

As with CH4, lakes were not previously considered significant for the global carbon

budget, but several studies in the last decade have shown the opposite and there are now many indications that most freshwater environments are supersaturated relative to the atmosphere of CO2. Freshwater environments emit CO2 in amounts comparable to ocean carbon uptake, even

though CO2 is easily soluble in water and is mainly emitted via diffusive flux. Similar to CH4,

the measurements concerning CO2 in lakes have most often been performed in small systems

(Drake et al. 2018).

Inland waters play a big role in the processing and transport of organic carbon (Battin et al., 2009). The systems are a source of emissions but also as a CO2 sink, thus lakes have the ability to bury CO2in their sediment. The CO2 burial in lakes is small compared to the CO2 emissions, but it is still a burial rate that is greater than the ocean and represents a significant long-term carbon-decrease (Aufdenkampe et al., 2011; Mendonça et al., 2017).

Carbon compounds in lakes can be either organic or inorganic and consist of two main fractions; dissolved organic carbon (DOC) and particulate organic carbon (POC). Up to 90% of the total organic carbon (TOC) consists of DOC (Tranvik & von Wachenfeldt 2009, p. 754). DOC in inland water ecosystems can either originate from autochthonous primary production within the system, or from allochthonous production in the catchment reaching the lake through external inputs. The external sources, composed of terrestrial materials, results in a surplus of CO2 (Tranvik et al., 2009) and the largest share of this in lakes is in the form of DOC (Wetzel, 2001, p. 739-740). Globally it has been estimated that 5.1 Pg C is delivered to inland waters annually from terrestrial landscapes and 3.9 Pg is annually outgassed to the atmosphere as CO2

(Drake et al., 2018).

Mineralization is the process where organisms use the DOC as fuel in the cellular metabolism to DIC or CH4 (Gudasz et al., 2010). Mineralization is a complex process that is

affected by several factors such as the amount and quality of DOC, availability of electron acceptors, trophic conditions, depth, temperature, and pH (Cardoso et al., 2019). Temperature is a relevant factor as warmer water leads to more mineralization and less burial (Gudasz et al., 2015). Some of the DOC escapes mineralization and stays in the sediment to be stored (Gudasz et al., 2015; Mendonça et al., 2017).

The DIC consists of three constituents: dissolved CO2, the bicarbonate ion (HCO3−) and

the carbonate ion (CO32 −). The most important regulators of DIC in the lakes are the soils and

(16)

15 between photosynthesis and respiration changes, which creates variability over the day and with the seasons (Cole & Prairie, 2009, p. 30-32). In the surface waters DIC exchanges with CO2 in

the atmosphere via diffusive fluxes (see section 2.1.2. and Eq 1). Most of the world's lakes have a higher concentration of surface-water CO2 than can be explained by equilibrium. Such systems are said to be ‘oversaturated’ in CO2 with respect to the atmosphere. If the surface water is oversaturated with CO2, it becomes a net source for CO2 to the atmosphere(Cole &

Prairie, 2009, p. 30-32).

By measuring pCO2aq (the partial pressure of CO2 according to Henry’s law

corresponding to the concentration in water) a good insight into the supersaturation is given. The CO2 flux is often estimated from surface water concentrations in aquatic environments

(pCO2aq). An important regulator for pCO2aq in inland waters is the aquatic primary production

(Nydahl, 2020). Photosynthesis causes the CO2 concentration to be reduced and the pH to

increase, when respiration dominates at night the concentrations increase and pH decrease. This is tightly linked to the carbonate system which is partially driven by changes in pH (Cole & Prairie, 2009, p. 30-33).

2.3. Spatiotemporal variability of CH

4

and CO

2

There are many factors that contribute to the large spatial and temporal emission variability that exist between and within lakes. The seasonal variability in weather that regulates temperature, wind speed and precipitation has been recognized for being important (Fernandes-Sanches et al., 2019; Natchimuthu et al., 2017). Short term variabilities in the CH4 fluxes that

change rapidly have been seen to be driven by wind speed in other lake ecosystems (Jansen et al., 2019). The lake characteristics, including environmental conditions (e.g. location regarding climate zone and proximity to industries, agriculture or surroundings with a lot of organic material, and size - both depth and area) and climate related variables (e.g. temperatures and wind), have been shown to have a major impact on CO2 concentrations in the water and the

CH4 production, which result in spatial differences (Fernandes-Sanches et al., 2019;

Weyhenmeyer et al., 2015).

