CARBONDIOXIDE FLUXES FROM A CONTROLLED BOREAL RIVER

Master’s thesis

Two years

ENVIRONMENTAL SCIENCE

CARBON DIOXIDE FLUXES FROM A CONTROLLED BOREAL RIVER

MID SWEDEN UNIVERSITY

Ecotechnology and Sustainable Building Engineering

Examiner: Anders Jonsson, anders.jonsson@miun.se

Supervisor: Andreas Andersson, andreas.andersson@miun.se Author: Frank Arthur, frar1601@student.miun.se

Degree programme: International Master's Programme in Ecotechnology and Sustainable Development,120 credits

Main field of study: Environmental Science Semester, year: VT, 2018

Table of Contents

ABSTRACT………....i

ACKNOWLEDGEMENT...ii

1 INTRODUCTION ... 1

1.1 PROBLEM, MOTIVATION AND PURPOSE OF THESIS ... 3

1.2 AIR-WATER GAS EXCHANGE ... 4

1.3 GAS TRANSFER VELOCITY ... 5

1.4 PROCESSES THAT INFLUENCE THE TRANSFER VELOCITY ... 5

1.4.1 ESTIMATION OF K BASED ON WIND SPEED ... 6

1.4.2 STREAM VELOCITY ... 6

1.4.3 BUBBLES ... 8

1.4.4 TURBIDITY ... 8

1.4.5 WATER SIDE CONVECTION ... 9

1.4.6 RAIN ... 9

2 METHOD ... 11

2.1 EDDY COVARIANCE METHOD ... 11

2.2 DATA ANALYSIS / MEASUREMENT ... 13

2.3 OTHER METHODS USED TO MEASURE GAS FLUXES ... 13

2.4 MEASUREMENT SITE ... 14 2.5 INSTRUMENTATION/ LI-7500A ... 15 2.6 DATA COMPUTATION/PROCESSING ... 15 3 RESULTS ... 16 3.1 METEOROLOGY ... 16 3.2 HEAT FLUXES ... 17

3.3 CARBON DIOXIDE FLUXES ... 18

3.4 CARBON DIOXIDE FLUXES AND WIND SPEED ... 19

3.5 CARBON DIOXIDE FLUXES AND STREAM DISCHARGE ... 20

4 DISCUSSION ... 22

5 CONCLUSION ... 24

i

ABSTRACT

River, lakes and streams account for more carbon dioxide emissions than all other freshwater reservoirs together. However, there is still lack of knowledge of the physical processes that control the efficiency of the air-water exchange of CO2 in these aquatic systems. In the more turbulent water sections of a river, the gas transfer is thought to be governed by the river’s morphology such as bottom topography, slope and stream flow. Whiles for wider sections of the river, the gas transfer could potentially be influenced by atmospheric forcing (e.g. Wind speed). The main purpose of this project is to study the fluxes of carbon dioxide and how (wind speed and stream discharge) influence the CO2 fluxes in the river. In this study, direct and continuous measurements of CO2 emission was conducted for the first time in a

controlled boreal river in Kattstrupeforsen (Sweden) from 18th April to 10th May 2018. A unique measurement setup which combines eddy covariance techniques, general meteorology and in situ water variables (for high accuracy emission measurements) was used. The results show that in the late winter, an upward directed CO2 fluxes measured in the river was approximately 2.2 μmol m−2 s−1. This value agrees with many other small and large rivers where CO2 fluxes has been studied. The river can be said to serve as source of CO2 to the atmosphere in the day due to the dominant upward fluxes recorded during the daytime. The results also show that carbon dioxide fluxes increase with increasing wind speed notably at wind speed above 2 m s-1. There was no relation between CO2 fluxes and stream discharge. This indicates that wind speed could be one principal factor for air- river gas exchange. The findings in this work on river gas exchange will provide a basis for a regional estimate and be applicable for many river systems on a global scale.

KEY WORDS: Carbon dioxide fluxes, Gas transfer Velocity, Turbulence, Eddy covariance,

ii

ACKNOWLEDGEMENT

I appreciate the Almighty God for His divine strength, love, grace, wisdom and favor He bestowed on me concerning the success in the completion of this project. My cherished thanks goes to my supervisor Dr. Andreas Andersson and course coordinator Prof. Anders Johnson for their marvelous, uncommon support in this project. All Staff of the Ecotechnology Department are also acknowledged for their support in various ways.

I would also like to thank the Department of air, water and landscape Science at Uppsala University and Jämtkraft for their enormous support in setting up the eddy covariance instrument at the site.

Finally i would like to thank Judith Waller, Marcus Wallin and Erik Sahlée for their moral support in this project. I say that God should replenish anything lost and God richly bless you all.

1

1 INTRODUCTION

The Fifth Assessment Report (AR5) by the Intergovernmental Panel on Climate Change IPCC, (2013) concludes that, the global mean surface temperature has increased significantly with a warming trend of 0.85 [0.65 to 1.06] °C since the start of the industrial revolution, resulting in a rise of the global sea level. This rising temperature is due to the increasing greenhouse gas concentration in the atmosphere caused by human activities. This for example has altered the climate of the earth resulting in an acidification of aquatic and marine systems such as oceans, rivers, lakes, and estuaries (IPCC, 2013). Carbon dioxide (CO2) is known to be one of the most important anthropogenic greenhouse gases and hence plays a major role in the climate and climate change. The carbon dioxide concentrations was found below 280 ppm before the start of the industrial revolution. However this global CO2 concentration level has been increasing steadily during the last century up to today’s level of 407 ppm (National Oceanic and Atmospheric Administration NASA, 2018). For example, the analysis of the report from the IPCC (2013) predicts that, the carbon dioxide concentration will double and is expected to result in an increased global mean temperature of 2°C - 4°C by the year 2100.

Cole & Caraco (1998) described carbon dioxide outgassing from rivers and streams to the atmosphere as a part of the global carbon cycle. Fundamental knowledge of the processes that occur in both regional and global carbon cycle is necessary in order to make an accurate forecast of the future climate and study the effect of the increased CO2 emissions. The IPCC (2013) in their AR5 report added freshwater outgassing from surface water to the carbon cycle model as seen in Fig 1. The updated carbon cycle with fluxes also acknowledges the role that freshwater plays as sources and sinks of CO2 to the atmosphere. This gives a much clearer understanding that carbon dioxide evasion from inland waters plays a more significant role than previously thought (Ciais et al., 2013; Cole and Caraco, 2000).

Lakes have been estimated to cover about 3% of the global land surface but cover approximately 7% of the land surface in the boreal zone (Downing et al., 2006). According to Huotari et al. (2010), one unique characteristics of boreal lakes is their brown watercolour which implies high amount of dissolved organic carbon. In Sweden, inland waters (lakes and rivers) occupy almost 9.8% of the total land area (Lindegarth et al., 2016). Though these rivers occupy a little portion of the land surface, they play a significant role in regional carbon cycling (Rasera et al., 2008). The global emission rate of carbon dioxide from lakes and reservoirs is estimated to be approximately 0.32 PgC yr−1 and 1.8 PgC yr−1 from rivers and streams respectively (Raymond et al., 2013). Rasera et al. (2008) estimated the global surface area of small rivers to be approximately 0.3 ± 0.05 million km2 with the potential to emit 70 ± 42 TgC yr−1 as carbon dioxide to the atmosphere.

2

Fig 1. The carbon cycle; The numbers next to the arrows represent fluxes in PgC yr−1, Figure taken from IPCC (2013) ipcc.ch.

