The influence of the proximity of
exhaust fumes on the ammonia
deposition around dairy farms
Student: Linde Kloks Student number: 12387770
Date: 6-6-2021
Abstract
Human activity has greatly influenced the global nitrogen cycle by producing excessive emissions of reactive nitrogen (Nr). This alteration of the natural cycle is mainly caused by the agricultural and transport sector. The agricultural sector, especially the livestock, is emitting Nr predominantly in the form of ammonia (NH3) into the atmosphere with the
production of manure. The transport sector is emitting nitrogen oxides (NOx) and sulfur
oxides (SOx). The deposition of these emissions has severe consequences for the
environment such as acidification, eutrophication, and biodiversity loss. NOx and SOxcan
also influence the deposition of NH3; when NOxand SOx concentrations in the atmosphere
are higher, the deposition of ammonia is increased due to the formation of aerosols. In addition, the soil becomes more acidic through the deposition of NOx and SOx, which also
results in increased ammonia deposition due to increased adsorption. This project will compare two dairy farms and their relative impact on local ammonia deposition. The main difference between both sites is that one of both sites is situated next to a highway where the other site has no other major N sources nearby. The influence of the proximity of exhaust emissions on the ammonia deposition around these farms was investigated. In this research, statistical analysis with a linear mixed-effects model was used to analyze which variables significantly affected the ammonia deposition. The distance to the highway was added to the farm near the highway to investigate the significance. It was expected that the distance to the highway would affect the ammonia deposition. However, the results did not show an effect. Also, the correlations between NH4/NOx, NH4/SOx, and NOx/SOx were done to
compare the co-deposition rates per farm. It was expected that the farm near the highway would have higher co-deposition rates which was also not the case. To conclude, the hypotheses were not confirmed by the results which could have been because of the external factors influencing the deposition. One important factor could have been the climate variables. Other factors may include manure application on the farms and surrounding farms, differences in housing which includes the manure storage, the feed, and the number of cows.
Keywords: NH3, co-deposition, emission, nitrous oxide, sulphuric oxide, highway,
Table of Contents
Abstract 2 Introduction 4 Methods 8 Study sites 8 Data points 8 Bulk deposition 8 Data clean-up 8 Data calculations 9 Distance to highway 10 Statistical analyses 10 Results 12Spatial patterns of ammonia deposition 12
Relations between NH4-N and NOx-N 13
Relations between NH4-N and SOx 14
Relations between NOx-N and SOx 15
NH4, NOx-N, and SO4 deposition in relation to the distance to the highway 17
Discussion 18
Influences on ammonia deposition 18
Co-deposition 20
The difference in NH4-N, NOx-N, and SOx deposition per farm 20 NH4-N, NOx-N, and SOx deposition in relation to the distance to the highway 21
Conclusion 22
Bibliography 23
Acknowledgments 26
Appendices 27
Appendix S1: Formulas 27
Appendix S2: Molar masses 27
Appendix S3: Distance to the highway from every measuring point 27
Introduction
Human activity has influenced the global nitrogen cycle to such an extent that it is the most disturbed biogeochemical cycle today (Fowler et al., 2013). Before there was human involvement in the cycle, the only possible routes for unreactive atmospheric nitrogen (N2;
which makes up 78% of the atmosphere (Guthrie et al., 2018)) to become available for the environment (as reactive nitrogen; ‘Nr’) was through biological nitrogen fixation (BNF) and the production of NOxby lightning (Fowler et al., 2013).
Nowadays, there are various sources and methods that add Nr to the nitrogen cycle. The disturbances in the cycle are caused by additional fluxes of Nr, which are driven mainly by the agricultural, transport, and industrial sector (Guthrie et al., 2018). For instance, the largest contributor to the release of ammonia (NH3) into the atmosphere is the agricultural
sector; in Europe, the agricultural sector makes up 80% of the total NH3emissions (Velthof
et al., 2012). The NH3emissions from livestock are caused by combining faeces and urine,
which together compose manure (Dai & Karring, 2014). This manure is stored and applied around the dairy farms and thus emits NH3into the atmosphere (Kurvits, 1998; Mosquera,
Monteny & Erisman, 2005). Another way of bringing NH3 into the cycle is by applying
inorganic fertilizer which is produced through industrial fixation via the Haber-Bosch process (Fowler et al., 2013). This addition of fertilizer is necessary to support the growing food demand driven by population growth (Fowler et al., 2013). The Haber-Bosch invention is responsible for feeding 48% of the world population in 2008 (Erisman et al., 2008). Consequently, by applying manure and inorganic fertilizer to the land, more NH3 will be
emitted and eventually deposited (Kurvits, 1998). Furthermore, the transport and industrial sector also emit reactive nitrogen in the form of, predominantly, nitrogen oxides (NOx) which
are released by fossil fuel combustion in engines (Fernando et al., 2001; Fenn et al., 2003). Another emission product of the transport and industry sector is sulfur oxides (SOx)
(Geilenkirchen, ten Broeke & Hoen, 2016). Eventually, these emissions of NH3, NOx,and SOx
can be deposited onto the soil and surface water (Galloway et al., 2008). This Nr can be returned to the atmosphere by the process of denitrification. However, the denitrification process can not denitrify all the deposited Nr because there is an excessive amount of Nr which overrides the natural denitrifying capacity (Fowler et al., 2013). This results in a cumulative Nr deposition which leads to acidification, eutrophication and biodiversity loss (Kurvits, Marta, 1998).
Reactive nitrogen can deposit on the surrounding land through wet and dry deposition. Wet deposition comes down in the form of rain, snow, hail, and/or fog and can be measured by collecting and analyzing precipitation. Wet deposition consists of, among other compounds, ammonium, nitrate, and sulfate (van der Swaluw et al., 2011). Dry deposition occurs when plants and soil directly absorb and/or adsorb nitrogen from the atmosphere (Wichink Kruit et al., 2017). This dry deposition mainly consists of ammonia and nitrogen oxides.
There have been policies in the Netherlands to reduce the Nr emissions, for example via direct manure injection into the agricultural soil, a reduction in fertilizer usage, adequate coverage of manure storage systems, and the use of three-way catalysts in cars (Erisman et al., 2001). The latter method resulted in a decrease in NOx concentrations (Figure 1A,
sulfur-free (Geilenkirchen, ten Broeke & Hoen, 2016). As a result of this policy, SO2
emissions have been reduced (Figure 1B).
Since the 90s there has indeed been a decline in NH3emissions. This decrease is
predominantly the result of injecting manure instead of spreading it over the land as this way the nutrients are used more efficiently and are less prone to volatilization (Erisman et al., 2001). However, between the years 2005-2017, the concentration is no longer decreasing (Figure 1C, Wichink Kruit et al., 2019).
This can be explained because the ammonia concentration is dependent on the NOx
and SOx concentrations. If there is less NOx and SOx in the atmosphere, there are fewer
aerosols formed which leads to a higher concentration of NH3in the atmosphere (Wichink
Kruit et al., 2019). Another form of interaction between NH3 and NOx and SOx is
co-deposition. This process is dependent on the concentrations of sulfur and nitrogen in the atmosphere. When there is an increase in NOx and SOx there will be more sulphuric and
nitric acid formed which results in more acidic soils and vegetation. Consequently, this increase in acidity will lead to more dry deposition of ammonia which leads to a lower concentration in the atmosphere (Wichink Kruit et al., 2017).