Climate zones with higher mean air temperature have reported generally higher emission rates (Fernandes-Sanches et al., 2019), since it increases methanogenesis ( Yvon-Durocher et al., 2014). The increase of freshly produced organic substrates from production of algae or aquatic plants in nutrient-rich sites appears also to increase methanogenesis and lead to higher emissions of CH4 (Fernandez et al., 2020; Fernandes-Sanches et al., 2019).

(17)

16 The lake size has also proved to be important partly because smaller lakes have less residence time (Bastviken et al., 2004), partly because there is a greater sediment‐to‐lake volume ratio (Natchimuthu et al., 2017). This possibly makes ebullition and sediment disturbances more common in smaller lakes as they are often also shallower (Bastviken, 2009, p. 792), i.e. less CH4 can be oxidized (Bastviken, 2009, p. 792). Large lakes are fewer in

number, but their total area reaches approximately half of the global lake area (Verpoorter et al., 2014). Although the fluxes per area are lower than observed in smaller lakes, this makes large lakes a big source of GHGs (DelSontro et al., 2018; Fernandez et al., 2020).

3. Method

As in Sawakuchi et al. (2014), a floating chamber method was used to enable measurements of CH4 flux, where the chambers drifted after the boat on open water. The drift

with the boat allowed the chambers to move with the water minimising risk for extra turbulence and enable continuous check of chamber integrity and reduce risk of lost samples due to flipping chambers – which is more likely on large and more wind-exposed lakes than on small lakes if chambers are moored. The aim was to measure fluxes on all three lakes during as similar weather and temperature conditions as possible – therefore selecting to visit different lakes adjacent days during a five-week period.

Sampling could not be performed in strong wind, for safety reasons, so the schedule was updated continuously according to weather forecasts. The dates sampling was conducted, and at what lake, is presented below in Table 1. A total of 13 sampling days were conducted. The plan was to sample three out of five days a week, which seemed realistic considering weather conditions, and initially it was. However, the later in the autumn, the more difficult the sampling became because of strong wind and cold temperatures in both air and water. Therefore, during the fifth week, only the calmest day was chosen for sampling. Some days were cut short because of unstable weather conditions and shorter daylight, and because of that the amount of data varies slightly from time to time.

(18)

17

Table 1. Table of which lakes were sampled at which dates during the autumn of 2020.

Lake Week 1 Week 2 Week 3 Week 4 Week 5

Glan Sep 24th _{Sep 30}th _{Oct 7}th _{Oct 14}th _-

Roxen Sep 21st _{Sep 28}th _{Oct 5}th _{Oct 15}th _{Oct 21}st

Vättern Sep 23d _{Sep 29}th _{Oct 6}th _{Oct 12}th _-

3.1. Study sites

The lakes selected (Figure 2) for this thesis are freshwater lakes classified as large lakes by having an area >10 km2 (Verpoorter et al., 2014). Furthermore, the lakes have different hydrological and trophic characteristics and represent different types of large lakes. This is a great advantage as more variability is covered and can to a greater extent demonstrate how environmental conditions regulate the fluxes.

(19)

18

3.1.1. Glan

Based upon the trophic classes in Carlson & Simpson (1996) and the phosphorus concentrations from VISS (n.d.,a), Glan is classed as eutrophic. Glan has an area of 73 km2_.

The average depth is 9.9 m, and the maximum depth is 22.8 m. It has a volume of 730.2 million m3_{and is located in Norrköping and Finspång municipalities (SMHI, n.d.).}

The lake was accessed through a boat ramp into the river mouth of Motala stream outside of Skärblacka, the sampling was performed further out in the south part of the lake. The surrounding environment consisted mainly of forest, some agriculture, and settlements. Figure 3 shows the sampling locations of different dates.

Figure 3. Map over dates and sampling locations in Glan.

3.1.2. Roxen

Based on trophic classes and the lake's phosphorus concentrations, Roxen is classed as mesotrophic (Carlson & Simpson, 1996; VISS, n.d.,b). It has an area of 95 km2. The average depth is 4.8 m and the maximum depth is 8 m. It has a volume of 457.7 million m3 and is located in Linköping and Norrköping municipalities (SMHI, n.d.).

The lake was accessed through a boat ramp into the river mouth of Stångån in Linköping, the sampling was performed further out in the southeastern part of the lake. The

(20)

19 surrounding environment consisted of open landscapes, agriculture and the city of Linköping including both factories and residential areas. There were always plenty of birds on Roxen. Figure 4 shows the sampling locations of different dates.

