Spatial variability of soil properties related to small-scale agriculture at
the Strandvliet Park Amsterdam South-east
A Thesis by: Bart Housmans UvAnetID: 10172769 barthousmans@gmail.com +316 3486 3848 Daalwijk 405 b 1102 AA Amsterdam Southeast Supervisor: Head supervisor: dr K. Kalbitz Second supervisor: dr. L.H. Cammeraat Date of submission 29-06-2014 Word count: 5563 Abstract:
Urban agriculture seems to be an upcoming trend. These small-scale projects generally encounter some problems with infertile or contaminated soils, due to urban pollutions. The foundation ‘Groene Vingers’ performs small-scale urban agriculture in the Strandvliet Park in Amsterdam Southeast, The Netherlands. This paper will focus on the plot of agricultural land that was appointed to ‘Groene Vingers’, and the Strandvliet Park surrounding it. An analysis on the spatial variability and mutual correlations of horizon thickness, soil organic carbon, pH, electrical conductivity and texture will be conducted. Statements will be made about possible spatial relations between these parameters with the use of geostatistics. Variations and correlations were investigated with the use of semi- and crossvariograms. The optimal grid cell size was determined through a first phase research at 4 meters. Crossvariograms proved a positive correlation between grain size and pH for the A horizon. The other parameter pairs did not prove to validly correlate. All soil parameters proved at acceptable levels for the purpose of agriculture, except the pH of the soil, which proved too acidic. Liming or precision agriculture could be applied to overcome this problem. Finally, the knowledge about geostatistical procedures and the soil parameters related to small-scale agriculture gained in this research, might contribute to methods for selecting potential future sites for urban agriculture. Keywords
Table of Content page
1 Introduction 3
1.1 Small-scale urban agriculture 3
1.2 Soil fertility 4
1.3 Overall aim 5
1.4 Research questions 6
1.5 Relevance and originality 6
2 Methodology 7
2.1 First phase research 7
2.2 Second phase research 8
3 Results 9
3.1 First phase research 9
3.1.1 Semivariograms 9
3.2 Second phase research 10
3.2.1 Interpolated maps 10 3.2.2 Semivariograms 11 3.2.3 Crossvariograms 12 3.2.4 Correlation coefficients 13 3.3 Interpretation results 13 4 Discussion 15 4.1 Discussion results 15 4.2 Methodological improvements 16 5 Conclusion 17 References 18
1 Introduction
The local government of the Dutch municipality Amsterdam South-east has indicated it desires more small-scale local initiatives in the district (‘District Amsterdam Southeast’, 2012). These initiatives should provide more awareness of the local environment and increase the social cohesion among inhabitants. Subsidies are made available in order to meet these wishes (‘District Amsterdam Southeast’, 2013). In line with this trend, the foundation ‘Groene Vingers’ (Green Fingers) was launched at the end of 2013. This foundation performs small-scale agriculture in the Strandvliet Park in Amsterdam Southeast, with aid of local volunteers. The local government has appointed them a plot of land of 600 m2 in the Strandvliet Park for the purpose of agriculture.
Figure 1. An overview of the Research area: the Strandvliet park is situated just below the city of Amsterdam, in the
southeast district.
1.1 Small-scale urban agriculture
Initiatives similar to ‘Groene Vingers’ seem to be an upcoming and popular trend, with a focus on increasing the quality of life of inhabitants of urban areas (Cisneros, 2012). The production of food near large consumer centres has multiple advantages; it increases food-security, social cohesion and is often economically viable (Wakefield, Yeudall, Taron, & Skinner, 2007). However, these small–scale projects encounter some challenges, mainly related to their location (Agrawal, Singh, Rajput, Marshall, & Bell 2003). Densely populated urban areas commonly encounter high concentrations of heavy metals in the soil, which are emitted by anthropogenic sources (such as industrial sites and waste disposal) (Romic & Romic, 2003). These high concentrations can be harmful for people consuming the crops, as well as people working on the contaminated soil (McClintock, 2012). Furthermore, air pollution caused by traffic – a common phenomenon in urban areas - can limit crop productivity (Agrawal et al, 2003). Additionally, urban soils are often disturbed and configured by anthropogenic forces related to construction work. These reconfigurations disturb the development of a soil and its horizons. Furthermore, extensive paved areas and a lack of natural vegetation can cause compaction of the underlying soil, causing problems concerning water drainage (Jim, 1998). All these topics regarding urban soils make the performance of urban agriculture a challenging affair; they all have potential negative effects on efficient growth of crops efficiency. Thus, for urban agriculture to be success, all these soil infirmities should be monitored and managed. In case of the foundation ‘Groene Vingers’ it is desired that this is performed in a sustainable way.