Rivers and lakes are in general supersaturated with CO2 (Cole & Caraco, 1998). A large number of lakes have a surface concentration of carbon dioxide higher than the equilibrium with the atmosphere and therefore serves as net sources of CO2 to the atmosphere (Cole & Caraco, 1998; Kling et al., 1991). The surplus of this carbon dioxide comes as a result of the in-lake heterotrophic respiration of organic carbon which originates from the earth (Johnsson et al., 2003; Huotari et al., 2011). It has been found that carbon can also enter freshwater bodies through the inflows of dissolved inorganic carbon and dissolved particulate organic carbon (DOC and POC) from the watershed (Lindsay, 2008). Another way carbon can enter the aquatic system is through the uptake of atmospheric carbon dioxide to be used for photosynthesis by the plants algae in the water. There has been several efforts from numerous studies to measure the carbon dioxide emission rate from these freshwater bodies.

3

The gas transfer velocity, k is the rate at which a gas is transferred across the air-freshwater interface. The flux of carbon dioxide is controlled by the direction of the CO2 concentration gradient between the air and the water surface (Cole & Caraco, 1998). This flux is highly controlled by the gas transfer velocity, k (e.g Wanninkhof, 1992). Meybeck (1993) suggests that knowing the CO2 flux of rivers and lakes will be important to understand the global CO2 flux. Understanding the concentration of CO2 in both water and air is needed in order to calculate accurately the magnitude of the carbon dioxide flux. The estimation of carbon dioxide fluxes also requires a knowledge of k (Wanninkhof, 1992). The most common parameter used to determine the value of k in freshwaters is wind speed (Cole and Caraco, 1998). However, there are still other factors that affect k which needs further investigation.

1.1 PROBLEM, MOTIVATION AND PURPOSE OF THESIS

There has been many attempt by scientists around the globe to investigate the exchange of CO2 between air and freshwater surface. However there is still lack of agreement and understanding on the factors that control k in rivers, streams and lakes (Cole & Caraco, 1998; Garbe et al., 2014). The possible parameters that could influence the resulting value of k in freshwaters could be wind speed, the channel morphology of the river, the stream flow, heat fluxes, the slope and even the different measurement methods used to estimate k. Berg and Pace, (2017) indicates that apart from the eddy covariance technique, none of the other methods shows a direct way of estimating the values of k because they rely on assumptions that are usually difficult to understand, Because of this large variability in k, there has been a large uncertainty in estimating the CO2 fluxes from rivers and lakes (Raymond and Cole, 2001; Raymond et al., 2012). Therefore, the research question to address in this study will then be; What are the parameters that influence the CO2 fluxes in rivers?

The purpose of this project is to study the processes that regulates CO2 fluxes in the air-water interface for a wide river section. The main objective is to study the fluxes of carbon dioxide and the relation between these two factors (wind speed and stream discharge). A unique measurement setup which will combine eddy covariance techniques, general meteorology and in-situ water variables will be used in this project. The specific aims are:

● To study the variations of CO2 emissions between air and streams. ● To identify the factors controlling the fluxes of CO2 in a river.

4

1.2 AIR-WATER GAS EXCHANGE

The evasion of gases (such as carbon dioxide, oxygen and methane) from the surface of water to the atmosphere is a significant component in the aquatic ecosystem, but the gas exchange rates are frequently estimated with vast uncertainties (Berg and Pace, 2017). This is probably due to the lack of understanding of the factors that govern the gas exchange in these aquatic systems.

Anderson (1999) suggest that, turbulent motions are mainly responsible for transporting gases between the atmosphere and the lake surface. A viscous sublayer (about 1 mm thick) as seen in (Fig 2) which lies between the turbulent air and the surface is where molecular diffusion occurs most in the gas exchange (Liss and Slater 1974). Hence the rate of the gas transport is described by Fick’s law which states that; the flux, F of any gas is proportional to the concentration gradient.

𝐹 = 𝐷𝛿(𝐴)/𝛿𝑍 (1)

where D represents the molecular diffusion expressed as a function of temperature and the gas. 𝛿𝑍 = Thickness of the viscous layer, 𝛿(𝐴) = concentration difference across the film. When the wind speed is increasing, it causes wind stress and contributes to the tearing of the viscous layer thereby reducing 𝛿𝑍 (Liss, 1973). As 𝛿𝑍 decreases, the flux of CO2 begins to increase across the air-water interface and vise versa (Liss, 1973).

Fig 2. Shows how turbulent motions transport gases between the atmosphere and the water-surface. Figure taken from Heiskanen (2015).

5

1.3 GAS TRANSFER VELOCITY

The flux of any gas (such as O2, CO2 and CH4) between the surface water and the atmosphere is mainly influenced by these two main factors; (The difference in partial pressure between the water and air. and the gas transfer velocity for a given gas at a given temperature, k (Wanninkhof, 1992). The solubility of the gas in the water also affects the flux. Hence the air-freshwater flux of CO2 can be expressed as:

𝐹_𝑐 = 𝑘𝐾_𝑜𝛥𝑃𝐶𝑂2 (2)

where k is the gas transfer velocity (m s−1), 𝑘𝑜 is the solubility (mol m−3 atm−1), and ΔpCO2 =PCO2 water - PCO2 air. The transfer velocity is the rate at which the gas is transferred across the air-water interface and it depends mainly on the turbulence in the water which is controlled by the wind speed and the Schmidt number (Sc) (Jähne et al., 1987).

The Schmidt number has no unit (dimensionless) and its the ratio of the kinematic viscosity, V (m2 s−1) to mass diffusivity of the gas, D (m2 s−1) (Wanninkhof, 1992). As seen in equation 3 below;

(3) Several experiments to estimate k in different lakes and freshwaters have been made (Cole and Caraco, 1998). Wanninkhof (1992) suggests that, the value of k varies by about a magnitude of two in these freshwaters and marine system. Different studies done on lakes and rivers conclude that k increases with increasing wind speed notably at wind speed above 3 m s−1 (MacIntyre et al., 1995; Wanninkhof 1992, 2001). This means it is possible to calculate a theoritical value of k from different parameterization if the wind speed is known (Cole and Caraco, 1998). Many studies have also found the lack of correlation between k and low wind speed approximately less than 3 m s−1 (Clark et al., 1995; Cole and Caraco, 1998; Crill et al., 1988). However when the wind speed is low, the contribution of wind to the overall turbulence in the water reduces and other factors become more important (Crusius and Wanninkhof, 2003).

1.4 PROCESSES THAT INFLUENCE THE TRANSFER VELOCITY

Crusius and Wanninkhof (2003) observes that the absence of a unique relationship to represent the transfer velocity and wind speed at low wind speed shows that, there are still other processes which can contribute significantly to gas exchange when the wind speed is low (Crusius and Wanninkhof, 2003). Several studies suggest that, k could possibly depend on factors such as stream velocity (Allin et al., 2010; Beaulieu et al., 2012; Borges et al.,

6

2004; Ho et al,. 2016; Zappa et al., 2003, 2007), water side convection (Podgrajsek et al., 2015), Bubbles (Crosswell, 2015; Woolf et al., 2005) and rain (Ho et al., 2007; Takagaki, 2007). These different number of factors clearly shows that further investigations is still needed in order to understand the processes governing the gas exchange (Garbe et al., 2014).

1.4.1 ESTIMATION OF K BASED ON WIND SPEED

Wind plays a significant role in the turbulence of surface water and it's easy to measure ( Crusius and Wanninkhof, 2003). In order to calculate gas fluxes using the bulk flux model (equation 2), 𝛥𝑃 must be measured whiles k is estimated from known parameters. Wind speed is normally used to parameterize and calculate k.

In freshwaters such as lakes and streams, k is usually parameterized with the equation suggested by Cole and Caraco (1998) as seen below;

𝑘 = 2.07 + 0.21𝑈101.7 (4)

where 𝑈10 denotes wind speed at 10 m,

However in the ocean community, The transfer velocity has been related to the wind speed in either linear, quadratic or cubic equations. The current most frequently cubic relation used in ocean and sea is the one suggested by Wanninkhof (2009) seen in equation (5).