Figure 4. Map over dates and sampling locations in Roxen.

3.1.3. Vättern

Vättern is classed as an oligotrophic lake (Carlson & Simpson, 1996; VISS, n.d.,c). It has an area of 1 756 km2, which makes it Sweden's second largest lake. The average depth is 40 m and the maximum depth is 120 m. It has a volume of 77 604 million m3 and is locatedin southern Sweden between the counties Östergötland and Västergötland. In the north it borders Närke county and in the south to Småland county (SMHI, 2021).

The lake was first accessed through a boat ramp in Hästholmen harbor in Ödeshög municipality but due to the need for calmer water, the access point was changed to Vadstena harbor in Motala municipality, where a large bay made sampling possible. The measurements were mainly conducted in the lake's northeast area and the surrounding environment consisted largely of open landscapes, rocks and deciduous trees. Figure 5 shows the sampling locations of different dates.

(21)

20

Figure 5. Map over dates and sampling locations in Vättern.

3.2. Sampling

3.2.1 Floating chambers

Floating chambers were used to measure the CH4 flux across the air-water interface. The

method is a well-established and common way to measure GHG flux from open water (Bastviken et al., 2004, 2010, 2011; Cole et al., 2010; Gålfalk et al., 2013; Jansen et al., 2019, 2020; Podgrajsek et al., 2014; Sieczko et al., 2020; Wik et al., 2016) and has been verified to not bias flux measurements (Cole et al., 2010; Gålfalk et al., 2013). Floating chambers is also an inexpensive method (in equipment) that allows replication and yields information about spatial variability by simultaneous measurements and multiple locations (Bastviken et al., 2015). This method was performed similarly to Sawakuchi et al. (2014) and was for this study the most robust way to perform measurements and covers a larger total area but provides less measurement time at each location.

(22)

21 The chambers were round plastic containers covered with aluminium tape to reflect light and minimize internal heating and were equipped with Styrofoam rods for flotation (Figure 6) (Gålfalk et al., 2013). There were two different

sizes of chambers used: the area and volume were 0.075 m2 and 7800 ml, and 0.075 m2 and 8600 ml. The chambers were placed carefully on the water with strings of 1-2 m between each other (Figure 7). With 10 chambers behind the boat the equipment (including the boat) measured at least 20 meters. The drift was done with the engine switched off. During the drift, water samples for

concentrations of CH4 and DIC were collected. Meanwhile, inside the chamber the CH4

concentration changes over time in proportion to the gas exchange across the water surface. The headspace samples were taken after 2-3 hours of drifting, chamber sampling details are developed in section 3.2.5.

Figure 7. Drifting chambers.

Moored transects (anchored, i.e., not drifting) for chambers were intended to be placed out to investigate more of the spatial and temporal variability between and within the lakes. The moored transect would have given measurements on the shallower depth and closer to the shoreline, after strategic placement. According to Cole et al. (2010) transect can be used to capture heterogeneity in the turbulence of the water which can affect the process of ebullition. This type of sampling was found not to work due to weather conditions. Therefore, this study focuses on drifting chambers.

(23)

22

3.2.2. Vial management

All samples were transferred to 20 ml glass vials, capped with a natural rubber cap that sealed airtight, kept in place with an aluminium crimp-seal cap. In the field all vials were marked with relevant information: lake, date, time, and type of sample, so that the samples could be linked to the correct information. The vials were sealed in the lab prior to sampling with lab air inside. The needles used to inject samples through the rubber were BD Microlance 0.4 x 13 mm injection needles. When inserting needles through the rubber cap, the aim was always to hit the thicker part of the rubber cap, avoiding the thinner middle part. When removing needles, a thumb grip was held firmly on the cap for at least 10 seconds to give time for the rubber to seal the hole after the removed needle. The

vials with liquid samples (see below) were stored upside down, making the liquid function as an extra gas barrier in the vial. All syringes used were fitted with 3-way Luer-Lock valves to enable an airtight closing mechanism to the syringes. The 3-way lock is X-shaped with one rod shorter than the other three, the short rod has no exit (hence, the direction of the short rod is locked). For instance, in Figure 8 the airway through the needle and into the barrel is open, and the third exit (out into the air, to the left) is closed. If set askew, all three paths through the valve are closed. Note that even though the locking mechanism has four rods, there are three valves available (hence the name).

