Although soil compaction is widely recognized as a soil threat to soil
resources, reliable estimates of the acreage of overcompacted soil and
of the level of soil compaction parameters are not available. In the
Netherlands data on subsoil compaction were collected at 128 locations
selected by stratified random sampling. A map showing the risk of
subsoil compaction in five classes was used for
stratification. Measurements of bulk density, porosity, clay content
and organic matter content were used to compute the relative bulk
density and relative porosity, both expressed as a fraction of
a threshold value. A subsoil was classified as overcompacted if either
the relative bulk density exceeded 1 or the relative porosity was below
1. The sample data were used to estimate the means of the two subsoil
compaction parameters and the overcompacted areal fraction. The
estimated global means of relative bulk density and relative porosity
were 0.946 and 1.090, respectively. The estimated areal fraction of
the Netherlands with overcompacted subsoils was 43
Soil compaction is recognized as one of the major soil
threats.
Soil compaction is estimated to be responsible for the degradation
of an area of about 33 million ha in Europe
Previous work by
The aim of this research was to design a sample for estimating the current means of subsoil compaction parameters and the areal fraction where subsoil compaction has exceeded a critical threshold. These means and areal fraction must be estimated for the Netherlands in their entirety, as well as for the five units of the soil compaction risk map. The estimates must be accompanied with estimates of their accuracies.
The soil compaction risk map for the Netherlands of
The risk of subsoil compaction is a function of wheel loads of
machines, which is related to land use, and soil mechanical strength, which is determined by various soil properties such as soil texture
and water content. The map of soil compaction risk was constructed
by combining information derived from the land use database of the
Netherlands
The land use database was used to determine typical agricultural
machinery and associated typical wheel loads, tyres and tyre
inflation pressures for agricultural areas in the Netherlands. In
this, an inventory of
The calculated soil stresses for 2010 were compared with the soil
strengths in the same way as presented in
Map of risk for subsoil compaction
In a second step factors that increase or decrease the risk of
subsoil compaction in the long term were taken into
account. Factors that improve the resilience and natural
recuperation of the compacted subsoil and in that way decrease the
subsoil compaction risk are the following:
The soil is well drained and in general dry, improving the resilience and the natural
recuperation. Clay content is Organic matter content is Coarse sand: hardly any increase in dry bulk density, water infiltration is never
a problem. Only a limited part of the parcel can be trafficked and so compacted, e.g. forests or
orchards. The soil is often wet. The typical wheel loads of the land use will cause compaction at depths
Factors that increase the risk of subsoil compaction are as follows:
All positive and negative factors are added together and the risk
class in the first step is increased or decreased by a maximum
of one class. The change in class is limited to one step to
account for the fact that overloading and compaction of the
subsoil are cumulative in time and recuperation by shrinkage and
biological processes is never complete; therefore, the risk
classification should be mainly determined as a function of the
exerted stresses at a certain depth and the strength of the soil
at that depth. Figure
For estimating means of subsoil compaction parameters, locations
were selected by probability sampling, i.e. by random sampling with
known inclusion probabilities which are
The total sample size was 128. The sample sizes in the provinces
Gelderland, Noord-Brabant and Zeeland were 20, 39 and 30, respectively,
leaving 39 for the remaining provinces. These sample sizes were
allocated proportionally to the area of the five risk map units
within the provinces. The total sample sizes in the risk map units
1 (“low risk”) to 5 (“high risk”) were 4, 5, 56, 44 and 19. The small sample sizes for the risk map units 1 and 2
reflect the small areas of these two units: the sum of their areas
is only 4.6
The target population consists of all soils in the Netherlands, both cultivated and uncultivated soils, except soils with a low compaction risk due to peat layers, naturally compacted soils (“knipkleigronden”) and soils in glasshouses.
The randomly selected locations were localized by differential GPS. If a randomly selected sampling location was unsuitable for collecting soil samples (no soil present, no permission, not part of the target population), the first point on a reserve list, in the same stratum as the omitted point, was added to the list of points to be visited.