1.2 Soil fertility
In order to assess the fertility of a soil for the purpose of (urban) agriculture, the condition of the three categories of soil properties - chemical, biological and physical properties – and their mutual interaction should be investigated. Chemical properties mainly constitute the presence and availability of nutrients in the soil. The quality and quantity of soil life are described as the biological properties of a soil. The physical properties include the texture and structure of a soil (Baldwin, 2001). The conditions of these categories of soil properties all have relevance for the agricultural potential of a soil. They can be assessed through various soil parameters, some more easily measureable than others.
This research will focus on physical and chemical properties of the soil in the Strandvliet Park. These two categories were chosen because they are relatively easily measured; yet still provide a clear view on the condition of the soil and its agricultural suitability (Baldwin, 2001). Moreover, the condition of these categories of the soil can be measured through parameters that provide information on multiple aspects of the condition of the soil (Carroll & Oliver, 2005). One of these parameters that is representative of multiple aspects of soil condition and is considered in this research is the Soil Organic Carbon Content (SOC) of the soil. SOC plays a major role in the physical properties of a soil. SOC has the potential to improve the structure of the soil, increase the cation exchange capacity (CEC) and permeability of the soil. Furthermore, the carbon provides trace elements vital for plant growth. Thus, ideally - in view of agricultural purposes - a soil has a high SOC content. (Jobbágy & Jackson, 2000). The second soil parameter under investigation is pH. pH is a measure for the acidity or alkalinity of a soil and largely affects the nutrient availability in the soil (Robson, 2012). Figure 2 displays the relation between nutrient availability and soil pH. It can be observed that heavy metals (Iron, Manganese, Boron, Copper and Zinc) are optimally available in strongly/medium acidic soils. Whilst the desired, and most important nutrients (Nitrogen, Phosphorus Potassium and Sulphur) are less available in these acidic ranges. Furthermore, pH can be a measure for the age of a soil. In general, older and more developed soils are acidic, compared to young undeveloped soils (Hajkowicz & Young, 2005). In view of nutrient availability a pH between 6.5 and 7.5 is considered optimal for most crops (Fernández, & Hoeft, 2009).
Figure 2. Nutrient availability in relation to pH (Lucas & Davis, 1961).
Another parameter that is considered in this research is Electrical Conductivity (EC). This parameter represents the concentrations of salts in the soil. High salt content constitutes high EC values. A high salinity can reduce the amount of water available for plant-uptake. Furthermore, the salt content
can alter the configuration of cations on the exchange surface of soil particles; this can influence the soil permeability (Corwin & Lesch, 2003). Thus, – in view of the agricultural potential of the soil - low salt concentrations are desired (thus low EC). Finally, texture is measured as an indicative soil parameter. This parameter was chosen because the soil particle size (texture) has great effects on soil hydraulic properties (Zak, 2003). A more fine-grained soil (clayey) having best effects on the water retention curve of a soil (Dexter, 2004). Moreover, the soil texture influences the CEC of the soil and – together with the pH – base saturation (the ratio between basic cations held on CEC sites and total number of available sites) (Cooper, 2009). This relationship is schematically shown in figure 2. In general, smaller sized particles (clayey) have a higher CEC - thus base saturation - than larger sized particles (sandy).
Figure 2. Relationship between soil pH, H+ and base saturation (Cooper, 2009).
In general, it is expected that some spatial correlations between these soil parameters will be present. Ritchie & Dolling (1985) proved that a negative correlation between Soil Organic Matter and pH is likely. Soil Organic Matter has an acidic character. Furthermore the acidic conditions can reduce micro-bacterial activity, which consumes Organic Matter/Carbon.
Because only a limited amount of time is available in this research, the previously enumerated soil parameters (SOC, pH, EC and texture) will have to be measured on regular points throughout the park (regular grid). Hereafter, statistics and interpolation will have to be conducted in order to be able to make founded statements on the correlations between the parameters. The field of geostatistics is most suitable for this aim; this branch of statistics deals with spatial patterns of certain data (such as the parameters under investigation) and provides the possibility to draw founded inferences and conclusions with accompanying uncertainties (Webster & Oliver, 2007). In this case, geostatistics will be used as a tool to determine the appropriate sampling interval of the regular grid, and to be able to draw inferences about spatial patterns.
1.3 Overall aim
The overall aim of this research is to obtain spatial relationships between the soil parameters EC, pH, SOC, and texture in the Strandvliet Park. In order to be able to investigate these relations, the spatial distributions throughout the park have to be known. When the first spatial distributions are mapped, the next objective will be to determine the appropriate sampling strategy (i.e. sampling distance). After interpolation, the research on the relations between the soil parameters will be conducted. Finally, the aim of this research is to make the outcome suitable for application for ‘Groene Vingers’ to increase their agricultural potential. A short overview of potential improvements will be presented in the thesis.
1.4 Research questions Main research question:
- Are the soil properties related to small-scale agriculture at the Strandvliet Park, Amsterdam Southeast spatially correlated?
