An evaluation of a newly developed method with required beneficial qualities for

An evaluation of a newly developed method with required beneficial qualities for

Department of Thematic Studies

Campus Norrköping

An evaluation of a newly developed method with required beneficial qualities for measuring pCO2 from fresh water

Test-study performed in a small boreal stream network, south west of Sweden during

March – October 2013 and 2014

Lunden Madelene

Bachelor of Science Thesis, Environmental Science Programme, 2015

Linköpings universitet, Campus Norrköping, SE-601 74 Norrköping, Sweden

Språk

Language

Svenska/Swedish

Engelska/English

________________

Institution, Avdelning

Department, Division

Tema Miljöförändring,

Miljövetarprogrammet

Department of Thematic Studies – Environmental change

Environmental Science Programme

Rapporttyp

Report category

Licentiatavhandling

Examensarbete

AB-uppsats

C-uppsats

D-uppsats

Övrig rapport

________________

URL för elektronisk version

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

Datum 2015-06-03

Date 2015-06-03

ISBN

_____________________________________________________

ISRN LIU-TEMA/MV-C—15/16--SE

_________________________________________________________________

ISSN

_________________________________________________________________

Serietitel och serienummer

Title of series, numbering

Handledare David Bastviken

Tutor David Bastviken

Title An evaluation of a newly developed method with required beneficial qualities for measuring pCO2 from fresh water Test-study

performed in a small boreal stream network, south west of Sweden during March - October 2013 and 2014

Author Madelene Lunden

Sammanfattning

Forskning har visat att små skogsbäckar släpper ut CO2 och på så sätt bidrar de till naturliga utsläpp av växthusgaser och klimatförändringar. Studiernas resultat är dock debatterade då de till att börja med används olika metoder för att dra dessa slutsatser, vilka pekar på både olika kvantiter av CO2 utsläpp och också på olika påverkan från hydrologiska och kemiska faktorer. På gru nd av detta har en alldeles ny metod utvecklats. Den grundar sig på en uppochnervänd kammare placerad i vatten som med hjälp av en inbyggd sensor fångar upp och mäter partialtrycket CO2 (pCO2) i vattnet. En fördel är att den är byggd av enkelt material til l lågt pris, vilket gör att metoden kan användas på många platser samtidigt, då man har råd med ett stort antal kammare. Målet för avhandlingen

är att utvärdera nyttan av denna nya metod genom att köra flera statistiska analyser på de insamlade uppgifterna och även gen om att jämföra utvecklingen från statistiska resultaten till andra metoder statistiska trender av pCO2. Studien kan visa att vattenhastigheten påverkar pCO2 från bäckarna, och korrelationens struktur är betydande på vattennivån i bäcken. Man kan även dra slutsatser om att det finns en dygnsrytm i hur CO2 släpps ut från bäckar med högsta utsläpp under förmiddagen och lägsta på eftermiddagen. Detta skulle kunna bero på att CO2 utsläpp beror på fotosyntes/respiration och/eller temperaturskillnader. Dessa slutsatser är väldigt intressanta för att bygga på kunskapen om hur kolemissioner från sötvatten påverkas av klimatförändringarna. Om man utökar studien med fler möjliga variabler för att studera hur andra miljöfaktorer påverkas pCO2 och modifiera metoden och datainsamlandet nå got så kan detta vara en metod värd att fortsätta användas.

Abstract

Studies have concluded that streams emit CO2, which indicates that natural sources of Greenhouse Gases can contribute to climate change feedback. Why this is of interest is to be able to make reliable climate models. These studies are however debated, since there are different methods to measure CO2 evasion from streams which conclude that different hydrological and chemical factors are affecting the gas exchange the most. It is based on an upside-down-placed chamber in the streams, containing a sensor which is able to directly measure the partial pressure of

CO2 (pCO2) in streams. An advantage with this method is that it is built on cheap equipment and therefore can be afforded to cover a big catchment with differing hydrological factors. The aim for the thesis is to evaluate the usefulness of this new method by running several statistical analyses on the collected data and also by comparing the trend from the statistical results to other methods statistical trends of pCO2. What can be concluded by this study is that discharge affects the pCO2 in streams and it often appears with a negative correlation. Also, diurn al patterns of pCO2 seem to appear, with a peak before lunch and minimum levels in the afternoon, which could indicate that pCO2 are dependent on photosynthesis/respiration and/or temperature. These conclusions are of interest to understand how C acts in freshwater and respond to the climate change. The study has to be extended with investigation of how more factors affect pCO2 and also some improvement for the method, before it can be fully used.

Nyckelord: pCO

2

, direkt mätning, ny kammarmetod, efterfrågade fördelar

Keywords: pCO

2

, direct measuring, new chamber method, required qualities

Abstract

Studies have concluded that streams emit CO2, which indicates that natural sources of

Greenhouse Gases can contribute to climate change feedback. Why this is of interest is to be able to make reliable climate models. These studies are however debated, since there are different methods to measure CO2 evasion from streams which conclude that different hydrological and chemical factors are affecting the gas exchange the most. It is based on an upside-down-placed chamber in the streams, containing a sensor which is able to directly measure the partial pressure of CO2 (pCO2) in streams. An advantage with this method is that it is built on cheap equipment and therefore can be afforded to cover a big catchment with differing hydrological factors. The aim for the thesis is to evaluate the usefulness of this new method by running several statistical analyses on the collected data and also by comparing the trend from the statistical results to other methods statistical trends of pCO2. What can be concluded by this study is that discharge affects the pCO2 in streams and it often appears with a negative correlation. Also, diurnal patterns of pCO2 seem to appear, with a peak before lunch and minimum levels in the afternoon, which could indicate that pCO2 are dependent on photosynthesis/respiration and/or temperature. These conclusions are of interest to understand how C acts in freshwater and respond to the climate change. The study has to be extended with investigation of how more factors affect pCO2 and also some improvement for the method, before it can be fully used.

Sammanfattning

Forskning har visat att små skogsbäckar släpper ut CO2 och på så sätt bidrar de till naturliga utsläpp av växthusgaser och klimatförändringar. Studiernas resultat är dock debatterade då de till att börja med används olika metoder för att dra dessa slutsatser, vilka pekar på både olika kvantiter av CO2 utsläpp och också på olika påverkan från hydrologiska och kemiska faktorer. På grund av detta har en alldeles ny metod utvecklats. Den grundar sig på en uppochnervänd kammare placerad i vatten som med hjälp av en inbyggd sensor fångar upp och mäter partialtrycket CO2 (pCO2) i vattnet. En fördel är att den är byggd av enkelt material till lågt pris, vilket gör att metoden kan användas på många platser samtidigt, då man har råd med ett stort antal kammare. Målet för avhandlingen är att utvärdera nyttan av denna nya metod genom att köra flera statistiska analyser på de insamlade uppgifterna och även genom att jämföra utvecklingen från statistiska resultaten till andra metoder statistiska trender av pCO2. Studien kan visa att vattenhastigheten påverkar pCO2 från bäckarna, och korrelationens struktur är betydande på vattennivån i bäcken. Man kan även dra slutsatser om att det finns en dygnsrytm i hur CO2 släpps ut från bäckar med högsta utsläpp under förmiddagen och lägsta på eftermiddagen. Detta skulle kunna bero på att CO2 utsläpp beror på fotosyntes/respiration och/eller temperaturskillnader. Dessa slutsatser är väldigt intressanta för att bygga på kunskapen om hur kolemissioner från sötvatten påverkas av klimatförändringarna. Om man utökar studien med fler möjliga variabler för att studera hur andra miljöfaktorer påverkas pCO2 och modifiera metoden och datainsamlandet något så kan detta vara en metod värd att fortsätta användas.

1. Introduction

Until 1750, the Greenhouse Gas (GHG) concentration in the atmosphere had been relatively stable for thousands of years (Denman et.al, 2007). Due to the industrial revolution, the atmosphere has been affected by anthropogenic activities, leading to an increase of GHGs and a changing climate. There are mainly three GHGs, carbon dioxide (CO

2

), methane (CH4) and nitrous oxide (N2O), affected by human activities (Denman et.al, 2007). The emissions of

CO

2

are critical as CO

2 is considered to account for most of the radiative forcing. Therefore it is needed to fully understand the global C cycle, its processes and the amount of C it includes

(Cole et.al, 2007). The climate models, used by IPCC 2007 (Denman et.al, 2007), are calculated with the assistance of the C cycle as in Figure 1. The figure shows how IPCC assumes that the freshwater system on earth was not considered active in the C cycle until

2007. Since anthropogenic emissions have been in focus for a long time natural emissions are poorly understood (Cole et.al, 2007). Today there is an ongoing discussion on how large the freshwater CO2 flux are and how they are regulated (Teodoru et.al, 2009). To further be able to determine if freshwater acts as a sink or a source of CO

2

, or just a neutral pipe delivering C from the terrestrial to the oceanic ecosystem, more research is needed (Tranvik et.al, 2009), especially when the natural sources of CO

2 are overseen (Cole et.al, 2007). C that reaches the freshwater system has basically three alternative fates: reaching the ocean, the great global C sink, being emitted to the atmosphere as CO

2 or become stored into the sediment (Tranvik et.al, 2009). The quantities of relative importance of these C fates are unknown and debated.

