TY - JOUR

T1 - Generating a Danish raster-based topsoil property map combining choropleth maps and point information.

AU - Greve, Mogens H.

AU - Greve, Mette B.

AU - Bøcher, Peder K.

AU - Balstrøm, Thomas

AU - Breuning-Madsen, Henrik

AU - Krogh, Lars

PY - 2007

Y1 - 2007

N2 - The Danish environmental authorities have posed a soil type dependent restriction on the application of nitrogen. The official Danish soil map is a choropleth topsoil map classifying the agricultural land into eight classes. The use of the soil map has shown that the maps have serious classification flaws. The objective of this work is to compile a continuous national topsoil texture map to replace the old topsoil map. Approximately 45,000 point samples were interpolated using ordinary kriging in 250 m x 250 m cells. To reduce variability and to obtain more homogeneous strata, the samples were stratified according to landscape types. Five new soil texture maps were compiled; one for each of the five textural classes, and a new categorical soil type map was compiled using the old classification system. Both the old choropleth map and the new continuous soil maps were compared to 354 independent soil samples. 48% of the 354 independent samples fell into the correct class in the old map; in the new map 60% fell into the correct class, which is a significant improvement. The verification also showed that 62% of the samples in the forest areas were correctly classified, although these areas were not classified in the old map. Furthermore, when both the old and the new map were compared cell-by-cell, 74% of the cells were classified as belonging to the same class. The new textural maps were tested against the textural properties of 631 independent samples, and the root mean squared error (RMSE) of this comparison was calculated and found to be in the range of 2.8 to 5.2.

AB - The Danish environmental authorities have posed a soil type dependent restriction on the application of nitrogen. The official Danish soil map is a choropleth topsoil map classifying the agricultural land into eight classes. The use of the soil map has shown that the maps have serious classification flaws. The objective of this work is to compile a continuous national topsoil texture map to replace the old topsoil map. Approximately 45,000 point samples were interpolated using ordinary kriging in 250 m x 250 m cells. To reduce variability and to obtain more homogeneous strata, the samples were stratified according to landscape types. Five new soil texture maps were compiled; one for each of the five textural classes, and a new categorical soil type map was compiled using the old classification system. Both the old choropleth map and the new continuous soil maps were compared to 354 independent soil samples. 48% of the 354 independent samples fell into the correct class in the old map; in the new map 60% fell into the correct class, which is a significant improvement. The verification also showed that 62% of the samples in the forest areas were correctly classified, although these areas were not classified in the old map. Furthermore, when both the old and the new map were compared cell-by-cell, 74% of the cells were classified as belonging to the same class. The new textural maps were tested against the textural properties of 631 independent samples, and the root mean squared error (RMSE) of this comparison was calculated and found to be in the range of 2.8 to 5.2.

KW - Faculty of Science

KW - GIS

KW - Top soil map

KW - GIS

KW - kriging interpolation

KW - soil database

M3 - Journal article

VL - 107

SP - 1

EP - 12

JO - Geografisk Tidsskrift

JF - Geografisk Tidsskrift

SN - 0016-7223

IS - 2

ER -