Sub-research questions:
(1) What is the appropriate grid cell size for the different soil parameters related to small-scale agriculture in the Strandvliet Park?
(2) What is the spatial distribution of SOC, EC, pH and soil profile at the studied location? (3) In which way can the agricultural potential of the soil in Strandvliet Park (especially around
agricultural plain) be increased in a sustainable way? 1.5 Relevance and originality
The studied location in Amsterdam South-east used to be part of the Bijlmer-lake, which was drained in 1626 (‘Archive Utrecht’, 2006). After this reclamation, the soil has been used for different purposes, and around 1970 large-scale construction activities took place surrounding the Strandvliet Park, which altered the environment and soils. This means the soil is relatively young at the studied location, and this research might give insight in spatial variability of the studied soil parameters in undeveloped and anthropogenically affected urban soils. Moreover, in-depth soil research has not been conducted in the Strandvliet Park, so a niche in data seems present. The only currently present soil data can be found in a drilling conducted by TNO in (1987), and is displayed in figure 4. It can be observed that part of the soil under investigation (top +- 1 meter max) consist mainly of peat, so it can be expected to encounter high levels of SOC and acidic conditions (Freeman, Evans, Monteith, Reynolds & Fenner, 2001). It must be noted, that this data should be considered purely indicative, because the depth resolution is not detailed enough for agricultural purposes.
Figure 4. Drill data up to 5-meter dept. Drill location displayed in Figure 1 (pink dot in centre of Strandvliet Park) (TNO, 1987). Furthermore, the outcomes might prove useful for the ‘Groene Vingers’ foundation as input for soil improvement to increase agricultural potential.
2 Methodology
In order to obtain the spatial distributions of the soil parameters, hand auger borings will be performed on regular intervals (regular grid) throughout the Strandvliet Park, and samples will be taken. Preceding these borings, the grid size has to be determined. This grid size will have to represent the optimal balance between two extremes. On the one hand, the - by the finite amount of time available - limited number of measurements that can be performed. Meanwhile, the distance between measurements needs to be small enough for relationships between the parameters to be present. This optimal balance is reached with the largest grid size for which a valid relationship between parameters is still present (Steffens et al., 2009). Seen as there has been no previous soil research in the Strandvliet Park, the frequency of the spatial variation will have to be determined through a preliminary research (first phase). This research will precede the main research (second phase) of the whole park.
2.1 First Phase
In the first phase research, measurements of SOC, EC, pH and Soil Profile will be conducted on a cross-grid. A cross-grid is chosen to measure both the latitudinal as longitudinal variations of the soil parameters. The centre of the cross-grid will be chosen, so that at least a part of the agricultural plain is included. In order to investigate more short distance variances, five drill holes will be conducted one meter apart from the centre in each wind direction. In order to review long distance variations, drillings will also be performed at a distance of 10 meter from the centre, in each wind direction. In total, 25 drillings will be performed. Measurements will be done, and samples taken at each bore hole, on at least three distinct soil horizons (A, B and C horizon), if present. These measuring depths (horizons) are chosen because they are all in the root zone of most crops. The thickness of the soil horizons and soil texture will be determined in the field with the use of sand ruler and the World Reference Base respectively (Deckers, Nachtergaele & Spaargaren, 1998). Samples will be collected and taken to the lab for analysis of SOC, pH and EC.
Figure 5. Schematic overview of cross-grid and the different drill hole distance intervals.
After the samples have been gathered, they will be processed in the lab. SOC will be analysed through loss of ignition. This method estimates the SOC by comparing the weight of a soil sample before and after intense heating though an oven. The difference in weight is a measure for the Soil Organic Matter (SOM). SOM is a measure for the SOC in a sample, and in general a conversion factor of about 0.5 is applied to obtain the SOC content (Pribyl, 2010). Because the carbon content can vary strongly over small spaces in a soil, these measurements will be conducted in duplo to obtain a representative overview (Ball, 1964).
The samples of EC and pH will firstly put into solution with demineralised water with a ratio of 1 : 2.5 (soil : demineralised water). Thereafter, the samples will be shaken by a vibration platform for two hours. Finally the EC will be measured with an EC-meter, and the pH with a pH-meter. The needed materials will be provided by the UvA lab. After the data has been processed, and the values for all the different parameters at the measured points are known, the variances can be analysed. This will be done using the program MATLAB. Semivariograms will be determined for each of the
parameters. A semivariogram is a scatter, which describes the relationship between distance and the variance of a parameter (Goovaerts, 1999) as defined by equation 1.
With y (h) = semivariance for lag vector h. N (h) number of data points separated by lag vector h. Ai(xi) and Ai(xi + h) = actual values of measured parameter A at certain points (xi) separated by h
(Steffens et al., 2009).