Therefore it is critical to study the C balance of the freshwater system further (Fiedler et.al,

2006) to be able to predict the climate change more precise.

Important to have in mind is that aquatic freshwater systems exists of components such as large rivers, lakes, reservoirs, groundwater and smaller streams (Cole et.al, 2007) and probably because the freshwater cover a relatively small area of the globe, less than 3% according to a study of Raymond et.al, (2013), that naturally C gas exchange have not yet received the attention to the global modeling as it perhaps should have had (Cole et.al, 2007).

Now it is assumed that the ecosystem of freshwater is not an inactive pipe and does both emit

GHGs as well as store C in the sediment (Billet and Harvey 2013). The remaining question is at what level and under which circumstances these processes are the greatest. Lakes and rivers in aquatic areas are often supersaturated with CO

2

(Kling et.al, 1991; Teodoru et.al, 2009), about 95% of the times (Raymond et.al, 2013). This could as well be the case in smaller streams and Cole et.al, (2007) suggest that streams are net sources of CO

2 into the atmosphere. Even if scientists have concluded that streams do release CO

2

(Tranvik et.al,

2009) and that it significantly affect the climate change (Abril et.al, 2013), it is still a frontline question since it is poorly understood. Even in spite of a lot of calculations from e.g. Cole et.al, (2007); Fiedler et.al, (2006) and measurements by e.g. Kling et.al, (1991); Billett and

Harvey (2013). More studies about C losses are requested, especially for small, boreal streams and this thesis will focus on that landscape, a study from southwest of Sweden.

To estimate the CO

2 flux from the stream into the atmosphere one need to know the concentration of the partial pressure of CO

2

(pCO

2

) and the piston velocity i.e. the gas exchange velocity (k) in the streams (Wallin et.al, 2011). This thesis will only have focus on the measuring of pCO

2

.

To make estimations of the pCO

2

there are mainly two different ways, using either an indirect calculation based method or a direct measuring method. There are complications known for both of these methods, and consequently a new direct chamber method is recently developed by Bastviken et.al, (2015). This method has never before been used in large-scales studies, but it has great potential to be a valuable method and details for both the established and the new

method are described further below in section 2. Background. The aims for the thesis are then to evaluate how this method is working, its advantages and what improvements can be done during running these chambers. Note that not a comparison between the different methods normally used can be performed in this thesis since lack of time and such information are available for this study. However there will be running statistical tests which trends are compared to other studies result to see if this new chamber method calculates the same trends and another type of validation of the new chamber method is also possible to make. The research questions are developed to be able to perform the aim of the thesis and are: i) How does discharge affect pCO

2

in streams and what correlations are there? ii) Which similarities or variability’s of spatial difference exists between and in-between locations at Skogaryd with pCO

2

exists? iii) Which similarities or variability’s of temporal difference as diurnal and seasonal aspects exists with pCO

2

at Skogaryd? iv) How do this study’s results differ or look alike earlier studies results with pCO

2 patterns in boreal streams and how can that reveal how this new method are useful?

The thesis contain a background with a comparison of existing theories and common methods for measuring pCO

2

. Later on a section contain a detailed description of the new method developed and the structure of measuring pCO

2

for this project. A result section which will present what findings this sampled data resulted in. Continuing with a discussion about the spatiotemporal patterns observed and how these can be compared with previous studies and earlier findings. In the final, the thesis contains a discussion about how this new method has been working out and how it can be improved for continued use.

2. Background

To determine the CO

2

gas exchange in rivers, a method developed by Kempe is common today. This method is indirect and based on calculating CO

2

concentrations from measured pH, total alkalinity (TA) and water temperature (WT) (Kempe, 1984 cited by Raymond et.al.

2013). This has been the most commonly used method until today, because pH, TA, and WT are commonly available from water monitoring programs (Raymond et.al. 2013). The basis is the carbonic acid equilibrium system in water, where alkalinity is assumed to represent hydrogen carbonate and carbonate, but several studies have developed these calculations further to include Dissolved Inorganic Carbon (DIC), ions, dissolved silica or suspended materials (SPM), to improve the result (Lauerwald et.al, 2013; Venkiteswaran et.al, 2014).

The other method used, is instead based on direct measurements of pCO

2

using the headspace-equilibration-technique (Raymond et.al. 1997). A commonly used approach of this technique is to fill a bottle with stream water, close it, make a gas headspace by adding air while removing some of the water, and shake it till the CO

2

in the water have equilibrated with the headspace. A small headspace to water volume ration is desired. CO

2

partial pressure measured in the headspace (by e.g. gas chromatography; Billett et.al. 2004) represent the pCO2 (Cole et.al. 1994). Complementary to the pCO

2

estimation, one also collect the atmospheric CO

2

concentration right above the water surface and calculate differences, to evaluate if the stream water are supersaturated or not (Cole et.al. 1994). The first direct measurements exist from mid-90 and did also indicate a result in line with indirect method, i.e. that freshwaters were generally supersaturated by CO

2

(Cole, et.al. 1994; Raymond et.al.

1997).

Needed for all methods are to make reliable estimations of pCO

2

in big datasets that cover up both spatial and temporal variability’s (Raymond et.al, 2013). Datasets collected today, often

contains a gap since the data hasn’t been sampling during the whole day and night-times data are consequently missing. The dilemma is that respiration and photosynthesis affect the pCO

2 differently and are occurring during different times of the day and the whole picture of the C cycle therefore might be missing (Bastviken et.al, 2015). As important is to collect data for differing locations as well to enable consideration of spatial variability. Without such measurements it is impossible to get a complete, reliable dataset, which is required for an upscaling of the C cycle (Fidler et.al, 2006). Due to this one can find problems with the methods being used today. Measuring of the pH has been proven often to be uncertain as well as TA has shown to be, especially in low alkalinity environment, and a overestimation of CO

2 evasion rate are likely from the indirect method (Abril et.al, 2015). Meanwhile the direct method is costly since one has to go out in field to collect data, each time pCO

2

estimation is performed and also the direct measuring require expensive equipment, resulting in fewer collections. Therefore Bastviken et.al, (2015) have established this new method. A direct method, based on the head-space idea, but built with an upside-down-placed chamber in the water, with an ability to be in the water measuring pCO

2

frequently for days. Since it is built with inexpensive equipment it will result in the ability to use a lot of chambers which consequently can cover up data collection for a great area to find spatial varieties (Bastviken et.al, 2015). Another reason why this could be a preferable method is also due to its possibilities to measure CO

2 during the whole day, which today is a common data-gap

(Bastviken et.al, 2015).

3. Affecting factors of pCO

2

in fresh water

Even if scientists obviously have discovered that streams are supersaturated and have a net

CO

2 flux into the atmosphere, opinions are divided in the quantification of the C.

Uncertainties due to what factors do affect the most therefore remains to be discovered.

3.1 Discharge

The possibility with a correlation with pCO

2

and discharge is discussed (Billett et.al, 2004;

Billett and Harvey 2013; Fiedler et.al, 2006) and findings indicates that higher turbulence probably is a main driver of CO

2

(Wallin et.al, 2013) and CO

2 losses from stream to atmosphere are concluded to be large during high flow events (Billet and Harvey 2013).

Scenarios that slow down the CO

2 losses are a slow water movement and especially with a combination of high presence of nutrient, Then photosynthesis could inhibit the CO

2 loss from the aquatic to the atmosphere (Kling et.al, 1991) and could also contribute to make the downstream lake into a carbon sink, for a short period (Tranvik et.al, 2009).

3.2 Spatial variability

CO

2 losses might vary much due to study site and its different ecosystems (Fiedler et.al, 2006;

Cole et.al, 2007) and soil composition (Fiedler et.al, 2006). If a stream is highly supersaturated, CO

2 loss will happen rapidly and hotspots of CO

2 degasing exists and consequently pCO

2

can differ even between a short stream section in the same catchment

(Billett and Harvey 2013; Fiedler et.al, 2006). CO

2 evasion rate seems to be dependent of hydrological characteristics of the stream and considerably spatial variability probably exit but is rarely quantified (Billet and Harvey 2013).

