The impact of water repellency on water infiltration in the Dutch dunes

(1)

University of Amsterdam

Faculty of Science

Institute for Biodiversity and Ecosystem Dynamics

Bachelor thesis

The impact of water repellency on water

infiltration in the Dutch dunes

Claudia Schwennen

Amsterdam, July 2016

Student number: 10655808

Supervisor: Dr. L. H. Cammeraat

Study programme: Future Planet Studies

Study track: Earth Science

(2)

List of Figures

Figure 1. Research area ... 6

Figure 2. Locations of sample points in the Hors ... 7

Figure 3. Setup of the rainfall simulator ... 9

Figure 4. The filterbox ... 10

Figure 5. Soil classifications of the 5 landscapes ... 11

Figure 6. The runoff and infiltration as a percentage of the rainfall intensity for each run ... 14

Figure 7. The erosion rate for each run ... 15

Figure 8. Standard infiltration curve when w exceeds fc ... 18

Figure 9. The amount of slopewash and SOM content of the slopewash for each run ... 23

List of Tables

Table 1. Repellency categories according to the Water Drop Penetration Time test ... 8

Table 2. Order of the different rainfall intensities ... 9

Table 3. The 3 rainfall intensities at the different landscapes ... 9

Table 4. The average dry bulk density of each landscape ... 12

Table 5. The SOM content at each landscape ... 12

Table 6. The repellency levels of the soil profiles of 25 cm at each landscape ... 12

Table 7. Overview of the rainfall simulation at 30 mm hr-1 ... 13

Table 10. Correlation coefficients between repellency categories and diverse variables ... 16

Table 11. Correlation coefficients between runoff and erosion ... 16

List of notations and abbreviations

𝒇_𝒄 Infiltration capacity 𝒇 Infiltration rate

θ Water content

𝑴_𝒎 Dry mass of the soil

OM Organic Matter

𝝆𝒃 Dry bulk density

SOM Soil Organic Matter

SWR Soil Water Repellency

𝒘 Water input

WDPT Water Drop Penetration Time

WEPP Water Erosion Prediction Project

(4)

Abstract

This study provides information on the effects of water repellency on the hydrological responses of dune soils along the succession stages of vegetation. The research was conducted in the Hors, on the southern part of Texel, which is one of the Wadden Islands of the Netherlands. Soil monoliths of 40x40x5 cm were taken along 5 different landscapes, starting with the highest succession stage of these landscapes and ending with a bare soil. The dry bulk density and the Soil Organic Matter (SOM) content were determined, as well as the level of water repellency at every 5 cm of a soil profile of 25 cm, according to the Water Drop Penetration Time (WDPT) test. Finally, the rainfall simulator was used at intensities of 30, 60 and 90 mm hr-1 in order to determine the runoff rate, the infiltration rate, the erosion rate and the amount of SOM present in the slopewash. The results show that water repellency is generally positively correlated with SOM, runoff and erosion and negatively correlated with infiltration. However, the results are not statistically significant, due to the small sample size. Several points can be made that need to be taken into account for future research. First, the sample size needs to be increased. Second, the role of SOM should be examined further, as well as the specific hydrophobic compounds that cause water repellency. Third, other methods and devices should be evaluated to determine which are best to use. Finally, the role of vegetation should be examined further.

(5)

1. Introduction

Coastal dunes exhibit many different functions, including ecosystem functions, recreational functions and protective functions. This is also demonstrated by the fact that they are used in different ways, such as for agricultural purposes, natural protection from the sea and as a source of naturally purified drinking water (Martínez et al., 2004). Their importance within the landscape is especially significant in the Netherlands. Since a large part lies below the sea level, the coastal dunes are one of the most important barriers that protect the inland from the seawater (Van Koningsveld et al., 2007). They have also provided a large part of the drinking water supply since the 19th century and today almost all of the mainland dunes contribute to the drinking water supply, as well as some of the islands of the Netherlands (Best & Oosterhaven, 2012).

The dunes on the Wadden islands in the north of the Netherlands are particularly important from an ecological perspective, as they serve as habitats for several protected species. As a result, they are protected under Dutch legislation. The dunes’ functions demonstrate why coastal management is important and why it should be adequate. Therefore, sufficient understanding of the dynamics that are at play in the dunes is essential for coastal management (Durkin, 2011).

Soil Water Repellency (SWR) is an example of the factors that influence the dynamics of the dunes. According to Dekker et al. (2000), this phenomenon occurs at almost the complete coastal dune area of the Netherlands. Examples of the sources of water repellent substances include Organic Matter (OM) and vegetation (Doerr et al., 2000). This could impact the hydrological pathways that exist in the dune area, which is important to understand for sufficient management strategies. This study aims to provide this understanding, with special regard to the infiltration of water into dune soils. The research question of this study is “How does water repellency of dune soils along different stages of vegetation succession affect the soil’s infiltration capacity?”. In order to answer this question, several sub-questions were formulated and are as follows:

• How are other soil characteristics and environmental factors related to the infiltration capacity?

• How is water repellency related to organic matter content?

• What could be the consequences of a reduced infiltration capacity on the amount of runoff and erosion?

• How is rainfall intensity related to the amount of runoff and erosion?

The hypothesis is that “the infiltration capacity of dune soils reduces along higher stages of vegetation succession as an effect of increased levels of water repellency”. The study area is the southern part of Texel, one of the Wadden islands, located in the north of the Netherlands. The study includes a literature review, fieldwork and lab work, which includes the use of a rainfall simulator and the Water Drop Penetration Time (WDPT) test.

First, some background information is provided in the literature review. This is followed by the methodology and the results. Subsequently, the results will be discussed and related to other literature, which is followed by the conclusion of this research.

