Harmonization of Soil Profile Data for 3D Spatial Prediction of Soil Properties in Denmark

Most environmental research which involve ecological and hydrological modelling require soil data as primary inputs to study flow, transport and other processes in the earth's critical zone. Information on the continuous vertical distribution of soil properties in a landscape could be very useful for such modelling activities. In our study, we harmonised soil texture and soil organic matter (SOM) profile data from the Danish soil database in order to map the vertical distribution of those properties to a depth of 1 meter from the soil surface. We fitted equal-area splines for texture and SOM data from about 2000 soil profiles and generated a continuous depth function of those properties. Digital soil mapping model based on data mining techniques was calibrated on the data and was used to predict the attribute values at un-sampled locations at a spatial resolution of 30m. Most of the terrain parameters for prediction were derived from the LiDAR based digital elevation model. These include elevation, slope, aspect, curvatures, topographic wetness index (TWI), stream power index (SPI), slope positions, vertical and horizontal overland flow distance, multi resolution indices of valley bottom and ridge top flatness (MRVBF & MRRTF) and altitude above channel network. Similarly, data on soil, geology and landscape types were extracted from the corresponding maps and land use information was derived from the corine land cover (CLC) data. Solar radiation, temperature, precipitation and evapo-transpiration data were also used specially in the prediction of SOM. The uncertainty of prediction was also accessed. The predicted maps showed the vertical distribution of clay, silt, sand and organic matter in Danish soils. We also believe these maps could be very useful for the farmers and decision makers especially in the sustainable use of soil resources in Denmark.