3.3 Temporal variations

A scenario could be that a slowdown of CO

2 losses are correlated with a high presence of nutrient and/or with a slow water movement and streams emissions of C could depend on their primary production (Kling et.al, 1991). Even occasionally make a lake or another reservoir, once again, into a C sink (Tranvik et.al, 2009).

In streams there are likely to be variations between the different seasons. Some correlations analysis indicates greater evasion rate of CO

2

during spring than during autumn (Fiedler et.al,

2006) and probably caused by metabolic activities. Billett et.al, (2004) concluded that during darker seasons, autumn and winter, there are lot more CO

2

losses from streams. This variations between the seasons might be a result of that the photosynthesis and respiration have changed patterns between summer and winter. pCO

2

as well seems to have positive correlation regarding to temperature (Kling et.al, 1991), which indicates there are likely to be a bigger quantity of pCO

2 in streams during these darker seasons. There could also be variations explained by diel patterns. It could be caused by photosynthesis and respiration

(Dawson et.al, 2001).

3.4 Upscaling

One problem with upscaling stream fluxes is that stream ecosystems vary considerable in time

(e.g. due to flooding or by periodically drying out), in contrast to lakes and big rivers that are more stable over time (Teodoru et.al, 2009). They have a smaller buffering capacity, which result in a complex field to model with highly differing CO

2

evasion patterns. Billett and

Harvey (2013) pointed out that site-specific pattern are easy to detect due to CO

2 degassing.

This confirms how an upscaling of the C cycle then is a difficult task, since pCO

2

data can be varying within the same stream system (Billett and Harvey 2013). They stress for the great importance of pCO

2

data collected for as many locations as possible. Since it seems to be very depending on the site how C are acting, there is a need to have realistic data, that can cover many different sites, temperatures and other hydrological factors. (Denman et.al, 2007).

To then get a more realistic picture of the gas exchange there is a need for more data (Wallin et.al, 2013) that can cover up gas losses patterns from different landscapes (Humborg et.al,

2010; Teodoru et.al, 2009).

4. Material and method

4.1 Study site

The samples of pCO

2

were collected from a stream network in south west of Sweden, at the

Skogaryd Research Catchment station, located 10 km north of Gothenburg. Samples were taken from several locations in a stream network, floating through the boreal landscape. The stream network is split into different sections due to intermediate lakes. The location mainly exists of spruces and pines and land consist of mires, lakes and streams, as are typical for

South of Sweden (Karlsson, 2015).

4.2 Sample collection

The chambers locations were differing slightly between the years (Figure 2 and Figure 3).

During 2013 the sampling took place during end of April – end of September with 8 different chambers and a sampling interval of 1 hour. In 2014 the numbers of chambers were increased to 20. The sampling interval remained to be 1 hour but the sampling collection were increased and lasted between mid of March – mid of October. The timespan of 1 hour was because the loggers in the chamber was capable of saving a maximum number of data and therefore sampling more frequently was impossible, to not risk that collected data would have been overwritten. If the intervals would have been set to be shorter more frequent visits to the sample locations would have been needed. During 2013 and 2014 the visit to the chamber has rd differed but has been around every 3 – 4 th

week. In the thesis the visiting periods are named after the Day of the year (DOY) with a starting date as 2013-01-01 due to convenience when making graphs.

During the period when the chambers were out in the stream, discharge data were collected as well. The discharge data were collected at 4 different stations at the catchment area (Figure 2 and Figure 3) and are measuring in the units of liters per second.

4.3 Chambers

The method used in this study is the new direct method, mentioned above in 2. Background, which sample pCO

2

by sensors in floating chambers, developed by Bastviken et.al, (2015).

The chambers were built in the lab to be able to make the method with as cost efficiency equipment as possible and therefore could be used at many locations at the same time in the catchment area. The chambers were built with a body made out of a PVC-bucket. Attached to the bottom of those chambers were battery boxes, which are waterproof and where the battery to the sensor and the connection to the computer were placed. Attached to the bottom was also the sensor box, slightly bigger. The sensor box contains, besides the sensor, drilled ventilation holes on one of the box’s short sides to protect the non-water-resistant sensors. In the sensor box is also a slanting plastic piece, protecting the logger from condensation drops and water splashes. One of the other prize-reducing products for this method was the sensors being used. These were in matter of fact constructed for indoor use and were therefore not waterproof. However they were protected by the sensor box and the slanting plastic piece, so they wouldn’t get destroyed. These loggers were able to measuring the pCO

2

in the streams, and air T and Relative Humidity (RH) as well. These were also capable of collect data frequently in a logging memory. The chambers were put into the stream upside-down and were floating by pieces of Styrofoam rods (Bastviken et.al, 2015). The chambers were tied to the shore at the location, so they had space to float freely, but still weren’t floating away.

When the chambers were put upside down, there was a headspace created. Because of the super-saturation in the stream, the chambers started to accumulate gas leaving the water and being logged by the sensors and in a while, this headspace was with equilibrium to the stream and then the pCO

2

in the stream was logged by the sensor.

The chamber was placed in the stream network in Skogaryd for 2013 (Figure 2) and 2014

(Figure 3) at several locations. The chambers were named by the area they were collecting pCO

2

data from, as can be seen in the Figures. Chambers before a lake were named B and the initial letter of the lake, chambers after a lake were named A and then the initial letter of the lake and side flow joining the main stream were named with an extra J and then the area it’s join in.

4.4 Data analyses

4.4.1 Necessary validation

After data collecting process by the chambers, the loggers were connected to the computer and collected data for pCO

2

, T and RH for each hour and chamber were transferred to excel.

Since the loggers were set to collect data every hour there was a potential to a great collection of data. Before it could be used, it had to go through a validation process, described in this section, which ensured that the collected data was trustworthy. Some problems occurred because the sensors were still exposed to condensation, even though the “sensor box” did a lot of necessary protection. If such condensation happened, the sensors did get unreliable results.

Why that happened was because the loggers not are meant to be in a damp climate and obviously can’t stand such environment for a longer time without getting an error. The error could occur more frequently during high flow event, when the chambers were placed in sunny locations because then condensation on the sensor more easily happened, when the sensors had been used in the field for a while or when it were a dry period and the chambers either were relocated by the water on the ground or in a small pool of non-moving water. Then it

wouldn’t measure a fair pCO

2

value due to streams and these different problems had to be detected. They could be found by different processes, described in coming section.

4.4.2 Validation process

The sensors had a maximum of measuring a value of 10 000 ppm pCO

2

. Therefore condensations on the sensor could easily be detected in the dataset due to high pCO

2

values. If pCO

2

then were at 10 000 ppm, these data points were cut away.

Condensations could appear for the T- and the RH- sensors as well. It appeared as the T or the

RH has a measured value of 376.62, which was the maximum level for these sensors. In these cases, the CO

2 sensor could anyhow had measured useable values of pCO

2

since the condensation only appeared at one part of the sensors (in this case at the T and RH part). The

RH should never exceed 100% and every time it did, correction of the value was performed by change it to 100 since it still symbolized a high humidity. The T correction process was slightly more complex because then data of T had to be known for the temperature before as well as after the time when the sensor started to fail measuring the T. If that were the case, a mean-value of these before and after measuring were taken and put into the missing data points. This could only be done if the T measuring had been down for only two or three times in a row. After that span, too many hours would have passed by and the air T would have been changed too much to just copy. Another backup was then to copy the same T from a chamber close by, if they were proven to have a correlation with each other’s remaining T value. If these two correction processes were unable to perform, the pCO

2

data were cut away as well. Those filtering process by expurgate unrealistic pCO

2

, T and RH values were performed in Microsoft Excel 2010 since it has valuable function system and as well gives an easy overview of the data.

4.4.3 Drift correction

Over time the sensor response may drift resulting in biased measurements since incorrect pCO

2

values are being collected. This could happen after a long-time run of the chambers in the water (Bastviken et.al, 2015). If this were the case, it had to be corrected, a process that were performed individually for each chamber. The drifting was possible to identify when the chambers start and end values of pCO

2

were compared. Since the sensors were activated in the air at the same time, they were collecting pCO

2

in the same air at the same places as each other which should result in the same pCO

2

value measured. They should as well, have measured the same end value of pCO

2

in the air, since when they were picked up from the stream the sensors weren’t closed off until reaching the lab. If one chamber tended to be measuring higher or lower pCO

2

values than the other chambers, then the chambers had been drifting. Since the pCO

2

value can occur as different at different locations in the stream, only start and end values from the air can be compared. This drifting correction process then supposes that there were existing whole datasets from the whole period out in field for the chambers, otherwise the comparisons of end values were impossible. Therefore this was the first step preformed after cutting away the 10 000 pCO

2

values (Figure 4). If some of the chambers were indicating a lower or higher value than the other chambers, it had been drifting. Then it was needed to be corrected by estimating when the start of the drifting had occurred by looking at the collected pCO

2

values in a scatter dot graph.