3.2.3. Background sample

This sample was taken to represent background conditions when the chambers were placed in the lake, and is later used to calculate the chamber headspace. Three 60 ml syringes were first held above our heads, to avoid breathing air, and rinsed with air three times, to avoid contamination. The syringes were then brought approximately 10-15 cm above the water surface and filled with sample air. The sample was transferred to a vial using two 10 cm long Ø 5 mm (outer diameter) polyurethane extension tubes. These had a 3-way valve in one end, and a standard bore needle fitting in the other. Both extension tubes were fitted with needles. Both injection needles were inserted in a vial, with the 3-way valves on the extension tubes still closed. One of the syringes containing sample air was fitted to one of the extension tubes, all

Figure 8. Syringe with 3-way Luer-Lock valves.

(24)

23 3-way valves were opened, and the vial was flushed with the sample. The 3-way valves on the extension tubes were closed, and the syringe removed. The same procedure was done for the second and third syringe. Which was done to ensure the vial only contained sample air. As the vials were 20 ml, this procedure involves rinsing the vial eight times its own volume. When the third syringe had approximately (≥) 5 ml sample left, the 3-way valve on the outflow extension tube was closed, and the remaining sample was pushed into the vial, creating approximately 5 ml overpressure inside the vial. The overpressure was created to ensure sample integrity, in case of leakage, and make possible leaks go out instead of air leaking into the vial contaminating the sample. The 3- way valve on the other extension tube was closed, and the extension tubes removed from the vial. The background sample was collected before and after chamber deployment.

3.2.4. Open-air sample

The open-air sample was retrieved in the same manner as the background sample, with one exception, the air was collected from the open air approximately 5 dm above our heads. The syringes were held above our heads to avoid contamination by exhalation. The open-air sample was collected before and after both chamber deploying and chamber elevation.

3.2.5. Flux chamber headspace sampling (CH

4

flux)

The headspace sample was treated in the same manner as the background sample with one exception, it was taken from the headspace of the flux chambers. A similar procedure is also described in Sieczko et al. (2020). The valve on the 60 ml syringe was connected to the 3-way valve on the sampling tube from the chamber, the syringe was filled and emptied with headspace gas phase three times to create turbulence inside the chamber with the goal to sample a homogeneous gas mixture from the chamber. The syringe was then filled a fourth time to collect the actual gas sample, both 3-way valves were closed to prevent contamination from the surrounding air, the syringe was removed, and the two remaining syringes were filled. Transfer to vial was performed the same way as for the above water-air sample. The headspace sample was collected at chamber elevation.

3.2.6. Water concentrations of CH

4

and DIC

The water sample was collected using a 10 ml syringe with a 3-way valve. The syringe was brought below the water surface, where approximately 7 ml of water sample was retrieved. The syringe was lifted in an upright angle (needle up, plunger down) and all water bubbles were

(25)

24 removed by firmly knocking on it. Once the bubbles were gone, the syringe was again brought below the water surface. The water was pushed out away from the sampling area, and without lifting the syringe above the surface, slightly more than 5 ml sample was collected. When brought above the surface the sample was adjusted to precisely 5 ml. For the water samples vials prepared with phosphoric acid and an overpressure of N2 were used. To transfer the sample to the vial a needle was first stuck in a vial, this to ensure the vial contained, and released, overpressure. The needle was then fitted directly on the valve, the valve opened, and the sample was pushed into the vial. The 3-way valve was closed, and the syringe removed. The water sample was collected before and after both chamber immersion and chamber elevation, and every 30th minute during drift.

3.2.7. pCO

2aq

The pCO2aq sample was collected using a 140 ml syringe with a 3-way valve and a 60

ml syringe with a 3-way valve. The 140 ml syringe was brought below the water surface, where approximately 110 ml of water sample was retrieved. The syringe was lifted in an upright angle (needle up, plunger down) and all air bubbles were removed by firmly knocking on it. Once the bubbles were gone, the syringe was again brought below the water surface. The syringe water was pushed out from the syringe, away from the sampling area, and without lifting the syringe above the surface, approximately 105 ml water sample was collected. This was done to avoid air entering the syringe. When brought above the surface the sample was adjusted to precisely 105 ml.