Commonly, a semivariogram describes an increasing variance, thus decreasing correlation, with an increasing distance. This correlation follows Tobler’s first rule of geography; "Everything is related to everything else, but near things are more related than distant things" (Moellering & Tobler, 1972). Subsequently, the correct distance that should be used for the second phase fieldwork will be situated before the variances in the semivariogram level out (i.e. range). These variances and corresponding distances will be determined arbitrarily.
2.2 Second Phase
When the first phase research has led to the appropriate grid size, the second phase research can be conducted. A regular grid will be placed on the park. Measurements will be conducted on the intervals and samples will be collected. The samples of SOC, pH and EC will be taken to the lab, and analysed similarly as described in the 2.1 first phase research chapter. When the data is obtained, an interpolation can be performed. Thereafter, Kriging interpolation will be conducted to interpolate the known data points. Kriging is chosen as interpolation technique, because is appoints weights to known data points based on their distance as well as their mutual arrangement (Goovaerts, 1999). The interpolation as the visualization will be done in MATLAB.
The mutual relations between the different soil parameters will be investigated through cross-variograms. A cross-variogram describes the spatial correspondance of two or more variables (equation 2).
With Bi(xi) and Bi(xi + h) = actual values of measured parameter B at certain points (xi) separated by h
(Steffens et al., 2009).
These cross-variograms can be used to describe the variance between the different soil parameters. The outputs can be analyzed and conclusions can be drawn. These conclusions will mostly have a qualitative characters, because cross-variograms do not allow thorough quantitative analysis (Wackernagel, 1998). Seen as the aim is to investigate spatial correlations, only parameters that display a spatial patern are relevant input parameters for crossvariograms. These relevant parameters will be selected through semivariograms of the second phase research data.
To be able to draw further conclusions on correlations between the soil paramters correlation coefficients wil be constructed. These coefficients provide information about non-spatial correlations between different parameters (Lawrence & Lin, 1989).
Further inferences about the agricultural potential of the soil and possible improvements will be made through literature research in combination with the obtained data from the park.
3 Results
3.1 First phase research 3.1.1 Semivariograms
Results of the first phase research: semivariograms:
A Horizon B Horizon C Horizon
EC
pH
LOI
GS
HT
Figure 6. Overview of obtained semivariograms from the first phase fieldwork. x-axis displays averaged distance between
observations in meters. y-axis displays semivariance in respectively: µS cm-1 for Electrical Conductivity (EC), pH for pH, percentage (%) of total soil for Loss Of Ignition (LOI), µm for Grain Size (GS) and centimetres for Horizon Thickness (HT). The dashed line represents the variance of the plotted parameter.
3.2 Second phase research 3.2.1 Interpolated maps
Results of second phase fieldwork: (Kriging) interpolated maps A Horizon B Horizon EC
pH
GS
HT
LOI Figure 7. Overview of obtained
interpolated maps from the second phase fieldwork. x-axis displays x-coordinate. y-axis displays y-coordinate. Parameters in following units: µS cm-1 for Electrical Conductivity (EC), pH for pH, µm for Grain Size (GS), centimetres for Horizon Thickness (HT) and percentage (%) of total soil for Loss Of Ignition (LOI).
3.2.2 Semivariograms
Results of second phase fieldwork: semivariograms.
A Horizon B Horizon EC pH GS HT LOI
Figure 8. Overview of the obtained
semivariograms from the second phase fieldwork. x-axis displays averaged distance between observations in meters. y-axis displays semivariance in respectively: µS cm-1 for Electrical Conductivity (EC), pH for pH, percentage (%) of total soil for Loss Of Ignition (LOI), µm for Grain Size (GS) and centimetres for Horizon Thickness (HT). The dashed line represents the variance of the plotted parameter.
3.2.3 Crossvariograms
Results of the second phase field work: crossvariograms.
A Horizon B Horizon EC & GS EC & pH GS & pH pH & HT
Inter-crossvariograms (between A and B horizon)
pH EC
Figure 9. Overview of obtained crossvariograms from the second phase fieldwork. x-axis displays averaged distance
between observations in meters. y-axis displays semivariance between: parameters µS cm-1 for Electrical Conductivity (EC), pH for pH, percentage (%) of total soil for Loss Of Ignition (LOI), µm for Grain Size (GS) and centimetres for Horizon Thickness (HT).
3.2.4 Correlation coefficients
Results of the second phase field work: correlation coefficients and corresponding p-values Correlation coefficients. Table 1. Overview of correlation coefficients of the second phase fieldwork. First section
displays correlation coefficients of A horizon with corresponding p-values. Second section displays correlation coefficients of B horizon with corresponding p-values. Third section displays correlation coefficients for inter-horizon crossvariograms (Thus, correlation between A and B horizon for each parameter individually).
3.3 Interpretation Results
The results of the first phase fieldwork are represented as semivariograms in figure 6. Spatial correlations are present, when a definite trend can be observed from semivariograms. Figure 6 shows that most parameters at most horizons show spatial patterns. Only pH at the B horizon and loss of ignition at the B and C horizons seem to lack a spatial distribution.