4.4.4 Last cutting

The data had to be cut for the first hours out in the field, since the loggers were started in the lab and they collected pCO2 data on the way to the stream. Once the chambers were placed in the stream, it took further some hours for them to be in equilibrium with the stream so the first data points were only an increasing of pCO2, which are useless for this study since the

interest lies within pCO2 in the stream locations. Therefore 5 hours of data in the beginning were cut away.

Further quality audit still had to be done, for the pCO2 data in the end of the series. This was performed by zooming into the scattered graphs, as the one in Figure 5, to see where data gaps seemed to have appeared for the measured pCO2. When data gaps were found, which occurred in Figure 5 after DOY 451, the data set were cut away from the that time until the end of that dataset out in field. Why the whole data set were cut away was because then the sensor probably gave unreliable measurements of pCO2. After the cutting, the dataset result to be as in Figure 6. All data for all chambers were cut away since then they were more easily compared to each other.

4.5 Statistical analysis

Correlation analyses are used to see possible relationships between different variables with a linear, normal distributed data (Grandin, 2003) In this study, the correlation were examine the relationship between and in-between different locations at Skogaryd. The correlation analysis could be between pCO

2 and factors as discharge (l/s), months, time of the day and between locations for the chamber in Skogaryd. To examine the diel patterns between different groups at Skogaryd an ANOVA test were executed. ANOVA test compare means of a dependent and an independent variable (Grandin, 2003). Groups were defined by every two hours of the day and these groups became the independent factor and pCO

2 as the dependent variable.

The analyses are performed in IBM SPSS Statistics 22. For all data a significance level at

0.05 was chosen and a confirmation so the data were normal distributed were performed.

5. Result

The result is based on data after the filtration, described in 4.4. Data analyses and the result is only referring to that useable data. Which chambers data were of use can be seen in Figure 7.

All data was normal distributed.

5.1 CO

2

evasion and discharge

In Skogaryd the discharge was strongly correlated between the four measurement stations

(Figure 8 and Table 1), even though they are in different levels. Since they correlate, this study uses discharge data for further correlation analysis from the Mineral station (Figure 2 and Figure 3).

Correlation tests from 2013 showed that all correlations are negative with pCO

2

and discharge, as interpret from Table 2. The correlations are strong, 0,6 in 3 times and 0,7 one time (out of 5 times) but any independent scatters for the correlations are not seen (Figure 9).

These independent scatters seems to be due to the low discharge since when the discharge are null, pCO

2

still can be measured (Figure 9 D-H). Same trend exists for discharge and pCO

2 scatters for 2014, Figure 10.

Correlations with discharge and pCO

2

during 2014 (Table 2) has both positive and negative correlations. Significant negative or significant positive correlations only occur at the same time out in field so there is no positive and negative correlation existing at the same time, however they could occur at same locations. This could be because different discharge rates generate different correlations patterns, as can be seen if comparison between Figure 10 and

Figure 11 are made. Figure 11 visualize independent scatters for a discharge between 2 – 15 l/s meanwhile a low discharge, around 0 – 0,2 l/s, gets scatters that are L shaped (Figure 10 c-

d) and a high discharge, 100 – 250 l/s, results in a positive correlation, and also with a break, but as an upside-down L pattern instead (Figure 10 e-f).

5.2 CO

2

evasion and spatial variations

Spatial comparison of pCO

2

has been for the 1 st

time out in field for both 2013, DOY 114-

127, and 2014, DOY 441-452, due to there were a lot of chambers working for different areas during that time, and that number decreased during the year (Figure 7). Similarities between different locations exists (Figure 12 and 13 and Table 3 and 4) but it is tricky to explain why these specific areas correlate to each other’s pCO

2

and why not other areas does. A comparison between Table 5 and 6 and Figures 12, 13, 14 and 15 indicate that a correlation with pCO

2

are of greater chance between different location than what it is in-between a location, at Skogaryd.

5.3 pCO

2

and seasonal patterns

A temporal examination is performed for two chambers during 2014, BF 8 and BS 20, since these were the only two chambers which were working for 4 respectively 5 measurements periods (Figure 7). There might be difference in the mean of pCO

2

in streams during different month since Figure 16 E-F indicates for a pCO

2

peak during the autumn.

5.4 CO

2

evasion and diel trends

One-way ANOVA tests indicate significant trends for a diel pattern for half of the chambers sampling pCO

2

. The daily pattern can have two possible shapes. The most common significant trend pattern as in Figure 16 A-D is for all chambers but two, BS - H in 2013 and

JAF 11 in 2014, and is visualized in Figure 16 E-F. In the Figures 16 an indication of a daily pattern got peaks of pCO

2

before noon between 6-9 and a through in the afternoon between

14-17. The other significant pattern from the ANOVA test only occurred at two chambers.

This pattern has no defined through, rather more of a slightly increasing during the day and reaches a peak in the afternoon. The peak appears almost the same time as the through appeared for the common significant pattern.

5.5 Cost-efficient method

Since pCO

2

data are being collected at several locations, every hour for a whole summer season, it resulted in a huge quantities of data. The data series, from the beginning contained almost 61 000 data points. However, some complications with this new methods data quantity have occurred along the work. As a result, some of the data has been omitted, by following the process described in section 4.4 Data analyses. After the expurgate, only around 25 % of all data points remained, but still a lot of data was remained and being usefull. The cut were mainly made after 11 days but could last for 17 days, as the most. Sometimes chambers with good pCO

2

results were cut away due to a short time span, 1-8 days, to be able to make the remaining datasets as long as possible (Figure 7).

One other result according the method is that for the first time out in field (every year) a result for plenty of the chambers were collected but every time out in the field, the more chambers had become inoperative. Consequently data had only been collected for a few days with more and more 10 000 peaks and had to be removed due to lack of reliable datasets and an acceptable length of the dataset. The cutting span for 11 – 17 days anyhow remained stable, so only the numbers of chambers were decreasing.

For the measuring in 2013 and 2014 correction of datasets because of condensations was necessary, for T 4% respectively 1% of the times/year and for RH 12% respectively 15% of the time/year. RH has as well been corrected to be at the 100% level when is has been

showing values slightly higher than 100% like 103% etc. This is rather a common phenomenon and has happened 51% and 32% for respectively year.

6. Discussion

6.1 Discharge

Discharge in streams at Skogaryd catchment seems to mainly have negative correlations with pCO

2

, even if positive correlations did occurred. However, negative and positive correlation did never occur during the same time. This then probably explain that correlation with pCO

2 and discharge are not site specific, but rather explained by something happening during the specific time. Studies which have examined this correlation pattern have got different results.

Some gets only positive correlations and some gets only negative correlations with discharge and pCO

2

, but no one has seen these two possible results for the same location. Wallin et.al,

(2011) found result indicating that pCO

2

are highest during spring and autumn and explain this from their findings that k is positive correlated to the slope of the stream. If assumed that a slope should increase the discharge, they got the same result as Campeau et.al, (2013). They both use a direct headspace method, but not a chamber. Other studies also using the same method are Billet et.al, (2004) and Fiedler et.al, (2006) and instead find negative correlations to discharge and pCO

2

in streams. Similar result, with negative correlations with pCO

2

and discharge has also Teodoru et.al, (2009) found in his study, but by using an indirect method.

Dinsmore et.al, (2013) suggest instead that pCO

2

due to discharge can be both negative and positive with discharge in streams, they however only found one pattern at one location, in opposite to this study. They explain the negative correlation with dilution of pCO

2

in the stream during high flows. But, there could as well be other hydrological factors that are affecting the pCO

2

in stream which this study isn’t able to consider. As well as the correlation could be affected by that, Dinsmore et.al, (2013) suggest that pCO

2

might also be dependent on chemical substances like pH, DIC and DOC. Such factors can as well this study don’t examine since lack of such data. What can be concluded by this study is that discharge affects

CO

2 concentrations in streams and it often appears as negative, but sometimes it can be positive. This could be due to the water level in the streams and also has some connection with k, the degasing constant. This seems to open on to conclusions that pCO

2

increases during the initial increase of discharge in the stream (Figure 10) since then probably big quantities of CO

2

-rich water is pushed out of the ground into the stream, but later on the pCO

2 decreases with an increasing discharge due to a correlation with degasing since k correlates positive, according to Campeau et.al, (2013) since then probably less CO

2

-rich water reached the stream, in a slower motions as well.