(6)

2. Literature review

2.1 Water repellency

During normal conditions, a force of attraction exists between water and the contact surface of soil particles, which is caused by the fact that the two bonds between an O2- ion and the H+ ions are positioned at an angle of 105° apart, giving the molecule a positively and a negatively charged side (Hem, 1985). As a result of this, water can be adhered to any contact surface that is either positively or negatively charged. This can only happen when that force is larger than the surface tension of water, which is 0.07197 N m-1 at 25 °C (Mortimer, 2008). This causes the water to lose its spherical shape, as the cohesiveness of the water decreases. Consequently, water can flow freely through the soil column along the contact surfaces of the soil particles (Dekker & Ritsema, 2000). However, when water repellent substances are present, this force of attraction is neutralized and water will have a spherical shape, so it cannot move freely through the soil.

As stated before, OM and vegetation are examples of sources of water repellent substances, but this also includes microorganisms and fungi (Doerr et al., 2000). The presence of these water repellent substances does not automatically mean that the soil is water repellent (Dekker & Ritsema, 2000). According to Ma’Shum et al. (1988), water repellent substances are usually absorbed as small globules, but a certain number of uniform monolayers are needed to make a soil particle water repellent. Moreover, the texture of the soil affects the chance of a soil being exposed to water repellency, as this influences the total surface area of the soil particles. A sandy soil for example has a lower total surface area than a clayey soil and has therefore a higher chance of being affected by hydrophobic substances (Doerr et al., 2006). The water content θ is also a factor that negatively influences the level of water repellency in the soil (Doerr & Moody, 2004; Hallett, 2007).

2.2 Coastal dune systems

According to Kim & Yu (2009) and Jungerius & Van der Meulen (1988), a coastal dune system consists of three parts at which different processes are at play. The first part of this system is the foredune or also called the yellow dune, located nearby the sea. Processes that are mainly dominant here are geomorphological processes, including Aeolian processes. The vegetation in this part is in the early stages of the succession process, due to the young dune age and the dominant processes in this part. Furthermore, yellow sand is present here. Yellow sand can easily be transported by wind erosion, but due to its high permeability, water erosion has a lower effect due to low runoff rates (Jungerius & De Jong, 1989).

The second part, adjacent to the foredune, are the grey dunes. Both geomorphological and biological processes are present here, including acidification. These determine the vegetation patterns in this part. However, it is unclear how acidification works in this part, whether it is caused by internal factors, such as the decomposition of OM (Berendse et al., 1998), or by external factors, such as the deposition of atmospheric acidic compounds (De Vries et al., 1994). Grey sand is present here, which is substantially affected by water erosion, but less by wind erosion. This is due to the fact that the higher OM amount in grey sand increases the cohesiveness of the sand (Jungerius & De Jong, 1989).

Finally, the part adjacent to the grey dunes are the inner dunes. Biological processes are mainly dominant in this part and this allows the vegetation to further stabilize. This also stabilizes the soil, so that geomorphological processes have a smaller influence than on other parts of the coastal dune system.

(7)

2.3 The Hors

As stated earlier, the research area is located on the southern part of the island Texel, which is part of the Wadden islands of the Netherlands. The exact location is the Hors area, nearby Mok bay. This is visualized in figure 1. This area is a quite young dune area that has been and still is under anthropogenic influences, and the dynamics that exist in this area play a significant role in the formation of small dune ridges, which prevent the dune slacks to be flooded regularly by the sea (Shahrudin et al., 2014). A study that describes the Moksloot area, located north of the Hors, explains that this area has become substantially drier due to the digging of a ditch, the drainage of polder areas around the Moksloot and the extraction of drinking water (Nienhuis & Gulati, 2002). This might also have been the case for the Hors. Furthermore, isolated plots in the Moksloot show a higher biodiversity than flooded plots, which is likely caused by dispersal actors, including wind and animals (ibid.). Since the Hors seems to be protected by the dune ridges from floodings, the dispersal actors might also be active in this area.

Figure 1. Research area

2.4 Factors influencing the infiltration capacity

Horton (1941) states that the infiltration capacity (ƒc) depends on a number of factors, but

what is mostly important to notice is that this variable can vary significantly over time, due to the fact that soil conditions significantly affect the infiltrability and that these also vary over time. Factors that are important to note within soil conditions include the mineral composition of the soil, the texture and the structure, which also includes the micro- and macro-structure and also the biologic structure. Other factors include temperature of the air, water and soil, the intensity of the rainfall and the initial moisture content (ibid.). These factors were also recognized by Dingman (2015). However, he mentions another variable that depends on the previously mentioned factors, which is the hydraulic conductivity. This can vary throughout the soil column, which also determines the infiltrability of the soil.

As mentioned before, the rainfall intensity also affects the infiltrability of the soil. This could also vary during a rainfall event, depending on the rainfall rate. Three situations can be distinguished at which the rainfall intensity equals or is below the infiltration capacity (𝑓_!) or exceeds the infiltration capacity (ibid.). These are clarified below.

𝑤 𝑡 ≤ 𝑓_! (𝑡) 𝑤 𝑡 > 𝑓_! (𝑡)

(2.1) (2.2)

(8)

The first equation shows the situation when the water input (𝑤) is lower than ƒc and the

situation when 𝑤 is equal to ƒc. This means that the infiltration rate equals 𝑤, as ƒc is not

exceeded. So, ƒc does not change during the first two situations, as ƒc has not been exceeded.

In contrast, the second equation shows the situation when 𝑤 exceeds ƒc, so the water that does

not infiltrate into the soil will accumulate above the soil surface, which results into ponding (Dingman, 2015). Only when the rainfall rate exceeds ƒc, ƒc could change during the event

(Horton, 1941). In this case, ƒc will decrease during the event until it reaches a minimum

value.

3. Methodology

3.1 Fieldwork

5 monoliths were taken at 5 different locations, which are visualized below. These were indicated through Google Earth. The locations will be discussed further in the following chapter together with the description of the soil profiles. Moreover, the soil at each location was classified according to the World Reference Base for soil resources (WRB, 2014).

Figure 2. Locations of sample points in the Hors

Wooden frames of 40x40x5 cm were used for the soil monoliths, which were used for the rainfall simulator. Besides the soil monoliths that were taken at the landscapes, samples for the bulk density, the determination of the SOM content and the WDPT test were taken as well. Two samples were taken for the bulk density and the SOM at each landscape by using metal rings with a volume of 100 cm3. Moreover, samples for the WDPT were taken using a soil probe of 25 cm. These samples were taken at every 5 cm of the soil profile.