Table 2 displays the optimal grid cell size as observed and arbitrarily determined from the semivariograms. Besides the optimal range a maximal and minimal value has been selected. This interval displays the borders for which the grid cell size is still valid, though not optimal.
Optimal grid cell sizes. Table 2. Arbitrarily determined optimal grid cell sizes in meters from first phase results.
Parameter and Horizon Optiamal grid cell size Maximal – minimal values
EC A Horizon 6 5.5 – 6.5
EC B Horizon 6 5.5 – 6.5
EC C Horizon 4 3.5 – 4.5
pH A Horizon 5 4.5- 5.5
pH B Horizon - -
Correlation coefficient between parameters in A horizon
A GS EC pH HT LOI GS - 0.5824 -0.6003 0.1042 0.1140 EC 0.5824 - -0.6598 0.2805 -0.0002 pH -0.6003 -0.659 - -0.4754 -0.2959 HT 0.1042 0.2805 -0.4754 - -0.0394 LOI 0.1140 -0.0002 -0.2959 -0.0394 -
Corresponding p-values (A-horizon)
A GS EC pH HT LOI GS - 0.0023 0.0015 0.6202 0.5875 EV 0.0023 - 0.0003 0.1744 0.9992 pH 0.0015 0.0003 - 0.0163 0.1510 HT 0.6202 0.1744 0.0163 - 0.8518 LOI 0.5875 0.9992 0.1510 0.8518 -
Correlation coefficient between parameters in B horizon
B GS EC pH HT
GS - 0.5515 -0.6106 0.3478
EC 0.5515 - -0.7344 0.3807
pH -0.6106 -0.7344 - -0.6333
HT 0.3478 0.3807 -0.6333 -
Corresponding p-values (B-horizon)
B GS EC pH HT
GS - 0.0043 0.0012 0.0885
EC 0.0043 - 0.0001 0.0604
pH 0.0012 0.0001 - 0.0007
HT 0.0885 0.0604 0.0007 -
Correlation coefficient and p-value for inter-horizon crossvariograms (between A and B horizon)
Soil Parameter Correlation coefficient p-value
GS 0.5494 0.0044
EC 0.3772 0.0631
pH 0.9715 0.0002
Parameter and Horizon Optiamal grid cell size Maximal – minimal values pH C Horizon 4 3.5 – 4.5 LOI B Horizon - - LOI C Horizon 5 4.5 – 5.5 GS A Horizon 4.5 5 – 5 GS B Horizon 5 4.5 – 5.5 GS C Horizon - - Thickness A Horizon 6 5-7 Thickness B Horizon 6 5.5 – 6.5
Seen as all the spatial correlations should be included in the measurement interval of the second phase research, the lowest range of table 2 must be selected. In this case a grid cell size of 4 meter should include all spatial correlations, and therefore be used as the grid cells size for the second phase fieldwork. With a grid cell size of 4 meter, the agricultural plain was selected as the eventual research area for the second phase fieldwork, because this matched its size and because this part of the Strandvliet Park is considered most important in light of the aims of this research.
The first results from the second phase fieldwork are visualized in figure 7 as (Kriging) interpolated maps. These maps provide a fist insight in possible spatial correlations between parameters and horizons. A slight increase in EC seems present in both horizons, along the north-south gradient through the research area. Furthermore, the spatial patterns of pH at the A and B horizon show strong similarities. Both horizons show a decrease in pH along a north south gradient. Moreover, the concentration of contour lines of both the A and B horizon seems to be lower in the centre of the research area. Grain size at the A and B horizon both show generally higher values in the south and lower values at the east of the research area. Finally, correlations between the soil horizon thickness (HT) A and B are not present.
Besides the distributions of the different parameters the A and B horizon respectively, spatial correlations between parameters are investigated trough crossvariograms. Before the crossvariograms are discussed, the input parameters need to be investigated for spatial patterns. This is done through semivariograms, and the results are displayed in figure 8. Again, most parameters at most horizons show distinct ranges (thus, spatial patterns). Only the B horizon thickness and loss of ignition at the A horizon lack a distinct range. The crossvariograms of the remaining parameters are displayed in figure 9 of the results. Unlike the semivariograms of the first and second research, the ranges are far less distinct. However, weak and averagely strong trends can be observed in most cases. The arbitrarily determined trends with there accompanying strengths are displayed in table 3.
Crossvariogram trends. Table 3. Arbitrarily determined trends with accompanying strengths of crossvariograms from figure 9.
Strength is defined as the tendency of crossvariograms to uniform unidirectional trend.
A horizon Trend Strength
GS and EC negative Weak
GS and pH positive Average
EC and pH positive Weak
B horizon
GS and EC - -
GS and pH positive weak
EC and pH positive weak
pH and HT positive weak
Inter A & B horizon correlations
EC A and B Horizon - -
Table 3 shows only that only the positive correlation between grain size and pH in the A horizon is of average strength. The other parameter pairs show weak correlation strengths and are considered to insignificant to prove actual correlations.
