The relationships between soil bulk density and penetration resistance, soil moisture, particle-size distribution, and soil organic carbon content were investigated for soddy-podzolic soils within the territory of the MSU Educational and Experimental Soil and Ecological Center (EE SEC) “Chashnikovo.” At six key sites with different land-use types, penetration resistance was measured using a penetrologger (a handheld electronic soil penetrometer with data logging) in areas adjacent to the walls of three soil pits. The same pits were used to collect soil samples for bulk density determination using the cutting-ring method with 100 cm3 rings, followed by the determination of soil moisture content, organic carbon content (Tyurin method), and particle-size distribution using a laser diffraction analyzer. A total of 24 regression models were developed to predict soil bulk density. The coefficients of determination (R2) ranged from 0.23 to 0.89, and the root mean square error (RMSE) ranged from 0.07 to 0.18 g cm–3. The modeling results show that organic carbon content contributed most strongly to bulk density prediction, followed by depth and penetration resistance. The inclusion of specific particle-size fractions in the models was more informative than the use of principal components derived from particle-size data as predictors. Field soil moisture was generally an insignificant predictor of bulk density in many models. Models with reduced predictor sets are proposed given the labor intensity associated with determining the full set of predictors: (1) penetration depth and resistance, (2) organic carbon content and penetration resistance, and (3) organic carbon content alone. Data for the first of these models are automatically collected during penetrologger operation, which makes it particularly suitable for monitoring surveys. Important limitations of the method are related to soil particle-size distribution and the method used for its determination. For example, in soils containing large stones or boulders, continuous measurement of penetration resistance from the surface downward is impossible, because the instrument encounters coarse fragments and cannot be inserted into the underlying horizons. The use of the proposed regression equations is therefore inappropriate when particle-size distribution is determined by methods other than laser diffraction. The choice of a specific particle-size fraction as a predictor should be based on correlation analysis for each individual study area.
Manakova et al. (Sun,) studied this question.