A
B
C
Figure 1a. Emission and concentration of nitrogen oxides (NOx) in the Netherlands. b.
Emission and concentration of sulfur dioxide (SO2) in the Netherlands. c. Ammonia (NH3)
concentration and emission in the Netherlands. Copied from
Local increases in reduced nitrogen (NHx) and oxidized nitrogen (NOx)
concentrations can affect human health as well as the natural environment (Guthrie et al., 2018). Soil acidification is caused by the accumulation of deposited Nr which then impacts biodiversity because it changes the composition of plants and soil (Wichink Kruit et al., 2017), ultimately leading to the loss of rare species and habitats (Guthrie et al., 2018). Biodiversity is providing ecosystem services that are valuable for human and societal benefits because they provide food, energy, clean water, and air. Through Nr deposition, biodiversity loss occurs which in turn impacts the ecosystem services which are crucial to the world (Guthrie et al., 2018).
Another effect of Nr deposition is the leaching of Nr to the surface water. This occurs when the soil is N-saturated. This leads to eutrophication. This eutrophication again leads to biodiversity loss (Kirchner et al., 2005).
To reduce these negative environmental effects, both the NH3 emissions from
agricultural practices and the NOx and SOx from the transport sector need to be reduced.
Moreover, the NOx and SOx exhaust emissions from the transport are influencing the
deposition of ammonia (Wichink Kruit et al., 2019). Therefore, it is important to study the interplay between these emissions. In this research, the local Nr deposition patterns of two dairy farms will be compared. One farm has a highway in the proximity where the other farm does not. The exhaust emissions (NOx and SOx) concentrations are especially present near
roads and highways (Kirchner et al., 2005). Therefore, mapping the NH3, NOx, and SOx
deposition around dairy farms gives insight into how the proximity of the road will influence the spread of NH3, NOx, and SOx deposition. Whether this proximity of the highway is of
influence will be researched in this paper by answering the question: ‘What is the effect of the proximity of exhaust emissions on the ammonia deposition around dairy farms?’.
The probable difference in concentration of NH3, NOx, and SOx between the two
farms will raise questions of where the concentrations came from. Determining whether these emissions came from the nearby highway, through dry deposition, or deposited from higher atmospheric layers, through wet deposition, will be important to policymakers.
To study the potential effects of the proximity of the highway, the ammonia deposition of two dairy farms will be investigated. Also, the difference in co-deposition will be analyzed.
Methods
Study sites
There are two sites where experimental data has been collected. Both sites are comparable (in terms of size and production) dairy farms surrounded by low vegetation (predominantly production grassland). The main difference between both sites is that one of both sites is situated next to a highway (400m west of the stable, hereafter called ‘FNH’ for ‘Farm Near Highway’) where the other site has no other major N sources in close proximity (< 1km, hereafter called ‘FWH’ for ‘Farm Without Highway’).
Both farms house around 200 cows that are kept indoors throughout the year. The manure is stored directly underneath the stable. Because of this, the source of ammonia emissions is mostly localized to the same building, except for fertilization practices and relatively low outside storage of straw-containing manure. Weather conditions are recorded every 5 minutes with a HOBO RX3000 Remote Monitoring Station (a meteo mast) and consist of wind direction, wind speed, rain (mm), temperature (ºC), and other variables. Another factor that influences the data is the use of manure and fertilizer. Both farms apply manure (from the stable) three or four times a year.
Data points
Bulk deposition
At both farms, deposition measurements are performed in four wind directions (NE, SE, NW, SW). At each wind direction, data at six set distances are collected. The bulk deposition dataset consists mainly of wet deposition that rains out and local dry deposition. These data points were measured using an HDPE funnel (Ø = 16cm) and a 2L HDPE bottle underneath it to catch the water and the dry deposition particles. The measurement distances from the stable are 15.6m, 31.25m, 62.5m, 125m, 250m, and 500m and are allocated in the four wind directions. The deposition of NO2, NO2+NO3, NH4, PO4, SO4, and DON are continuously
collected every two weeks from April and May onwards for the farm with and without a highway respectively. For this research, the concentrations of NO2, NO2+NO3, NH4,and SO4
are relevant because they are partly indicative of the amount of NOxand SOx from exhaust
emissions. The NOx and SOx could also have originated from higher atmospheric layers and
thus not directly from exhaust emissions. The NH4 concentrations indicate partly the
emissions from the stable. Here, the NH4 could also be deposited from higher atmospheric
layers and therefore not from the stable. Ammonium is the ionized form of ammonia (NH3)
which can be deposited. Therefore, from here on the term NH4will be used.
Data clean-up
The datasets were acquired from the laboratories and researchers of the Institute for Biodiversity and Ecosystem Dynamics (IBED), University of Amsterdam. The two files consist of measurements of the electric conductivity, the pH, and the concentrations of NO3+NO2, NO2, NH4, PO4, SO4,and DON. These measurements are done every 2 weeks. To
conduct statistical analyses, the data files were first corrected for faulty measurements and potentially polluted samples.
Firstly, many values were below measuring capacity which is indicated by ‘<’ and that is the lowest value that can be measured. For example, the lowest value of SO4 is 15 and is
indicated by ‘<15’. When encountering a value with a ‘<’, the value needs to be substituted with 0 because it was too low to measure.
Secondly, there are blank measurements which will ensure the measurements are valid. It investigates the Thymol substance that is present in the bottles before the 2 weeks begin. To ensure that the Thymol is not reacting with any particles, a closed bottle with Thymol is placed next to the measuring point. If the concentrations are below the detection limit, the used Thymol is valid because there are no concentrations found. However, if the measurements of the Thymol consist of concentrations higher than the detection limit, the Thymol inside the bottle somehow has reacted with the atmosphere and the concentration needs to be subtracted from the concentration in the real measurement.
Thirdly, notes were collected in the field about observations during sample collection. A frequent note is the detection of bird droppings. When this occurs, the concentrations of NH4,
PO4, SO4, DON are polluted and can not be used for the research. These values are
changed to Not Available (NA). Certain values deviated strongly and were located outside of the margin of 3 times the standard deviation. If there was a reason to suspect pollution, this value was substituted with NA.
Data calculations
The concentration of the data samples is measured in μmol/L. However, every water sample has a different weight which leads to incomparable values. Therefore, the μmol/L needs to be converted to grams. This is done with several formulas which can be found in the appendix (S1).
The first step is to calculate the water weight from grams to L by dividing it by 1000. Thereafter, the formula for molarity (S1) was used to calculate the moles. Then, the μmol was converted to mol. The NO3 moles need to be calculated to convert them to grams.
Therefore, the moles of the NO2was subtracted from the moles of NO3+NO2which gives the
amount of NO3.
To get to grams, the molar masses from NO3+NO2, NO2, NO3, NH4, PO4, SO4, DON
were obtained which are shown in table S2. Another factor that was included in the calculation is the surface area of the funnel. The surface area is determined by the radius (r=0.08m) of the funnel with the formula (S1). This surface area fits a number of times in 1 m2and then is multiplied by the amount of nitrogen [g] and gives the nitrogen in [g/m2]. The
steps taken have converted the grams to grams/m2/2weeks because there are measurements every 2 weeks. The last step was to obtain the actual weight from the individual nitrogen (N) molecules (S1). This will indicate the weight of the individual N atoms which is a more valid way of comparing the nitrogen in the molecules.