Next 35 ml open air was carefully pulled into the syringe, avoiding contamination of breath. The 3-way valve was closed, and the syringe was then shaken for 3 minutes to achieve equilibrium of CO2 between the water and the gas phase in the syringe. To transfer the gas

phase sample to a vial, the 140 ml syringe was held in an upright angle (plunger down, needle up) as the 60 ml syringe was connected to its valve. The 30 ml air sample was carefully retrieved by pulling the gas phase from the 140 ml syringe into the 60 ml syringe, this way the sample has overpressure and protection from leakage. A needle was attached to the second syringe so that the sample could be transferred to a vial.

Since there is only one syringe with 30 ml sample – which is not enough for flushing the sample vial as for the above gas samples – the sample vial was evacuated prior to

transferring the sample to it, so that no sample was wasted. This was done using another syringe

(26)

25 between the inside of the vial and syringe. When the syringe was pulled out, gas from the vial moved into the syringe creating under-pressure. When the syringe was fully protracted the 3-way valve on the extension tube was turned, allowing the gas to be expelled from the syringe without any gas getting into the vial (i.e. under-pressure remained in the vial, but the syringe was emptied) (Figure 9). Then the valve was turned to the first position and the syringe plunger was pulled out again removing even more of the remaining gas in the vial.

This process was repeated 3 times, meaning that 180 ml of syringe space was used to remove gas from 22 ml vial space, and hence very little gas remained in the vial after this evacuation process and the 30 ml sample could then be added to the vial with negligible dilution with original gas in the vial. The pCO2aq sample was collected before and after both chamber

deployment and chamber elevation.

Figure 9. Step by step sample vial being evacuated from gas, creating under-pressure.

3.2.8. Air- and water temperature, wind, coordinates, depth and air pressure

Beyond the air and water gas samples, background information was also collected. Depth, coordinates, wind speed, air temperature and water temperature were collected in connection with the samplings. Depth was measured with a digital, portable hand-held Echotest II sonar depth gauge, water temperature was measured with a kitchen thermometer and air temperature and wind speed was measured with a Uni-T UT363 anemometer and thermometer. The air pressure during the sampling days was collected afterwards from the Swedish Meteorological and Hydrological Institute which receives high-resolution real data from the national Swedish weather monitoring (smhi.se). The weather station was chosen as close as

(27)

26 possible to the sampling points. These parameters were needed for calculations and processing of the collected data.

3.3. Analysis

The collected samples were analysed at the laboratory at Linköping’s University by a Gas Chromatograph (GC) Aglient 7890A with a 1,8 m * 3.175 mm Porapak Q 80/100 Supelco column and a flame ionization detector (FID) for CH4 and a thermal conductivity detector

(TCD) for CO2, from Agilent Technologies, USA. The used carrier gas was helium. Before

analysis, all samples, and standards (10 ppm CH4, 2000 ppm CO2; single-point standards

adequate as verified by previous multi-point standard curves) were allowed to reach a similar temperature. Identical pressure was assured by pressure equilibration to ambient barometric pressure via controlled gas release from the vials through a needle. Preparations before the analysis was to temper all samples and gas standards (10 ppm CH4, 2000 ppm CO2;

single-point standards adequate as verified by previous multi-single-point standard curves) into the same temperature. Then the overpressure created in the field was removed so the pressure was equalized, not all samples had an intact overpressure.

3.3.1. GC separation - CH

4

and CO

2

The GC is an analytical technique used to separate the chemical components of a sample mixture to detect and determine how much is present. These chemical components are usually organic molecules or gases. To separate in a gas chromatograph a sample solution is injected into the instrument, which then enters a gas stream which transports the sample into a separation tube known as the "column." The various components are separated inside the column based on differences in “stickiness” to the surfaces in the column – more stickiness means longer transport time through the column. The detector measures the quantity of the components that leave the column. To measure a sample with an unknown concentration, a standard sample with known concentration is injected into the instrument. The standard sample is compared to the test sample to calculate the concentration (Bydén et al., 2003). The method is particularly suited to mixtures of gases. The sample is heated during the analysis to reduce the less volatile components. The remainder is transported to the detector which responds to the chemical components eluting from the column to produce a signal. The signal gives a result that is proportional to the amount of substance and is registered by the computer (TN, 2021).

(28)

27

3.3.2. CH

4

detection: Flame Ionization Detector (FID)

A FID produces a small flame from the reaction of two compounds, hydrogen and oxygen, from the air. The mobile phase carrier gas is not affected by this flame, but when a component of the analyte reaches the flame, it loses an electron and becomes ionized. These electrons create an electric current that is amplified and sent to a computer for display. The extent of the current is proportional to the amount of substance present. It is a non-selective detector which responds to almost all organic compounds. Although, if there is separation beforehand then the detection becomes selective based on time after injection when the sample reaches the detector (Houck & Siegel, 2010).

