Modelling Chlorine Transport in Temperate Soils

(1)

The Tema Institute Campus Norrköping

Master of Science Thesis, Environmental Science Programme, 2007

Olatunde Idris Ibikunle

Modelling Chlorine Transport in

Temperate Soils

(2)

5DSSRUWW\S Report category Licentiatavhandling Examensarbete AB-uppsats C-uppsats D-uppsats Övrig rapport x Master of Science ________________ 6SUnN Language Svenska/Swedish x Engelska/English ________________ 7LWHO 7LWOH 0RGHOOLQJ&KORULQH7UDQVSRUWLQ7HPSHUDWH6RLOV

)|UIDWWDUH Author

Olatunde Idris Ibikunle

6DPPDQIDWWQLQJ

Abstract

Microbes have been suggested to have a strong impact on the transportation of chlorine in soils. There are speculations about environmental factors limiting microbial effect on chlorine movement and retention. For this study, a numerical hydrochemical model was built to describe microbial transformation of chlorine in a laboratory lysimeter experiment. Undisturbed soil cores used to set-up the experiment were collected from a coniferous forest soil in southeast Sweden. The lysimeters were modelled in groups depending on their different water and chloride treatments. Microbial transformation of chlorine was better described under high water residence times and high chloride loads compared to low water residence times and low chloride loads. Microbial activity was also shown to properly account for a sudden shift from net-chlorine retention to net chlorine release in most of the lysimeters. Oxygen proved to be very important in accounting for the short-term shift from chloride retention to release in all the lysimeters. Model outcome revealed that 0.02– 0.10 mg Cl-_{could be available per day in a coniferous soil depending on} season and other soil conditions.This study shows that modeling enable a better understanding of chlorine biogeochemistry. It also confirms the speculated importance of microbial activities on chloride availability in soils.

,6%1 BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB ,651LIU-TEMA/ES-D--07/07--SE _________________________________________________________________ ,661 _________________________________________________________________ 6HULHWLWHORFKVHULHQXPPHU

Title of series, numbering

+DQGOHGDUH Tutor Per Sandén 1\FNHORUG Keywords

Hydrochemical model, chlorine biogeochemistry, microbial retention and oxygen

'DWXP Date 2007-06-15 85/I|UHOHNWURQLVNYHUVLRQ http://www.ep.liu.se/index.sv.html ,QVWLWXWLRQ$YGHOQLQJ Department, Division Insitutionen för tema Miljövetenskap The Tema Institute Environmental Science

(3)

ABSTRACT

Microbes have been suggested to have a strong impact on the transportation of chlorine in soils. There are speculations about environmental factors limiting microbial effect on chlorine movement and retention. For this study, a numerical hydrochemical model was built to describe microbial transformation of chlorine in a laboratory lysimeter experiment. Undisturbed soil cores used to set-up the experiment were collected from a coniferous forest soil in southeast Sweden. The lysimeters were modelled in groups depending on their different water and chloride treatments. Microbial transformation of chlorine was better described under high water residence times and high chloride loads compared to low water residence times and low chloride loads. Microbial activity was also shown to properly account for a sudden shift from net-chlorine retention to net chlorine release in most of the lysimeters. Oxygen proved to be very important in accounting for the short-term shift from chloride retention to release in all the lysimeters. Model outcome revealed that 0.02– 0.10 mg Cl- could be available per day in a coniferous soil depending on season and other soil conditions. This study shows that modeling enable a better understanding of chlorine biogeochemistry. It also confirms the speculated importance of microbial activities on chloride availability in soils.

Keywords: Hydrochemical model, chlorine biogeochemistry, microbial retention and oxygen.

(4)

ACKNOWLEDGEMENT

First, I want to appreciate the guidance and blessings I have been bestowed by the Almighty from cradle till present. He has provided me the wherewithal that have been used to prosecute existence through the hard and very difficult times fate’s put me through.

My sincere appreciation goes to my supervisor, Ass. Prof. Per Sanden who has always been there beyond the walls of the class room. I hope your shoulders will still be available anytime it’s requested. My sincere gratitude also goes to Dr. Frank Laturnus who has always provided ever-willing assistance in times of need. The same order of gratitude goes to all my tutors and guidance through the environmental science programme. I beg your forgiveness for my inability to equate thanks with your kindness.

My thanks also goes to the Swedish government, for providing the rare opportunity of studying on free tuition, your gesture has been a good pedestal on which my floating dreams have lived.

To Stephen, my project mate, the times we spent scrolling through those multi-rows of data will never be forgotten. I pray that the calibrated values of our friendship in Sweden will be enough for the validation of more future scenarios. This same wish goes to all my folks home and abroad, your support through the course of my studies will never be forgotten.

This space cannot be enough to appreciate the help, support and prayers of my parents and siblings, but I pray God’s blessings to enable me reciprocate your outstanding attention on me.

And to her, the very love of my life, what you gave me cannot be equated with words.

(5)

TABLE OF CONTENT

ABSTRACT ... 1

ACKNOWLEDGEMENT ... 2

TABLE OF CONTENT ... 3

LIST OF FIGURES AND TABLES ... 4

1.1 INTRODUCTION ... 5

1.1.1 STUDY OUTLINE ... 6

1.1.2 STUDY OBJECTIVES ... 6

1.2 BACKGROUND AND STATE OF RESEARCH ... 6

1.2.1 SOURCES AND CYCLING OF CHLORINE IN A FOREST ECOSYSTEM ... 6

1.2.2 CHLORIDE IMBALANCES IN THE SOIL ... 8

1.2.3 MICROBES AND CHLORIDE RETENTION ... 8

2.1 METHOD ... 10

2.1.1 EXPERIMENTAL DATA AND DESCRIPTION ... 10

2.1.2 SYSTEM DEFINITION ... 10

2.1.3 MODEL CONCEPTUALIZATION: Assumption and Calibration ... 11

2.2 HYDROLOGY... 12

2.2.1 Variables, Flows, Parameters and Parameterisation of the Modified HBV Model ... 13

2.3 BIOGEOCHEMICAL MODEL ... 15

2.3.1 Transformation ... 15

2.3.2 Microbial Biomass Assimilation (MBA) ... 16

3.0 RESULTS... 18

3.1 HYDROLOGICAL SUB-MODEL ... 18

3.2 BIOGEOCHEMICAL SUB-MODEL... 18

3.3 Biomass Carbon and Chloride Availability ... 28

3.4 Internal Dynamics of the Model... 28

4.0 DISCUSSION ... 29

4.1 Chloride Movement in Soils ... 29

4.2 Organic matter and microbial activity ... 30

4.3 Chloride loads and Microbial Retention ... 30

4.4 Microbes and the Retention-Release Pattern ... 30

4.5 Study Implications... 31

5.0 CONCLUSION ... 32

REFERENCES ... 33

(6)

LIST OF FIGURES AND TABLES

Figure1: Conceptual diagram defining the system as a whole ... 10

Figure 2: Conceptual diagram of the modified HBV model ... 13

Figure 3: Simulated accumulated outflow (mm) from the Lysimeters for a period of 122days... 18