𝑘 = 3 + 0.1𝑈10+ 0.064𝑈₁₀ 2+ 0.0011𝑈10 3

(5)

Most of these parameterization consider wind speed as the only process causing turbulence in the water and driving the gas exchange. Again, the different ways in the parameterization is problematic and shows the difficulty in the attempt to determine the transfer velocity in a more accurate way which can be relied on for different aquatic conditions. However many other parameters mentioned earlier on such as bubbles, water side convection etc have been left out of the models because of lack of knowledge.

1.4.2 STREAM VELOCITY

Ames, (2018) defines stream velocity as the speed of water, how quickly the water moves through its channel. It is usually measured in meters per second or feet per second. Several factors could affect the velocity of water in a river, such as the shape of its channel, the gradient of the slope that the river flows along, the volume of water contained in the river and also the friction caused by rough edges within the riverbed (Ames, 2018). Different terms are usually used to represent the stream velocity, such as; water current, tidal current or current velocity depending on the type of aquatic system. The stream velocity can generate turbulence in the water at high flow speed as the water molecules follow an unparalleled path. The water flow becomes unstable and changes from laminar to turbulent flow as the stream velocity decreases suddenly. There have been few studies on how this stream velocity driven

7

turbulence in the water could control or affect k and the outgassing of carbon dioxide in the air-water interface.

Stream driven turbulence generated in the water surface is known to contribute significantly to the gas transfer velocity (Allin et al., 2010; Beaulieu et al., 2012; Borges et al., 2004; Ho et al., 2016; Zappa et al., 2003, 2007). For instance the findings by Zappa et al. (2007) in the Parker river suggest that, the river’s turbulence is controlled mainly by tidal flow velocity at low wind speed below 2.0 m s−1 but controlled by wind at high wind speed above 5.5 m s−1. The correlation between wind speed and the gas transfer velocity in large rivers and between stream velocity and the gas transfer velocity in small rivers and streams shows that, the significance of wind speed declines upstream as water stream velocity becomes a more dominant driver of gas exchange (Allin et al., 2010). This confirms the findings from an earlier study that, the turbulence driven by water currents enhances k but decreases with increasing wind speeds (Borges et al., 2004; Zappa et al., 2003). Even though stream velocity is expected to contribute to the turbulence in the water surface no matter the wind speed, their effect is felt most if the contribution to turbulence from wind speed is low (approximately less than 4 m s−1 (Borges et al., 2004). For example over a micro-diurnal tidal river estuary, Zappa et al. (2003) shows that, the gas transfer velocity at low wind speeds is controlled by water flow-induced near-surface turbulence.

Allin et al. (2010) suggests that, the air-water gas exchange in the outmost upstream section of most river systems is controlled by factors linked to the stream velocity such as the shape of its channel and the friction in the river bed which induce turbulence in the water. Wind speed becomes more significant in the downriver section as the channel broadens and protection from the forest cover dwindle, but still turbulence induced on the water surface is dominated by factors related to the stream flow velocity (Allin et al., 2010). Thus, the relative importance of wind to enhance the gas transfer velocity diminish to zero whiles the effect from stream velocity and river bed friction continuously increase as one moves upriver within the river system (Allin et al., 2010).

Borges et al. (2004) established an experimental relationship to parameterized the gas transfer velocity of carbon dioxide by taking into consideration the contribution of both wind speed and stream velocity. The relationship suggest that stream velocity contributes greatly to the gas transfer velocity. The equation modeled in the study is shown below (equation 6);

𝑘₆₀₀ = 1.0 + 1.719𝑤0.5ℎ−0.5_{+ 2.58𝑢}

10 (6)

Where k represents the gas transfer velocity of CO2 (cm h−1) normalized to Sc at 600, w and h denotes the stream velocity (cm s−1) and the depth (m) respectively whiles 𝑢10 (m s−1) is the

wind speed at 10 m height.

A tracer release experiment in the Shark river, an estuary located in the largest mangrove forest in North America was conducted (Ho et al. 2016). The studies has parameterized the gas transfer velocity k by taking into account the two main sources of turbulence generated by the stream velocity and the wind stress on the water surface. (Zappa et al., 2007; Allin et al., 2010) concludes that, a more pragmatic model of the gas transfer velocity in a river that is controlled by wind speed, stream velocity and depth will allow scientist to refine

8

measurement of the carbon dioxide fluxes in other freshwater systems. The addition of wind speed and stream velocity to the parameterization of the gas transfer velocity is expected to modify earlier estimates of regional and global carbon dioxide fluxes from freshwaters such as streams and rivers.

1.4.3 BUBBLES

Contrary to air-water outgassing, where the transfer is totally driven by the turbulence in the surface water and the molecular diffusion coefficient of the gas, the emission through bubbles is also a function of the gas solubility and bubble size (Wanninkhof et al., 1995). Hare et al., (2003) observes that the transfer of a gas across the air water interface may also be governed by the mediation of bubbles. stream velocity, wind speed and wind direction are known to control the structure of the bubbles in the air-water interface (Crosswell, 2015). The bubbles form and accumulate some materials when the air has been entrapped in water as a results of the breaking waves seen as a whitecap coverage (Nightingale & Liss, 2003; Anderson, 2017). The materials carried on the surface of the bubbles may be ejected on the water spray during the crushing process (Woolf, 2000).

There is a stronger enhancement of the gas transfer velocity in the air water interface for low soluble gases such as carbon dioxide (Crosswell, 2015). Therefore the efficient means of gas exchange is highly controlled by the water-side transport and injection of bubbles by breaking waves (Crosswell, 2015). This confirms previous studies on the relation between solubility and gas exchange which suggest that smaller bubbles have a significant effect on the gas exchange rate in freshwaters (Wanninkhof et al., 1995). However, the process involved in bubble-mediated transport are complex and largely depend on the mechanism of bubble formation, the dynamic behavior of bubbles, and the time evolution of bubble plumes (Crosswell, 2015). Woolf et al. (2005) concludes that gas transfer induced by bubbles solely depends on the molecular properties of the dissolved gas.

According to Crosswell (2015), bubbles can also enhance the gas transfer velocity by increasing the area of the air water interface which results in an unbalance exchange that supersaturate the surface water with gas relative to the atmosphere. (Asher et al., 1997; Crosswell 2015) shows that surfactants (which enables a gas to diffuse by acting as surface layers) could affect wave breaking and bubbles properties and further influence the gas exchange rate. Asher et al. (1997) proved that when surfactants are present in a wave-breaking simulation tank it could decrease the magnitude of CO2 flux in seawater but increase CO2 fluxes in freshwaters. The reasons for the authors conclusion was the fact that freshwaters are rich with surfactants and leads to significant bubble formation. However there is still lack of understanding of how surfactants influence gas exchange in the natural environment (Crosswell, 2015).

1.4.4 TURBIDITY

Turbidity means how clear a water is. It usually refers to the amount of total suspended solids (TSS) in the water, Some of the sources of turbidity in most open streams and lakes are from algae, clays and dust from erosion, sediments from the bottom of the stream and discharge from wastewater (Moore, 1989).

Abril et al. (2009) studied some river estuaries in France and found out that the gas transfer velocity was significantly lower at high turbidity. Their studies also shows that the presence

9

of suspended particles (more than 0.2 g L−1) has a great influence on the turbulence in the water and therefore affects the gas fluxes between the air-water interface.

High amount of suspended solid materials could affect k by altering the density and viscosity of the water which further affects the Schmidt number (Abril et al., 2009). This is because the Schmidt number (equation 3) is highly dependent on the viscosity and density of the water. The density of water may change from 1000 kg m−3 to 1001 kg m−3 in just 1 g L−1 value of suspended solid particles in the water (Abril et al., 2009). However some studies do not attribute this enhancement of the transfer velocity to the change in the schmidt number as the change looks insignificant (Abril et al., 2009).