3.3.3. CO

2

detection: Thermal Conductivity Detector (TCD)

The TCD is used to detect changes in thermal conductivity of the column eluent, comparing it to a reference flow of carrier gas. The detector has two parallel tubes, with heating coils, the reference gas is in one of the tubes and the sample gas passes through the other tube. The TCD can detect all molecules and is often used for CO2. The thermal conductivity, when

different gases pass the detector, affects a voltage field which gives an electric signal in proportion to deviations in the heat conduction, i.e., how much gas passes (Sevcik, 2011).

3.4. Calculations

Initially all information from the field protocols and the raw data from the GC analysis were compiled in an Excel-file. Prior to all calculations (CH4 flux, aquatic concentrations and

pCO2aq) the following units are converted: Celsius at extraction in the field was converted to

kelvin, partial pressure from Pa to atm and the Henry's law constant is calculated (M/atm) at the current temperature (Weiss, 1974; Wiesenburg & Guinasso, 1979).

For the sample types (except CH4 flux, and samples collected during drift) duplicate

samples were taken at each sampling time, and the mean from the duplicates were used in the calculations (unless any of the samples had to be discarded due to e.g. obvious contamination by breath or leakage). The mean value from these samples was obtained in ppm.

3.4.1. CH

4

flux

The GC results in ppm were converted to µatm through multiplication of the air pressure (atm) with the conversion constant 9.869*10-6 (conversion from Pa to atm) - in this step hPa is also converted to Pa by multiplying with 100. Which then was multiplied with the CH4 ppm value of the background sample and the chamber sample, respectively:

(29)

28 (hPa*100*0.000009869) * CH4 ppm air/chamber, (Eq 2)

Then the background sample is subtracted from the final sample (chamber) to produce the difference. Chamber volume measured in ml was converted to litre by dividing by 1000.

The change in amount of compound (n) for the initial and final values were calculated with the common gas law:

PV = nRT, in this case: 𝛥n=𝛥P*V / R*T, (Eq 3)

Where 𝛥P is the change in gas partial pressure (atm), V is the gas volume (l), n is the amount of compound (mol), R is the common gas constant (0.082056 L atm K-1 mol-1) and T is the water temperature (K). 𝛥n were then converted to mmol. The next step is dividing𝛥n with the chamber area (m2) and with the time the chambers drifted (day), which results in the CH4 flux

in the unit mmol/m2/d. Following this equation:

𝐹 = 𝛥𝑛 / (A∗𝛥𝑡), (Eq 4)

F is flux, Δn is the CH4 change in mmol, A is chamber area (m2) and Δt is the flux chamber

deployment time, in days.

3.4.2. Aquatic concentrations of CH

4

and DIC

The GC results in ppm were converted to µatm (Eq 2). No correction for the background presence of CH4 or CO2 was needed because the background gas was N2. The gas from the

water was distributed partly in gas phase and partly in water phase. Based on analysis of the gas phase, the number of moles in the gas phase was calculated via the general gas law, and the number of moles left in the liquid phase via Henry's law. Then these two quantities were summed to get the total amount in the original water sample.

The amount of compound in gas form is calculated with the common gas law:

n in headspaceg (µmol) = ng = (Px * Vg) / (R * T), (Eq 5)

Where Px is Ptot*(ppm/106) e.g. the partial pressure of concerned gas (atm), Vg is the headspace

volume and T is the water temperature. The amount of compound in aquatic form is calculated by:

(30)

29 Px is Ptot*(ppm/106) (atm), Vaq is the water sample volume and KH is the selected Henrys’ law

constant dependent on the gas and temperature (Weiss, 1979; Wiesenburg & Guinasso, 1979). In the last step, the total n in the vessel (µmol) is calculated by addition (Eq 7). Then the concentration (µM) is calculated by dividing the total n (subtracted with the amount of background in vial, nbkg, in this case zero) with the water sample volume (l) (Eq 8):

Total n in vessel (µmol) = ntot = ng + naq, (Eq 7)

The concentration in sample (µM) = Caq = (ntot – nbkg) / Vaq, (Eq 8)

3.4.3. pCO

2aq

The GC results in ppm were converted to µatm (Eq 2), which results in CO2 in the 35

ml extraction syringe headspace. The DIC concentration (µM) (Eq 9) is used to compute the total IC in the syringe, by multiplying DIC with the water volume. The next step can be seen in Eq 6 and Eq 7. The total IC is calculated by adding the IC in the water sample (C*V) with the CO2 in the headspace (from common gas law), which is the basis for Eq 9. Followed by these

equations:

Total IC (in 140 ml extraction syringe) = DIC (retrieved from Eq 8) * Vsample + PCO2air *

Vheadspace / (R * T), (Eq 9)

CO2headspace = PCO2headspace * Vheadspace / (R * T), (Eq 10)

CO2aq after extraction = CO2aq-ex = PCO2headspace * KH, (Eq 11)

DIC after extraction = DICex = Total IC – CO2headspace, (Eq 12)

The temperature corrected carbonic acid constants (Ka1 and Ka2) of the carbonic acid system is computed, after which these equations are used (the hydrogen ion concentration [H+], expressed as H in below equations, is the wanted factor):

(DICex - CO2aq-ex) * H2 - CO2aq * Ka1 * H - CO2aq * Ka1 * Ka2 = 0, (Eq 13)

DIC / CO2aq = 1 + Ka1 / H + Ka1 * Ka2 / H2, (Eq 14)

Eq 13 produces the pH-values via H (H = 10-pH_{) by iteration or resolving the quadratic equation,}

(31)

30 carbonic acid system equilibration equation, which uses the pH to calculate the CO2aq (µM)

from DIC before the extraction (Wallin et al., 2010). Lastly, the result of the equilibration equation is divided with KH, computing the pCO2aq (uatm) from CO2aq.

3.5. Statistics

Statistical analysis was performed in IBM SPSS Statistics to analyse potential differences within and between the three lakes. As the data was not normally distributed non-parametric tests were used with a significance level at 5%. At interpretation, the adjusted statistical significance (p-value) was used. Kruskal-Wallis tests were performed (with pairwise comparisons as post-hoc) to detect if the parameters median significantly differs between the groups, in this case the three lakes are the grouping variables. This test is non-parametric and does not assume normal distribution nor same variance. The Kruskal-Wallis test does however assume that the samples are random and independent of each other. The null hypothesis assumes that all groups have the same median and the alternative hypothesis assumes that at least one of the groups median differs from the others (McKight & Najab, 2010).

Correlation analysis was performed to find covariation to different variables. When analysing covariation to environmental variables across lakes, the dependent parameters in the dataset was normalised through Z-score to give the lakes the same mean and exclude differences between lakes (Cheadle et al., 2003). Correlation analysis was performed with Kendall rank correlation coefficient, or Kendall’s Tau. Correlation analysis examines whether there is covariation between two variables, and how strong it is. It gives a correlation coefficient from -1 to 1 which indicates the strength, where 0 is no covariation and -1 or 1 is a perfect relationship. If it is 1, it means that when the value of one variable increases, the values in the other variable also increase. If it is -1, the values in one variable decrease when the value of the other variable increases. Kendall’s Tau is a non-parametric test which does not require monotonous or linear data. It notes how many times the values are raised and lowered – and calculates the difference between them. The test also produces a p-value, where the null hypothesis is that there is no covariation (correlation coefficient = 0) and the alternative hypothesis is that the variables do covariate (correlation coefficient ≠ 0) (Newson, 2002).

3.6. Geographical information system (GIS)

The ArcMap software from ArcGIS Esri Inc was used to visualize geographical data. GIS is a map that by projection reflects the spherical globe. It is a crucial tool to help interpret real world phenomena, specifically concerning spatial data (Burrough et al., 2015). The dataset

(32)

31 had a geographic anchoring through the coordinates that were sampled in the field with a Garmin eTrex 10 GPS. The GPS sampled coordinates in WGS 1984 degrees decimal minutes (DDM), which were converted to WGS 1984 decimal degrees (DD) to be compatible with ArcMap. A new Feature class was created in ArcCatalog using XY data from the file. The coordinate system was projected to SWEREF 99 and a background map was downloaded from ArcMaps’ Add Basemap. The visualization of data has taken place through these steps: Layer properties → Symbology → Categories → Unique values and Layer properties → Symbology → Quantities → Graduated symbols.

4. Results

4.1. CH

4

flux

4.1.1. Within-lake variability

The CH4 fluxes measured in the lakes are shown in Table 2.

Table 2. Presents the lakes minimum, maximum, median and mean value of the CH4 flux, in unit mmol/m2_/d.