3.2 Experiments

Before conducting any experiment, the monoliths and the samples were dried in the oven at 30 °C. Before each repetition of the rainfall simulator, the samples needed to be dried again at

(9)

30 °C in order to standardize the soil conditions. Secondly, the samples that are used for the determination of the SOM content needed to be sieved first.

3.2.1 Dry bulk density

First, the density of the soil was determined in order to determine the volume of the soil that was transposed into containers. Approximately 5 g of each sample was transposed twice to containers and these were put in the oven at 105 °C for 24 hours. These samples were weighed again after they were dried. The dry bulk density was calculated according to the equation proposed by Dingman (2015), where 𝜌! is the dry bulk density in g/cm3, 𝑀! is the

dry mass of the soil samples in g and 𝑉_! is the volume of the soil samples before drying in cm3.

𝜌! =

𝑀_! 𝑉_!

3.2.2 Determination of the SOM content of the samples from Texel

The method proposed by NEN 5754 (1992) was used for the determination of the SOM content. Firstly, the samples were sieved with a 2 mm sieve. Next, 5 g was taken twice from these samples and put into containers. These samples were dried at 105 °C for 24 hours as well. Lastly, they were put in the oven at 375 °C for 16 hours. The soil loss after the drying at 375 °C is equal to the SOM content.

3.2.3 The Water Drop Penetration Time test

There are different methods that can determine the level of water repellency of the soil. The method that was used in this research is the WDPT test, due to its simplicity (Leelamanie et al., 2008; Hallett, 2007). This test also gives an indication of the erosion, since the time to infiltration directly influences the amount of overland flow (ibid.). The principle of the test is that a water droplet is put onto the soil samples and the time to infiltration is recorded. The time to infiltration and the corresponding water repellency categories are given in table 1.

Repellency categories

Water Drop Penetration Time (WDPT) Repellency Category

≤ 1 s Non-repellent

1-60 s Slightly repellent

60-600 s Strongly repellent 600-3600 s Severely repellent ≥ 3600 s Extremely repellent

Table 1. Repellency categories according to the Water Drop Penetration Time test (WDPT) (Leelamanie et al. 2008; Bisdom et al. 1993; Chenu et al. 2000; King 1981)

3.2.4 The rainfall simulator

According to Cerdà et al. (2015), the use of the rainfall simulator is a very common method to observe hydrological responses of the soil. The rainfall simulator consists of several parts, which include a water tank, a metal frame with a dripping plate on top and a mesh grid underneath the dripping plate. The water tank is connected to the dripping plate through a tube. In order to keep a constant rate of the water transport, the water tank is closed off at the top with a lid. The rainfall intensity is set manually with a tap located on the water tank.

(10)

For the setup of the simulator, the plot that is being used should have a slight angle in order to direct the water flow into the bottles in which the water and slopewash will be collected. This was set at an angle of 4°. After this, a metal ring should be placed on the plot, which makes sure that the water will flow into one direction. The metal frame with the dripping plate and the mesh grid should be placed perpendicular to the metal ring and horizontally. The complete setup is visualized in figure 3.

The three rainfall intensities that were used in this research are (1) 30, (2) 60 and (3) 90 mm hr-1. Due to the fact that the rainfall intensity cannot be set automatically, the intensity was determined manually, which caused some issues. Therefore, the three rainfall intensities were used in different orders, which can be seen in table 2.

Figure 3. Setup of the rainfall simulator

The runoff and erosion that were produced during the rainfall simulations were collected in 200 ml polyethylene bottles for each minute of a simulation of 20 minutes. The runoff and erosion rates can be calculated using the volume, the area of the plot and the simulation time. The infiltration rate is determined by the rainfall intensity minus the runoff rate (Toy et al., 2002).

3.2.5 Determination of the SOM content of the slopewash

The SOM content was also determined for the slopewash. Before putting the samples into the oven, the slopewash needed to be separated from the runoff, which was done by using a filterbox. The filterbox is visualized in figure 4. Both the runoff and erosion were weighed together, before the vacuum filtering. In this way, the amount of runoff can be calculated after

Order of the rainfall intensities

Plot 1.1 Plot 1.2 Plot 2 Plot 3 Plot 4 Plot 5

3 2 2 3 1 1

1 3 3 1 2 2

2 1 1 2 3 3

Table 2. Order of the different rainfall intensities

Rainfall intensities

Intensity 1 Intensity 2 Intensity 3

Plot 1.1 32.61 mm hr-1 47.63 mm hr-1 94.46 mm hr-1 Plot 1.2 34.42 mm hr-1 59.98 mm hr-1 98.99 mm hr-1 Plot 2 35.98 mm hr-1 45.51 mm hr-1 95.45 mm hr-1 Plot 3 17.99 mm hr-1 70.43 mm hr-1 102.08 mm hr-1 Plot 4 23.28 mm hr-1 _{67.73 mm hr}-1 _{84.46 mm hr}-1 Plot 5 23.28 mm hr-1 50.80 mm hr-1 77.47 mm hr-1

(11)

the weight of the slopewash has been determined. Next, the samples were filtered on the filterbox. Afterwards, the filters and the slopewash were dried for 24 hours at 70 °C to determine the dry weight of the slopewash. Finally, these samples were dried for 16 hours at 375 °C and weighed again to calculate the weight loss (NEN 5754, 1992).

Figure 4. The filterbox

3.3 Statistical analyses

All the statistical analyses were conducted in MATLAB ®. Firstly, the data needed to be tested for normality, as this determines which tests should be used. This was done by using the Lillie test. The parametric substitute of the ANOVA test was used to compare non-normally distributed data, which is the Kruskal-Wallis test (Burt, 2009). The ANOVA test is used for the comparison of normally distributed data. These tests return a p-value and if this is below or equal to the significance level, the differences between the different distributions are statistically significant.