Figure 4: Simulated Chloride concentration for Group 1 Lysimeters over a period of 122days. ... 21

Figure 5: Simulated Chloride concentration for Group 2 Lysimeters over a period of 122days... 23

Figure 8: Simulated Oxygen, Biomass Carbon and Modelled and Observed Chloride concentration over a period of 122 days ... 28

Table1: Different groups of the lysimeters ... 11

Table 2: Parameter values used for the hydrological sub-model ... 18

Table 3: Parameter values used for the biogeochemical model ... 19

Table 4: Comparison of the available modelled and observed chloride amount for group 1 lysimeters ... 20

Table 8: Comparison of the modelled available chloride amount and Biomass carbon accumulated during the retention period. ... 28

(7)

1.1 INTRODUCTION

Chlorine is one of several elements that have been given little attention in early studies of forest soils. There are less available knowledge as regards its interaction, movement and different species in the soil environment. The negligence to study chlorine and its behaviour in forest soils is borne out of the textbook conservative hypothesis widely accepted by earlier researchers. It was believed that chlorine move through the soil unreactive, and that, the exact amount deposited can be recovered as leachates from the soil (Schlesinger, 1997). This idea informed the use of chlorine as a mere conservative tracer of water and other solutes movement in the soil.

However, the last two decades have produced various researches into the study of chlorine and its interactions in various ecosystems. The outcome of such published works have led to the understanding of the existing species of chlorine; their movement in the soil; their interaction with other components in the soil; possible factors that influence their behaviour; and the overall consequence of their turnover in various parts of the biosphere. These developments have led to the birth of a new non-conservative paradigm of chlorine and its different forms. The evolving paradigm has informed the review of various old data and researches in order to validate the descriptions presently offered about chlorine. One of such attempts was published by Lovett et al (2005), where the researchers re-evaluated their findings about chlorine over a period of almost four-decades in Hubbard Brook USA. Another attempt was a review of old research findings about chlorine and its presence in different biological life by Öberg (2002). Öberg fitted scattered research pieces (that were neglected because of the conservative hypothesis) into a knowledge that was perfectly consistent with the recently discovered non-conservative hypothesis. In her research work, there was a special spotlight on the importance of chlorine in microbes and their influence on the transformation of chlorine into other forms hitherto un-noticed. Microorganisms are important components of all ecosystems. In forest ecosystems, microorganism comprise mainly of fungi and bacteria. Hence, they are referred to as soil microbes. Soil microbes play major roles in the sustenance and regeneration of the environment they live in. The interaction of microbes with other soil components such as major elements (e.g. carbon and nitrogen) is well documented (Wardle, 1992). Their impact on the transformations of these major elements have also been understood and under continuous research. Amongst various tools that have led to the continuous understanding of microbial interaction with elements (such as carbon) has been the use of numerical modelling. Modelling is a viable tool often used to integrate scattered pieces of research knowledge (Jorgensen, 2001). It is usually very potent to gain a better understanding of new areas of research that have gaps of knowledge to bridge. Thus, an evolving area of research such as the biogeochemistry of chlorine can be well understood with the use of modelling as a tool. This will help understand the interaction between important factors influencing the transportation, movement and availability of chlorine in forest soils.

Having referred the possible impacts of microbial activity on chlorine biogeochemistry, it is imperative to employ modelling as a tool to understand the influence of microorganism on the movement of chlorine in soils. This will enable a proper description of the influence of microbial activity on chloride movement under environmental factors that have been reported by Bastviken et al. (2006).

(8)

1.1.1 STUDY OUTLINE

Here, I started with a short synopsis on chlorine: its description and available forms. I went ahead to explain the two chlorine forms of particular interest (i.e. inorganic chlorine and organic chlorine) in this study, bringing to light their sources and cycling in the environment. I went further to describe the current state of research in chlorine biogeochemistry and the particular challenge that led to the objectives of this study.

1.1.2 STUDY OBJECTIVES

§ To evaluate the extent to which microbial activity explains the biogeochemistry of chlorine.

§ To evaluate the extent to which microbial activity can account for the shift from, chloride retention to chloride release.

§ To evaluate the extent to which microbial retention can control the availability of chloride in soil pore water.

1.2 BACKGROUND AND STATE OF RESEARCH

The description, nomenclature and categorisation of the chlorine have witnessed various dimensions through several decades of scientific research. Elemental chlorine is a very reactive gas leached out to the environment after its volatilisation from the hot magma that made up the earth’s crust (Johansson, 2000). Though a minor constituent of igneous, metamorphic and sedimentary rocks, chlorine is available in the ocean, with only hydrogen and oxygen having larger pools than it (Lovett. et al. 2005). Because of its reactivity, chlorine is rarely found in its elemental form in natural environments, but rather as chloride - its ionic form. Chlorine with other halogens such as, fluorine, bromine, iodine and astatine has a very high tendency to form salts due to their high electronegativity and high bonding energies (Lovett. et al. 2005). Out of the Halogens, Chlorine is regarded as very important and the most common after Fluorine (Johansson, 2000).

Chlorine exists in different forms and species in water, air and soil compartments. It can exist in its more common form as inorganic chloride, as volatile or non-volatile chlorinated compounds. Chlorine has been found to exist in these forms in different part of the biosphere. Common examples of chlorine species are chloroform, trichloroacetic acids, tetrachloroacetic acids, and chloromethane. (Laturnus, 2003; Johansson, 2000; Svensson, 2006). Chlorine in its different forms, affects the health and survival of various parts of the ecosystem, it threatens plant/animal life in soil and water, and on a global scale destroys the ozone - the layer protecting the earth from dangerous ultraviolet radiation (Laturnus, 2003).

1.2.1 SOURCES AND CYCLING OF CHLORINE IN A FOREST ECOSYSTEM

This section aims to bring forth the source and cycling of chlorine (with particular attention on organic chlorine and inorganic chlorine), its interaction, presence and occurrence in the vegetation and soil of a coniferous forest ecosystem.

(9)

CHLORIDE

Wind, water, weathering, precipitation, pH, ion exchange and soil organisms are among the geochemical forces that characterise the cycling and transformation of chloride ions (Öberg, 2003). Lovett et al. (2005) reviewed that, the movement of chloride between the ocean and the atmosphere accounts largely for the global chlorine cycle, with chloride from the ocean being the source of those found in both terrestrial and fresh water ecosystems. Chloride is generated from the ocean as salt aerosols through the continuous breaking of the ocean surface by waves. The ions in the aerosols are transported by wind action and evaporation to the atmosphere; they leave the atmosphere to the terrestrial environment either as wet or dry depositions (Johansson, 2000). The amount of chloride ions deposited to an environment depends largely on the depth of precipitation, the concentration of chloride in the precipitation (Öberg, 1998), the distance of the environment from the ocean, topography and wind direction (Öberg, 2003).