Distance to highway
The points of measurements are visualized in Google Earth (Fig. 2). To indicate the distances from the measuring points to the highway (on the left), the google earth ruler was used. From every measuring point, a straight horizontal line was drawn to the edge of the highway. These distances are shown in table S2. These values are transferred to the excel file in a new column with the name ‘Distance_to_highway’.
Figure 2. The measuring points of the farm with a highway present. (Google Earth)
Statistical analyses
Firstly, the subquestion: ‘Which variables are causing how the ammonia deposits around the farm without the highway?’ was analyzed. The dependent variable is the ammonia deposition and the independent variables are the distance from the stable to the measuring points, the wind direction, and the day of the research. To analyze this, linear mixed effect models were utilized (function ‘lme’, R-package ‘nlme). The potential effects of the distance to the stable (in m), wind direction (NE, NW, SW, and SE), and the day of the year that was measured plus all potential interactions on ammonia deposition were investigated. To account for the repeated measure design, a random variable with the bulk sampler location ID was included.
To visualize potential effects, scatter plots of the NH4deposition around the two farms
were made using Excel (Microsoft 365).
Secondly, the subquestion: ‘Which variables are causing how the ammonia deposits around the farm with the highway?’ was analyzed. Again, linear mixed effect models are utilized (function ‘lme’, R-package ‘nlme). However, this time the independent variable, distance to the highway (in m), is added to see the potential effects.
Thirdly, an analysis of the two patterns was done. This is to show the possible influence of one form of deposition on the other. Therefore, potential correlations between NH4-N/NOx-N, NH4-N/SOx, and SOx/NOx-N were investigated using a two-sided Spearman.
Thereafter, the difference between the NH4-N, NOx-N, and SOx between the two
farms was investigated with a boxplot and the median. A Wilcoxon test was used to see if the medians significantly differ.
Fourthly and lastly, the sub-question: ‘How is the ammonia deposition affected by the distance from the highway?’ was analyzed. The distance from the highway to the measuring points was analyzed to see if there is a pattern between the distance to the highway and the ammonia deposition. To analyze this, linear mixed effect models are utilized (function ‘lme’, R-package ‘nlme). The potential effects of the distance to the stable (in m), wind direction (NE, NW, SW, and SE), the day of the year that was measured, and the distance to the highway (in m) plus all potential interactions on NH4-N, NOx-N and SOx depositions were
investigated.
To test for the normality of the model, histograms and QQ-plots were used. The data for the ammonia deposition was square-root transformed. Whether there is a normal distribution is of importance to the method of the correlation test. The Pearson method is used if there is a normal distribution and the Spearman method is used when there is not a normal distribution. The statistical analyses are conducted in R (version 1.1.463).
Results
Spatial patterns of ammonia deposition
The deposition of NH4-N is visualized relative to the distance to the stable at which it was
measured (Fig 3a/b). There seems to be a small decrease in NH4-N deposition with
increasing distance at both farms. However, when reducing non-significant potential interactions between the variables, increasing distance to the farm showed a significant decrease in the ammonia deposition (numDF=1, denDF=18, F=3.5593, p=0.0004). Time of the year showed to significantly increase the ammonia deposition at FWH (numDF=1, denDF=473, F=5.46287, p=0.0198).
At the FNH (Fig. 3b), the measurements are more equally divided over the distances. This could mean that the distance to the stable is not an explanatory variable for the FNH. Consequently, there were no variables that significantly increased the ammonia deposition at FNH. Then, the distance to the highway was added to see if this makes a difference in the significance. There are still no variables that have a significant effect on ammonia deposition. The distance to the stable is almost significant (numDF=1, denDF=480, F=12.8205, p=0.0754) for the ammonia deposition.
A
B
Figure 3. a. The NH4-N deposition (all data) of FWH plotted against the distance to the
stable from 13-05-2020 until 17-3-2021 and b. The NH4-N deposition (all data) of FNH
Relations between NH
4-N and NO
x-N
There is a strong correlation at FWH (r=0.70, p=2.2e-16, Fig 4a). There is also a strong correlation at FNH (r=0.80, p=2.2e-16, Fig 4b). The correlation between NH4-N and NOx-N
is stronger at FNH. The relationship between NH4-N and NOx-N is dependent on the amount
of deposition. Therefore, the medians between the locations were compared because they can be an explanatory factor of the given relationship. The median NH4-N of location FNH is
23.51 mg/m2/2weeks and is significantly lower (w=124120, p=0.0007) than for the location
FWH, which had a median NH4-N of 25.83 mg/m2/2weeks during the measured period (Fig.
7a). The NOx-N median of location FNH is 11.9 mg/m2/2weeks and significantly lower
(w=111640, p=3.989e-09) than for FWH, which had a median NOx-N of 13.32
mg/m2/2weeks (Fig. 7b).
A
B
Figure 4. Correlation between the NH4-N and NOx-N deposition (all data) a. FWH and b.
Relations between NH
4-N and SO
xHere, the correlation between NH4-N and SOxis investigated. There is a correlation between
NH4-N and SOx at FWH (r=0.39, p=2.2e-16, Fig 5a). Likewise, the FNH location is
investigated. There is a weak correlation (r=0.17, p=7.874e-5, Fig. 5b). There is a stronger correlation at FWH between NH4-N and SOx.
Again, an explanatory factor for the difference is the amount of deposition. The NH4-N is mentioned above. The SO4median of FNH is 0 mg/m2/2weeks and is significantly
lower than FWH (w=95821, p=2.2e-16), which had a median of 53.84 mg/m2/2weeks (Fig.
7c).
The SO4data of FNH contains many measurements of 0 mg/m2/2weeks which results
in a median value of 0 mg/m2/2weeks. Another value that can be addressed is the mean.
The SO4mean of FNH is 27.59 mg/m2/2weeks and is significantly lower than the SO4mean
of FWH (w=95821, p=2.2e-16), which is 65.45 mg/m2/2weeks.
A
B
Figure 5. Correlation between NH4-N and SO4deposition (all data) a. FWH and b. FNH. The
Relations between NO
x-N and SO
xNow, the correlation between NOx-N and SOxat the different farms is looked at.
First, the FWH location will be investigated. In the figure below a scatterplot is drawn of the SOxvalues against the NOx-N values. Notably, many SO4values have the value 0.
There is a significant correlation at FWH between NOx-N and SOx(r=0.3588, p=2.2e-16, Fig
6a).
Second, the FNH location will be investigated. There is a weak correlation in FNH (r=0.1333, p=0.0024, Fig 6b). The correlation is stronger at FWH.
A
A
B
C
Figure 7. a. The difference in NH4-N, b. The difference in NOx-N and, c. The difference in
SO4deposition between the FNH and FWH (all data). The data is from 15-04-20 until
NH
4, NO
x-N, and SO
4deposition in relation to the distance to the
highway
The distance to the highway does not significantly increase the NH4-N, NOx-N, or SO4
deposition. Even when eliminating the potential interactions between the variables, there still is no significant effect visible of the distance to the highway (p > 0.05 for all comparisons) (Fig. 8).