Herskanen et al. (2015) studied some small lakes in Finland and observed that water clarity was an important factor which defines the thermal stratification of the lakes. They argue that turbid water could reduce the water column temperature as a results of shorter stratification period and thereby have a significant influence on the lake - atmosphere gas exchange rates. According to Abril et al. (2009), the presence of suspended solid materials in the water enhance the viscous dissipation of turbulent kinetic energy. The materials collide, interact and exchange momentum which suppress the turbulent in the water column and thereby affecting the gas exchange (Abril et al., 2009). More research on how turbidity actually affects the turbulence in the water is still needed to bridge the gap and understanding on how it affects the transfer velocity.

1.4.5 WATER SIDE CONVECTION

Convective mixing occurs from an increase in the density of surface water as a results of surface water cooling (Norman, 2013). Podgrajsek et al. (2015) studied some Swedish lakes using the eddy covariance method and concludes that water side convection could affect the transfer velocity and enhance CO2 fluxes. The studies further show that high nighttime CO2 fluxes could be attributed to the enhanced transfer velocity caused by water side convection. Additionally Podgrajsek et al. (2015) emphasized that, the parameterization of k with both water side convection and wind speed model fits better to measurement when compared to parameterization considering only wind speed.

1.4.6 RAIN

Takagaki, (2007) shows that rain can enhance carbon dioxide outgassing in freshwater environment. Ho et al. (1996) analysed the process involved in this enhancement in the laboratory with the SF6 evasion experiment and field observations made during a tropical rainstorm in Miami and Florida. Their results show that low and heavy rain events accelerated gas exchange in the lakes and established a correlation between k and natural rain rate. This finding is supported by a study on carbon dioxide exchange in a low wind oligotrophic lake by Cole and Caraco (1998) which asserts that the rain rate was better related to the gas transfer velocity in the lake than the other parameters tested. The influence of rain on carbon dioxide evasion is probably more dominant in lakes or streams where turbulence driven by wind is not a major factor (Cole and Caraco 1998; Ho et al., 2004). But it appears insignificant at higher wind speed (Harrison et al., 2012). The enhancement of k by rain is

10

dependent on the generation of turbulence and other secondary motions (Ho et al., 2004, 2007). Rain bubbles can contribute about 0 to 20% of the total gas exchange but it depends on the size of the raindrops and the rain rate (Ho et al., 2000

Several processes occurs in the surface layer of the water during rainfall. Turk et al. (2010) points out that rain can lead to chemical dilution in the surface water which reduces the surface salinity, the dissolved inorganic carbon (DIC), the total PH and affects the magnitude of the concentration of CO2 in the water.

11

2 METHOD

2.1 EDDY COVARIANCE METHOD

The eddy covariance method (EC) is one of the most reliable, accurate and direct approaches used to measure gas emissions and fluxes across the air-water interface (Burba, 2013). The in-situ micro-meteorology eddy covariance uses a sonic anemometer with a gas analyzer (Fig 4) to produce a half-hourly flux averages from 10-hz data (Aubinet, 2013; Baldocchi, 2012; Burba, 2013). The EC method works by measuring the vertical turbulent transport of a gas across the water surface (Burba, 2013). That is, the covariance between the gas concentration and the vertical wind speed (Burba, 2013). Burba (2013) explains that, the general principle involved in the eddy covariance method is to measure the number of molecules which moves downward and upward over a certain period of time and how fast they travel. This means the gas fluxes of any gas such as carbon dioxide can be calculated by measuring the vertical transportation of the gas via eddies (Fig 3) created by both the vertical and horizontal turbulence in the surface water (Burba, 2013).

Mathematically, the flux, 𝐹 of any gas is the product of the air density (ρ), the vertical wind speed (w) and the mixing ratio of the gas (s) also known as the dry mole fraction as represented below:

𝑭 = 𝜌_𝒔𝒘𝒔 (7)

The Reynolds decomposition is often used to break down the terms into (overbars) and deviation from mean (prime) for the equation to be simplified to;

(8)

Which represents the product of the mean air density and covariance of the vertical velocity fluctuation and concentration fluctuation.

However there are a number of assumptions involved in the use of the eddy covariance method. (Aubinet et al., 2012; Baldocchi, 2012; Burba, 2013; Lee and Law, 2004) points out that; (a) Measurements at a point are assumed to represent an upwind area, (b) the flux is assumed to be fully turbulent, (c) the instruments can detect very small changes at very high frequency, and (d) the air density fluctuations are negligible. But if these assumptions will hold true, it will depend if the site was selected properly for the experimental setup, and also the weather conditions (Aubinet et al., 2012; Burba, 2013).

The main disadvantage in the use of the eddy covariance technique is the fact that the flux measurements may not be perfect. This is probably due to some of these assumptions and other physical phenomena, problems in the instrument and setup errors (Burba, 2013;

12

Massmon and Lee, 2002). Time response errors (which occur because of the instruments inability to catch all the rapid changes that result from the eddy transport) can also affect the measured flux value (Burba, 2013; Massmon and Lee, 2002;). Fortunately these errors can be minimized or corrected during the data processing.

The significant advantage of the eddy covariance method is that it has the ability to provide both direct and continuous measurement of 30 minutes fluxes throughout the year both day and night (Burba 2013).

Fig 3 shows a conceptual, framework for atmospheric transport of eddies, Figure taken from (Burba, 2013)

Fig 4; Photograph showing the set-up of the eddy covariance instrument in Kattstrupeforsen (study area), With the sonic anemometer marked by the red arrow and with the LI-7500 to the right

13

2.2 DATA ANALYSIS / MEASUREMENT

Aubinet et al. (2012) observes that certain conditions such as horizontal homogeneity and stationarity needs to be fulfilled when using the eddy covariance technique. In this case the the scalar flux (eqn 9) is expressed by using the Reynolds decomposition;

(9)

where x represents the concentration of any scalar (CO2); w is the vertical wind component. The overbars stands for the mean value whiles the primes denotes the deviation from the mean value (Anderson et al., 2014).

The first term in the right hand side of the equation represents the turbulent vertical flux (that is the covariance of the scalar and the vertical wind deviation). However the second component is usually neglected because after the double rotation or tilt correction, the mean vertical wind speed becomes zero (Anderson et al., 2014). Crosswind influence from the sonic anemometer must be corrected (Burba 2013). Tilt correction or coordinate rotation is done where the coordinate system is rotated to ensure that the vertical wind velocity is zero over a certain averaging period (Aubinet et al., 2013). Tilt correction is needed to reduce the errors associated with instrument leveling deficiencies which introduces a horizontal component in the actual measured vertical wind speed (Liu et al., 2012).

Density fluctuations due to temperature and humidity variations from measurement made by the open path instrument also affect the mean vertical velocity (Aubinet et al., 2013; Burba, 2013). The correction can be done by using the Webb - Pearman Leuning (WPL) correction (Webb, Pearman and Leuning, 1980) in the post data processing. This correction is very important because the temperature and humidity variations can cause contribution to the measured CO2 fluxes (Aubinet, 2013; Burba, 2013).

2.3 OTHER METHODS USED TO MEASURE GAS FLUXES

There are other different type of methods apart from the eddy covariance which can be used to estimate CO2 emissions and the transfer velocity between air and water interface. There are both direct and indirect techniques. For example the eddy covariance measures the gas flux directly whiles the indirect methods usually relies on model values of k from carbon dioxide concentration in the surface water (eg Boundary layer model, BLM) to calculate the gas fluxes. Livingston and Hutchinson, (1995) indicates that the floating chamber is an indirect method because it measures the amount of gas concentration gain or loss within a given fixed time volume. Another way to indirectly estimate the gas emissions commonly referred to as the Gradient flux technique is to measure the gas concentrations at the air-water interface together with different estimates of k (Clark et al., 1994). The Gradient flux techniques quantifies the CO2 flux between water and the atmosphere indirectly (Cole and Caraco, 1998).