Furthermore, the only hypothesis that was made in the introduction did not prove effective for the research area. As noted, a negative correlation was expected between Soil Organic Carbon and pH. Seen as Soil Organic Carbon did not show any distinct spatial patterns, it was irrelevant for the crossvariograms to be constructed. Thus, this expected correlation did not prove itself valid for the research area.
4 Discussion
4.1 Discussion results
The first point of discussion is the only correlation that was proven valid through the crossvariograms; GS and pH at the A horizon. This correlation between grain size and pH was not expected from the literature discussed in the introduction. The correlation could possibly be explained through the mechanism in which an increase in grain size decreases the number of basic ions bound to the soil surface, thus increasing the pH levels of the soil. However, more research is needed on this topic to be able to make founded inferences.
As noted in the results section, the absence of an apparent negative spatial correlation between SOC and pH did not match the hypothesis. However the constructed correlation coefficients of table 1 can provide information about the non-spatial correlation between these parameters. With a correlation coefficient of -0.2959 and a p-value of 0.1510, the correlations weakly negative (and reasonably significant). Thus, the correlation between SOC and pH is only non-spatially present, and not as strong as expected. Possibly this could be explained by configurations of the soil. When horizons are mixed - and deeper layers reach the soil surface - the present organic carbon deteriorates faster than in the oxygen low deeper soil zones (Hartnett, Keil, Hedges & Devol, 1998). Thus, disturbing the development of a possible relation between pH and SOC.
Furthermore, the kriging-interpolated maps from figure 7 seem to suggest a strong spatial correlation between the pH of the A and B horizon. However, the crossvariograms did not prove this positive correlation in the slightest. Though, the correlation coefficient for this parameter pair is 0.9715 with a p-value of 0.0002 (table 3). Thus, a highly significant strong positive correlation. The reason for the absence of the hypothesised spatial correlations between parameters could be caused by two things.
Firstly, seen as the research area was drained from the Bijlmer lake early 1600, and configured for agricultural purposes later, the soil is relatively young of age. Correlations as investigated in this research develop over time.
Secondly, another reason that could explain the absence of correlations is the large construction projects active over the past 40 years in the research area. It is possible that some spatial correlations were present in the soil, but that large-scale reconfigurations of the soil mixed different soil layers, so that the stratification and development was disturbed.
The maps of figure 7 give us a first insight in the levels of the different parameters at the research area. Thus, the agricultural potential of the soil. The highest values for EC are present in the south with about 400 µS cm-1. According to Rhoades & Loveday (1990) these maximal values are – for most crops – at acceptable levels. pH levels are rather low. The optimal range of pH for agricultural soils is between 6.5 and 7.5 (figure 2). The observed values from the research area range from 5.5 in the south to 6.5 in the north. Thus, only the top north reaches acceptable levels. This point could be overcome in two ways. Firstly, the pH could be increased. Normally this is done through liming of the soil. Secondly, precision agriculture could be applied. This technique implies that the distribution of different crops is modified to the properties of the soil. In this case acidic crops should be planted in the south of the research area, and alkaline crops in the north. The soil texture of the research area does not imply any problems. The texture is not to sandy, so it maintains a decent water retention in
the soil. Moreover, having positive effects on CEC and base saturation. Finally, the SOC proves to be at acceptable levels. Only the top north-east shows a definite sink with levels of only 2%. Some organic material could be applied to increase SOC levels there.
When the results from this research concerning grid size are compared with other projects investigating spatial correlations, differences are present. Research was conducted on spatial variations of horizon thickness, grain size and SOC in a young alluvial site in Venezuela by Burrough (1989). Although the environmental conditions vary strongly between the two sites, both are of recent age, and a comparison can be made. The optimal grid size for horizon thickness, grain size and SOC for the Strandvliet Park varied between 4.5-6 meters. The grid sizes for the Venezuelan sites were significantly higher varying between 200-450 meters. Burrough (1989) stated that the short-distance variations superimposed the general pattern. This phenomenon could also apply for the Strandvliet Park. This could be investigated through further research with an exploratory phase that contains a cross-grid with maximal measurement intervals in the order of 50-100 meters.
Gaans & Hootsmans (1997) investigated an area near Albert, Canada at an old lacustrine basin. Seen as the soil at their research area was relatively old compared to the Venezuelan and Strandvliet location, it can be expected that soil parameter correlations were stronger, considering the soil had more time to develop. However, the grid cell size appropriate for their research was 60 x 60 meters; still a significantly higher sampling distance compared to the one of the Strandvliet Park.
Other research performed on a plot of 250 x 250 meter of strongly agricultural altered 14 000 yr old soil in Central Iowa also resulted in much higher ranges of the semivariogram for SOC; 100 meters (Cambardella, Moorman, Parkin, Karlen, Novak, Turco, & Konopka, 1994). So, even very old soils had significantly larger ranges than the ‘Groene Vingers’ plain.