Different ages of research have reported various properties and behaviour of chloride in the soil. Earlier researches favoured the conservativeness of chloride, while recent studies have discovered chloride transformations in the soil. Earlier works reviewed by Lovett et al. (2005) described chloride as only reactive in the atmosphere but unreactive in the soil. It was assumed that chloride is poorly absorbed and do not participate in the formation of secondary minerals. Most of the referred works were measurements predominantly carried out to account for ion balances with little or no study of other forms, properties and behaviour of chloride in the soil.

The conservative assumption about chloride informed its use as a tracer in the study of most forest ecosystems. Johansson (2000) complemented this by reporting that the conservative assumption made chloride together with other halides like bromide to be used as mere tracers of water movement in hydrological modelling. They were used primarily to describe the sources, transportation and transformation of other ions in the soil. This was because the sources and sinks of chloride were understudied and assumed negligible in the soil. On the contrary, Öberg (2002) reported recent studies that confirmed chloride as not only participating in complex biogeochemical reactions, but also transforms into organic forms in the soil.

The movement of chloride ions in the soil is associated with that of soil water. Because of its solubility (Lovett et al, 2005) and negative charge, chloride ions move at a very fast rate through the soil medium. This fast movement is aided by the repel action between negatively charged chloride ion and solid structures (mostly clay and organic matter) of the soil (Öberg, 2003)

ORGANIC CHLORINE

This specie of chlorine exists by bounding to carbon atoms (organic compounds), with specific examples like fulvic and humic acids (Öberg 1998; Svensson 2006). The main source of organo-chlorines in the environment was initially thought to be anthropogenic - with suggested point sources from pulp and paper industry cum the use of pesticides, herbicides and chlorinated solvents. However, recent studies have confirmed natural sources of organo-chlorines. Johansson (2000), in her doctoral thesis, reviewed recent researches that confirm detectable quantities of organo-chlorines in soil samples collected from remote areas all over the world. Results of the researches revealed that organo-chlorines could be present as volatile, non-volatile, high or low molecular weight compounds. The result also showed that organic matter

(10)

content of the analysed soils was related to the quantity of the discovered compounds. In addition to this, Öberg (1998) also reported chloride amount in organic matter being close to that of phosphorus but slightly less than sulphur and nitrogen.

Öberg (2003), revealed that in addition to the natural formation of organic chlorine in soil, other major sources could be litter and through fall from trees as a result of precipitation. Precipitation has been reported to wash down chloride particles stuck on tree branches, needles and crowns.

The formation of organic chlorine in the soil has been strongly forwarded to be biotic (Johansson et al., 2003) but Keppler et al. (2000) however suggested that their formation could also be due to abiotic factors. A typical biotic mechanism is the production of organic chlorine as a bye-product during the biodegradation of organic matter. The biodegradation is a resultant effect of organic matter oxidation by reactive chlorine e.g. Hypochlorite (HOCL). Specialised microbial enzymes were reported to form this reactive chlorine in the presence of hydrogen peroxide and chloride ions (Öberg, 2002).

1.2.2 CHLORIDE IMBALANCES IN THE SOIL

Chloride is the main source of chlorine that cycles the soil. As earlier explained, it is majorly deposited to the soil with precipitation and through-fall (Öberg, 2003). Chloride imbalance results when the amount of chloride deposited to the soil is not completely recovered in form of leachate from the soil medium. This clearly contradicts the original knowledge of chloride behaviour, hence focusing many research works on the factors that could be responsible for chloride retention in soil. Amongst major factors that have been identified to cause chloride imbalance in forest soil are: vegetative uptake (Likens, 1995), ion exchange (Viers et al., 2001), heterogeneous chloride movement in soil pores, evaporation and microbial uptake. Among all these factors, Bastviken et al. (2007) reported microbial uptake as a major and significant cause of chloride retention over long and short time periods. Thus, in order to establish the relationship between chloride retention and microbial activity; it is important to explain the processes and environmental factors that characterize the existence of the microbes in forest soils.

1.2.3 MICROBES AND CHLORIDE RETENTION

Recent researches have shown that soil microbes could retain chloride through short-term uptake for metabolism and long-short-term chlorination of soil organic matter. Bastviken et al. (2007) showed that soil microbes could retain as much as 24 and 4 percent of the initially available chloride in soil water on short and long terms respectively. Important factors that have been identified to affect the retention of chloride by soil microbes are temperature, available organic matter, and oxygen (Bastviken et al., 2006). The availability of easily digestible organic matter fractions will increase microbial activity and hence an increase in chloride assimilation/retention. Due to the fact that most microbial activities are enzymatic, temperature optimum for enzymatic activities will increase microbial retention. Microbes have also been reported to retain more chloride in oxic (i.e. abundant

(11)

oxygen) than anoxic (reduced/no oxygen) soil conditions (Thomsen, 2006). Bastviken et al. (2006) also reported that microbial activities could decrease with a reduction in the amount of oxygen available to oxidise refractory organic materials, when the easily available ones have been depleted.

Bastviken et al., (2006) investigated the effect of water residence and nitrogen and chloride loads on chloride retention and release in a forest soil. The retention noticed was suggested to be more of a microbially induced process than any other means of retention, considering the conditions under which the experiment was conducted. However, the chloride retention rates in terms of microbial interaction with those factors were not documented. Hence, it is imperative to quantify how much microbial activity (with its limiting conditions like oxygen and organic matter) could explain chloride retention under different water and chloride conditions. In a similar research by Bastviken et al. (2007), to quantify microbial retention in a forest soil, it was established that an increase in microbial biomass carbon in the orders of 10 percent could reduce chloride in soil water by 25 percent, but the effect of this retention rate in relation to fluctuations in environmental conditions was not elaborated.

(12)

2.1 METHOD

2.1.1 EXPERIMENTAL DATA AND DESCRIPTION

The data used for the model was obtained from a lysimeter experiment (Bastviken et al., 2006) conducted on soil from the Stubbetorp catchment (58°, 44’ N, 16°, 21’ E) in southeast Sweden. The catchment is a 0.87km2 coniferous forest predominantly consisting of Norwegian spruce and pine trees. Undisturbed soil cores (of 15cm depth and 80cm2 cross-sectional area) were collected and used to set-up a lysimeter experiment according to Rodhstedt et al. (2003) specifications. They were incubated in a climate chamber for 127 days at a temperature of 10oC and humidity of approximately 90%. The Lysimeters were treated with three factors (chloride, nitrogen and water) relative to the objective of the experiment and the factors were made to two levels of high and low amounts. The experimental design resulted into eight combinations and at three replicates each; there were 24 lysimeters in all. The lysimeters were treated twice a week with artificial rain containing ion SO42-, Ca2+, Mg2+, Na+, K+ and H+ corresponding to what obtains at the Stubbetorp catchment.

Percolation were measured weekly for the duration of the experiment.

2.1.2 SYSTEM DEFINITION

The natural soil environment represents the general system to be modelled. It has a general outer boundary that describes inlets and outlets of driving variables

determining the condition of state variables. The system has two distinct internal mechanisms (Fig.1) (Hydrology and Biogeochemistry), which are translated into different models with their respective driving and parametric values.