14

The tracer gas method is usually used in inland water to measure the gas emissions from the water surface (Cole and Caraco, 1998). For example (Raymond et al., 1997) studied the freshwater portion of the Hudson River estuary from Albany, New York, In their work, they estimated k by using an experimental tracer gas technique (SF and He) to the river. Cole and Caraco (1998) also measured directly the gas transfer velocity, by adding chemically inert gas, sulfur hexafluoride to a low wind oligotrophic mirror Lake.

2.4 MEASUREMENT SITE

The measurement was made in river Kattstrupeforsen which is situated at Latitude 63.32°N

and longitude 14.58°E. Kattrusperforsen is a hydropower station which is located in Indalsälven downstream of Hissmofors (Fig 5). River Indalsälven is a wide river section with relatively slow flowing water which has a length of approximately 430 km (HELCOM, 2011). According to the Swedish Meteorological and Hydrological Institute SMHI, 2010), the average annual precipitation in the area is 800 mm while the average mean flow is approximately 455.3 m3 s−1. With the highest flow occurring in july when the snow melts in the mountains. This is the first time direct measurement of carbon dioxide fluxes is taking place in the study area. The eddy covariance tower was mounted on a brick wall around the water high above the ground and 2.2 m high above the surface of the water in an upwind distance (Fig 4). The data of CO2 fluxes with wind from the sector (240°<WD<300°) are suitable for the measuring site because it represent measurements in the water but not the land.

15

2.5 INSTRUMENTATION/ LI-7500A

The eddy covariance instrument (LICOR; LI-7500A open path CO2 gas analyzer) was mounted on the 18th April 2018 at Kattstrupeforsen, Indalselven. The instrument was installed on a strong platform (Fig 4) to ensure stability and prevent vibrations which will be unnessassary for the experiment. Collection and recording of data started right after the installation. The LI-7500A gas analyzer has a high precision of 0.11 ppm of CO2 RMS at a frequency of 10 HZ. It measures densities of carbon dioxide and water vapour (LICOR Bioscience). The instrument consist of a Gill windmaster 3D sonic anemometer (Fig 4 shown in red arrow) which measures the wind speed. The LI-7500A also consist of a removable USB data storage and an ethernet for world wide connectivity (LICOR Bioscience). The instrument has an LI-7500A pc communication software. The software is needed to communicate with the computer to do data logging setting (LICOR Bioscience). The installation is done by placing the gas sensor head very close to the sonic anemometer and at equal height as well (LICOR Bioscience). The recommended orientation is through vertical mounting to provide 360 degrees direction.

2.6 DATA COMPUTATION/ PROCESSING

The eddyPro software was used to process the raw data collected by the instrument into full time period output spreadsheet data. The default settings in the software was used to run the raw data into output data. Tilt correction and air density fluctuations were compensated before processing the data. Only data corresponding to the wind direction from 240 - 300N was used in the computation. The rest of the data was discarded. Because that direction represent CO2 exchange over the river but not the land (Fig 6).

Fig 6. Figure showing the wind direction (240-300) at the site where the exchange of gases in the river was computed

16

3 RESULTS

3.1 METEOROLOGY

The collection of data started from 18th April to 8th May. In general, the mean average temperature for the period of measurement was estimated to 275 K (3°C). There was not much variation in the temperature as seen in Fig 7 below. Temperatures were generally below 290 K for the entire period notably lower in the night and higher in the daytime. The relative humidity was above 50% in most of the days which indicates the presence of more clouds in the atmosphere throughout the period of measurement. The wind speed for the entire measurement period was considerably weak (Fig 8 ) with an average of 1.28 m s−1 and a range of 0.1-7.9 m s−1. The maximum wind speed (7.90 m s−1) was recorded in late April whiles winds below 3 m s−1 were observed in early May (Fig 8).

17

Fig 8. Time series of wind speed for the measurement period

3.2 HEAT FLUXES

The heat fluxes across the air water interface in the lake consist of both sensible and latent heat. There seems to be a large variations between the sensible and the latent heat (Fig 9 up) and a balance in both heating and cooling in the water column due to the positive and negative flux values. The negative sensible heat fluxes is likely because the water was colder than the underlying air. The average sensible and latent heat fluxes were -6.70 W m−2 and 10.75 W m−2 respectively. The highest LE value (149 W m−2) occured on the 26th April whiles the lowest sensible heat flux (-148 W m−2) was recorded on 6th May (Fig 9). The wind directions was highly domiated with Northwesterly winds (Fig 9 down)

18

3.3 CARBON DIOXIDE FLUXES

In this study direct CO2 fluxes were measured on a half hourly basis for more than three weeks. In the late winter, an upward average directed CO2 fluxes measured in the river is about 2.2 μmol m−2 s−1. Throughout the measurement period, the highest CO2 flux from the river was observed on 19th April at dawn (03:30) with a value of 222.80 μmol m−2 s−1. The flux of CO2 from the river shows a peak in early May (Fig 11). The data of the CO2 flux shows that the fluxes increase with increasing wind speed and decrease when the wind speed is decreasing (Fig 11 top and down). The trend in the carbon dioxide fluxes looks similar to the trend in the latent heat fluxes (Fig 11). There is a variation of CO2 fluxes on half hourly basis with both negative and positive fluxes (Fig 10). In general, the results shows a dominant upward directed (positive) carbon dioxide fluxes which range from 0.1 -10.0 μmol m−2 s−1 (Fig 11). The diurnal (day and night) variation in the fluxes also indicates higher CO2 flux values in midday as compared to the night (Fig 10). This probably can be connected to the higher wind speed in daytime. For example all flux values below -8 μmol m−2 s−1 were recorded in the midnight whiles most of the positive flux values above 4 μmol m−2 s−1 occurred during the day (Fig 10).

19

Fig 11. Comparison between carbon dioxide fluxes (Top), heat fluxes (Middle) and wind speed (Down)

3.4 CARBON DIOXIDE FLUXES AND WIND SPEED

The carbon dioxide fluxes from the river (which represent only wind directions from 240°-300°N appears to increase with increasing wind speed (Fig 12 ). The results also indicate that, low wind speed below 2 m s−1 has insignificant influence on the carbon dioxide fluxes. This is because the highest and lowest fluxes were recorded at wind speed below 2 m s−1. For example at wind speed of about 1 m s−1, both positive and negative fluxes of about -8 and 12 μmol m−2 _s−1 _{were recorded. Even at wind speed below 1 m s}−1 _{the carbon dioxide fluxes} increased to about 14 μmol m−2 s−1 (Fig 12). Which shows lack of correlation between wind speed and carbon dioxide fluxes at low wind speed. However, at wind speed above 2 m s−1 a relation between carbon dioxide fluxes and wind speed can be observed where the relation 𝐹𝐶𝑂₂ = 1.43 + 0.05𝑈2 _{shows the best fit to data (black dashed line). Though wind speed} were generally weak in the period, as the wind speed peaks up gradually the carbon dioxide outgassing from the river begins to increase. For instance an increasing trend can be observed even as the speed increase above 4 m s−1 (Fig 12) with only positive upward carbon dioxide fluxes. It was rare to see negative fluxes at wind speed above 2 m s−1 as compared to wind speed below 2 m s−1 which shows that the exchange of carbon dioxide between the air-water interface is strongly dependent on high wind speed.