On the contrary, Ingvarsson (2008) couldn’t find any relationships between discharge and pCO

2

. She has been using a chamber method and suggests that correlations aren’t found since the chamber might disturb the water-physics like turbulence and discharge. Other researchers as well claim that this chamber method are disturbing the discharge and wind effect (Cole and

Raymond 2001) and pCO

2

won’t be able to measure correct. This won’t get backed up from everyone, Bastviken et.al, (2015) concludes that this is an advantageous way to measure pCO

2

. I think that the result of this study and as well e.g. Dinsmore et.al, (2013) and Billet et.al, (2004) could prove that pCO

2

and discharge are correlated, positive or negative, and sometimes even strongly correlated. After this study, I think you can say that it is despite if using the chamber method or not, as the results indicates. This could lean towards a conclusion that the method used in this study should not be accused for not be able to measure pCO

2

since it seems to get same results as other studies. It anyhow needs some more studies with longer and mainly more datasets that indicates for the same result. This is due to the big

filtration processes that were needed during this study. Improvement suggestions can be found in section 6.5 Method improvements.

6.2 Spatial variability’s discussion

The result indicates that pCO

2

could be correlated between areas (table 6 and 7). Researchers who have found spatial variability’s among CO

2 evasion (e.g. Fiedler et.al, 2006; Humborg et.al, 2010; Lauerwald et.al, 2013; Teodoru et.al, 2009) has slightly different theories of which parameters affect the most. As earlier studies indicates, probably chemical factors as

DOC and Si etc. (e.g. Humborg, et.al, 2010; Lauerwald et.al, 2013; Teodoru et.al, 2009) and hydrological conditions as slopes or groundwater inlet etc. (Teodoru et.al, 2009; Fiedler et.al,

2013) can explain this. Wallin et.al, (2011) explains 83% of the spatial variations to hydrological factors as slopes. Since this study has a lack of result for these suggested factors

(besides discharge as a hydrological factor) no clear links to specific explanatory factors can be presented for the correlations between the areas. However, spatial patterns could be observed although sometimes irregular. Correlation in-between locations (Table 5 and Table

7) do not seems to be as common and strong as the correlation between locations (Table 4 and

Table 6) seems to be. This could tend to that area constructions might affect pCO

2

more than specific soil types etc. Like in the in-between areas, maybe a waterfall or a groundwater inlet exist which could affect the pCO

2

a lot, meanwhile a mean for these different areas compared to each other are much more fluently and look-a-like. Interestingly, the area EF is never correlating to any other location but has kind of strong in-between correlations meanwhile the locations AF, JBS and BS are correlating to each other but these locations in-between correlations seems rather absent.

This could consequently be because the quality of the lake Erssjön, separating the EF areas, affect pCO

2

trends due to others areas different. This theory is likewise to Ingvarsson (2008) who got a result with differing pCO

2 in the same stream but at different locations. She has theories that it could depend on what kind of lake the water are flowing from. Suggested that streams are generally very sensitive to hydrological changes and to chemical loads due to its small size (Teodoru et.al, 2009) They found that CO

2 evasions decreases further down in a reach which could also explain this study’s result with differing correlations patterns of pCO

2

.

6.3 Temporal patterns

As only two chambers provided consistent long term data (Figure 8) the seasonal patterns are based on limited data. Pattern from both chambers were consistent with higher pCO2 during autumn (Figure 17). In support of this results, Dawson et.al, (2001) found higher pCO

2

during autumn as well. They write that this could be because of the total amount of carbon that are fixed to the plants are higher during the summer and spring than autumn and winter.

Photosynthesis may then probably play a role, not only for the diel trend, but as well for temporal trends, since during the autumn there are a lot of material in the water, due to the primary production going on during summertime, which increases the respiration. Discharge could as well appear with seasonal trends, and result in seasonal trends with pCO

2

(Teodoru et.al, 2009). This study are however only performed for a third of a year and it is hard to tell if this just a coincidence or not.

6.4 Diel patterns

According to the location of the chambers, two diel patterns were found. The pCO

2

seems therefore to be significant dependent to hours. Dawson et.al, (2001) tried to establish diel variations finding other factors which could be affecting pCO

2

and found no such pattern with discharge, DOC, pH and stream metabolism. As well did not this study find significant diel patterns for discharge.

The two differing significant diel variation might as well be explained by different shadowy pattern. Studies concludes pCO2 with seasonal variations and explains it to be because of photosynthesis as well as they found positive correlations with pCO2 due to WT (Kling et.al,

1991) and probably, WT, could be a possible variable to explaining the diel pattern for pCO2 since the surface water probably has different WT during the day, which could correlate to the time. This can’t however be concluded for this study, since the WT-measuring station was broken.

6.5 Method discussion

Even though there are different methods used to calculate and measure pCO

2

, researchers still discuss about site specific patterns and a returning matter among everyone are that more sampling are needed (e.g. Billett et.al. 2004; Cole et.al. 2007; Teodoru et.al. 2009) Even though data today are over or underestimated, even though the one or the other method are most reliable researchers agrees that there need to be more studies to determine a correct estimation of the C budget (Cole et.al, 2007; Wallin et.al, 2011). Teodoru et.al, (2009) define the importance of streams in boreal ecosystem have a greater exposure of exploring their CO

2 losses. They argue that known for sure that boreal lakes has a significant CO

2 losses in a global scale but there still are uncertainties for the boreal areas. They say, as well as Cole et.al, (2007), that there are a lack of samples collected to make a trustworthy model for the stream-ecosystem, they however conclude that the C losses from streams are far higher than expected and cannot be negligible and have to be including in the aquatic system of carbon process in regional scales.

6.5.1 New method suggestions

Since this are the first big-scaled study performed by this new developed chamber. Some complications with the data has occurred which resulted in some improvement suggestions to get longer, more reliable datasets for next time these chambers are being used. Also to make the data easier to handle and the time-consuming validation process can be reduced. The cutting of the datasets was performed after 11 – 17 days, which indicates that this is the preferable length of time that the chambers can be out in field, as well as Bastviken et.al,

(2015) suggests. As written in the section 5.5 cost-efficient method, a lot of data points were removed, almost 45 000 points. This could absolutely have been valuable data that now have been thrown away. For thesis tests, it was however necessary to have equal length on the datasets and therefore, I estimate, a lot of good data were thrown away. If following suggestions can be applicate in future studies, this big quantity of good data points might not have to be thrown away like in this case but instead can be used for further examinations of the C cycle on the globe.

During this test period of the method, there were chambers which never worked, maybe due to bad sensors or really inappropriate locations. There were chambers which consistently worked for one or two days and there existed chambers which the cutting always was dependent on. Maybe because these sensors needed to be out of the streams for one day and become dried. There were as well chambers that always had datasets longer than the period with data used after filtration. The longest datasets that could be found were for the chambers located at the shadowy places, and notable was that at these places these normally as well existed foam in the water. The shadow could be explained to be reducing the heating inside the chamber and consequently the condensation of water at the sensor, which obviously seems to be a problem. The foam itself may not affect the pCO

2

in the water, but it could be an indications that the water movement have been reduce which might stable the chamber and consequently reduces the potential water splashing on the sensor, which also are indicated as a condensation problem in the data file. That the chambers in the sunny locations gets

unreliable datasets quicker than these in the shadowy places could as well indicate that heating might be the problem. However, this is probably not the case, since the sensors has been tested in lab to see whether they reacts and collect strange measurements during higher temperature, which they indicated not to do (Bastviken et.al, 2015). Therefore, the problem that the chambers in the sun measure fewer reliable pCO

2

data is most likely caused by the condensation from the water caused by the heating, than that the heating itself are affecting the sensors.

Another interesting pattern is that the longer the chambers were deployed, the fewer reliable data were delivered probably because they got affected by the humidity environment they were extended for. This resulted in that the spatial variation analysis could only be tested for the 1 st

time out in field. This has probably to do with the sensors sensitivity to water. The same sensors and chambers were used at the same location for a whole season (season in this case March – October) and never were replaced. This might make them more affected by the unappropriated damp environment these indoor chambers were executed for. Chambers which measured good datasets for the whole year might have had better sensors and/or a more appropriate location which didn’t expose the sensor for the same inappropriate environment.

This also lead to a more demanding filtration and correction of the data being used, which as well might affect the result slightly. This is an important part to improve, especially when one reaches that level that are needed to quantify the amount of pCO

2

and CO

2

evasion from streams, and not only looking for patterns, as in this study. If a more frequently visited routine should be scheduled it could result in more consistent datasets for other locations where the chambers never worked this time as well.