Figure1: Conceptual diagram defining the system as a whole

The precipitation that enters the soil system has a certain concentration of chloride. Precipitation and chloride witness different dynamics that are regulated by hydrology

Biogeochemistry Hydrology Precipitation Chloride Precipitation with Chloride Chloride Outflow Percolation LEACHATE Boundaries

(13)

and biogeochemistry respectively. These two-subsystems are interconnected to properly describe retention of chloride in soils. The above represents a simplification of chloride transformation and the details covered in the model are as accurate as: the state of research in biogeochemistry of chlorine and data availability.

2.1.3 MODEL CONCEPTUALIZATION: Assumption and Calibration

The two internal mechanisms of the system are made into a single-larger model (The Hydrochemical Model). The hydrological sub-model is a modification of the HBV model (Bergström et al., 1985) while the biogeochemical model was built to achieve the objective of this study. The biogeochemical model represents a simplified system with the inclusion of vital components influencing the transformation of chlorine primarily by soil microbial community. This is in particular reference to the justification offered as background to this study.

The main assumption of the model is that, the effect of nitrogen is not important on the retention of chloride in the soil. This assumption is informed by a report by Bastviken et al. (2006) – the particular research effort that motivated this present study. With this assumption, the lysimeters are grouped as shown in Table 1.

Table1: Different groups of the lysimeters.

GROUP Precipitation Chloride Load Lysimeter replicates

1

LOW LOW _{1a, 1b, 1c}

2a, 2b, 2c

2

HIGH _{3a, 3b, 3c}

4a, 4b, 4c

3

HIGH LOW _{5a, 5b, 5c}

6a, 6b, 6c

4

HIGH _{7a, 7b, 7c}

8a, 8b, 8c

One lysimeter from each group is calibrated and its parameter values subsequently used to run input data for the other lysimeter replicates in the group. For example, in group 1 Lysimeter 1(a) will be calibrated and its parameter values used to run lysimeters 1(b) to 2(c). This will represent the general calibration procedure for the groups unless otherwise stated. The general idea behind this assumption is to fulfil the set aims and objective for this study.

(14)

The model’s qualitative performance will be tested through a sensitivity analysis and the evaluation of model efficiency (R2) according to Nash and Sutcliffe (1970). The efficiency ranges between a perfect fit value of 1 and -∞. If the model efficiency is lower than zero, then, the mean of the observed time-series will be a better predictor than the model (Krause et al., 2005). Important model outputs will also be compared with the observed values to evaluate the quantitative performance of the model. The R2 are calculated as:

Q com: Modelled value Q obs: Observed value

2.2 HYDROLOGY The HBV Model

The original HBV model has been in use for more than thirty years (Lindström et al.1997). It has undergone series of modification informed by the objective of accurately estimating run-off conditions in catchments of varying sizes across different regions of the world. The HBV-model is a semi-distributed conceptual model that continuously calculate run-off for hydrological forecasting. It uses sub-basins as the primary units and takes into account characteristic conditions (such as land-use, area, and elevation) of the sub-basin for proper calculation of run-off conditions over diverse land areas. It is divided into four sub-routines of snow accumulation and melt; soil moisture accounting procedure; routines for run-off generation; and simple routing procedure. Each of these routines has specific parameters that define the characteristics of the sub-routine. The parameters are calibrated accordingly during simulation procedures. Input data for the model are precipitation, air temperature and estimates of potential evapotranspiration. The model is usually run on daily time-step, but higher resolutions can be achieved (Lindström et al.1997).

Use and Modification of the HBV Model

The HBV model has been adapted for numerous hydrological and hydrochemical studies on various scales and dimensions. Andersson et al. (2005) adapted it for use in estimating nitrogen and phosphorus flow in a catchment, Tonderski et al. (2005) used it to estimate phosphorus retention in soils while Sanden (1991) also adapted it to monitor metal transport in an old-mining area.

For this study, the original HBV model was simplified and adapted for our purpose using the STELLA® software (HPS, Hannover, USA). The software has been used for numerical modelling of various ecosystem studies, though, it has reduced capacity for large data set, but it proved sufficient for this task. The model was simplified to having one sub-routine. i.e. the soil moisture accounting procedure (see Fig 2). The aim of this model is to simulate three hydrological inputs needed to estimate chloride

) 1 ( ) ) ( ( )) ( ) ( ( 1 2 1 2 1 2 2 ≤ < −∞ − − − =

∑

= = _R Qobs i Qobs i Qobs i Qcom R _n i n i

(15)

Accumulated outflow Precipitation Percolation/Run-off Actual Evapotranspiration SOIL

concentration in the lysimeters. The hydrological inputs needed in the biogeochemical sub-model are:

1. Soil moisture balance: This is needed to quantify the chloride amount or concentration in the soil of each lysimeters.

2. Percolation/run-off: This is also important for estimating the chloride amount or concentration in the outflows/leachates from the lysimeters.

3. Actual evapotranspiration

In order to simulate the above, the model is used to primarily simulate percolation/run-off for the soil lysimeters. Once, the modelled percolation is consistent with observed outflow from the experiment, we can be sure that other hydrological inputs needed to estimate chloride concentration in the soil and leachates were accurate.

.

Figure 2: Conceptual diagram of the modified HBV model.

2.2.1 Variables, Flows, Parameters and Parameterisation of the Modified HBV Model

§ Precipitation: This is the main driving variable of the model. In the lysimeter experiment, it represents the artificial rain treatment at two levels of high and low as explained above i.e 1449mm and 483mm per annum, respectively. These amounts correspond to the precipitation on the west and east coasts of southern Sweden. The lysimeters were treated twice a week resulting in 38 data points, but the data were adjusted to the length of the experiment (127 days) by allotting zero values for days of the week without precipitation. This was done to generate a

(16)

continuous data set, which was essential for the numerical method used for the simulations.

In order to estimate the gravimetric equivalent of the precipitation treatments (1449mm and 483mm per annum), the values were measured in grams with the assumption of 1 g ml-1 as density of water. The resulting millilitre (the equivalent of grams) value was converted back to precipitation depth in millimetres (mm) with respect to the cross-sectional area of the lysimeters (80 cm2).

122 observed data points (i.e. the duration of the experiment excluding the first four days) were used to run the model. This was so done because the water conditions for the first four days in the lysimeters were not stable for use by the numerical method in STELLA®. The first amount of water added to the lysimeter to achieve field capacity was too large.

§ Initial Moisture Content: This is an initial state parameter estimated for all lysimeters. At the start of the experiment, the following were measured in grams: A: weight of the lysimeter,

B: weight of the lysimeter with wet soil.

The wet weight of the soil was estimated by subtracting A from B, and then the estimated dry weight (which was measured after the termination of the experiment) was subtracted from the calculated wet weight to get the initial moisture content.

The above estimate was primarily used as a guide to set the range for the different initial moisture conditions in the lysimeters. It differed from the one used in the model because of the wide differences noticed in the initial moisture conditions in each of the lysimeters. These differences were reflected in the different dry weights measured for all the lysimeters. The choice of a single calibrated moisture value was to set the same initial soil conditions for the lysimeters.