-10 -5 0 5 10 CO 2 f lu x ( u m o l m -2 s -1 ) -100 0 100 H ,L e ( W m -2 s -1 ) H Le 24/040 26/04 28/04 30/04 02/05 04/05 06/05 08/05 10/05 5 10 date U ( m /s )

20

Fig 12. The relationship between carbon dioxide fluxes and wind speed

3.5 CARBON DIOXIDE FLUXES AND STREAM DISCHARGE

The average stream discharge during the studies was estimated to be 130.70 m3 s−1 with a range of 58-3 - 273 m3 s−1. This indicates that, in general the flow rate of the river for the entire measurement period was high with an average water level of 273.1 m. There is no relation between the stream discharge and the carbon dioxide fluxes (Fig 13),

21

Fig 13 The relationship between carbon dioxide fluxes and stream discharge

-60 -40 -20 0 20 40 60 80 0 50 100 150 200 250 300 CO 2 flux μmol m −2 s −1 Stream discharge m3 _s−1

22

4 DISCUSSION

Carbon Dioxide fluxes - The emission of carbon dioxide from the river was compared to

wind speed from the north (240-300) direction which represents the exchange from the river but not the land. Other meteorological parameters such as temperature, relative humidity, sensible and latent heat were also measured. The effect of these parameters especially wind speed on CO2 fluxes has been analysed in this study. The results show that the average mean CO2 flux from the river (Kattstrupeforsen) during the late winter is 2.2 μmol m−2 s−1. This value is consistent with several other studies. For example Anderson et al. (1999) estimated the carbon dioxide fluxes in a lake with the eddy covariance technique during the springtime to be 2.3 - 2.7 μmol m−2 s−1. Previous direct measurement of CO2 fluxes is between 1.87- 10.62 μmol m−2 s−1 in the lower amazon river (Allin et al., 2011; Sawakuchi et al., 2017). Alin et al. (2011) reported again the carbon dioxide fluxes in the Tapajös river to be 0.04 - 6.36 μmol m−2 _s−1_{. which is quite comparable to the fluxes in Kattstrupeforsen. The positive flux} values which were more dominant than the negative flux values in the results suggest that the river was constantly outgassing carbon dioxide to the atmosphere. The negative values also show that there was still an uptake of CO2 by the river. The positive fluxes dominates during the daytime whiles the negative fluxes occurs most often in the nighttime (Fig 10). This confirms an earlier study by Lafleur et al, (1997) which concludes that boreal rivers serve as net source of CO2 to the atmosphere during mid-day.

The highest fluxes were recorded in the late April. It is not quite clear the reason for this, but one may suggest that, there was high concentration of PCO2 in the river at that time or the release of CO2 through respiration could also be a factor. The fluxes begun to decline after April and peak up later again (Fig 11). It is important to note that the possible uptake of carbon dioxide for photosynthetic activities could probably have an influence on the fluxes at that time. The plants growing on the sides of the lake can also use the carbon dioxide. (Kling et al., 1991; Miller et al., 1986) indicate that the amount of CO2 normally decrease before the ice-out from photosynthetic activities. This is because net primary production can reduce if there is little snow cover (Kling et al., 1992 as cited in Miller et al., 1986). However there was no ice or snow during that time. According to Johnson et al (2007), ice cover can prevent gas exchange across the air water interface in rivers and lakes. Again, if the river

(Kattstrupeforsen) is rich with nutrients, then its possible that any increase in nitrogen could accelerate primary production in the water and reduce the PCO2 in the water. The flow regulation from the hydropower that has been installed on the site is likely to affect the properties of the river or the friction in the river beds and have influence on the CO2 fluxes from the river as well.

Wind Speed/Stream discharge and CO2 Fluxes - The results suggest that the exchange of carbon dioxide across the air-water interface is dependent on wind speed and increase with increasing wind speed. There is a relation between carbon dioxide fluxes and wind speed

23

notably at high wind speed (Fig 12). The vertical mixing in the water which is caused by the wind is likely to have generated a lot of turbulence which increased the rate of the gas transfer. The results agrees with the theoretical background that wind speed plays a major role in the exchange of gases across the air-water interface. It is necessary to note that at high wind speed above 2 m s−1 wind speed proves to be a dominant factor that controls the exchange of CO2 fluxes in the river. It is not clear if wind is the only dominant factor controlling the gas fluxes in this river when the wind speed is high. However, at low wind speed, the results in the studies suggest that the importance of wind in the influence of the gas exchange becomes very negligible and its possible that other factors such as stream velocity may becomes relevant than wind speed. The results is in agreement with previous other studies in small and large lakes and rivers. Harrison et al. (2012) studied the combined effect of wind speed and rain on carbon dioxide evasion within the air-water interface and concludes that, the influence of rain on CO2 fluxes at high wind speed is insignificant but when the wind speed is low then rain becomes a more important dominant factor controlling the CO2 fluxes and the effect of wind speed on the fluxes becomes negligible. Cole and Caraco (1998) also tested rain as an important factor controlling the gas exchange in a low wind oligotrophic lake. The amount of precipitation during this study period was quite low and was not texted for in this studies, hence it is not possible to suggest if rain could have had influence on the fluxes when the wind speed was below 2 m s−1. Stream velocity has also been suggested to affect the turbulence in most boreal rivers when wind speed is low. For example (Alin et al 2010; Borges et al 2004) concludes that the stream velocity becomes a dominant driver of CO2 fluxes in rivers when the wind speed is below 2 m s−1 but the study was quick to point out further that, stream velocity becomes less significant when wind speed is above 5 m s−1.

In this studies, the results show that wind speed has an influence in the exchange of CO2 fluxes in the river. There is no relation between stream discharge and the carbon dioxide fluxes in this river. It is therefore unlikely that when wind speed was low, stream discharge became the driver for the CO2 fluxes. Long term studies in the area will provide better clue if the turbulence generated in the water surface due to the high flow rate will influence the transfer velocity of the CO2 fluxes in the river. Again, one may notice that, the influence of stream discharge on the carbon dioxide fluxes was very negligible especially at low water flow (Fig 13) which may probably suggest that wind speed may be the dominant driver of the gas exchange in this river. However, it is also possible that some other factors which is not known yet could have also influence the CO2 fluxes in the river when the wind speed was calm.

Sensible and Latent Heat fluxes- The results show much more positive trends in the latent

heat which may indicate that, there was much more rising of warm air upwards from the river. The sensible heat flux is seen to be clearly negative or below zero in most cases. In Fig 9, an opposite pattern between the sensible and latent heat fluxes can be seen. This pattern is

24

similar to a boreal lake in southern Finland which was studied by Vesala et al, (2006). The latent heat also increase with wind speed which agrees with previous studies in two boreal lakes in Sweden. Heikinheimo et al. (1999) studied lake Tämnaren and Räksjö and concluded that the variation in the latent heat flux was strongly related to wind speed. The low positive latent heat flux in this study is most likely due to a combination of dry air and low water temperature (1- 5°C). The average mean latent heat was 10.75 W m−2 which is quite low when compared to (60-80 W m−2 ) in two adjacent boreal lakes studied by Venälalneh et al. (1999). The latent heat was higher in the daytime than in the night. This is likely because, during the daytime the water warms up due to the incoming amount of solar radiation and the river is able to store up high amount of energy fluxes. However according to Venäläinen et al. (1999) during the night when the solar radiation is decreasing, the lake begins to lose some of the energy it has stored or accumulated during the day thereby reducing the latent heat fluxes.