Thus, in general the spatial variations and sampling distances were much higher for comparable researches. Both in relatively young and old soils, optimal values were at least ten times higher than the sampling distances used at the Strandvliet park. The relatively small-scale variations found at the Strandvliet Park could be induced by configurations of the landscape. However, it seems more likely they find their origin in the statement made by Burrough (1989); the short-distance variations superimposed the general pattern.
4.2 Methodological improvement
The main deficiency in this research is the sample size. The first and second phase research both consisted of only 25 drill sites of the three horizons. Thus, the semivariograms and crossvariograms were constructed from merely 25 values. In general a sample size of at least 30 is advised for – variograms (Goovaerts, 1998). A larger sample size could increase the strengths of the trends of the semivariograms and crossvariograms. Thus, shedding more light on possible correlations between the parameters under investigation in this research.
Furthermore, the statistical tool of crossvariograms did not prove very effective. The interpolated maps of figure 6 showed clear spatial similarities between the pH of the A and B horizon. However, the crossvariograms of this parameter pair failed to reproduce these spatial likenesses. This problem could be partly caused by the small sample size, but not completely; seen as the sample size of 25 is close to the advised 30, and still no correlation seems valid from the crossvariogram of pH at A and B horizon.
Moreover, the method for determining the grain size is not considered optimal. Grain size was determined in the field with the use of a sand ruler. Ideally, grain size is determined with a sedigraph. However, in order to analyse each sample with the use of a serigraph would take an ample amount of time, which was not available.
Furthermore, further research with similar aims should focus on more soil than the ones included in this research. A very important factor in the agricultural potential of a soil is the presence of abundant nutrients. This research only takes into account the availability of nutrients as influenced by pH, but does not provide information on actual presence of these nutrients.
Lastly, despite of the first phase exploratory research included in this project, optimal grid cell sizes proved significantly lower than found in comparable researches. Steffents et. al. (2009)
already pointed out the large influence of scale (scale dependency) on the results of this type of projects. Further research should aim to investigate small scale variations as well as larger scale variations. The first phase fieldwork included sampling intervals of 1 – 20 meters. Additional research should at least include maximal sampling intervals of 100 meters; the fivefold of current intervals.
5 Conclusion
In this research the spatial variations and correlations between horizon thickness, grain size, SOC, pH and EC were investigated, at the Strandvliet Park, Amsterdam Southeast. The variations for the parameters related to small-scale agriculture proved to be much lower than found in comparable researches. This point could be induced by configurations of the landscape in the recent past, thus young soils. But it is considered more likely to be caused by scale dependency of the research.
The hypothesised expected negative correlation between SOC and pH did not prove valid for the research area. However, the only correlations that was confirmed by the crossvariograms was a positive correlation between grain size and pH in the A horizon.
Geostatistics proved a useful method for investigating spatial distributions and correlations of the soil parameters. However, further research should investigate the effectiveness and precision of crossvariograms
Furthermore, all the values of the different soil parameters under investigation proved to be at acceptable levels for the purpose of agriculture. Except for the pH of the soil; this was considered too acidic for the desired function of the area, with levels ranging from 5.5 – 6.5. Liming or precision agriculture should be applied to overcome this problem.
Concluding, despite the negative effects literature (see introduction) suggests the location of the ‘Groene Vingers’ plain should have on its agricultural potential; this research proves the otherwise.
Literature
Agrawal, M., Singh, B., Rajput, M., Marshall, F., & Bell, J. N. B. (2003). Effect of air pollution on peri-urban agriculture: a case study. Environmental Pollution, 126(3), 323-329.
Amsterdam South-east (2013): Local initiatives. Consulted on
http://www.zuidoost.amsterdam.nl/buurten/aanpak-focuswijken/bewonersinitiatieven/ Archives of Utrecht (2006). Consulted on
http://www.utrechtsarchiefnet.nl/zoeken?miview=inv2&mivast=1535&mizig=210&miadt=142&miaet=1&mico de=0501&minr=1323674&milang=nl
Baldwin, K. R. (2001) Soil Fertility on Organic Farms. Centre for environmental farming centres.
Ball, D. F. (1964). Loss of ignition as an estimate of organic matter and organic carbon in non‐calcareous soils. Journal of Soil Science, 15(1), 84-92.
Burrough, P.A., 1989. Fuzzy mathematical methods for soil survey and land evaluation. J. Soil Sci. 40, 477-492.
Cambardella, C. A., Moorman, T. B., Parkin, T. B., Karlen, D. L., Novak, J. M., Turco, R. F., & Konopka, A. E. (1994). Field-scale variability of soil properties in central Iowa soils. Soil Science Society of America Journal, 58(5), 1501-1511.