The distribution of the data also determines which correlation coefficient should be calculated. The Pearson correlation coefficient was used for the normally distributed data, while the non-parametric substitute was used for the data that is not, which is the Spearman correlation coefficient (Trauth, 2007).

4. Results

4.1 Description of the plots

Plot 1 was located at the oldest dunes, more inland. The dunes of this part of the Hors are approximately 275 years old. The upper part of the soil is the first Ah horizon with a depth of 1.5 cm. This horizon had a 10 Y/R 6/2 Munsell colour. The next horizon is the second Ah horizon with a depth of 8.5 cm and had a 10 Y/R 5/2 Munsell colour. Both horizons are weakly crumbly. The last horizon is the C horizon with a 10 Y/R 7/3 Munsell colour. A sandy texture was found throughout the whole profile.

The next plot was located slightly further towards the sea, with slightly less vegetation than the first plot. The first horizon of the soil profile was the Ah1 horizon with a Munsell colour of 10 Y/R 6/2 and had a depth of 1 cm. The soil in this horizon was weakly crumbly. The next horizon was the Ah2 horizon with a Munsell colour of 10 Y/R 5/2 and a depth of 4 cm. The last horizon was the C horizon with a 10 Y/R 7/2 Munsell colour.

(12)

The third plot was a semi-bare plot. The soil profile consisted of two Ah horizons and a C horizon. The first Ah horizon had a depth of 1.5 cm and a Munsell colour of 10 Y/R 5/2. The second Ah horizon had a depth of 1 cm and had a 10 Y/R 6/2 Munsell colour. In both horizons the soil was weakly crumbly. The last horizon had a 10 Y/R 6/3 Munsell colour and it had a sandy texture.

Only one Ah horizon was found throughout the soil profile of the second last plot, which had a Munsell colour of 10 Y/R 5/2 with a depth of 1.5 cm. The soil was very weakly crumbly in this horizon. The next horizon had a Munsell colour of 10 Y/R 6/3 and had a sandy texture. The last plot was located on one of the youngest dunes with yellow sand. The soil profile had no A horizon and only a C horizon, which had a Munsell colour of 10 Y/R 7/2. Shell fragments were found on the surface of the soil.

Due to the texture, the fact that there was no reaction with HCl and that the soils are likely to have a pH > 5 (Shahrudin et al., 2014), all soils were classified as Eutric Arenosols. All of them received the suffix qualifier Hydrophobic, except for plot 5, which received the suffix qualifier Aeolic.

Eutric Eutric Eutric Eutric Eutric Arenosol Arenosol Arenosol Arenosol Arenosol Hydrophobic Hydrophobic Hydrophobic Hydrophobic Aeolic

Figure 5. Soil classifications of the 5 landscapes

4.2 The dry bulk density

The following table includes the average dry bulk density of each landscape in g/cm3_{. This}

was calculated from 4 samples of each landscape. The dry bulk density was used to convert the slopewash from g to mm, in order to have a rate of mm hr-1.

0 5 10 15 20 25

Plot 1 Plot 2 Plot 3 Plot 4 Plot 5

D ept h (c m ) C Ah C Ah1 Ah2 C Ah1 Ah2 C Ah1 Ah2 C

(13)

Dry bulk density

Plot Average dry bulk density (g/cm3) Standard deviation

1 1.174 0.062

2 1.009 0.011

3 1.197 0.007

4 1.299 0.023

5 1.550 0.033

Table 4. The average dry bulk density of each landscape

4.3 The SOM content

Table 5 gives an overview of the SOM content that was found in the samples of each plot. Again, the average was calculated from 4 samples of each landscape. Generally, the SOM content seems to decrease from plot 1 to plot 5, but more data is needed to confirm this.

SOM content

Plot Average SOM content (%) Standard deviation

1 1.845 0.204

2 2.081 0.501

3 0.977 0.213

4 1.266 0.180

5 0.076 0.007

Table 5. The SOM content at each landscape

4.4 The Water Drop Penetration Time test

The water repellency levels of every 5 cm of the soil profiles of the landscapes are given in table 6. As can be seen in the table, plot 5 is completely non-repellent, while plot 1 shows considerably high levels of water repellency. In addition, the levels of water repellency seem to increase with depth from plot 4 to plot 2.

Water repellency categories of the soil profiles

Depth Plot 1 Plot 2 Plot 3 Plot 4 Plot 5

0 - 5 cm Extremely repellent Strongly repellent Strongly repellent Strongly repellent Non-repellent 5 - 10 cm Extremely

repellent Strongly repellent Strongly repellent Strongly repellent Non-repellent 10 - 15 cm Extremely repellent Strongly repellent Strongly repellent Slightly repellent Non-repellent 15 - 20 cm Strongly

repellent Strongly repellent Slightly repellent Slightly repellent Non-repellent 20 - 25 cm Slightly repellent Slightly repellent Slightly repellent Slightly repellent Non-repellent

Table 6. The repellency levels of the soil profiles of 25 cm at each landscape

4.5 Rainfall simulator experimental data

The following tables show the summary of the data that was retrieved from the rainfall simulations, which contains the time to runoff, the rainfall intensity, the average runoff, the runoff as well as the infiltration as a percentage of the rainfall intensity, the maximum runoff (%), the final infiltration (%), the maximum erosion rate and finally the average erosion rates. Generally, the time to runoff seems to increase from plot 1 to 5 for each run, as well as the amount that infiltrates. Also, plot 1 seems to produce the highest slopewash for run 1 and 2, but a linear decrease from plot 1 to 5 is not very clear for run 3.