A
B
C
Discussion
Influences on ammonia deposition
This research aims to examine the NH4-N, NOx-N, and SOx deposition around dairy farms
and how the proximity of the road influences the spread of this deposition.
The day of the research has shown to be a variable to significantly affect the ammonia deposition. This can be explained because the day of the research involves other variables. For example, wind speed, the amount of rainfall, and the temperature are all hidden in the day. All these climate variables are known to affect both ammonia emission and deposition (Sutton et al., 2013). Therefore, it was likely that the day of the research would indeed predict the measured ammonia deposition.
Unexpectedly, the wind direction did not have a significant impact on the ammonia deposition at both locations, which was not according to predictions. Namely, wind direction normally strongly influences the spread of the ammonia deposition (Sommer et al., 2009). The wind direction changes frequently and different variables influence the deposition. Moreover, the gas measurements at FNH did show a clear effect of the wind direction on the deposition around the farm (Barmentlo et al., personal communication). This indicates that the bulk measurements are not suitable for measuring dry deposition. The, mainly wet, deposition is strongly dependent on the intensity and frequency of the rain.
Research by Sommer et al. (2009) concludes that the distance to the farm and the wind direction are related to the NH4-N deposition. Yet, the distance to the stable, day of the
research, and distance to the highway all did not have a significant effect on the ammonia deposition at FNH. Again, the wind direction is of influence here. The significance of the variables differs between the two farms. This difference between the two farms was unexpected because the farms were compared because of their similarity in size, production, and low vegetation. Differences in climate variables (wind speed, temperature, amount of rainfall) could have affected this difference in significance. High wind speeds can result in rapid dispersion and dilution of the ammonia resulting in low NH3concentrations and low dry
deposition in downwind areas (Shen et al., 2013).
For follow-up research, these climate variables need to be looked at more closely. Conveniently, the climate variables have been collected as long as the research runs but they were not incorporated in this research. Follow-up research can implement periods with similar conditions (wind direction, rainfall, temperature, and wind speed) which are then compared. Also, wet-only data is available which can be used to quantify the dry deposition.
Additionally, at FWH, the distance to the stable showed to significantly increase the ammonia deposition. Other researchers conclude that only a part of the NH3 emissions
deposit in the vicinity of the source (Shen et al., 2013). For example, in one research 10.4% of the total NH3emissions were deposited within 500m (Walker et al., 2008), and in a similar
research a value of 16% was found (Hao et al., 2006). For this reason, the distance effect can be explained. Unexpectedly, at FNH, the distance to the stable did not predict the ammonia deposition. This means that there is no distance effect. This could be the result of differences in emission between the farms which will be discussed below.
The fertilization rate could differ per farm. It is known that the farmers of both dairy farms inject the slurry manure into the soil as is standard per law (Bouma, 2016). However,
the manure application methods from neighboring farms are not established. The measurement setup of FWH is almost exclusively surrounded by grasslands. In contrast to the FNH setup which is also surrounded by bulb, potato, and sugar beet patches. These different crops have different fertilization regimes, for example, differences in the amount of fertilizer. This is related to the amount of slurry and fertilizer that is applied. The surrounding patches may receive higher N concentrations for optimal growth and therefore more fertilizer could be added which leads to more emissions (Velthof & Mosquera, 2011). There could also be a difference between injecting manure, which is mostly applied on grasslands and spreading manure, which may apply to the crop patches. Emission rates differ between these application techniques (Lagerwerf et al., 2019). As said earlier, the injecting of the manure leads to less volatilization and therefore fewer emissions (Erisman et al., 2001).
Since there is no data of the exact measurements taken on the surrounding farms, the exact effect may be unclear. Nonetheless, all these differences in emissions surrounding the farms could have led to the differences in deposition on the farms.
The volatilization rate of NH3 from the stable is another factor that could have
influenced the deposition measurements. This volatilization rate is influenced by the nitrogen content, the application rate, and the conditions. When there is an increase in any of these factors, the volatilization rate will increase (Ross et al, 2002).
There could be a difference in feed for the cows which can influence the N content of the manure. According to Frank et al. (2002), the protein in the feed is related to the nitrogen content in the manure. The degradation of the protein causes higher excretion of nitrogen. Therefore, when there is more protein in the feed, the nitrogen content is higher. The FWH could have fed the cows with a higher protein level feed which would cause a higher nitrogen content in the manure and therefore could lead to a higher NH3deposition around the farm.
The housing structure is an important factor to look at when discussing NH3
emissions. Namely, the NEMA model states that in 2012, 50% of the NH3 emissions
originate from housing (Velthof et al., 2012). The FNH had several methods to separate the manure from the urine. For example, a manure vacuum that actively cleans the floor and a slider that goes across the floor to remove the manure. Additionally, there is a low-emission floor present which has a low emission factor. Yet, at the FWH, only a slider is used to clean the floor. This will lead to a difference in emission from the stable. Furthermore, my fellow student Pieter de Beaufort established the difference in emission factor per housing. The emission factor of the FNH is 0.26 and that of the FWH is 0.69. Here, the difference that the low emission floor has on the emission factor becomes apparent. This has resulted in fewer NH3emissions from the stable at the FNH (Beaufort, personal communication), which could
have led to less deposition.
As a result of housing, storage, and manure application differences between the farms, different NH3 emissions may have caused a different NH4 deposition pattern than
Co-deposition
The phenomenon of co-deposition is the interplay between NH3, NOx,and SOx. When there
is an increase in NOx and SOx there will be more sulphuric and nitric acid formed which
results in more acidic soils and vegetation. Consequently, this increase in acidity will lead to more dry deposition of ammonia. That there is a correlation between NH4/NOx-N, NH4/SOx
and NOx-N/SOx is not surprising because there is always an effect of the co-deposition.
There are almost always concentrations of NH3, NOx, and SOx in the atmosphere which can
deposit and therefore lead to co-deposition. However, the difference between the farms is interesting because it indicates where the co-deposition is stronger. The prediction was that there would be a higher correlation at the FNH because of the local exhaust fumes. Indeed, the NH4/NOx-N correlation was stronger at FNH. However, it was surprising to find that the
NH4/SOxand NOx-N/SOxcorrelation were less strong at the FNH.
The NH4/SOx correlation was stronger at FWH. Moreover, there are no roads in the
proximity that could influence the SOxconcentrations. Possible theories could involve other
sources of SOx emissions at FWH that caused a higher concentration of SOx in the
atmosphere. For example, the SOxemissions from tractors could have influenced the local
deposition data (Lovarelli & Bacenetti, 2019). Or other local emissions of SOx.
A link can be made between this higher correlation of NH4-N/SOxand the difference
in SOxper farm. Specifically, the FWH showed higher SOxdeposition which could have led to
the higher co-deposition and therefore to a higher correlation. That the FWH had more SOx
was however an unexpected result. Factors influencing this high SOxdeposition are the wet
deposition of SOxor other SOxinputs. The lower SOxdeposition at FNH can be explained by
the high volatilization rates of SOx. Again, the climate factors such as wind speed and
amount of rainfall could also be an explanation (Sutton et al., 2013). Furthermore, the SOx
output from exhaust fumes has been reducing since 1990 (Fig 1b). This research indicates that the reduced output of SOx is no longer problematic for the local ammonia deposition.