If measuring won’t continue and be more preformed and cover a more various sites, modeling about global carbon budget exposes a great chance to be overestimated. This is why this new chamber method should be used more widely, since it really has potential to cover up these criteria’s, but it still needs to get some more evaluations done.

6.6 Climate change

Major knowledge gaps of the global C budget have been noticed and it is highly debated and therefore a relevant field to study (Cole et.al, 2007). The knowledge gaps for inland waters exist because e.g. data of CO

2 isn’t measured during a whole day (Bastviken et.al, 2015) and without that kind of data an upscaling of the global C cycle is impossible (Cole et.al, 2007).

Small lake or streams do have great potential to have a greater net CO

2 evasion in the world than earlier estimated (e.g. Fiedler et.al, 2006). Inland waters tend to have high spatiotemporal variability in their CO2 dynamics, which makes measurements supporting efficient modelling important (Wallin et.al, 2011).

As already pointed out in the report, the C cycle on the globe has a great deal to do with the climate change (Denman et.al, 2007). If the climate change will occur as predicted the mean temperature on earth will rise, water level will increase, precipitation rate will increase at some parts on the globe and other hydrological factors will have changes patterns A land use change will take place (Denman et.al, 2007) which will affect the total area of inland water on the globe (Raymond et.al, 2013). If today only 0.47% of the land surface are streams and can affect the climate as much as assumed, then what will happen when the fresh water on the globe increases? To continue studies to find out more of the C-cycle on the globe are important so one can be able to make more realistic prediction of the future climate than what today’s climate models are able to tell us, and highly important are then to find out the best possible way to make these estimations.

7. Conclusions

Patterns of pCO

2

have been studies using data from a new method by data collected of this new method. Spatial similarities seem to be stronger between areas than to in-between areas.

This might be because the environment can be very variable on a local basis, e.g. with water falls, while average conditions in larger areas can be more similar. As well correlation due to discharge and pCO

2

are found, and it seems to be very dependent on the discharge rate. Low discharge indicted for positive correlation and vice versa. A diel trend could be observed for pCO

2

in streams, probably because of the balance between photosynthesis and respiration.

The advantages of this new method are clearly the measuring capabilities from the chambers which can measure a lot of pCO

2

during a long time span. If more frequent visits to the catchment area are performed, it will probably turn out to give even longer datasets due to more chambers might be working longer. Another advantage is the simplicity to collect the data, to handle the chambers and to get the data transferred to the computer. An improvement that needs to be done should be the protection of the sensor, especially the CO2 sensor, since it seems to be highly vulnerable to condensations and easily destroy series of data. If that could be done, this method has potential to be a key method to improve the understanding of the C-cycle and GHG emissions from stream water into the atmosphere. Anyhow, I am still encouraging more complementing studies where these improvements have been performed before one could know if this is the most appropriate method for pCO

2

measuring in this study field and also studies which aims will focus on CO

2

flux instead of this primary step of pCO

2

.

8. Acknowledgement

This study has been a project with the aim to evaluate a new method for the highly interesting subject of climate change. I would like to forward a special thanks to my fantastic tutor David

Bastviken who has given me the opportunity to do this study, to handle his data and his chambers. Thanks for coming with great feedback during the writing of this thesis as well. I would also give majors of thanks to the intelligent woman Sivakiruthika Natchimuthu and all the time she has given to me when struggle with data handling had occurred. Also to all her great advices of how to move forward. Also, thanks to Linnea Melander and Razan Koj for been inspiring colleges and have given me the energy to write the thesis when no energy were existing.

9. References

Abril, G., Bouillon, S., Darchambeau, F., Teodoru, R., Marwick, T.R., Tamooh, F., Ochieng

Omengo, F., Geeraert, N., Deirmendjian, L., Polsenaere, P. and Borges, A.V. 2015.

Technical Note: Large overestimation of pCO2 calculated from pH and alkalinity in acidic, organic-rich freshwaters. Biogeosciences. 12:67-78.

Bastviken, D., Sundgren, I., Natchimuthu, S., Reyier, H. and Gålfalk, M. 2015. Technical

Note: Cost-efficient approaches to measure carbon dioxide (CO2) fluxes and concentrations in terrestrial and aquatic environments using mini loggers.

Biogeosciences Discuss. 12:2357–2380

Billett, MF., Palmer, SM., Hope, D., Deacon, C., Storeton-West, R., Hargreaves, K.J.,

Flechard, C. and Fowler, D. 2004. Linking land-atmosphere-stream carbon fluxes in a lowland peatland system. Global Biogeochemical cykles. 18:

Billett, MF. and Harvey, FH. 2013. Measurements of CO2 and CH4 evasion from UK peatland headwater streams. Biogeochemistry. 114:165-181

Campeau, A., Lapierre1, J-F., Vachon, D. and del Giorgio, P. 2013. Regional contribution of

CO2 and CH4 fluxes from the fluvial network in a lowland boreal landscape of Québec.

Global Biogeochem. Cycles, 28:57–69

Cole, J. and Raymond, A. 2001. Technical Notes and Comments Gas Exchange in Rivers and estuaries: Choosing a Gas Transfer Velocity. Estuaries 24(2):312-317

Cole, JJ., Prairie, YT., Caraco, NF., McDowell, WH., Tranvik, LJ., Striegl, RG., Duarte, CM.,

Kortelainen, P. and Downing, JA. 2007. Plumbing the Global Carbon Cycle: Integrating

Inland Waters into the Terrestrial Carbon Budget. Ecosystems. 10:171-184

Dawson, J., Billett, M.F. and Hope, D. 2001. Diurnal variations in the carbon chemistry of two acidic peatland streams in north-east Scotland. Freshwater Biology 46:1309-1322

Denman, K.L., G. Brasseur, A. Chidthaisong, P. Ciais, P.M. Cox, R.E. Dickinson, D.

Hauglustaine, C. Heinze, E. Holland, D. Jacob, U. Lohmann, S Ramachandran, P.L. da

Silva Dias, S.C. Wofsy and X. Zhang, 2007: Couplings Between Changes in the

Climate System and Biogeochemistry. In: Climate Change 2007: The Physical Science

Basis. Contribution of Working Group I to the Fourth Assessment Report of the

Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z.

Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)]. Cambridge

University Press, Cambridge, United Kingdom and New York, NY, USA

Dinsmore, J., Wallin, M., Johnson, M., Billett, F., Bishop, K., Pumpanen, J. and Ojala A.

(2013). Contrasting CO2 concentration discharge dynamics in headwater streams: A multi-catchment comparison. Journal of Geophysical Reserch. Biogeoscience. 118:445–

461

Fiedler, S., Höll, B. and Jungkunst, H. 2006. Discovering the importance of lateral CO2 transport from a temperate spruce forest. Science of the Total Environment. 368:909-

915

Grenadin, U. 2003. Dataanalys och hypotesprövning för statistikanvändare. Swedish environmental protektion Agency.

Humborg, C., Mörth, C-M., Sundbom, M., Borg, H., Blenckner, T., Giesler, R. and Ittekkot,

V. 2010. CO2 supersaturation along the aquatic conduit in Swedish watersheds as constrained by terrestrial respiration, aquatic respiration and weathering. Global Change

Biology 16:1966–1978

Ingvarsson, M. (2008). Quantifying CO2 evasion from a headwater stream - A multidimensional study. Master’s thesis, 38 pp. Swedisch Univ. for Agric. Sci. Uppsala.

Karlsson, R. 2015. Skogaryd Research Catchment [hämtad: 2015-03-09 http://gvc.gu.se/english/research/skogaryd]

Kling, GW., Kipphut, GW. and Miller, MC. 1991. Arctic Lakes and Streams as Gas Conduits to the Atmosphere: Implication for Tundra Carbon Budget. Sience. 251:298-301

Lauerwald, R., Hartmann, J., Moosdorf, J., Kempe, S. and Raymond, P. 2013. What controls the spatial patterns of the riverine carbonate system? — A case study for North

America. Chemical Geology 337-338:114–127

Raymond, P., Hartmann, J., Lauerwald, R., Sobek, S., McDonald, C., Hoover, M., Butman,

D., Striegl, R., Mayorga, E., Humborg, C., Kortelainen, P., Dürr, H., Meybeck, M.,

Ciais, P. and Guth, P. 2013. Global carbon dioxide emissins from inlands waters.