(14)

Overview of run 1

Run 1 Plot 1 Plot 2 Plot 3 Plot 4 Plot 5 Time to runoff (s) 163 470 - - - Rainfall intensity (mm hr-1) 34.43 35.98 17.99 23.28 23.28 Runoff (mm hr-1) 11.691 0.149 0 0 0 Runoff (%) 33.96 0.41 0 0 0 Max. runoff (%) 76.19 3.67 0 0 0 Infiltration (%) 66.04 99.59 100 100 100 Final infiltration (%) 52.46 100 100 100 100 Max. erosion (mm hr-1) 0.020 ~ 0 0 0.001 0.001 Average erosion (mm hr-1) 0.001 ~ 0 0 ~ 0 ~ 0

Table 7. Overview of the rainfall simulation at 30 mm hr-1

Run 2 Plot 1 Plot 2 Plot 3 Plot 4 Plot 5 Time to runoff (s) 103 410 203 - - Rainfall intensity (mm hr-1) 59.99 45.51 70.43 67.73 50.80 Runoff (mm hr-1) 9.497 0.149 4.718 ~ 0 ~ 0 Runoff (%) 15.83 0.33 6.70 ~ 0 ~ 0 Max. runoff (%) 64.42 4.39 24.42 ~ 0 ~ 0 Infiltration (%) 84.17 99.67 93.30 ~ 100 ~ 100 Final infiltration (%) 84.08 100 83.86 100 ~ 100 Max. erosion (mm hr-1) 0.035 0.028 0.048 ~ 0 0.004 Average erosion (mm hr-1) 0.002 ~ 0 0.001 ~ 0 ~ 0

Run 3 Plot 1 Plot 2 Plot 3 Plot 4 Plot 5 Time to runoff (s) 36 181 97 1250 - Rainfall intensity (mm hr-1) 98.99 95.45 102.08 84.46 77.47 Runoff (mm hr-1) 71.067 21.702 19.415 1.317 0 Runoff (%) 71.79 22.74 19.02 1.56 0 Max. runoff (%) 95.72 66.95 55.63 9.51 0 Infiltration (%) 28.21 77.26 80.98 98.44 100 Final infiltration (%) 31.44 75.02 87.18 100 100 Max. erosion (mm hr-1) 0.707 1.056 6.346 0.001 0 Average erosion (mm hr-1) 0.097 0.064 1.702 ~ 0 0

The graphs containing both the runoff and infiltration as a percentage of the rainfall intensity are visualized in figure 6. As can be seen, plot 1 produces the highest runoff and has the lowest infiltration rate in general. This is followed by plot 2 and 3, with a moderately lower runoff rate and higher infiltration rate. Finally, plot 4 and 5 seem to produce almost no runoff and thus have an infiltration rate of almost 100% during the simulations. It also seems that higher rainfall intensities generally lead to higher amounts of runoff. What also can be seen is the typical response of water repellent soils, which is that the runoff rate is quite high in the start of the simulation, but decreases after some time has passed by until it reaches

(15)

equilibrium. This is due to the fact that the level of water repellency decreases after being in contact with water (Doerr & Moody, 2004), as explained in earlier sections.

Figure 6. The runoff and infiltration as a percentage of the rainfall intensity for each run

Furthermore, the erosion rates are visualized in figure 7. A closer look should be taken at the y-axes, as these differ for the different runs. It also seems that plot 1 produces the highest slopewash in general. However, this is not so clear for run 3, as stated before. Moreover, after 10 minutes the erosion rate of plot 3 is unusually high for run 3, but it is not clear what could have caused these high values, as the runoff rate is not unusually high. For a clearer comparison, another graph was included in which plot 3 was excluded.

(16)

The slopewash in grams and the SOM present in the slopewash are included in Appendix A. As will be explained in the following section, some issues were encountered when determining the amount of slopewash, which resulted in some negative values that were corrected according to the method described in the next section. Due to this, the SOM content exceeded 100% for a number of the samples and is therefore not considered further.

Figure 7. The erosion rate for each run

4.6 Statistical results

Firstly, negative values were recorded for the weight of the slopewash and needed to be corrected in order to do any statistical analysis. These were treated as missing values and several methods that deal with such missing values were reviewed by Sifford (1996). Considering these methods, it seemed most appropriate to estimate the values by pchip interpolation, as this preserves the high frequency of the data and the shape of the data (Trauth, 2007).

The Lillie test showed that most of the data was not normally distributed. As a result, the ANOVA test was only used for a small amount of the data. The test results of the Kruskal-Wallis test and ANOVA test show that most of the data is statistically different, except for plot 4 and 5 in different runs, as this produced almost no runoff and erosion.

The Pearson correlation coefficient was only calculated for the relationship between the repellency categories and the SOM content, as these were normally distributed. This resulted in a correlation coefficient of 0.7914 with a p-value of 0.1107, which is not statistically significant. The results of the Spearman correlation coefficients are summarised in table 10 and 11, which show strong negative relationships between repellency levels and infiltration and positive relationships between repellency levels and erosion, although they are not

(17)

statistically significant. Moreover, some positive relationships are found between runoff and erosion, indicating that runoff could contribute to erosion. No correlation coefficient was calculated between rainfall and runoff and between rainfall and erosion, as the sample size was too small.

Table 11. Correlation coefficients between runoff and erosion

5. Discussion

The statistical analyses show some interesting results, such as the strong relationships found between diverse variables, which were also found in other studies. However, they are not statistically significant due to the p-values that exceed the confidence levels. In addition, the accuracy of the methods seems to be inadequate.

5.1 Evaluation of the methods and devices

First of all, the rainfall simulator was supposed to produce a constant rainfall intensity throughout the complete simulation, which was not the case. This could have been caused by a leak in the water tank. Due to this, the tap was slightly further opened so that the intensity would reach the initial intensity again for some of these simulations. Moreover, a setup that includes a tap on which the intensity can be set at an equal value would also improve the accuracy of the results, as these were not equal for all plots during the different runs.