Carroll, Z. L., & Oliver, M. A. (2005). Exploring the spatial relations between soil physical properties and apparent electrical conductivity. Geoderma, 128(3), 354-374.
Cisneros M. G. (2012) Assessing the potential of small-scale urban agriculture in Havana. University of Amsterdam
Cooper, T.H. (2009) pH and CEC. University of Minnesota
Corwin, D. L., & Lesch, S. M. (2003). Application of soil electrical conductivity to precision agriculture. Agronomy journal, 95(3), 455-471.
Deckers, J. A., Nachtergaele, F., & Spaargaren, O. C. (Eds.). (1998). World reference base for soil resources: Introduction (Vol. 1). Acco
District Southeast (2012) Sustainability Plan for Amsterdam Southeast 2012-2014, Strategy and Policy Division of Policy Development.
Dexter, A. R. (2004). Soil physical quality: Part I. Theory, effects of soil texture, density, and organic matter, and effects on root growth. Geoderma, 120(3), 201-214.
Fernández, F. G., & Hoeft, R. G. (2009). Managing soil pH and crop nutrients. Illinois agronomy handbook. 24th ed. Univ. of Illinois, Urbana, 91-112.
Freeman, C., Evans, C. D., Monteith, D. T., Reynolds, B., & Fenner, N. (2001). Export of organic carbon from peat soils. Nature, 412(6849), 785-785.
Gaans, P. F., & Hootsmans, R. (1997). Continuous classification in soil survey: spatial correlation, confusion and boundaries. Geoderma, 77(2), 115-135.
Goovaerts, P. (1999). Geostatistics in soil science: state-of-the-art and perspectives. Geoderma, 89(1), 1-45. Hajkowicz, S., & Young, M. (2005). Costing yield loss from acidity, sodicity and dryland salinity to Australian agriculture. Land Degradation & Development,16(5), 417-433.
Hartnett, H. E., Keil, R. G., Hedges, J. I., & Devol, A. H. (1998). Influence of oxygen exposure time on organic carbon preservation in continental margin sediments. Nature, 391(6667), 572-575.
Jim, C. Y. (1998). Urban soil characteristics and limitations for landscape planting in Hong Kong. Landscape and Urban Planning, 40(4), 235-249.
Jobbágy, E. G., & Jackson, R. B. (2000). The vertical distribution of soil organic carbon and its relation to climate and vegetation. Ecological applications, 10(2), 423-436.
Lawrence, I., & Lin, K. (1989). A concordance correlation coefficient to evaluate reproducibility. Biometrics, 255-268.
Lucas, R. E., & Davis, J. (1961). Relationships between pH values of organic soils and availabilities of 12 plant nutrients. Soil Science, 92(3), 177-182.
McClintock, N. (2012). "Assessing lead contamination at multiple scales in Oakland, California: Implications for urban agriculture and environmental justice". Applied Geography 35: 460–473.
Moellering, H., & Tobler, W. (1972). Geographical variances. Geographical Analysis, 4(1), 34-50.
NWO (2013) Financial accountability. retrieved from: http://www.nwo.nl/financiering/hoe-werkt-dat/Salaristabellen
Pribyl, D. W. (2010). A critical review of the conventional SOC to SOM conversion factor. Geoderma, 156(3), 75-83.
Rhoades, J. D. & Loveday. J. (1990) Salinity in irrigated agriculture. Agronomy Monograph 30:1089-1142. Americal Society of Agronomy, Madison, WI.
Ritchie, G. S. P. & Dolling, P. J. (1985) The Role of Organic Matter in Soil Acidification.
Romic, M., & Romic, D. (2003). Heavy metals distribution in agricultural topsoils in urban area. Environmental Geology, 43(7), 795-805.
Robson, A. (Ed.). (2012). Soil acidity and plant growth. Elsevier.
Steffens, M., Kölbl, A., Giese, M., Hoffmann, C., Totsche, K. U., Breuer, L., & Kögel‐Knabner, I. (2009). Spatial variability of topsoils and vegetation in a grazed steppe ecosystem in Inner Mongolia (PR China). Journal of Plant Nutrition and Soil Science, 172(1), 78-90.
TNO (1987) consulted on: http://dinoloket.nl/
Wackernagel, H. (1998) Multivariate geostatistics. 2nd, completely revised edn., Springer Berlin
Wakefield, S., Yeudall, F., Taron, C., & Skinner, A. (2007). Growing urban health: Community gardening in South-East Toronto. Health Promotion International, 22 (2).
Webster, R., & Oliver, M. A. (2007). Geostatistics for environmental scientists. John Wiley & Sons.
Weijers, J. W., Schouten, S., van den Donker, J. C., Hopmans, E. C., & Sinninghe Damsté, J. S. (2007). Environmental controls on bacterial tetraether membrane lipid distribution in soils. Geochimica et Cosmochimica Acta, 71(3), 703-713.