5 CONCLUSION

The eddy covariance method which combines general meteorology has been used in a three week long study of carbon dioxide fluxes in Kattstrupeforsen (Sweden) during the late winter. An upward mean directed CO2 fluxes measured in the river is approximately 2.2 μmol m−2 s−1 This value is in agreement with many small and large rivers where CO2 fluxes has been studied including the lower amazon river. The river can also be said to serve as source of CO2 to the atmosphere in the day due to the dominant upward fluxes recorded during the daytime. The results also show that carbon dioxide fluxes were increasing with increasing wind speed notably at wind speed above 2 m s−1. There was no influence of stream discharge on the carbon dioxide fluxes in the river. The conclusion that can be drawn in the studies is that, the evasion of carbon dioxide fluxes from a controlled boreal river is highly controlled by wind speed. Other factors which are not known yet are likely to be an important factor driving the CO2 fluxes when the wind speed is weak. The CO2 fluxes shows a strong dependence at high wind speed and the stream discharge shows no correlation with the CO2 fluxes. It is still not clear how other factors would have an influence on the rate of carbon dioxide emissions in the river if wind speed is calm or weak with a very low stream flow. Long term studies in the site will give a clearer understanding of the other parameters that may influence the exchange of carbon dioxide fluxes in the river.

25

6 REFERENCE

Abril, Commarieu, Sottolichio, Bretel, & Guérin. (2009). Turbidity limits gas exchange in a large macrotidal estuary. Estuarine, Coastal and Shelf Science, 83(3), 342-348.

Alin, S., Fátima F. L. Rasera, M., Salimon, C., Richey, J., Holtgrieve, G., Krusche, A., & Snidvongs, A. (2011). Physical controls on carbon dioxide transfer velocity and flux in low-gradient river systems and implications for regional carbon budgets. Journal of Geophysical Research. Biogeosciences, 116(1), Journal of Geophysical Research. Biogeosciences, 2011, Vol.116(1).

Ames, H. (2018). Factors Affecting a River's Velocity. Sciencing. Retrieved from http://sciencing.com/factors-affecting-rivers-velocity-8223150.html. (assessed 05.07.18) Anderson, D. E., R. G. Striegl, D. I. Stannard, C. M. Micherhuizen, T. A. McConnaughey, and J. W. LaBaugh (1999), Estimating lake-atmosphere CO2 exchange, Limnol. Oceanogr., 44, 988– 1001.

Andersson, A. (2017). Air-sea exchange of O2 and CO2. Processes controlling the transfer aquatic metabolism. Mar. Freshwater Res., 52, 101–110.

Andersson, A., A. Rutgersson, and E. Sahlée, (2014). Using a High-Frequency Fluorescent Oxygen Probe in Atmospheric Eddy Covariance Applications. Journal Of Atmospheric And Oceanic Technology, 31(11), 2498-2511.

Asher, W.E.; Karle, L.M.; Higgins. (1997) B.J. On the differences between bubble-mediated air-water transfer in freshwater and seawater. J. Mar. Res. 1997, 55, 813–845.

Aubinet, M., T. Vesala, and D. Papale (Eds.), (2012). Eddy Covariance: A Practical Guide to Measurement and Data Analysis. Springer, Dordrecht, Heidelberg, London, New York, 442

Baldocchi, D., (2012). Advanced topics in biometeorology and micrometeorology. Department of Environmental Science, UC- Berkeley, California

Beaulieu, J. J., W. D. Shuster, and J. A. Rebholz (2012), Controls on gas transfer velocities in a large river, J. Geophys. Res., 117, G02007, doi:10.1029/2011JG001794

Berg, P., & Pace, M. (2017). Continuous measurement of air-water gas exchange by underwater eddy covariance. Biogeosciences, 14(23), 5595-5606.

Borges, A., Vanderborght, V., Schiettecatte, J., Gazeau, L., Ferrón-Smith, F., Delille, S., & Frankignoulle, B. (2004). Variability of the gas transfer velocity of CO 2 in a macrotidal estuary (the Scheldt). Estuaries, 27(4), 593-603.

26

Burba, G., (2013). Eddy Covariance method for Scientific, Industrial, Agricultural, and Regulatory Applications: A Field Book on Measuring ecosystem Gas Exchange and Areal emissions Rate. LI-COR Biosciences, Lincoln, NE, USA, 331 pp.

Ciais, P. in Climate Change (2013): The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change 465–570 (Cambridge Univ. Press, 2014)

Clark, J. E., Schlosser, R. Wanninkhof, H. J. Simpson, W. S. E Schuster, And D. T Ho. (1995). Gas transfer velocities for SF, and ‘He in a small pond at low wind speeds. Geophys. Res. Lett. 22: 93-96.

Clark, J . F., R. Wanninkhof P. Schlossera, Nd H. J. Simpson. (1994). Gas exchange rates in the tidal Hudson river using a dual tracer technique. Tellus 46B:264-285.

Cole, J. J., and N. F. Caraco (2001), Emissions of nitrous oxide (N2O) from a tidal, freshwater river, the Hudson River, New York, Environ. Sci. Technol., 35(6), 991–996, doi:10.1021/es0015848.

Cole, J. J., N. E Caraco, G. W. Kling, And T. K. Kratz. (1994). Carbon dioxide supersaturation in the surface waters of lakes. Science 265: 1568-1570.

Cole, J., & Caraco, N. (1998). Atmospheric exchange of carbon dioxide in a low‐wind oligotrophic lake measured by the addition of SF6. Limnology and Oceanography, 43(4), 647-656.

Crusius, J., & Wanninkhof, R. (2003). Gas transfer velocities measured at low wind speed over a lake. Limnology and Oceanography, 48(3), 1010-1017.Cycles, 25, GB3025, doi:10.1029/2010GB003975

Downing, J. A., Prairie, Y. T., Cole, J. J., Duarte, C. M., Tranvik, L. J., Striegl, R. G., McDowell, W. H., Kortelainen, P., Caraco, N. F., Melack, J. M. & Middelburg, J. J. (2006). The global abundance and size distribution of lakes, ponds, and impoundments. Limnol. Oceanogr. 51: 2388–2397

Garbe, C. S., A. Rutgersson, J. Boutin, B. Delille, C. W. Fairall, N. Gruber, J. Hare, D. Ho, M. Johnson, G. de Leeuw, P. D. Nightingale, H. Pettersson, J. Piskozub, E. Sahlee, W.-T. Tsai, B. Ward, D. K. Woolf, and C.J. Zappa (2014). Transfer across the air-sea interface, in Ocean-Atmosphere Interactions of Gases and Particles, edited by P. S. Liss and M. T. Johnson, pp. 87pp

Grill, I M., And Others. (1988). Tropospheric methane from an Amazonian floodplain lake. J. Geophys. Res. 93: 1564-1570.

Hare, J., C. Fairall, W. McGillis, J. Edson, B. Ward, and R. Wanninkhof (2004), Evaluation of the NOAA/COARE air‐sea gas transfer parameterization using GasEx data, J. Geophys. Res., 109, C08S02, doi:10.1029/ 2003JC002256.

27

Harrison, E. L., F. Veron, D. T. Ho, M. C. Reid, P. Orton, and W. R. McGillis (2012), Nonlinear interaction between rain‐ and wind‐induced air‐water gas exchange, J. Geophys. Res., 117, C03034, doi:10.1029/2011JC007693.

Heikinheimo, M., M. Kangas, L. Tourula, A. Vena¨la¨inen, and S. Tattari (1999), Momentum and heat fluxes over lakes Ta¨mnaren and Ra˚ksjo¨ determined by the bulk-aerodynamic and eddy-correlation method, Agric. For. Meteorol., 98 – 99, 521 – 534.

