Pedotransfer Functions for Estimating Soil Bulk Density Using Image Analysis of Soil Structure

Soil bulk density is one of the most important soil properties. When bulk density cannot be measured by direct laboratory methods, prediction methods are used, e.g., pedotransfer functions (PTFs). However, existing PTFs have not yet incorporated information on soil structure although it determines soil bulk density. We aimed therefore at development of new PTFs for predicting soil bulk density using data on soil macrostructure obtained from image analysis. In the laboratory soil bulk density (BD), texture and total organic carbon were measured. On the basis of image analysis, soil macroporosity was evaluated to calculate bulk density by image analysis (BDim) and number of macropore cross-sections of diameter ≥5 mm was determined and classified (MP5). Then, we created PTFs that involve soil structure parameters, in the form BD~BDim + MP5 or BD~BDim. We also compared the proposed PTFs with selected existing ones. The proposed PTFs had mean prediction error from 0 to −0.02 Mg m−3, modelling efficiency of 0.17–0.39 and prediction coefficient of determination of 0.35–0.41. The proposed PTFs including MP5 better predicted boundary BDs, although the intermediate BD values were more scattered than for the existing PTFs. The observed relationships indicated the usefulness of image analysis data for assessing soil bulk density which enabled to develop new PTFs. The proposed models allow to obtain the bulk density when only images of the soil structure are available, without any other data.