A different factor that could have influenced the overall results is the measuring methods used in this research. The method of capturing the bulk deposition (wet+dry) with a plastic funnel could have affected the results. The co-deposition of NH4-N with NOx-N and
SOxis enhanced by the acidity of surface area (Wichink Kruit et al., 2017). The plastic funnel
may not have represented the acidity the same as the soil. Therefore, the ammonia deposition could be an underestimation.
The difference in NH
4-N, NO
x-N, and SO
xdeposition per farm
The expectation was that there would be higher NH4-N deposition at FNH. However,
the results of this study show a different outcome. This can partially be linked to the difference of NOx-N and SOxemissions located closely to the farms. In this study, it is shown
that there is less NOx-N and SOxdeposition at FNH (Fig 7 b&c), which could lead to less
NH4-N deposition through the lower co-deposition. This can be a reason for the difference in
NH4-N deposition between the farms. The FWH has a somewhat higher NH4-N deposition
which is an unexpected result because the FNH was expected to have higher values of NOx-N and SOxbecause of the proximity of the highway.
The overall NOx-N and SOxdepositions were higher at FWH which was unexpected.
Especially because the SOxwas thought to come from exhaust fumes which are not present
in the local setting of FWH. Therefore, it can be argued that there are NOx-N and SOx
possible reason for this could be the NOx-N and SOxfrom higher atmospheric layers that are
deposited through wet deposition.
The data that was used in this research is mainly from 2020 which had several lockdowns due to Covid-19. In these periods there was a decline in traffic which could have affected the NOx-N and SOxemissions from the highway and therefore the deposition at the
FNH (Lovarelli et al., 2020). Therefore, the data from 2020 is not a good representation of the regular deposition.
NH
4-N, NO
x-N, and SO
xdeposition in relation to the distance to the
highway
The expectation was that there would be more NOx-N and SOxpresent closer to the highway.
Also, the NH4-N deposition could be higher closer to the highway because of the
co-deposition. However, the results did not back up this hypothesis. Shorter distances to the highway did not significantly increase NH4-N, NOx-N, and SOxdeposition. In the research of
Kirchner et al. (2005), the distance to the highway did affect the NH3and NO2concentrations
in the air. Namely, the concentrations near the highway were 300% compared to 500m from the highway. However, the deposition of these concentrations can occur far away from the sources (Kirchner et al., 2005). After the emission of the exhaust fumes, they are dispersed into the atmosphere where they react with other particles (Cape et al., 2004). Here, variables such as wind speed and direction influence the dispersion and the range of the particles. Accordingly, there could have been high emissions of exhaust fumes but the deposition occurred somewhere else (Cape et al., 2004).
Further research could look into the variables affecting highway depositions. It is concluded that climate factors affect the spread of deposition (Sutton et al., 2013). Therefore, looking at the wind direction around the highway could give insight into where the emissions go and if there is a correlation between the wind direction and the NOx-N and SOx
deposition. Again, when using the same data, the influence of Covid-19 could be present. There could be less NOx-N and SOxdeposited at FNH because of the several lockdowns
Conclusion
This research has investigated the effects of the proximity of a highway on the ammonia deposition around dairy farms. The exhaust fumes emissions from the highway were expected to influence this NH4-N deposition because of co-deposition. However, the results
show that the distance to the highway was not significantly increasing the ammonia deposition at the farm with the highway. A possible explanation for this unexpected result is that the wind direction was not implemented in the analysis. This should be implemented in follow-up research. Another reason could be that the measuring method did not sufficiently measure the dry deposition. Other climatic variables such as the amount of rain or wind speed also may have influenced the ammonia deposition.
The distance to the highway also did not significantly increase the NOx-N and SOx
depositions. However, a stronger correlation between NH4-N/NOx-N at the FNH was found,
which can be indicative of co-deposition. Yet, the short distances from the highway may not be important to this co-deposition. It is more likely that this co-deposition is something that happens at the site. In addition, the correlations between NH4-N/SOx and NOx-N/SOxwere
less strong at the farm located near the highway which was not expected. Other inputs of SOx could have affected these results. Also, the strongly reduced SOx is no longer
problematic for the local deposition of ammonia. In addition, wet deposition may contain NOx
and SOxfrom higher atmospheric layers.
Another outcome relates to the fewer NOx-N and SOxdepositions at the FNH. An
explanation could be the influence of climatic variables, for example, that the high wind speeds may have caused dispersion to higher atmospheric layers.
The distance to the stable and the day of the research significantly increased the ammonia deposition at the FWH. The wind direction did not show to significantly increase the ammonia deposition. This could be a result of the variance of this variable. The wind direction changes frequently and other variables are of influence. At the FNH, the variables, distance to the stable, wind direction, and day of the research did not show a significant effect on the ammonia deposition. In addition, the gas measurements did show a clear effect of the wind direction. This concludes that again, the bulk measurements are not suitable for measuring dry deposition.
The difference in distance effect to the highway can be explained by the climate variables. Similarly, the ammonia concentrations close to the highway may be negligible which could mean that the effect of the highway may not be that interesting to look at when explaining ammonia deposition. After all, the climate factors may have had the greatest impact on the spreading patterns. Furthermore, it can be concluded that the impact of co-deposition is low at the FNH. Possible reasons for these unexpected results can be led back to the climate variables that have affected the deposition. Especially, the weather conditions, such as wind direction, wind speed, and the amount of rain are important to the deposition. In addition, the differences in housing, manure application, and manure storage could also have influenced the emissions from the stable. These factors are influencing the volatilization rate of nitrogen which could have affected the results. To further investigate the influence of the proximity of the highway, the climate variables need to be taken into account more profoundly. Moreover, the climate variables are crucial for the deposition patterns. Therefore, looking at the wind direction and the wind speed around the highway could give insights into their effect on the ammonia deposition around the farm.
Bibliography
Bouma, J. (2016). The importance of validated ecological indicators for manure regulations in the Netherlands. Ecological indicators, 66, 301-305.
CLO indicator 0081, 2018.
https://www.clo.nl/indicatoren/nl0081-relatie-ontwikkelingen-emissies-en-luchtkwaliteit Cape, J. N., Tang, Y. S., Van Dijk, N., Love, L., Sutton, M. A., & Palmer, S. C. F. (2004). Concentrations of ammonia and nitrogen dioxide at roadside verges, and their contribution to nitrogen deposition. Environmental Pollution, 132(3), 469-478.
Dai, X., & Karring, H. (2014). A determination and comparison of urease activity in feces and fresh manure from pig and cattle in relation to ammonia production and pH changes. PLoS
One, 9(11), e110402.
Erisman, J. W., de Vries, W., Kros, H., Oenema, O., van der Eerden, L., van Zeijts, H., & Smeulders, S. (2001). An outlook for a national integrated nitrogen policy. Environmental
Science & Policy, 4(2-3), 87-95.
Erisman, J. W., Sutton, M. A., Galloway, J., Klimont, Z., & Winiwarter, W. (2008). How a century of ammonia synthesis changed the world. Nature Geoscience, 1(10), 636-639. Fenn, M. E., Haeuber, R., Tonnesen, G. S., Baron, J. S., Grossman-Clarke, S., Hope, D., ... & Sickman, J. O. (2003). Nitrogen emissions, deposition, and monitoring in the western United States. BioScience, 53(4), 391-403.