§ Field Capacity (FC): This is also an initial state parameter used to regulate percolation and evapotranspiration in the model. It was estimated as equal to the initial moisture content because excess water was drained from the lysimeters at the start of the experiment The same calibrated value was used for all the lysimeters (irrespective of the treatments) because the same hydrological conditions were assumed for them all.

§ Soil Curve Parameter (BETA): This model parameter was also set for all the lysimeters irrespective of their treatments. It was used to define the distribution of soil particles in order to regulate percolation in terms of the soil moisture. § Potential Evapotranspiration (Evapo): This was also calibrated and validated

for use in all the lysimeters. The unit is also in mm. It is one of the variables used to regulate the actual evapotranspiration for all the lysimeters.

§ Alfa: The same Alfa values were set for all the lysimeters irrespective of their treatments. It is a model parameter used to regulate the rate of

Evapotranspiration from the model.

The Model simulated three OUTPUT VARIABLES: Actual evapotranspiration, soil moisture balance and percolation. Only two of these - soil moisture balance and

(17)

percolation were used as hydrological inputs in the biogeochemical model. All these have units in mm.

2.3 BIOGEOCHEMICAL MODEL

As earlier justified in the background for this study, the soil microbial community will represent the main agent of chloride transformation in the biogeochemical model. Once again, the main aim for building this model is to simulate chloride outflows using the same calibrated values for all the lysimeters irrespective of the different chloride and water loads. In order to achieve this, the biogeochemical model was further divided into two sub-models: transformation and microbial biomass assimilation.

2.3.1 Transformation

This describes chloride transformation in the lysimeters in terms of mobilisation and immobilisation. The former is synonymous to chloride release, while the latter is the same as chloride retention. With the assumption that the transformation of chloride in the lysimeters is primarily due to microbial activity, the model was conceptualised such that, chloride retention and release are driven by the second sub-model - microbial biomass assimilation.

2.3.1.1 Variables, Flows, Parameters and Parameterisation of Chloride Transformation Model

Chloride with precipitation: This is the main driving variable for the transformation model. Chloride (given as mg Cl-1) enters the model with precipitation and these represents the two treatment levels of chloride as earlier explained. The experimental treatments are equivalent to 1449 and 4346 mg m-2 yr-1 of chloride which represents the moderate load on the Stubbetorp catchment and loads along the west coast of Sweden respectively. Since the chloride treatment came with precipitation, it also corresponds to 38 data points (at two treatments per week), but the data were adjusted to the length of the experiment (127 days) by allotting zero values for days of the week without precipitation.

With specific reference to the number of data points (122) used to run the hydrological sub-model (which serves as an input into this sub-model) the first four days of chloride input were also excluded to assure the required internal logic for the model.

§ Initial chloride amount in the soil: This initial state parameter was estimated from the data. It corresponds to the initial chloride amount measured prior to any treatment plus the amount in the first artificial rain minus the first amount leached from the lysimeters. This estimation was done to return lysimeters with the same chloride treatment to the same initial chloride amount, owing to the fact that they might have different initial chloride contents. This also served as a means of accounting for the effect of the first chloride load on the initial conditions in the lysimeters. The hydrological input (i.e. the moisture balance) was used here to calculate the concentration of chloride in the soil and the outflow from the soil.

(18)

The soil moisture balance for the hydrological sub-model was converted back to litres before use for uniformity.

The estimated amount from the above calculation differed from the ones used in the model because of the excluded first four days from the data. However, the estimates provided a range of initial values appropriate to start the calibration for the different lysimeter treatments. The same initial chloride amounts were set for lysimeters with specific reference to the model assumption and calibration procedure earlier stated.

§ Mobilisation: This is one of the main processes that drive the objective of this model. It is classified an input because it is an internal flow process that returns the immobilised chloride in the microbes back to the soil. In theory, Öberg (1998) reported that the death of soil microbes would result in a release of the chloride in their biomass. Thus, this process is directly connected to the outflow from the other model - microbial biomass assimilation (see explanation of this sub-model below), and the release of chloride from the microbes is proportional to the chloride in their biomass.

§ Immobilisation: This represents the output process that retains chloride in the soil through microbial accumulation. As justified above, Öberg (2003) and Bastviken et al., (2007) reported that soil microbial community retains chloride on long and short-term periods. In this model, the process is connected to an estimated chloride amount in biomass and the amount of microbial biomass accumulated with time.

§ Chloride Outflow: This is also an important output variable from the soil. Its concentration is equivalent to the chloride concentration after the initial amount has undergone temporal or permanent transformation in the soil.

§ Chloride in Biomass: This was calculated from reported (Luria, 1960) estimates of chloride amounts in cell structures of most living organisms.

2.3.2 Microbial Biomass Assimilation (MBA)

This sub-model drives the transformation of chloride in soil. Both immobilisation and mobilisation processes are directly related to the rate of biomass carbon accumulation. In theory, microbes have been reported to use chloride as transport electrolyte and for maintaining osmotic balance (Öberg, 1998). Thus, with the availability of vital resources such as oxygen and available organic matter, the assimilation of chloride will continue with biomass growth. For this particular study, oxygen alone was considered the main resources limiting the accumulation of microbial biomass for the following reasons:

§ There was no data available for different organic matter fractions that could be used to limit the growth of microbes with time. Data on different fraction of organic matter is needed because microbes reduce in biomass growth when the easily degradable organic matter fractions of the soil have been exhausted. § Oxygen has been reported to limit the rate of chloride retention (Thomsen,

(19)

(Raubuch, 1999) of oxygen consumption by microbes in coniferous soil will serve as an appropriate literature guide for the calibration of oxygen consumption in our model.

2.3.2.1 Variables, Flows, Parameters and Parameterisation of MBA sub-model Initial Microbial Biomass Carbon: This is an initial state parameter for the MBA sub-model. The initial biomass content reported (Bauhus et al., 1999; Raubuch and Beese1995,2005; Friedel et al., 2006; Klose et al. 2004) for coniferous soils informed the starting value for its calibration. The unit of the biomass content was converted to milligram for consistency with other mass units in the biogeochemical model. For the model, the same initial biomass carbon content of 100 mg (which is equivalent to 12.5 g m-2) was used for all lysimeters irrespective of the different treatments.

§ Microbial Growth: This is the main inflow into the MBA. It is an increase in microbial biomass C with time at a growth rate calibrated for all the lysimeters. Microbial accumulation was regulated with oxygen through an oxygen sub-model built primarily for this. The objective of building the oxygen sub-model was to regulate microbial growth with time and resource depletion as noticed in the experiment. Though, the primary limitation here was that, the oxygen content of the soil at the start of the experiment could not be estimated because of lack of data.