At each sampling location three volumetric subsoil samples were
collected using a cylinder with a diameter of 7.6
We used the relative bulk density
and relative porosity as subsoil compaction parameters. The relative bulk density is defined as the
actual bulk density when seen as a fraction of the threshold value of the
bulk density
The relative porosity is defined as the actual porosity when seen as a fraction of the threshold value of the porosity, which is 0.4 as
determined in the ENVASSO project
If either the relative bulk density
The global means of the relative bulk density and relative porosity
were estimated by design-based inference, more specifically by the
usual estimator for stratified simple random sampling:
These estimators were also used to estimate the means of the two subsoil compaction parameters and the overcompacted areal fraction for the five units of the soil compaction risk map and for the three provinces. These sub-areas are unions of complete strata; i.e. they do not contain one or more strata which only partly belong to the sub-area, so that estimation is straightforward.
In all four strata of risk map unit 1 and the three strata of risk map
unit 2, only 1 point was selected. This complicates the estimation
of the sampling variance of the estimated means. Following the
approach of
Box plots of relative bulk density and relative porosity per risk map unit (1: low risk; 5: high risk)
The sampling variance of the estimated areal fractions was
estimated by
Figure
Estimated means of subsoil compaction parameters and areal fractions of overcompacted subsoils for the five risk map units. The error bars indicate the standard error of the estimated means or areal fraction
Design-based estimates of means of two subsoil compaction parameters and of overcompacted areal fraction for the low-risk groups (map units 1 and 2) and high-risk groups (map units 3, 4 and 5). In brackets: standard error.
The estimated global mean of relative bulk density was 0.946 with
an estimated standard error of 0.012. The estimated global mean of
relative porosity was 1.090 with an estimated standard error of
0.020. The estimated areal fraction of the target population with
overcompacted subsoils was 0.446 with an estimated standard error
of 0.053. Correcting the estimated areal fraction for the peat
soils that were excluded from the target population (covering about
4.5
Design-based estimates of the means of the two subsoil compaction
parameters and of the overcompacted areal fractions per risk map
unit are shown in Fig.
For map units 1 and 2 the estimated areal fractions of overcompacted
subsoils were both 0 (in both units no sampling points had
a relative bulk density
As differences between map units 1 and 2 and between the map units
3, 4 and 5 were small, we also estimated means and areal fractions
for these two groups (Table
Estimated means of subsoil compaction parameters and areal fractions of overcompacted subsoils for high-risk group of map units (units 3, 4 and 5) in the three provinces. The error bars indicate the standard error of the estimated means or areal fraction. The red dots are the estimated means or areal fraction for the Netherlands.
Estimated means of subsoil compaction parameters and estimated areal fraction of overcompacted subsoils for the risk classes in the field.
Finally, we estimated means of the two subsoil compaction
parameters and overcompacted areal fraction for the high-risk
group of map units in the three provinces to check the
assumption that in these provinces the problem of subsoil
compaction was more serious (Fig.
The aggregated map unit high risk covers 95.4
A possible explanation is the poor quality of the soil compaction
risk map. The soil compaction risk class as depicted on the map
will not correspond everywhere with the risk class in the field,
i.e. the risk class as based on the soil profile characteristics
observed in the field. We estimated the purity of the five map
units, i.e. the areal fractions of the map units where the soil
compaction risk class as depicted on the map corresponds with the
risk class in the field
A second explanation could be a poor performance of the SOCOMO model. However,
comparisons between modelled and measured stresses showed good agreement
A third possible explanation is the lack of time for the natural recuperation of
subsoil compaction. Due to the intensive agricultural land use, the subsoil is
overloaded every second or third year, so considering a recuperation time of
about 10 years of the upper subsoil up to a depth of 40
The 47
About 43
The map of risk of subsoil compaction of
In terms of the subsoil compaction parameters relative bulk density and relative porosity and in terms of the areal fraction of overcompacted subsoil, only two risk classes and risk map units can be distinguished: low risk (risk classes/map units 1 and 2) and high risk (risk classes/map units 3, 4 and 5).
The lack of time for natural recuperation can be an explanation of the fact that, despite the good quality of the risk map in terms of map unit purity and class representation, no differences in subsoil compaction can be distinguished between the map units 3, 4 and 5.
Data are available in the Supplement.
The supplement related to this article is available online at:
The authors declare that they have no conflict of interest.
The authors acknowledge the Interprovinciaal Overleg (IPO, Inter-Provincial Consultation) for funding the research in the PRISMA program. Edited by: Olivier Evrard Reviewed by: two anonymous referees