Fernando, H. J. S., Lee, S. M., Anderson, J., Princevac, M., Pardyjak, E., &
Grossman-Clarke, S. (2001). Urban fluid mechanics: air circulation and contaminant dispersion in cities. Environmental Fluid Mechanics, 1(1), 107-164.
Fowler, D., Coyle, M., Skiba, U., Sutton, M. A., Cape, J. N., Reis, S., ... & Voss, M. (2013). The global nitrogen cycle in the twenty-first century. Philosophical Transactions of the Royal
Society B: Biological Sciences, 368(1621), 20130164.
Frank, B., Persson, M., & Gustafsson, G. (2002). Feeding dairy cows for decreased ammonia emission. Livestock Production Science, 76(1-2), 171-179.
Galloway, J. N., Townsend, A. R., Erisman, J. W., Bekunda, M., Cai, Z., Freney, J. R., ... & Sutton, M. A. (2008). Transformation of the nitrogen cycle: recent trends, questions, and potential solutions. Science, 320(5878), 889-892.
Geilenkirchen, G., ten Broeke, H., & Hoen, A. (2016). Verkeer en vervoer in de Nationale Energieverkenning 2015. PBL-Planbureau voor de Leefomgeving, Den Haag.
Hao, X., Chang, C., Janzen, H. H., Clayton, G. & Hill, B. R. Sorption of atmospheric ammonia by soil and perennial grass downwind from two large cattle feedlots. J Environ Qual 35, 1960–1965 (2006).
Kirchner, M., Jakobi, G., Feicht, E., Bernhardt, M., & Fischer, A. (2005). Elevated NH3 and NO2 air concentrations and nitrogen deposition rates in the vicinity of a highway in Southern Bavaria. Atmospheric Environment, 39(25), 4531-4542.
Kurvits, T., & Marta, T. (1998). Agricultural NH3 and NOx emissions in Canada.
Environmental Pollution, 102(1), 187-194.
Lagerwerf, L. A., Bannink, A., Van Bruggen, C., Groenestein, C. M., Huijsmans, J. F. M., Van der Kolk, J. W. H., ... & Vonk, J. (2019). Methodology for estimating emissions from
agriculture in the Netherlands: Calculations of CH4, NH3, N2O, NOx, NMVOC, PM10, PM2. 5 and CO2 with the National Emission Model for Agriculture (NEMA), Update 2019 (No.
148). Statutory Research Tasks Unit for Nature & the Environment.
Lovarelli, D., & Bacenetti, J. (2019). Exhaust gases emissions from agricultural tractors: State of the art and future perspectives for machinery operators. Biosystems Engineering,
186, 204-213.
Lovarelli, D., Conti, C., Finzi, A., Bacenetti, J., & Guarino, M. (2020). Describing the trend of ammonia, particulate matter and nitrogen oxides: The role of livestock activities in northern Italy during Covid-19 quarantine. Environmental research, 191, 110048.
Mosquera, J., Monteny, G. J., & Erisman, J. W. (2005). Overview and assessment of techniques to measure ammonia emissions from animal houses: the case of the Netherlands. Environmental Pollution, 135(3), 381-388.
Sommer, S. G., Østergård, H. S., Løfstrøm, P., Andersen, H. V., & Jensen, L. S. (2009). Validation of model calculation of ammonia deposition in the neighbourhood of a poultry farm using measured NH3 concentrations and N deposition. Atmospheric Environment, 43(4), 915-920.
van der Swaluw, E., Asman, W. A., van Jaarsveld, H., & Hoogerbrugge, R. (2011). Wet deposition of ammonium, nitrate and sulfate in the Netherlands over the period 1992–2008.
Atmospheric Environment, 45(23), 3819-3826.
Reijs, J. W. (2007). Improving slurry by diet adjustments: a novelty to reduce N losses from
grassland based dairy farms.
Ross, C. A., Scholefield, D., & Jarvis, S. C. (2002). A model of ammonia volatilisation from a dairy farm: an examination of abatement strategies. Nutrient Cycling in Agroecosystems,
64(3), 273-281.
Shen, J., Li, Y., Liu, X., Luo, X., Tang, H., Zhang, Y., & Wu, J. (2013). Atmospheric dry and wet nitrogen deposition on three contrasting land use types of an agricultural catchment in subtropical central China. Atmospheric Environment, 67, 415-424.
Sutton, M. A., Reis, S., Riddick, S. N., Dragosits, U., Nemitz, E., Theobald, M. R., ... & De Vries, W. (2013). Towards a climate-dependent paradigm of ammonia emission and deposition. Philosophical Transactions of the Royal Society B: Biological Sciences,
368(1621), 20130166.
Velthof, G. L., & Mosquera, J. (2011). The impact of slurry application technique on nitrous oxide emission from agricultural soils. Agriculture, Ecosystems & Environment, 140(1-2), 298-308.
Velthof, G. L., Van Bruggen, C., Groenestein, C. M., De Haan, B. J., Hoogeveen, M. W., & Huijsmans, J. F. M. (2012). A model for inventory of ammonia emissions from agriculture in the Netherlands. Atmospheric environment, 46, 248-255.
Walker, J., Spence, P., Kimbrough, S. & Robarge, W. Inferential model estimates of ammonia dry deposition in the vicinity of a swine production facility. Atmos Environ 42, 3407–3418 (2008).
Wichink Kruit, R. J., Hoogerbrugge, R., Sauter, F. J., de Vries, W. J., & van Pul, W. A. J. (2019). Ontwikkelingen in emissies en concentraties van ammoniak in Nederland tussen 2005 en 2016.
Wichink Kruit, R.J., J. Aben, W. de Vries, F. Sauter, E. van der Swaluw, M.C. van Zanten, W.A.J. van Pul, 2017. Modelling trends in ammonia in the Netherlands over the period 1990-2014. Atmospheric Environment 154, 20-30.
Acknowledgments
I would like to thank my supervisor Dr. S. H. Barmentlo for his guidance during this project. His expertise was valuable to understand the complex nature of the subject. The extensive feedback that I received made me work harder trying to improve my work. I would also like to thank my supervisor A. G. Uilhoorn for her support during the project. Her expertise in statistical analysis was very helpful to me. I would also like to thank my father, Dre Kloks, and my mother, Wilma Wijnhoven, for supporting me during the project. They helped me by listening and asking critical questions about the subject.
Appendices
Appendix S1: Formulas
(1) M [mol/L] = mol [mol] / Volume [L]
(2) mass [g] = moles [mol] / molar mass [g/mol] (3) A = π * r² → π * 0.08^2 = 0.0201
(4) 1 m2/ surface area → 1 / 0.0201 = 49.7359.
(5) The molar mass from N (= 14.0067) is divided by the molar mass of the substance for example, NO2(=46.0055). 14.0067/46.0055 is the part of the individual N atom.
This number is multiplied by the g/m2/2weeks.