§ Initial Oxygen content: The same initial oxygen content was set for all lysimeters with specific reference to the general model assumption and calibration procedure earlier stated. This will elucidate the synergetic effect of different chloride and water loads on initial oxygen contents. This will also ensure same initial oxygen states necessary to evaluate the efficiency of the model on these sets of lysimeters. The oxygen depletion was regulated by a metabolic quotient (rate of oxygen depleted per time per biomass C which was started with a range as reported by Raubuch and Beese 1999.

§ Microbial Biomass Increase and Decrease rates: These two model parameters were calibrated for use in the model. They were set for lysimeters with the same chloride and water treatments in order to ensure the same initial conditions necessary to compare the model efficiency on these lysimeters

§ Metabolic Quotient: The metabolic quotient was calibrated for all the lysimeters irrespective of their different treatments.

(20)

3.0 RESULTS

3.1 HYDROLOGICAL SUB-MODEL

The modelled accumulated outflow from all the lysimeters was in good agreement with the observed accumulated outflow using the parameter values shown in Table 2. Figure 3 represents the general picture of the simulations for all the lysimeters. The general pattern was confirmed when a sensitivity procedure was performed for the model by altering the field capacity and initial moisture content to extreme values, there was no difference in the accumulated outflow pattern for all the lysimeters after the sensitivity test. The lysimeters returned an R2 value of approximately 0.99 calculated according to Nash and Sutcliffe (1970). This indicates that the model accounted well for the hydrological conditions in all the lysimeters.

Table 2: Parameter values used for the hydrological sub-model All lysimeters

Soil curve parameter Beta 20

Field Capacity (mm) 83.5

Initial Soil Moisture (mm) 83.5

Potential Evapotranspiration (mm) 0.38

Alfa 2

Figure 3: Simulated accumulated outflow (mm) from the Lysimeters for a period of 122days.

3.2 BIOGEOCHEMICAL SUB-MODEL

Table 3 shows the parameter values used to simulate chloride concentration and amounts over a period of 122 days in the lysimeters. The parameter values were determined with specific reference to the model conceptualization earlier explained and shown in Table 1. The lysimeters returned different parameter values for the same starting biomass carbon content (100 mg) and metabolic quotient (0.045 mg O2 d-1

0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 0 5 0 1 0 0 1 5 0 D a y s a c c u m u la te d o u tf lo w s ( m m ) o b s e r v e d m o d e l l e d R2 = 0.99

(21)

mg-1 C). The biomass increase rate ranged from 0.06 to 0.14 represented by group 3 and group 2 lysimeters respectively. The biomass decrease rate ranged from 0.009 to 0.02, for group 2 and 1 together with 3 lysimeters respectively. The model produced the best fit with two initial oxygen conditions of 500 and 650 milligrams. Groups 1 and 3 had 500 mg while groups 2 and 4 had 650 milligrams of oxygen.

In order to evaluate the implications of the parameter values used, a sensitivity analysis was carried out on the model by doubling the starting biomass carbon content. The test returned a model with a lower R2 value, though the differences in the values did not affect the interpretation of the efficiency. i.e. the reduced values were still in the same interpretation range with the new values. Specific examples were lysimeters 7b and 8a where the efficiencies were reduced from 0.603 to 0.4 and 0.644 to 0.46 respectively.

Table 3: Parameter values used for the biogeochemical model

* The simulation was done for the three replicates of each lysimeter treatment.

Figures 4-7 describe the general picture of the simulation and observed chloride concentration for all groups of lysimeters while Tables 4-7 describes a comparison of the modelled and observed chloride amount. Chloride amounts in the tables are accumulated amount of chloride available in the soil during the retention and release period of the simulation. This estimate was done to evaluate the quantitative performance of the model. The retention represents the period when the chloride concentration in the leachate decreased temporarily before a subsequent increase (period of release) as observed in almost all the lysimeters. The lysimeters have different times of shift from retention to release, and this is shown as the retention period in days on the tables. This estimation is done to compare the simulated output with the observed at both periods. The model R2 values for the different lysimeters were calculated according to the plotted figures using the estimate of Nash and Sutcliffe (1970). The R2 estimate was done to evaluate the qualitative performance of the model. The figures describe the different behavioural patterns of the lysimeters that resulted in values (in amounts) as shown on the tables.

Group Precipitation Chloride Load *Lysimeter treatment Biomass Increase Rate Biomass Decrease Rate Initial Chloride (mg) Initial O2 (mg)

1

2

3

4

LOW LOW ₁ 0.1 0.02 1.3 500 2 HIGH ₃ 0.14 0.009 5 650 4 HIGH LOW ₅ 0.06 0.02 0.6 500 6 HIGH ₇ 0.1 0.01 1.0 650 8

(22)

The pattern of the observed and modelled concentration of chloride for group 1 lysimeters (i.e. those under low precipitation and low chloride load) is presented as shown in Figure 4. For all the lysimeter treatments, the observed retention period ranged between 24-38 days (Table 4). The model (as shown in Table 4) for this group of lysimeters had chloride retained for 24 days with an available amount of 0.23 and 2.07 mg during the retention and release periods respectively. However, the observed available chloride amount during retention ranged from no value during retention in lysimeter 2c, to a maximum value of 0.47 mg in lysimeter 2b. The observed available amount during chloride release for all the lysimeters also ranged between 1.64 mg in lysimeter 2b to a maximum value of 2.96 mg in lysimeter 1c. This range did not include the value for 2c because of the exceptional pattern observed in this replicate. Model efficiency (R2)values ranged from as low as -2.57 for lysimeter 2c to a very strong value of 0.826 for lysimeter 1c.

It can be seen from Figure 4 that the model performed considerably better for all the treatment replicates of lysimeter 1. This was however different for lysimeter 2 replicates, where only 2a was fairly described with an efficiency of 0.668. Lysimeter 2c was exceptionally low most probably because of the observed behaviour of no retention during the course of the experiment. For all the lysimeters, it can also be discerned that the model accounted less for the observed values during the release period than the retention period. The model did not account for the noticed drops in chloride concentration during the release of chloride (see Figure 4).

Table 4: Comparison of the available modelled and observed chloride amount for group 1 lysimeters Lysimeter Observations Model 1a 1b 1c 2a 2b* 2c Available amount during chloride retention (mg) 0.23 0.25 0.12 0.12 0.10 0.47 No distinct retention Available amount during chloride release

(mg)

2.07 2.26 2.27 2.96 2.16 1.64 0

Retention period (days) 24 31 24 24 31 38 0

* Values are to be compared with an alternative model calculation of 0.13 mg as the available amount during chloride retention. This is because an initial observed value was specifically missing for this lysimeter at the start of the simulation.

(23)

Lysimeter 1a 0 1 2 3 4 5 6 0 50 DAYS 100 150 C h lo ri d e c o n c ( m g /L ) Lysimeter 1b 0 1 2 3 4 5 6 7 8 0 50 DAYS 100 150 C h lo ri d e c o n c ( m g /L ) Lysimeter 1c 0 1 2 3 4 5 6 7 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) Lysimeter 2a 0 1 2 3 4 5 6 7 0 50 DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 2b 0 1 2 3 4 5 6 7 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) Lysimeter 2c 0 2 4 6 8 10 12 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) Lysimeter 2c 0 1 2 3 4 5 6 7 8 9 10 11 12 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) R2 = 0.826 R2 = 0.644 R2 _{= 0.752} R2 _{= 0.012} R2 _{= 0.668} R2 _{= - 2.57}

Figure 4: Simulated chloride concentration for Group 1 Lysimeters over a period of 122days. Plotted line represents the simulated concentration while the squares represent the observed.

(24)

Model behaviour and outputs in relation to the observed scenarios for lysimeters in group 2 (i.e. those under low precipitation and high chloride treatment) are presented in Figure 5 and Table 5. The model for this group of lysimeters retained chloride for the first 24 days of the simulation period. The model simulated an available chloride amount of 1.00 mg for the retention period and 5.91 mg during the release period. The observed available chloride amount during retention ranges from the lowest value of 1.12 mg in 3a to the highest value of 1.97 mg in 3c. The observed available chloride amount during release also ranged between 3.74 mg and 7.46 mg for all the lysimeters. The lysimeters had two retention periods of 31 and 38 for lysimeter 3a and all other lysimeters respectively. The model efficiency also ranged from as low as -1.127 in lysimeter 3c to a good performance of 0.924 in lysimeter 3a (Figure 5). The model can be said to fairly represent the overall observed behaviour for this group of lysimeters. The model accounted better for the observed pattern during the release period for this group of lysimeters as compared with group 1’s model performance during this period.

Lysimeter 3c can be said to record the least similarity with the model because of its exceptional behavioural pattern during the release period. It can be seen that this replicate showed some peculiar fluctuations during the release period. It had the least available amount during chloride release for all the lysimeters in the group.

Table 5: Comparison of the available modelled and observed chloride amount for group 2 lysimeters

Lysimeter Observations

Model 3a* 3b 3c 4a 4b 4c*

Available amount during chloride retention

(mg)

1.00 1.12 1.48 1.97 1.62 1.67 1.86

Available amount during chloride release

(mg/l)

5.91 7.37 6.18 3.74 7.14 7.46 7.28

Retention period (days) 24 31 38 38 38 38 38

* Values are to be compared with an alternative model calculation of 0.58 mg as the available amount during chloride retention. This is because an initial observed value was specifically missing for this lysimeter at the start of the simulation.

(25)

Lysimeter 3a 0 2 4 6 8 10 12 14 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 3b 0 2 4 6 8 10 12 14 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L )) Lysimeter 3c 0 2 4 6 8 10 12 14 16 18 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 4a 0 2 4 6 8 10 12 14 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 4b 0 2 4 6 8 10 12 14 16 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 4c 0 2 4 6 8 10 12 14 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) R2 _{= 0.924} R2 _{= 0.831} R2 _{= 0.717} _R2 _{= 0.722} R2 _{= -1.127} R2 = 0.682

(26)

The model description and simulation for group 3 lysimeters (i.e. lysimeter treatments under high precipitation and low chloride input) is presented as shown in Table 6 and Figure 6. The model for this group of lysimeters retained chloride for the first 24 days of the simulation period. The model simulated an available chloride amount of 0.41 mg for the retention period and 3.17 mg during the release period. The observed available chloride amount during retention ranges from the lowest value of 0.31 mg in 5b to the highest value of 1.52 mg in 6c. The observed available chloride amount during release also ranged between 1.17 mg and 3.98 mg for all the lysimeters. The observed retention days for the lysimeter treatments were 31, 38 and 52 days with 5a, 5c and 6a having a retention of 31 days, 5b and 6c retained chloride for 52 days while 6b retained for 38 days (see Table 6). The model efficiency also ranged from as low as -1.901 in lysimeter 5b to 0.643 in lysimeter 5a (Figure 6).

Thus, it can be discerned that the model did not properly describe the observed chloride behaviour for the group. From Figure 6, it can be seen that the model made a general overestimation for most of the lysimeters. The model could not account well for the low chloride outflows observed in the lysimeters especially during the release period of the simulation. The model stopped the retention earlier and initiated release earlier, faster and higher than what was observed in the lysimeters. The model for this group of lysimeters performed less than the initially noticed fair performance of group 1’s model.

Lysimeter Observations Model 5a 5b 5c 6a 6b 6c Available amount during Chloride retention (mg) 0.41 0.54 0.31 0.67 0.42 0.39 1.52 Available amount during Chloride release

(mg)

3.17 3.98 1.75 3.50 2.88 2.82 1.17

(27)

Lysimeter 5a 0 0.5 1 1.5 2 2.5 0 50 DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 5b 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 50 DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 5c 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 6a 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 6b 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 6c 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysim eter 5a 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) Lysimeter 6c 0 0.2 0.4 0.6 0.81 1.2 1.4 1.6 1.82 2.2 2.4 2.6 2.83 3.2 3.4 3.6 3.84 4.2 4.4 4.6 4.85 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) R2 = 0.643 R2 _{= -0.379} R2 _{= -1.901} R2 _{= 0.331} R2 = -0.614 R2 _{= 0.311}

(28)

Also presented in Figure 7 and Table 7 is the model outcome in comparison with the observed patterns for lysimeters in group 4 (i.e. lysimeter treatments under high precipitation and high chloride input). The model for this group of lysimeters retained chloride for the first 24 days of the simulation period. The model simulated an available chloride amount of 0.63 mg for the retention period and 8.03 mg during the release period. The retention period of the model was similar to most of lysimeters as all of them but lysimeter 7c also had 24 days for the temporal retention period of chloride. The observed available chloride amount during retention ranges from the lowest value of 0.36 mg in 8a to the highest value of 0.83 mg in 8c. The observed available chloride amount during release also ranged between 6.68 mg and 8.97 mg for all the lysimeters. The model efficiency also ranged from as low as -0.057 in lysimeter 7c to an average performance of 0.644 in lysimeter 8a (Figure 7).

The model can be said to perform considerably well for this group in terms of having the general pattern observed for most of the lysimeters. This can be asserted with the accuracy level depicted in the retention day modelled for the lysimeters. However, the performance of the model was reduced, because of its inability to account for the irregular increase pattern and outlier values noticed during the release period of most of the lysimeters. The outlier effect is mostly evident in lysimeters 7c, 7a and 8c. Thus, the model did a slight overestimate for some of the lysimeters.

Lysimeter Observations

Model 7a 7b 7c 8a 8b 8c

Available amount during chloride retention

(mg)

0.63 0.46 0.53 0.58 0.36 0.41 0.83

Available amount during chloride release

(mg)

8.03 6.68 8.71 7.18 8.50 6.87 8.97

(29)

Lysimeter 7a 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 7b 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 7c 0 1 2 3 4 5 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 8a 0 1 2 3 4 5 6 0 50 DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 8b 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysimeter 8c 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 _DAYS 100 150 C h lo ri d e C o n c . (m g /L ) Lysim eter 7c 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) Lysim eter 8a 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 0 50 100 150 DAYS C h lo ri d e C o n c . (m g /L ) R2 _{= -0.057} R2 _{= 0.644}

(30)

3.3 Biomass Carbon and Chloride Availability

Table 8 below represents the average modelled estimate of the available chloride amount in the soil pore water during the active growth of the microbes. It is defined by the retention period of 24 days as modelled for all the lysimeters. It is assumed that they are retaining chloride due to their uptake for growth. As presented in the Table 8, the maximum biomass carbon represents the modelled amount accumulated as at the retention day (24) for all the lysimeter groups.