Nature. 503:355-359

Teodoru, C., del Giero, P., Prairie, Y. and Camire, M. 2009. Patterns in pCO2 in boreal streams and rivers of northern Quebec, Canada. Global Biogeochimical Cycles. 23:

Tranvik, L., Dowing, J., Cotner, J., Loiselle, S., Striegl, R., Ballatore, T., Dillon, P., Finaly,

K., Fortino, K., Knoll, L., Kortelainen, P., Kutser, T., Larsen, S., Laurion, I., Leech, D.,

McCalloster, S., McKnight, D., Melack, J., Overholt, E., Porter, J., Sobek, S., Tremblay,

A., Vanni, M., Verschoor, A., Wachenfeldt, E., and Weyhenmeyer, G. 2009. Lakes and reservoirs as regulators of carbon cycling and climate. Limnol Oceanography. 54:2298-

2314

Venkiteswaran, J., Schiff, S. and Walling, M. (2014) Large Carbon Dioxide Fluxes from

Headwater Boreal and Sub-Boreal Streams. PLoS ONE. 9(7):e101756

Wallin, M., Öquist, M., Buffman, I., Billett, M., Nisell, J. and Bishop, K. 2011.

Spatiotemporal variability of the gas transfer coefficient (K

CO2

) in boreal streams:

Implications for large scale estimates of CO2 evasion. Global Biogeochemical Cycles.

25:GB302

10. Figures

Figure 1. IPCC performed 2007 calculations on the C-cycle on the earth and in what quantities it appeared to be able to estimate the climate change. Black numbers indicate the preindustrial era and the red numbers represent the anthropogenic emissions of C. One can talk about the C cycle beeing active within three boxes, the Terrastrial box (around the green area in the Figure), the oceanic box (in the blue are in the Figure) and the atmospheric box (in the sky of the Figure). As can be seen, no C-evasion are considers from streams and rivers. This Figure are illustrated in the report of: Denman, K.L., G. Brasseur, A. Chidthaisong, P. Ciais, P.M. Cox, R.E. Dickinson, D. Hauglustaine, C. Heinze, E. Holland, D. Jacob,

U. Lohmann, S Ramachandran, P.L. da Silva Dias, S.C. Wofsy and X. Zhang, 2007: Couplings Between Changes in the Climate System and Biogeochemistry. In: Climate Change

2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M.

Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA

Ersjön-Följesjön

ER

EF - A

MI

After Följesjön

AF - C

AF - B

MIN BJS - D

BJS - E

BJS - F

Join in Before Skottenesjön

KR

BS - G

BS - H

Before Skottenesjön

Skottenesjön

Figure 2. Map describing the catchment area of Skogaryd Research Catchment station and show the different areas with a dotted line. Red dots indicate a chamber location and its name are written just beside the dots. The name of the chamber has the structure that their last letter is the initial letter of the name to the lake most close by. The B or A expose if the chamber are Before or After the lake, and the possible J in the name indicate if the chambers are located in a side flow that are Join in to the main stream. The letters between A-H follow the chamber through the stream and indicate in what down-stream order they are, starting with an A and ending with an H. The back squares indicate where the hydrological measuring station are located. They are able to measure discharge (l/s), water temperature ( o

C) and precipitations (mm). They are named MI for

Mire station, ER for Ersjön station, KR for Krondike station and MIN for Mineral station. These chambers are for 2013.

Join in After Följesjön

EF 6

JAF 9

JAF

10

JAF

11

BF 7

JAF

BF 8

Before Följesjön

12

Sp23

AF 15

AF 13

AF 14

ER

EF 3

Ersjön-Följesjön

EF 5

EF 4

BE 2

BE 1

Before Följesjön

MI

After Följesjön

AF 16

MIN

JBS 18

JBS 17 1717

Join in Before Skottenesjön

BS 19

KR

BS 20

Before Skottenesjön

Skottenesjön

Figure 3. Map describing the catchment area of Skogaryd Research Catchment station and show the different areas with a dotted line. Red dots indicate a chamber location and its name are written just beside the dots. The name of the chamber has the structure that their last letter is the initial letter of the name to the lake most close by. The B or A expose if the chamber are

Before or After the lake, and the possible J in the name indicate if the chambers are located in a side flow that are Join in to the main stream. The number the chamber has indicates in what down-stream order they are, starting with number 1 and ending with number 20. The back squares indicate where the hydrological measuring stations are located. They are able to measure discharge (l/s), water temperature ( o

C) and precipitations (mm). They are named MI for Mire station, ER for Ersjön station, KR

for Krondike station and MIN for Mineral station. This is chambers for 2014.

A

B

Figure 4. pCO2 scattered due to day of year (DOY, with starting day 2013-01-01) in x axis. As a second y axis is discharge (l/s) for the catchment of Skogaryd. Figure A visualizes how the collected data looked before the data were corrected due to condensation on the sensors which resulted in a misleading data with pCO2 measured to be 10 000 ppm. This figure shows as well before the drift corrections are performed. The result for the cuttings and drift corrections are shown in Figure B.

Figure 5. A scattered graph of pCO2 in the left y-axis due to day of the year (DOY, with starting day 2013-01-01) in the x-axis. The graph contains a second y-axis for discharge (l/s) also due to DOY. The Figure indicates where the data series seems to have gaps. When the first longer gaps appear, a cut of the remaining data should be done. This is to get a reliable dataset for the whole day. In this case, it appears to be reliable datasets until DOY 451, all resisting data should be deleted.

Figure 6. pCO2 scattered on the y-axis and discharge (l/s) on the second y-axis due to what day of year (DOY, with starting day 2013-01-01) they have been measured on. These pCO2 values have been corrected so they only contain data with a continuous series without any longer gaps of the data. The cuts are due to when the sensor measuring pCO2 has been out of order, data has been cut away and these data are as well deleted in this Figure. This scatter is the result of the datasets left during first time out in field of 2014 and indicates that a time was necessary during DOY 452 this time and there are pCO

2

for 8 different chambers.

2013

Chamber ID\Out in field 114-160 161-188 189-209 210-258

EF - A

AF - B

AF - C

JBS - D

JBS - E

JBS - F

BS - G

BS - H cut performed after 13 days 11 days 14 days 13 days

2014

Chamber ID\Out in field

441-501 513-553 553-594 595-630 631-679

JAF - 11

JAF - 12

AF - 13

AF - 14

AF - 15

AF - 16

JBS - 17

JBS - 18

BE - 1

BE - 2

EF - 3

EF - 4

EF - 5

EF - 6

BF - 7

BF - 8

JAF - 9

JAF - 10

BS - 19

BS - 20 cut performed after 11 days 17 days 11 days 11 days 11 days

Figure 7. A scheme of how many chambers that has been cut away. Orange cells indicates data sets that has been cut away because of condensation on the sensor measuring pCO

2

Pink cells indicate which chamber never were working

The remaining Green cells indicate which chamber the data were reliable on, and which chamber did collect as much pCO

2

data which started to contained gaps which forced the dataset to be cut. The day out in field tells when the chambers were visited due to Day of the Year, with start from 2013-01-01. The number of days written in the bottom tells for how long the dataset were okay and after how long the cut for all data were needed to be performed. The chambers are in the order depending on which area they were located at in the stream. The thick black lines, formed into squares around the colored cells, indicate for which chambers are at the same area in the stream.

Figure 8. Two scattered graphs shown how discharge (l/s) (y-axsis) for the four different locations in the stream network of

Skogaryd follow each other’s pattern with high and low flows at the same time, only with differing levels. The graph are due to what day of the year (DOY, with starting day 2013-01-01) in the x-axis. This graph indicates that there probably are correlations of the discharge at Skogaryd. The graph to the left are for values measured during 2013 and the graph to the right are from values measured during 2014.

Table 1. Correlations between discharge (l/s) at different locations for hydrological stations measuring discharge at the catchment of Skogaryd. It is a fusion of both 2013 and 2014. The different locations has a significant and strongly correlation to each other.