Secondly, the filterbox method might have caused some errors in the slopewash values, of which approximately 10% had negative values. However, it has also been used by other researchers, such as Brennan et al. (2011) and Battany and Grismer (2000). It might be useful to conduct an evaluation of different methods for the separation of runoff and slopewash and determine which is the most accurate method. This study could contribute to such evaluations. Finally, the fourth decimal of the weight was not considered, as this was unstable during the weighing of the slopewash. Since very small amounts of slopewash were found in some samples, it is recommended to use a balance with more decimals. It is also recommended to

Correlation between repellency levels and diverse variables Relation with repellency levels

Variable Run ρ p-value

Infiltration 1 -0.7500 0.3000 2 -0.8030 0.2000 3 -0.8944 0.1000 Erosion 1 0.6708 0.3000 2 0.4472 0.6000 3 0.6708 0.3000

Correlation between runoff and erosion

Run Plot ρ p-value

1 1 0,6055 0.0047 2 0.8747 4.5390e-07 3 - - 4 - - 5 - - 2 1 0.7431 1.7389e-04 2 0.5900 0.0062 3 -0.2610 0.2664 4 0.6167 0.0038 5 0.5368 0.0147 3 1 0.5070 0.0255 2 -0.0090 0.9699 3 -0.2535 0.2809 4 0.9844 4.8676e-15 5 - -

Table 10. Correlation coefficients between repellency categories and diverse variables

(18)

weigh both the slopewash and the runoff separately instead of discharging the runoff after using the filterbox, as the values of the slopewash weights determined the amount of runoff. 5.2 The role of OM

The results show that there is a strong positive correlation between the levels of SWR of each location and the OM content in the soil. However, this correlation is not statistically significant due to the sample size of this research. Still, the relationship that was found in this research has also been found in other studies. It must be noted though that other researchers have also found a negative relationship or even no relationship between the repellency category and the OM content in the soil. A possible explanation for this could be that the amount of hydrophobic substances in the soil might not be proportional to the amount of OM in the soil. Future research could observe which specific hydrophobic substances could cause water repellency and relate the amount of hydrophobic compounds to the repellency category and see whether this results in a stronger positive relationship (Doerr et al., 2000).

Moreover, the processes behind the production of water repellent substances are still unclear. Possible results have been provided by a number of studies. For example, some studies claim that it is the decomposition of OM in the soil that produces water repellent substances (McGhie & Posner, 1981). Others claim that the root activity determines the amount of hydrophobic compounds in the soil (Dekker & Ritsema, 1996). This is also supported by Mao et al. (2015), who also state that certain hydrophobic compounds have a stronger effect on the level of SWR than other hydrophobic compounds and that these are produced by root activity. 5.3 The hydrological responses

The statistical results suggest that water repellency is negatively related to infiltration and positively related to runoff. Water repellency also seems to be positively related to erosion, which is likely caused by the amount of runoff and the different impact of water erosion on grey and yellow sand, as explained before. Other researchers also found that water repellency leads to decreased infiltration rates and thus also to increased runoff and erosion rates. For example, Leighton-Boyce et al. (2007) found that runoff on repellent soils is 16 times higher than on wettable soils and that erosion is 23 times higher. Furthermore, they found that the hydraulic conductivity of repellent soils is lower than wettable soils. Doerr et al. (2006) found that the infiltration capacity and the storage capacity of repellent soils are significantly lower on repellent soils than on wettable soils. But in order to obtain significant results, the samples size needs to be increased.

Doerr and Moody (2004) state that many researchers found the linear relationship between hydrological processes and water repellency, but that the linear relationship might not be so clear anymore when other spatiotemporal scales are considered. For example, when looking at different temporal scales, the level of water repellency might vary significantly. This is due to the fact that the water content affects the level of water repellency. After long wet periods, soil might have a considerably lower level of water repellency. In contrast, the level can greatly increase again after long dry periods (ibid.). This is a very typical phenomenon of water repellency even on shorter terms. During a rainfall event, the level of water repellency may decrease after the first contact with water. This is also visible in the results of this research, where the runoff rate is considerably high in the start of the simulation. After a certain amount of time, the runoff rate decreases until it reaches equilibrium. Leighton-Boyce et al. (2007) also found this result, as a wetting front appeared after some time during their simulations on the repellent soils, which indicates a breakdown of the water repellency. This is especially evident when this is compared to a normal infiltration rate curve. As explained earlier, 3 situations can be distinguished where 𝑤 exceeds, equals or is lower than 𝑓_!. When it

(19)

exceeds 𝑓_!, 𝑓_! decreases during the rainfall event until it is approximately equal to the saturated hydraulic conductivity of the soil surface (Dingman, 2015). This is also visualized in the following figure. It shows that 𝑓 decreases until it reaches equilibrium, which is equal to the saturated hydraulic conductivity. Compared to figure 6, this is the opposite response. It shows that 𝑓 (%) is significantly low in the start after which it increases and finally reaches equilibrium, due to the breakdown of water repellency.

Figure 8. Standard infiltration curve when 𝑤 exceeds 𝑓_! (Şen, 2014)

5.4 Relation between rainfall and hydrological responses

No correlation coefficient could be calculated between the hydrological responses and the different rainfall intensities, due to the small sample size. Still, without a statistical analysis, the graphs suggest that erosion and runoff increase when the rainfall intensity increases. Similar results were found by others. Nearing et al. (2005) concluded that for each percentage increase of the rainfall intensity, the runoff and erosion would increase even further. Moreover, they state that the impact on erosion is larger than the impact on runoff. Pruski and Nearing (2002) found slightly different results, but they used the Water Erosion Prediction Project (WEPP) model to simulate different rainfall intensities. They concluded that the increase of runoff is larger than the increase of erosion when the rainfall intensity is increased. 5.5 The effect of vegetation

Before conducting the rainfall simulations, the vegetation was removed to see the relation between water repellent soils and hydrological responses. However, it might be interesting to observe how this relationship can be altered by the presence of vegetation. Cèrda (1997) compared the hydrological responses of cultivated and abandoned soils and concluded that the hydrological responses (runoff and ponding) were delayed on cultivated soils compared to abandoned soils, but they were also delayed on soils that have been abandoned for a longer time, as vegetation has developed, which increased the infiltration capacity. What they also found is that the infiltration capacity of soils with the vegetation type Pinus Halpensis was lower than the soils with different types of vegetation, due to the development of water repellency. This also suggests that some plant types are associated with water repellency, which has been found by other studies as well (Doerr et al., 2000).