Lake Minimum Maximum Median Mean

Glan 0.10 2.43 0.61 0.80 Roxen <0.10 1.00 0.16 0.26 Vättern <0.10 5.54 0.23 0.46

Glan

The flux in Glan was generally variable from day to day and Glan had the largest variability within the same day, sampled the furthest from shore (Figure 3 and 10). Glans’ smallest variability was from the day the sampling took place at the most central part of the lake (Figure 3 and 10). The spatial variability is visible in Figure 11A and 11B. A Kruskal-Wallis test showed that the null hypothesis was rejected, thus the fluxes within Glan’s dates differed significantly. The 7th of October differed from the 24th of September (p=<0.001), 30th September (p=0.016) and 14th of October (p=0.015). There was not a significant difference between the other dates.

(33)

32 Roxen

Also in this case, the highest flux variability was measured the day the drift was closest to the shoreline (Figure 4 and 10). The smallest variability was sampled much further from the shoreline (Figure 4 and 10). The spatial variability is visible in Figure 11A and 11B. A Kruskal-Wallis test showed that the null hypothesis was rejected, thus the fluxes within Roxen’s dates differed significantly. The 5th of October differed significantly from 21st of September (p=<0.001), 28th of September (p=<0.001) and 15th of October (p=0.008). Also, 21st of October differed from 28th of September (p=<0.001). There was not a significant difference between the other dates.

Vättern

Vättern did not have as much temporal variability as the other lakes but it had three extreme values (Figure 10) which can affect how one estimates the flux in the graphs, these may be due to ebullition. The smallest variability was from the date the sampling took place outside of Hästholmen at greater depth (compared to the other measurements from Vättern which were sampled at a shallower depth outside of Vadstena) (Figure 5 and 10). The spatial variability is visible in Figure 11A and 11B. A Kruskal-Wallis test showed that the null hypothesis was rejected, thus the fluxes within Vättern’s dates differed significantly. The 23rd

of September differed significantly from 29th_{of September (p=<0.001) and 12}th_{of October}

(p=0.001). There was not a significant difference between the other dates.

(34)

33

Figure 11A and 11B. CH4 flux (mmol/m2/d) in Roxen, Glan and Vättern. The amount of CH4 flux is presented through graduated symbols where increasing size means increasing value. These maps do

not picture drift, only location at chamber sampling.

4.1.2. Between-lake variability

In Figure 12, the CH4 flux is plotted per lake. It indicated higher values in Glan, while

Roxen and Vättern displayed lower, similar fluxes. Vättern had three extreme values, which could be due to ebullition bubbles. Roxen had five outliers which were all sampled the same day (28th of September), on that day the sampling was conducted closest to the shoreline. A Kruskal-Wallis test showed there was a statistically significant difference in CH4 fluxes

between the lakes. Specifically, between Glan and Roxen (p=<0.001), and between Glan and Vättern (p=<0.001); there was not a significant difference between Roxen and Vättern.

Greenhouse gas emissions from three large lakes during the autumn 2020

Anja Anderö Nordqvist & Veronica Bohlin

Greenhouse gas emissions from

three large lakes during the

autumn 2020

Abstract

Sammanfattning

Preface

Table of contents

1. Introduction

1.1. Aim and research questions

2. Background

2.1. CH

2.1.1. Gas properties

2.1.2. CH

flux from lakes to the atmosphere

2.2. CO

2.2.1. Gas properties

2.2.2. CO

flux from lakes to the atmosphere

2.3. Spatiotemporal variability of CH

and CO

3. Method

3.1. Study sites

3.1.1. Glan

3.1.2. Roxen

3.1.3. Vättern

3.2. Sampling

3.2.1 Floating chambers

3.2.2. Vial management

3.2.3. Background sample

3.2.4. Open-air sample

3.2.5. Flux chamber headspace sampling (CH

flux)

3.2.6. Water concentrations of CH

and DIC

3.2.7. pCO

3.2.8. Air- and water temperature, wind, coordinates, depth and air pressure

3.3. Analysis

3.3.1. GC separation - CH

and CO

3.3.2. CH

detection: Flame Ionization Detector (FID)

3.3.3. CO

detection: Thermal Conductivity Detector (TCD)

3.4. Calculations

3.4.1. CH

flux

3.4.2. Aquatic concentrations of CH

and DIC

3.4.3. pCO

3.5. Statistics

3.6. Geographical information system (GIS)

4. Results

4.1. CH

flux

4.1.1. Within-lake variability

4.1.2. Between-lake variability