Open access publications in the SLU publication database

With an aim to develop an improved soil map of topsoil in Swedish arable land, Swedish Board of
Agriculture conducted a soil survey collecting soil samples in the topsoil of arable land in southern
and central Sweden. Eurofins and The Rural Economy and Agricultural Societies performed this soil
sampling in 2011 and 2012.
The soil samples were then analyzed by Eurofins for texture (i.e. clay, silt, sand), pH, P-AL, K-Al, FeAl,
Al-Al, Mg-Al, Ca-Al, soil type and the content of sol organic material.
The main purpose of the data processing was to compare the results from this data set with an older
data set (data from environmental monitoring program " The land and crop inventory "), and to get a
general picture of parameter variations with nutrient leaching regions as a base. Other important parts
of the study were to produce maps showing the spatial variation of analyzed parameters, and to develop
a new soil map of Swedish arable land.
A total of 12,598 points is included in the new data set. However, the number of soil samples with
existing values for different parameters can differ between different variables, as some parameters
may not be analyzed for all soil samples.
In summary, the differences between the two studied data sets were small, but some of these differences
were statistically significant, for certain production regions and / or for the country as a whole.
The most likely explanations differ for the various parameters and include differences in analysis
methodology (texture), small differences in the data base (humus content) and over time changing crop
management, such as lower phosphorus fertilization (P-AL).
Statistical evaluation of comparison between the new and the old soil map show that both user accuracy
(B) and coefficient of agreement (κ) are higher for the new map. Uncertainty in analytical precision
of laboratory texture determination was high, reaching for example, 20% for clay content and 30% for
sand content. However, the applied soil mapping methodology, where soils where grouped based on
their texture class, reduces the risk of wrongful textural classification and is limited to the nearest texture
classes in the texture triangle. In that way, the risks for wrongful determination of soil types that
are far from each other in the textural triangle (e.g. sand and clay) is low. Another improvement between
the old and the new map is the map resolution. The old map consists of a 10x10-km grid while
the new is produced in a 25 x 25-m grid. In summary, the statistical evaluation parameters (B and κ)
show that improvement of the soil map was possible with the new soil survey points, at the same time
as the resolution was increased. This is particularly important for local applications of the soil map.