Discharge (l/s) Ersjön

Discharge (l/s) Krondike

Discharge (l/s) Mineral

Pearson Correlation

Sig. (2-tailed)

N

Pearson Correlation

Sig. (2-tailed)

N

Pearson Correlation

Sig. (2-tailed)

N

Discharge (l/s)

Krondike

,9

,000

8204

Discharge (l/s)

Mineral

,8

,000

8202

,9

,000

8197

Discharge (l/s)

Mire

,7

,000

8210

,9

,000

8203

,9

,000

8201

Table 2. Correlation for which chambers in the catchment of Skogaryd 2013 (1 st

row) and 2014 (2 nd

and 3 rd

row) had a significant strong correlation due to pCO

2

and discharge (l/s). The numbers in the header indicates for the Day Of

Yeas (DOY, with starting day 2013-01-01) out in field this correlation occurred

Dischagre

(l/s)

Dischagre

(l/s)

AF - C

(CO2 ppm)

1

JBS - D

(CO2 ppm)

1

JBS - E

(CO2 ppm)

1

EF - A

(CO2 ppm)

3

BS - H

(CO2 ppm)

3

BS - G

(CO2 ppm)

4

Pearson Correlation

Sig. (2-tailed)

N

-,540

,000

-,711

,000

-,598

,000

-,774

,000

-,510

,000

-,585

,000

224

BF 7

(CO2 ppm)

5

236 236 455 586 260

JAF 9

(CO2 ppm)

5

AF 16

(CO2 ppm)

5

JBS 18

(CO2 ppm)

5

BS 20

(CO2 ppm)

5

BF 8

(CO2 ppm)

7

Pearson Correlation

Sig. (2-tailed)

N

-,723

,000

-,670

,000

-,540

,000

-,524

,000

-,564

,000

246 224 239 253 255

BS 20

(CO2 ppm)

7

BF 8

(CO2 ppm)

8

BS 20

(CO2 ppm)

8

BF 8

(CO2 ppm)

9

BS 20

(CO2 ppm)

9

-,508

,000

255

,514

,000

223

,701

,000

236

-,522

,000

238

-,790

,000

257

-,764

,000

228

Dischagre

(l/s)

Pearson Correlation

Sig. (2-tailed)

N

1. DOY_in_field = 114-127

3. DOY_in_field = 189 -183

4. DOY_in_field = 210-223

5. DOY_in_field = 441-452

7. DOY_in_field = 553-564

8. DOY_in_field = 595-606

9. DOY_in_field = 631-642

A

C

E

B

D

F

Figure 9. In the graphs x-axis is discharge (l/s) and in the y-axis are the pCO

2

, measured for each chamber location. The scatter are to see whether the correlation scattered for pCO

2

in Skogaryd are independent and all examples visualized had a high correlation. For Days Of the Year (DOY, with starting day 2013-01-01) 114-128, graphs A, B and C, an independent pattern are shown. For the rest, D, E and F pCO

2 which had a high correlation, there does not exist an independent scatter, both for DOY189-204 and 210-223. These last graphs has very small numbers of the discharge, and are probably causing these L-shaped patterns.

Figure 10. In the graphs x-axis is discharge (l/s) and in the y-axis are the pCO2. The Scatters are to see if the correlation for pCO2-discharge in Skogaryd 2014 are independent. These scatters are for the different periods, measured from which Days

Of the Year (Start with 2013-01-01). Graph A and B are from 553-564, C and D from 595-606 and graph E and Fare from

631-642. The graphs to the left are pCO2 data from chamber BF 8 and to the right are measured from chamber BS 20. As can be seen the correlations independency seems to be very much due to what level of the discharge there are. A extremely low level of discharge can even be reviewed as in A and B. When discharge are extremely high, the scatters seems to appear a shape like un upside-down L as in C and D and if the discharge are low the data are shaped as an L.

Figure 11. In the graphs x-axis is discharge (l/s) and in the y-axis are the pCO

2

. The Scatters are to see if the correlation for pCO2-discharge in Skogaryd 2014 are independent. These scatters are for the Days Of the Year (Start with 2013-01-

01) 441-452, which all has negative independent strong correlations.

R

2

=0,78

Table 3. The correlation rate between different areas in the stream water system indicates that the areas After

Följesjön, Before Skottenesjön and Join in before Skottenesjön has strong correlation but the area between Ersjön and Följesjön does not correlate to these areas at all. The correlation test is performed for the day of the year 114-127,

1 st

time out in field, 2013.

EF (CO2 ppm)

AF (CO2 ppm)

JBS (CO2 ppm)

Pearson Correlation

Sig. (2-tailed)

N

Pearson Correlation

Sig. (2-tailed)

N

Pearson Correlation

Sig. (2-tailed)

AF

(CO2 ppm)

,238

,001

184

JBS

(CO2 ppm)

,005

,947

189

,882

,000

224

BS

(CO2 ppm)

,083

,261

186

,793

,000

220

,762

,000

N 230

Tabele 4. The correlation test are du to find out if it exist in between the same location during 2013. High correlation for chamber JBS - E and JBS - D, meanwhile chamber JBS - F hardly correlate with any of the chambers during first period Days out in field (DOY, with starting day 2013-01-01) 114-127. JBS - F has a high correlation with chamber

JBS - D the second time out in field, DOY 161-188.

JBS - D (CO2 ppm)

JBS - E (CO2 ppm)

1. DOY_in_field = 114-127

2. DOY_in_field = 161-188

Pearson Correlation

Sig. (2-tailed)

N

Pearson Correlation

Sig. (2-tailed)

N

JBS - E (CO2 ppm)

1

,859

,000

236

JBS - F (CO2 ppm)

1

,373

,000

236

,130

,046

236

JBS - F (CO2 ppm)

2

,802

,000

640

Table 5. The correlation rate between different areas in the stream water system indicates that the areas After

Följesjön – 16, Join in Befor Skottenesjön and Before Skottenesjön has correlations to each other. The area between

Ersjön – Följesjön and After Följesjön -13 has no strong correlations to any of the areas at Skogaryd. The correlation test is performed for the day of the year (with starting day 2013-01-01) 441-452, 1 st

time out in field 2014.

EF (pCO2) Pearson Correlation

Sig. (2-tailed)

N

BF (pCO2) Pearson Correlation

Sig. (2-tailed)

N

JAF (pCO2) Pearson Correlation

Sig. (2-tailed)

N

AF 13 (pCO2) Pearson Correlation

Sig. (2-tailed)

N

AF16(pCO2) Pearson Correlation

Sig. (2-tailed)

N

JBS (pCO2) Pearson Correlation

Sig. (2-tailed)

BF JAF

(pCO2) (pCO2) (pCO2)

,104

,122

223

,201

,004

208

,718

,000

216

,174

,011

215

-,197

,003

225

,278

,000

209

(pCO2)

,451

,000

225

,599

,000

234

,681

,000

218

,151

,024

222

JBS

(pCO2)

BS

(pCO2)

-,025

,711

226

,743

,000

245

,359

,000

223

-,386

,000

227

,222

,001

237

,230

,000

,678

229

,584

,000

239

,656

228

,851

,000

246

,672

,000

224

,028

,000

N 253

Table 6. Correlation due to pCO2 in between the areas occurred one out of two times during 2014. These two times were the only time during 2014 that two chambers in the same area were working, so the only possible correlations due to inbetween sreas with pCO2 could be performed during these Days Of the Year (DOY, with starting day 2013-

01-01) 441-452 and 513 – 530.

EF 4 (pCO2) Pearson Correlation

Sig. (2-tailed)

N

1. DOY_in_field = 441-452

2. DOY_in_field = 513-530

EF 5

(CO2 ppm)

1

,868

,000

226

JBS 17 (pCO2) Pearson Correlation

Sig. (2-tailed)

N

JBS 18

(CO2 ppm)

2

,379

,000

342

Figure 12. Scatters of the correlation between pCO2 due to different areas in the stream water system of Skogaryd. The graphs indicates that the areas After Följesjön, Before Skottenesjön and Join in before Skottenesjön has independent correlations.

Figure 13. The scatter shows some of the independent correlations scatter due to see if there are any correlations existing between the areas of the catchment of Skogaryd, 2014, according to pCO2 between the areas. All in total there were 10 locations that had correlations to each other and 7 scatters were as well independent.

Figure 14. The scatters are plotted for chamberslocated at the same area of the catchment which has strong correlation to each other during the same time period out in the field.

Figure 15. Scatter of the correlation confirms that only one independent scatter occurs 2014 with chambers correlation due to pCO2 in the same are. The area which has strong correlation in the area are Ersjön – Följesjön.

A B

C

E F

Figure 16. In the y-axis are the mean of the pCO

2

measured for each station due to the time of the day in the x-axis. A

ANOVA resulted in two diel pattern during 2013 and 2014. The pattern looks mainly like in this A-D with peaks in the forenoon and troughs in the afternoon. A-B is gleaners of the test for 2013 and C-D are gleaners of the tests representing

2014. The other diel patterns, visualized in E and F have a peak in the afternoon. E are from 2013, chamber EF - A and F is from 2014, chamber JAF 11. As can be told, these chambers are from different areas in the catchment of Skogaryd.

Figure 17. Temporal variability’s can be assumed, since it seems to be a higher pCO

2

value during later on in the year than in the beginning. It also seems to decrease as closer the measuring comes to the winter.

Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement