Estimating anthropogenic subsoil compaction in Germany using data-driven reciprocal modelling

Aim: Anthropogenic compaction is typically assumed to be major threat to soil health in agriculture. Compaction of the subsoil is considered irreversible and therefore more severe than topsoil compaction. To-date, quantitative estimates about the extent and severity of soil compaction are hardly available. This study aims to quantify anthropogenic subsoil compaction in German croplands by reanalyzing data from the German Agricultural Soil Inventory (BZE-LW). Grassland sites, which are assumed to exhibit negligible traffic-induced compaction below 30 cm depth, serve as a reference for the prediction of bulk density in cropland sites before anthropogenic compaction.

Methods: A data-driven reciprocal modelling approach is employed to estimate human-induced increases in bulk density at 1477 cropland sites scattered in a regular 8 x 8 km grid across Germany. The model is trained on data from ~400 grassland sites using information about soil texture, organic C content, soil pH, climate, and geological parent material. The model is then applied to the cropland sites to predict the bulk density of the upper subsoil in 30-50 cm depth prior to anthropogenic compaction. The disparity between modelled and observed bulk density represents the trafficking induced changes in soil compactness. To explain the drivers of this change, another data-driven model, incorporating soil and climate information as well as cropland management data, is trained and interpreted.

Results: Traffic-induced compaction has significantly increased the median bulk density of subsoils under cropland by 0.055 g cm⁻³, corresponding to a 4% increase. The modelled effects ranges from -0.07 g cm-3 to 0.180 g cm⁻³ (10th and 90th quantile), with the largest increases in subsoil compaction observed in eastern Germany. For the 20% most severely affected sites in Germany, the median increase in bulk density was 0.180 g cm⁻³ (0.142 g cm⁻³ – 0.267 g cm⁻³, 10th and 90th quantile), which corresponds to a 12% increase in subsoil bulk density. The anthropogenic increase in soil bulk density was most pronounced in loamy soils with relatively low soil organic carbon content.

Conclusion: This study represents a significant advancement in our ability to quantitatively assess the extent and severity of anthropogenic subsoil compaction at a national scale. The data-driven reciprocal modelling approach employed is promising for broad application in relation to soil health monitoring initiatives across Europe.