Appendix S2: Molar masses
Molar mass [g/mol] NO3+NO2 108.01 NO2 46.0055 NO3 62.0049 NH4 18.0386 PO4 94.9714 SO4 96.0626 DON (C15H20O6) 296.316
Appendix S3: Distance to the highway from every measuring point
Location Distance NO1 417.33 m NO2 435.98 m NO3 455.3 m NO4 487.13 m NO5 767.35 m
ZO2 438.34 m ZO3 469.51 m ZO4 503.13 m ZO5 577.62 m ZO6 668.17 m ZW1 329.46 m ZW2 309.91 m ZW3 261.95 m ZW4 234.79 m ZW5 165.79 m ZW6 101.72 m ZW6(2) 133.89 m NW1 391.96 m NW2 372.61 m NW3 343.08 m NW4 374.56 m NW5 278.01 m NW6 101.3 m
Appendix S4: R-script
# Final Script - Linde Kloks # Set working directory
setwd("~/FPS/scriptie/datasets") # Open data files
# jacobi - without highway
jacobi <- read.csv("jacobi_Final.csv", header = TRUE, sep = ";") jacobi
jacobiclean <- na.omit(jacobi) # almere - with highway
almere
almereclean <- na.omit(almere) # almere and jacobi merged
almerejacobi <- read.csv("jacobialmeresamengevoegd.csv", header = TRUE, sep = ";") almerejacobi
almerejacobiclean <- na.omit(almerejacobi)
#non lineair of lineair #random factor toevoegen
#lme mixed effect (+ random variabele) install.packages("nlme")
library(nlme) help(nlme) # test jacobi NH4
lmejacNH4 <- lme(NH4N_mg ~ Distance_to_stable * Direction * Day, data = jacobiclean, random = ~1| Sampler) #uitvoeren lme schoon dataframe
lmejacNH4
summary(lmejacNH4)
hist(resid(lmejacNH4)) #right skewed histogram qqnorm(resid(lmejacNH4))
qqline(residuals(lmejacNH4)) #not normally distributed anova.lme(lmejacNH4, type="marginal", adjustSigma = F) # because it is not normally distributed --> transform with sqrt
sqrtlmejacNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable * Direction * Day, data = jacobiclean, random = ~1| Sampler)
hist(resid(sqrtlmejacNH4)) #normally distributed
anova.lme(sqrtlmejacNH4, type="marginal", adjustSigma = F) # instead of multiplying the independent variables --> adding up
testlmejacNH4 <- lme(NH4N_mg ~ Distance_to_stable + Direction * Day, data = jacobiclean, random = ~1| Sampler)
hist(resid(testlmejacNH4))
transtestlmejacNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable + Direction * Day, data = jacobiclean, random = ~1| Sampler)
pluslmejacNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable + Direction + Day, data = jacobiclean, random = ~1| Sampler)
hist(resid(pluslmejacNH4))
anova.lme(pluslmejacNH4, type="marginal", adjustSigma = F) # another order of adding up
keerlmejacNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable * Direction + Day, data = jacobiclean, random = ~1| Sampler)
hist(resid(keerlmejacNH4))
anova.lme(keerlmejacNH4, type="marginal", adjustSigma = F) plot(ranef(lmejacNH4))
plot(lmejacNH4)
#exploring the data
boxplot(jacobi$NH4N_mg~jacobi$Direction) #spread over the 4 wind directions boxplot(jacobi$NH4N_mg~jacobi$Day) #spread over the days of measuring boxplot(jacobi$NH4N_mg~jacobi$Sampler) #spread over all the measuring points
#test almere NH4
lmealNH4 <- lme(NH4N_mg ~ Distance_to_stable * Direction * Day * Distance_to_highway, data = almereclean, random = ~1| Sampler) #uitvoeren lme schoon dataframe
lmealNH4
summary(lmealNH4)
boxplot(almere$NH4N_mg~almere$Direction) #spread over the 4 wind directions boxplot(almere$NH4N_mg~almere$Day) #spread over the days of measuring boxplot(almere$NH4N_mg~almere$Sampler) #spread over all the measuring points
boxplot(almereclean$NH4N_mg~almereclean$Distance_to_highway) #spread in relation to distance to the highway
hist(resid(lmealNH4)) #right skewed histogram --> transform qqnorm(resid(lmealNH4))
qqline(residuals(lmealNH4))
qqnorm(lmealNH4, ~ranef(.,level=1)) # punten volgen diagonaal dus normaal verdeeld! #als de punten diagonaal volgen -- normaal verdeeld - van de random variabele
#anova zonder distance to highway
sqrtlmealNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable * Direction * Day, data = almereclean, random = ~1| Sampler)
hist(resid(sqrtlmealNH4)) qqnorm(resid(sqrtlmealNH4)) qqline(residuals(sqrtlmealNH4))
anova.lme(sqrtlmealNH4, type="marginal", adjustSigma = F) #anova met de distance to highway
sqrtdistancelmealNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable * Direction * Day * Distance_to_highway, data = almereclean, random = ~1| Sampler)
anova.lme(sqrtdistancelmealNH4, type="marginal", adjustSigma = F) #anova met + in plaats van * zonder distance to highway
sqrtpluslmealNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable + Direction + Day, data = almereclean, random = ~1| Sampler)
anova.lme(sqrtpluslmealNH4, type="marginal", adjustSigma = F)
#anova met + in plaats van * met distance to highway
sqrtplusdistancelmealNH4 <- lme(sqrt(NH4N_mg) ~ Distance_to_stable + Direction + Day + Distance_to_highway, data = almereclean, random = ~1| Sampler)
anova.lme(sqrtplusdistancelmealNH4, type="marginal", adjustSigma = F)
#correlation between NH4/NOx #jacobi
plot(jacobiclean$NH4N_mg~jacobiclean$NOxN_mg,xlab="NOxN [mg/m2/2weeks]",ylab="NH4N [mg/m2/2weeks]")
hist(jacobiclean$NH4N_mg) #niet normaal verdeeld
hist(sqrt(jacobiclean$NH4N_mg)) #normaal verdeeld na transformatie hist(jacobiclean$NOxN_mg)
shapiro.test(jacobiclean$NOxN_mg) #niet normaal verdeeld #spearman gebruiken omdat het normaal verdeeld is
cor.test(jacobiclean$NH4N_mg, jacobiclean$NOxN_mg, method = c("spearman")) plot(jacobiclean$NH4N_mg~jacobiclean$NOxN_mg)
title("NH4/NOx interaction Jacobi")
abline(reg = lm(jacobiclean$NH4N_mg~jacobiclean$NOxN_mg)) #almere
plot(almereclean$NH4N_mg~almereclean$NOxN_mg,xlab="NOxN [mg/m2/2weeks]",ylab="NH4N [mg/m2/2weeks]")
#dus spearman gebruiken
shapiro.test(sqrt(almereclean$NH4N_mg)) shapiro.test(almereclean$NOxN_mg)