Table 8: Comparison of the modelled available chloride amount and biomass carbon accumulated during the retention period.

Group 1 Group 2 Group 3 Group 4

Available chloride amount (mg) 0.02 0.10 0.04 0.06

Biomass Carbon (mg)

420 1300 210 580

3.4 Internal Dynamics of the Model

With oxygen as the limiting factor for microbial growth, the model revealed some internal dynamics of chloride transformation in terms of retention and release, oxygen depletion, and the increase and decrease of the biomass carbon in soil microbes. The dynamics showed in Figure 8 represents a description of the model behaviour for almost all the lysimeters. The figure shows plotted oxygen and biomass carbon amount in milligrams with modelled and observed chloride concentration (in mg/l) for one of the lysimeters. The model described a drop in biomass carbon content of the microbes at around the time the oxygen content dropped to zero. The initial period (increase in biomass carbon with oxygen availability) also describes the retention of chloride from the soil water while the subsequent period (decrease in biomass carbon at zero oxygen level) describes the release of chloride back to the soil water.

Figure 8: Simulated oxygen, biomass carbon and modelled and observed chloride concentration over a period of 122 days 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 1 8 1 5 2 2 2 9 3 6 4 3 5 0 5 7 6 4 7 1 7 8 8 5 9 2 9 9 1 0 6 1 1 3 1 2 0 D A Y S 0 1 2 3 4 5 6 Cl -(mg/L) O2 (mg)

(31)

4.0 DISCUSSION

4.1 Chloride Movement in Soils

This is a first attempt to model the biogeochemistry of chlorine in soil. Thus, it brought different challenges in terms of comparison with published result. The main challenge is fulfilling the required details appropriate to build a biogeochemical model with limited available information. For this study, the building of the model involved careful selection of internal and external factors reported to have influenced the biogeochemistry of chlorine in forest soil. It also entailed the careful selection of reported parametric values that are relevant to the ecosystem of coniferous soils. In the face of the aforementioned challenges, the model outcome was able to complement and provide better understanding of the existing paradigms in chlorine biogeochemistry. Simulation patterns showed in Figures 4-7 for all the lysimeters clearly described the reported (Öberg and Sanden 2005; Öberg et al., 2005) non-conservative hypothesis. It revealed an initial retention of the deposited chloride and a subsequent release into the soil under different water and chloride loads. In general, the model made a considerable description of the reported chloride movement in the soil, even with the underlying assumption.

The model could be considered adequate to support the non-conservative hypothesis because; the important parametric values that were assumed to regulate the transformation of chloride (such as initial biomass carbon, metabolic quotient, and biomass increase and decrease rate) were consistent with reported values for forest soils. The modelled biomass carbon content ranged from 100 to 922.64 mg (which is equivalent to 12.5 -115.33 g m-2). This range was consistent with the reported biomass content (12 - 422 g m-2) of forest soils as reported by Bauhus and Khana (1999), Raubuch and Beese (1995; 2005) and Friedel et al. (2006). The metabolic quotient modelled for all the lysimeter (0.045 mg O2 d-1 mg-1 C) was also in the same order of magnitude as reported by Raubuch and Beese (1999).

Though, the model had different strength of efficiencies and outcome comparison with the observed (Tables 4-7), Figures 4-7 still defines it appropriate because of its ability to repeat the same trend of scenarios for all the lysimeters irrespective of the interaction of the different water and chloride treatment under varying soil conditions. The biogeochemical part of the model described the retention of chlorine in terms of mobilisation and immobilisation through microbial activity alone with the inclusion of oxygen as the main limiting factor for microbial growth. These are parts of the model’s simplifications that may have limited its performance. The present consideration of microbial activity as the only factor responsible for chloride retention could be part of the reasons why the model had low efficiencies for lysimeters under high water loads (short residence time) compared with those under low water loads (long residence time). It can be hypothesised that high water residence in group 1 and 2 lysimeters encouraged both chemical and biological activities of the soil – a situation that will enhance the surface reaction of soil particles and the microbes. Thus, these conditions could be speculated to have brought the best performance from the model for this group, since the retention of chloride was primarily assumed microbial.

(32)

For lysimeters (group 3 and 4) under short water residence time (situations that witnessed low model efficiency), it could be hypothesised that microbial transformation was not completely favoured, because there is a reduced reaction as a result of the short water residence. Hence, a microbial driven model may not completely describe chloride transformation in the lysimeters under this condition. It could be speculated here that, other factors not incorporated in the model may have predominate under unfavourable conditions for biological reactions.

4.2 Organic matter and microbial activity

Another possible reason that may have affected the model performance under varying water conditions is that, oxygen alone may not have completely limited microbial activity under different water residence time. The performance of the model could have been enhanced for both water residence periods if organic matter had been included in the model. Availability of different fractions of organic matter was forwarded by Bastviken et al. (2006) to influence the activity of soil microbes at different levels of oxygen availability, most especially during the immobilisation of chloride ions. The oxygen - organic matter - microbe interaction could have led to a better description of chloride retention by microbial activity alone. For this study, there was no data available for the different fractions of organic matter in the lysimeters. Thus, the reported efficiencies and amounts of modelling microbial transformation of chloride in the lysimeters may have been underestimated.

4.3 Chloride loads and Microbial Retention

Result of this study also revealed that, the model had better performance under high chloride loads compared with low chloride loads. In earlier studies (Bastviken et al. 2006, Johansson et al. 2003) it was reported that higher chloride loads could influence the net-retention of chloride ions, however, there was no report suggesting that this could be largely described by microbial action. Figures 4-7 showed a better model efficiency for lysimeters treated with high chloride loads compared to those treated with low chloride loads under the same water treatment. This suggests that an increase in chloride deposition could generally enhance the activity of soil microbes and their consequent ability to retain chloride.

4.4 Microbes and the Retention-Release Pattern

One of the characteristic behaviour of the non-conservative hypothesis is the sudden shift from net chloride retention to net release as showed in Figures 4-7. Bastviken et al. 2006 noticed this behaviour and they speculated possible soil reactions that could be responsible. They suggested microbial activity amongst other factors that maybe responsible. The model described the net retention release pattern for almost all the lysimeters and it described the shift at around the same period (see Tables 4-7) as reported by Bastviken et al. 2006. Thus, it can be affirmed that the limitation of microbial activity by oxygen or other conditions is an important factor that could be responsible for the shift noticed and described for the non-conservative behaviour of chloride in soils. The Figures (4-7) also confirms the substantial impact of microbial activity on chloride retention amongst other speculations.