Another study observed the effects of deforestation on the hydrological consequences and used water repellent soils. They concluded that the vegetation decreases the runoff and

(20)

certainly the erosion, but that the influence of water repellency can still be seen after long dry periods. According to them, the amount of runoff exceeded the expected amount when considering the vegetation cover. However, runoff was still low during wet periods (Benito et al., 2003).

Depending on the vegetation cover at the 5 different locations, the presence of the vegetation could alter the hydrological responses that were observed in this research. For the first two plots this means that erosion could be considerably lower, as well as the runoff, although this impact could be less than it is for erosion. For plots 3 and 4, the impact of vegetation could be less evident, due to the lower vegetation cover.

5.6 Management strategies

Water repellency could result in certain disadvantages, but it can also be advantageous for the soil. Examples of this are the positive impact on aggregate stability, the prevention of water loss through evaporation and the sequestration of organic carbon (Müller & Deurer, 2011; DeBano, 1981). However, it is important to recognize the disadvantages that are caused by water repellency, which include the negative impact on the Water Use Efficiency (WUE) of plants, on water retention, on the ability to prevent floodings through infiltration and finally on the provision of naturally purified drinking water (ibid.). Depending on the function of the area, whether this is for agriculture or nature conservation for example, attention should be given to certain impacts of water repellency and remediation methods should be chosen accordingly.

In addition, water repellency has a positive impact on the cohesion of the soil and therefore wind erosion does not really affect the soil, as mentioned earlier. This could have an impact on the biodiversity in coastal dune systems, as environments where Aeolian processes are dominant have a positive impact on the biodiversity (Jungerius & Van der Meulen, 1988; Kooijman et al., 2005), which is especially interesting for the Hors area, since this area is protected.

6. Conclusion

This research aimed to provide more insights into the hydrological consequences of water repellency of dune soils on the southern part of Texel, with special regard to the infiltration. The results are promising and seem to correspond with results found by other researchers, namely that the infiltration capacity reduces as an effect of water repellency and that in turn the runoff and erosion increases on repellent soils compared to wettable soils. However, the results of this research are not statistically significant. It is strongly recommended that larger sample sizes are used in future research, both for the number of landscapes and the amount of different runs at different rainfall intensities. Furthermore, it is recommended to evaluate different methods that separate the slopewash from the runoff and, if needed, to use different methods. This will considerably increase the accuracy of the results. Finally, it is also recommended to explore the role of OM in more detail, as it is unclear how this exactly influences the level of SWR. Specific hydrophobic compounds should be taken into account as well, as the amount of hydrophobic compounds might not be proportional to the amount of SOM. This caused diverse results in the existing literature, namely that there could be no relationship between OM and repellency or a negative relationship, which was not found in this research. Also, the role of vegetation could be examined further, as this might result into different hydrological responses. The knowledge obtained in this research could be used for coastal dune management strategies, which is highly recommended.

(21)

References

Battany, M.C. and Grismer, M.E., 2000. Rainfall runoff and erosion in Napa Valley vineyards: effects of slope, cover and surface roughness. Hydrological processes, 14(7), pp.1289-1304.

Bisdom, E.B.A., Dekker, L.W. and Schoute, J.T., 1993. Water repellency of sieve fractions from sandy soils and relationships with organic material and soil structure. Geoderma, 56(1-4), pp.105-118.

Benito, E., Santiago, J.L., De Blas, E. and Varela, M.E., 2003. Deforestation of water‐repellent soils in Galicia (NW Spain): effects on surface runoff and erosion under simulated rainfall. Earth Surface

Processes and Landforms, 28(2), pp.145-155.

Berendse, F., Lammerts, E.J. and Olff, H., 1998. Soil organic matter accumulation and its implications for nitrogen mineralization and plant species composition during succession in coastal dune

slacks. Plant Ecology, 137(1), pp.71-78.

Best, P.H. and Oosterhaven, J. eds., 2012. Netherlands-Wetlands: Proceedings of a Symposium Held

in Arnhem, the Netherlands, December 1989 (Vol. 88). Springer Science & Business Media.

Brennan, R.B., Fenton, O., Grant, J. and Healy, M.G., 2011. Impact of chemical amendment of dairy cattle slurry on phosphorus, suspended sediment and metal loss to runoff from a grassland soil. Science of the Total Environment, 409(23), pp.5111-5118.

Burt, J.E., Barber, G.M. and Rigby, D.L., 2009. Elementary statistics for geographers. Guilford Press. Cerdà, A., 1997. Soil erosion after land abandonment in a semiarid environment of southeastern

Spain. Arid Land Research and Management, 11(2), pp.163-176.

Cerdà, A., Pelayo, Ó.G., Pereira, P., Novara, A., Iserloh, T. and Prosdocimi, M., 2015, April. The wildgeographer avatar shows how to measure soil erosion rates by means of a rainfall simulator. In

EGU General Assembly Conference Abstracts (Vol. 17), p. 15878.

Chenu, C., Le Bissonnais, Y. and Arrouays, D., 2000. Organic matter influence on clay wettability and soil aggregate stability. Soil Science Society of America Journal, 64(4), pp.1479-1486.

DeBano, L.F., 1981. Water repellent soils: a state-of-the-art.USDA For. Serv. Gen.Tech. Rep. PSW-46. pp. 21

Dekker, L.W. and Ritsema, C.J., 1996. Variation in water content and wetting patterns in Dutch water repellent peaty clay and clayey peat soils. Catena, 28(1), pp.89-105.

Dekker, L.W. and Ritsema, C.J., 2000. Wetting patterns and moisture variability in water repellent Dutch soils. Journal of Hydrology, 231, pp.148-164.

Dekker, L.W., Ritsema, C.J. and Oostindie, K., 2000. Extent and significance of water repellency in dunes along the Dutch coast. Journal of Hydrology, 231, pp.112-125.

De Vries, W., Klijn, J.A. and Kros, J., 1994. Simulation of the long-term impact of atmospheric deposition on dune ecosystems in the Netherlands. Journal of applied ecology, pp.59-73. Dingman, S.L., 2015. Physical hydrology. Waveland press.