Heiskanen, J.J., Mammarella, I., Haapanala, S., Pumpanen, J., Vesala, T., MacIntyre, S. and Ojala, A. (2015). Lake-atmosphere greenhouse gas exchange in relation to atmospheric forcing and lake biogeochemistry. Tellus Ser. B-Chem. Phys. Meteorol., 66, 22827

Ho, D. L., W. E. Asher, L. F. Bliven, P. Schlosser, and E. L. Gordan. (2000). On the mechanisms of rain-induced air-water gas exchange. Journal of Geophysical Research 105:24045–24057

Ho, D. T., C. J. Zappa, W. R. McGillis, L. F. Bliven, B. Ward, J. W. H. Dacey, P. Schlosser, and M. B. Hendricks (2004), Influence of rain on air-sea gas exchange: Lessons from a model ocean, J. Geophys. Res., 109, C08S18, doi:10.1029/2003JC001806.

Ho, D. T., L. F. Bliven, R. Wanninkhof, and P. Schlosser (1997), The effect of rain on air-water gas exchange, Tellus, Ser. B, 49, 149–158, doi:10.1034/ j.1600-0889.49.issue2.3.x. Ho, D. T., N. Coffineau, B. Hickman, N. Chow, T. Koffman, and P. Schlosser (2016), Influence of current velocity and wind speed on air-water gas exchange in a mangrove estuary, Geophys. Res. Lett., 43, 3813–3821, doi:10.1002/ 2016GL068727

Ho, D.T., Law, C.S., Smith, M.J., Schlosser, P., Harvey, M., Hill, P., (2006). Measurement of air–sea gas exchange at high wind speeds in the Southern Ocean: implications for global parameterizations. Geophys. Res. Letter, 33:L16611.

Huotari, J., A. Ojala, E. Peltomaa, A. Nordbo, S. Launiainen, J. Pumpanen, T. Rasilo, P. Hari, and T. Vesala (2011), Long-term CO2 flux measurements over a boreal lake: Five years of eddy covariance data, Geophys. Res. Lett., 38, L18401, doi:10.1029/2011GL048753.

IPCC, Climate Change (2013): The Physical Science Basis. Contribution of Working Group 1 to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1535pp. ISBN 9789155498061.

Jähne, B., K. O. Münnich, R. Bösinger, A. Dutzi, W. Huber, and P. Libner (1987), On the parameters influencing air‐water gas exchange, J. Geophys. Res., 92(C2), 1937–1949, doi: 10.1029/JC092iC02p01937

28

Johnson, M. S., J. Lehmann, E. G. Couto, J. P. N. Filho, and S. J. Riha, (2006): DOC and DIC in Jonsson, A., Karlsson, J. & Jansson, M. (2003). Sources of carbon dioxide supersaturation in clearwater and humic lakes in northern Sweden. Ecosystems 6: 224–235, doi:10.1007/s10021-002-0200-y.

Joseph R. Crosswell. (2015). Bubble Clouds in Coastal Waters and Their Role in Air-Water Gas Exchange of CO2. Journal of Marine Science and Engineering, 3(3), 866-890.

Kling, G. W., G. W. Kipphut, And M. C. Miller. (1991). Arctic lakes and streams as gas conduits to the atmosphere: Implications for tundra carbon budgets. Science 251:298–301. Lafleur, P., Mccaughey, J., Joiner, D., Bartlett, P., & Jelinski, D. (1997). Seasonal trends in energy, water, and carbon dioxide fluxes at a northern boreal wetland. Journal Of Geophysical Research-Atmospheres, 102(D24), 29009-29020.

Lee, X., W. Massman, and B. Law (Eds.), (2004). Handbook of Micrometeorology: A Guide for Surface Flux Measurement and Analysis. Kluwer Academic Publishers, Dordrecht, The Netherlands, 252 pp.

LI-COR Biosciences, (2009). LI-7500A Open Path CO2/ H2O Analyzer Instruction Manual. Publication No.984-10563, 127 pp.

Lindegarth M, Carstensen J, Drakare S, Johnson RK, Nyström Sandman A, Söderpalm A, Wikström S A (Editors). (2016). Ecological Assessment of Swedish Water Bodies; development, harmonisation and integration of biological indicators. Final report of the research programme WATERS. Deliverable 1.1-4, WATERS report no 2016:10. Havsmiljöinstitutet, Sweden.

Lindsay. Y (2015). Estimating the Carbon Footprint of Lakes and Reservoirs, Retrieved from

http://z0ku333mvy924cayk1kta4r1-wpengine.netdna-ssl.com/wp-content/uploads/LakeLine/35-3/Articles/35-3-5.pdf (assessed 15.06.18)

Liu, L., T. J. Wang, Z. H. Sun, Q. G. Wang, B. L. Zhuang, Y. Han, and S. Li, (2012): Eddy covariance tilt corrections over a coastal mountain area in South-east China: Significance for near-surface turbulence characteristics. Adv. Atmos. Sci., 29(6), 1264–1278, doi: 10.1007/s00376-012-1052-9.

Macintyre, S., R. Wanninkhof, And J. I? Chanton. (1995). Trace gas exchange across the air-water interface in freshair-waters and coastal marine environments, p. 52-97. In P A. Matson and R. C. Harriss reds.], Biogenic trace gases: Measuring emissions from soil and water. Blackwell.

Massman, W., and X. Lee, (2002). Eddy covariance flux corrections and uncertainties in long-term studies of carbon and energy exchanges. Agricultural and Forest Meteorology, 113(4): 121-144

29

Meybeck, M., (1993). Riverine transport of atmospheric carbon: source, global typology and budget. Water Air and Soil Poilu.70: 443-463.

Moore, M.L. (1989). NALMS management guide for lakes and reservoirs. North American Lake Management Society, P.O. Box 5443, Madison, WI, 53705-5443, USA (http://www.nalms.org).

National Oceanic and Atmospheric Administration. Retrieved from

https://climate.nasa.gov/climate_resources/24/graphic-the-relentless-rise-of-carbon-dioxide/ (assessed 05.05.18)

Nightingale, P. D., and P. S. Liss (2003), Gases in seawater, in The Oceans and Marine Geochemistry, vol. 6, A Treatise on Geochemistry, edited by H. D. Holland, and K. K. Turekian, pp. 49–82, Elsevier, Amsterdam

Norman, M. (2013). Air-Sea Fluxes of CO2: Analysis Methods and Impact on Carbon Budget. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology, 2013. of Science and Technology 1471. 42 pp. Uppsala: Acta Universitatis Upsaliensis.

P. & Vesala, T. (2011). Long-term direct CO2 flux measurements over a boreal lake: Five years of eddy covariance data. Geophys. Res. Lett. 38, L18401, doi:10.1029/2011GL048753.

Podgrajsek, E. (2015). Lake Fluxes of Methane and Carbon Dioxide. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology, 2015. Podgrajsek, Sahlée, & Rutgersson. (2015). Diel cycle of lake‐air CO2 flux from a shallow lake and the impact of waterside convection on the transfer velocity. Journal of Geophysical Research: Biogeosciences, 120(1), 29-38. Q. J. R. Meteorol. Soc., 106, 85–100.

Rasera, M. F. F. L., et al. (2008), Estimating the surface area of small rivers in the southwestern Amazon and their role in CO2 outgassing, Earth Interact., 12(6), 1–15.

Raymond, P., & Cole, A. (2001). Gas exchange in rivers and estuaries: Choosing a gas transfer velocity. Estuaries, 24(2), 312-317.

Raymond, P., Jens Hartmann, Ronny Lauerwald, Sebastian Sobek, Cory Mcdonald, Mark Hoover., Peter Guth. (2013). Global carbon dioxide emissions from inland waters. Nature, 503(7476), 355-9.

Raymond, P., Caraco, A., & Cole, N. (1997). Carbon dioxide concentration and atmospheric flux in the Hudson River. Estuaries, 20(2), 381-390.

Rutgersson, A., Smedman, A., (2010). Enhanced air–sea CO2 transfer due to waterside convection. J. Marine Syst., 80:125–134.