cor.test(almereclean$NH4N_mg, almereclean$NOxN_mg, method = c("spearman")) plot(almereclean$NH4N_mg~almereclean$NOxN_mg)
title("NH4/NOx interaction Almere")
abline(reg = lm(almereclean$NH4N_mg~almereclean$NOxN_mg))
#correlation between NH4/SOx #jacobi
plot(jacobiclean$NH4N_mg~jacobiclean$SO4_mg,xlab="SO4 [mg/m2/2weeks]",ylab="NH4N [mg/m2/2weeks]")
hist(jacobiclean$NH4N_mg) #niet normaal verdeeld
hist(sqrt(jacobiclean$NH4N_mg)) #normaal verdeeld na transformatie shapiro.test(sqrt(jacobiclean$NH4N_mg)) # niet normaal
hist(jacobiclean$SO4_mg) # niet normaal verdeeld #spearman gebruiken omdat het niet normaal verdeeld is
cor.test(jacobiclean$NH4N_mg, jacobiclean$SO4_mg, method = c("spearman")) plot(jacobiclean$NH4N_mg~jacobiclean$SO4_mg)
title("NH4/SOx interaction Jacobi")
abline(reg = lm(jacobiclean$NH4N_mg~jacobiclean$SO4_mg))
#almere
plot(almereclean$NH4N_mg~almereclean$SO4_mg,xlab="SO4 [mg/m2/2weeks]",ylab="NH4N [mg/m2/2weeks]")
hist(almereclean$NH4N_mg) #niet normaal verdeeld
hist(sqrt(almereclean$NH4N_mg)) #normaal verdeeld na transformatie hist(almereclean$SO4_mg) # niet normaal verdeeld
hist(sqrt(almereclean$SO4_mg)) #niet normaal verdeeld #dus spearman gebruiken
shapiro.test(sqrt(almereclean$NH4N_mg)) shapiro.test(almereclean$SO4_mg)
cor.test(almereclean$NH4N_mg, almereclean$SO4_mg, method = c("spearman")) plot(almereclean$NH4N_mg~almereclean$SO4_mg)
title("NH4/SO4 interaction Almere")
#correlation between SOx/NOx #jacobi
hist(sqrt(jacobiclean$SO4_mg)) # not normally distributed --> transform --> not pearson --> spearman
hist(jacobiclean$NOxN_mg) # wel normaal verdeeld
cor.test(jacobiclean$NOxN_mg,jacobiclean$SO4_mg, method = c("spearman")) #scatterplot plot(jacobiclean$NOxN_mg,jacobiclean$SO4_mg,xlab="NOxN [mg/m2/2weeks]",ylab="SO4 [mg/m2/2weeks]") abline(reg = lm(jacobiclean$SO4_mg~jacobiclean$NOxN_mg)) #almere plot(almereclean$SO4_mg~almereclean$NOxN_mg) hist(almereclean$SO4_mg) #not normally distributed
hist(sqrt(almereclean$SO4_mg)) #try to transform but still not normally distributed hist(almereclean$NOxN_mg) # not normally distributed
hist(sqrt(almereclean$NOxN_mg)) # still not normally distributed # use spearman in stead of pearson
cor.test(almereclean$NOxN_mg, almereclean$SO4_mg, method = c("spearman")) plot(almereclean$NOxN_mg, almereclean$SO4_mg,xlab="NOxN
[mg/m2/2weeks]",ylab="SO4 [mg/m2/2weeks]")
abline(reg = lm(almereclean$SO4_mg~almereclean$NOxN_mg))
#difference in NH4 per farm
almerejacobiclean #dataset where jacobi and almere merged --> usage of boxplot easier boxplot(almerejacobiclean$NH4N_mg~almerejacobiclean$ï..)
title("Difference in NH4-N deposition", y = "NH4-N [mg/m2/2weeks]") median(almereclean$NH4N_mg) #23.51096
median(jacobiclean$NH4N_mg) #25.82984 mean(almereclean$NH4N_mg) #29.81841 mean(jacobiclean$NH4N_mg) #31.689 #difference in NOx per farm
median(almereclean$NOxN_mg) #11.89694 mean(jacobiclean$NOxN_mg) #13.84304 mean(almereclean$NOxN_mg) #12.59699 #difference in SOx per farm
boxplot(almerejacobiclean$SO4_mg~almerejacobiclean$ï..) title("Difference in SO4 deposition", y = "SO4 [mg/m2/2weeks]") median(jacobiclean$SO4_mg) # 53.83581
median(almereclean$SO4_mg) #0 mean(jacobiclean$SO4_mg) #65.44744 mean(almereclean$SO4_mg) #27.59317 #significance test for mean
#Conditions checken hist(almereclean$NH4N_mg) hist(jacobiclean$NH4N_mg) hist(almereclean$NOxN_mg) hist(jacobiclean$NOxN_mg) hist(almereclean$SO4_mg) hist(jacobiclean$SO4_mg) wilcox.test(almerejacobiclean$NH4N_mg ~ almerejacobiclean$ï..) wilcox.test(almerejacobiclean$NOxN_mg ~ almerejacobiclean$ï..) wilcox.test(almerejacobiclean$SO4_mg ~ almerejacobiclean$ï..)
#lme almere NOx
lmealNOx <- lme(NOxN_mg ~ Distance_to_stable * Direction * Day * Distance_to_highway, data = almereclean, random = ~1| Sampler)
lmealNOx #exploration boxplot(almereclean$NOxN_mg~almereclean$Direction) boxplot(almereclean$NOxN_mg~almereclean$Day) boxplot(almereclean$NOxN_mg~almereclean$Sampler) boxplot(almereclean$NOxN_mg~almereclean$Distance_to_highway) hist(resid(lmealNOx)) qqnorm(resid(lmealNOx)) qqline(residuals(lmealNOx))
#kijken of resid normaal verdeeld is
qqnorm(lmealNOx, ~ranef(.,level=1)) # volgt niet de diagonaal -- wat nu? qqnorm(resid(lmealNOx))
qqline(resid(lmealNOx)) #data looks normal
anova.lme(lmealNOx, type="marginal", adjustSigma = F)
sqrtlmealNOx <- lme(sqrt(NOxN_mg) ~ Distance_to_stable * Direction * Day * Distance_to_highway, data = almereclean, random = ~1| Sampler)
anova.lme(sqrtlmealNOx, type="marginal", adjustSigma = F)
sqrtlmealNOxplus <- lme(sqrt(NOxN_mg) ~ Distance_to_stable + Direction + Day + Distance_to_highway, data = almereclean, random = ~1| Sampler)
anova.lme(sqrtlmealNOxplus, type="marginal", adjustSigma = F)
#test almere SO4
lmealSO4 <- lme(SO4_mg ~ Distance_to_stable * Direction * Day * Distance_to_highway, data = almereclean, random = ~1| Sampler)
lmealSO4
is.numeric(lmealSO4)
boxplot(jacobiclean$SO4_mg~jacobiclean$Direction) #verspreiding 4 windrichtingen boxplot(jacobiclean$SO4_mg~jacobiclean$Day) #verspreiding over het jaar
boxplot(jacobiclean$SO4_mg~jacobiclean$Sampler) #verspreiding in alle meetpunten boxplot(almereclean$SO4_mg~almereclean$Distance_to_highway)
hist(resid(lmealSO4)) # hist(sqrt(lmealSO4)) qqnorm(resid(lmealSO4)) qqline(residuals(lmealSO4))
sqrtlmealSO4 <- lme(sqrt(SO4_mg) ~ Distance_to_stable * Direction * Day * Distance_to_highway, data = almereclean, random = ~1| Sampler)
hist(resid(sqrtlmealSO4)) #normally distributed
anova.lme(sqrtlmealSO4, type="marginal", adjustSigma = F)
sqrtlmealSO4plus <- lme(sqrt(SO4_mg) ~ Distance_to_stable + Direction + Day + Distance_to_highway, data = almereclean, random = ~1| Sampler)