Doerr, S.H., Shakesby, R.A. and Walsh, R., 2000. Soil water repellency: its causes, characteristics and hydro-geomorphological significance. Earth-Science Reviews, 51(1), pp.33-65.

(22)

Doerr, S.H. and Moody, J.A., 2004. Hydrological effects of soil water repellency: on spatial and temporal uncertainties. Hydrological Processes,18(4), pp.829-832.

Doerr, S.H., Shakesby, R.A., Dekker, L.W. and Ritsema, C.J., 2006. Occurrence, prediction and hydrological effects of water repellency amongst major soil and land‐use types in a humid temperate climate. European Journal of Soil Science, 57(5), pp.741-754.

Durkin, C., 2011. Coastal Dune Management at Selected Sites in the Netherlands: South Holland and

Zeeland: Report of the Study Tour of the Netherlands 1-5 March 2010. Liverpool Hope University

Press.

Hallett, P.D., 2007, August. An introduction to soil water repellency. Proceedings of the 8th

International Symposium on Adjuvants for Agrochemicals (Vol. 6), p. 9.

Hem, J.D., 1985. Study and interpretation of the chemical characteristics of natural water (Vol. 2254). Department of the Interior, US Geological Survey.

Horton, R.E., 1941. An approach toward a physical interpretation of infiltration-capacity. Soil Science

Society of America Journal, 5(C), pp.399-417.

IUSS Working Group WRB, 2014. World reference base for soil resources 2014. World Soil Resources Reports No. 106. FAO, Rome.

Jungerius, P.D. and Van der Meulen, F., 1988. Erosion processes in a dune landscape along the Dutch coast. Catena, 15(3), pp.217-228.

Jungerius, P.D. and De Jong, J.H., 1989. Variability of water repellence in the dunes along the Dutch coast. Catena, 16(4-5), pp.491-497.

Kim, D. and Yu, K.B., 2009. A conceptual model of coastal dune ecology synthesizing spatial gradients of vegetation, soil, and geomorphology. Plant Ecology, 202(1), pp.135-148.

King, P.M., 1981. Comparison of methods for measuring severity of water repellence of sandy soils and assessment of some factors that affect its measurement. Soil Research, 19(3), pp.275-285. Koningsveld, M. van, Otten, C.J. and Mulder, J.P.M., 2007. Dunes: the Netherlands soft but secure sea

defences. Global dredging proceedings, pp. 1-16.

Kooijman, A.M., Besse, M., Haak, R., Van Boxtel, J.H., Esselink, H., ten Haaf, C., Nijssen, M., Van Til, M. and Van Turnhout, C., 2005. Effectgerichte maatregelen tegen verzuring en eutrofiëring in open droge duinen. Eindrapport fase, 2.

Leelamanie, D.A.L., Karube, J. and Yoshida, A., 2008. Characterizing water repellency indices: Contact angle and water drop penetration time of hydrophobized sand. Soil Science & Plant

Nutrition, 54(2), pp.179-187.

Leighton‐Boyce, G., Doerr, S.H., Shakesby, R.A. and Walsh, R.P.D., 2007. Quantifying the impact of soil water repellency on overland flow generation and erosion: a new approach using rainfall simulation and wetting agent on in situ soil. Hydrological Processes, 21(17), pp.2337-2345. Martínez, M.L. and Psuty, N.P. eds., 2004. Coastal dunes. Berlin: Springer.

Mao, J., Nierop, K.G.J., Rietkerk, M. and Dekker, S.C., 2015. Predicting soil water repellency using hydrophobic organic compounds and their vegetation origin. Soil, 1(1), p.411.

(23)

Ma'Shum, M., Tate, M.E., Jones, G.P. and Oades, J.M., 1988. Extraction and characterization of water–repellent materials from Australian soils. Journal of Soil Science, 39(1), pp.99-110. McGhie, D.A. and Posner, A.M., 1981. The effect of plant top material on the water repellence of

fired sands and water repellent soils. Crop and Pasture Science, 32(4), pp.609-620. Mortimer, R., G., 2008. Physical Chemsitry. Academic Press.

Müller, K. and Deurer, M., 2011. Review of the remediation strategies for soil water repellency. Agriculture, ecosystems & environment, 144(1), pp.208-221.

Nearing, M.A., Jetten, V., Baffaut, C., Cerdan, O., Couturier, A., Hernandez, M., Le Bissonnais, Y., Nichols, M.H., Nunes, J.P., Renschler, C.S. and Souchère, V., 2005. Modeling response of soil erosion and runoff to changes in precipitation and cover. Catena, 61(2), pp.131-154.

NEN 5754, 1992. Bepaling van het gehalte aan organische stof in grond volgens de

gloeiverliesmethode. Het Nederlandse Normalisatie-instituut, Delft.

Nienhuis, P.H. and Gulati, R.D. eds., 2002. Ecological restoration of aquatic and semi-aquatic

ecosystems in the Netherlands (NW Europe) (Vol. 166). Springer Science & Business Media.

Pruski, F.F. and Nearing, M.A., 2002. Runoff and soil-loss responses to changes in precipitation: A computer simulation study. Journal of soil and Water Conservation, 57(1), pp.7-16.

Shahrudin, R., Van de Craats, A., Grootjans, A.P., Kooijman, A., 2014. Window of opportunity for

Liparis loeselii loeselii populations during coastal dune-slack succession. In: Shahrudin, R. ed. Do we really need management to preserve pioneer stages in wet dune slacks?. Energy, 13, pp.07.

Sifford, M.A., 1996. The Role of Nondetects in Statistical Analysis. Şen, Z., 2014. Practical and applied hydrogeology. Elsevier.

Toy, T.J., Foster, G.R. and Renard, K.G., 2002. Soil erosion: processes, prediction, measurement, and

control. John Wiley & Sons.

Trauth, M.H., Gebbers, R., Marwan, N. and Sillmann, E., 2007. MATLAB recipes for earth

(24)