Modeling compaction effects on hydraulic properties of soils using limited information

Soil compaction leads to an increase in bulk density ( ρ b ) and a shift in the pore-size distribution towards smaller pores. This in turn changes the soil hydraulic properties (SHP), i.e., the water retention curve (WRC) and the hydraulic conductivity curve (HCC). Up to now, attempts to model the effect of altered ρ b on SHP has been limited to SHP models that account only for capillary water, neglecting water stored and transmitted in adsorbed films (non-capillary water). We combine a recently developed model for compaction effects on SHP with a SHP model system, which accounts for both capillary and non-capillary water (Peters-Durner-Iden model system - PDI). Due to a plausible course of the PDI-WRC towards oven dryness and a physically based prediction scheme of the PDI-HCC based on the WRC, the new combined approach can fully predict both soil hydraulic functions of compacted soils, even with limited information. The new approach is analyzed via a sensitivity analysis and tested with a large dataset from a silty arable soil. A comparison with an established prediction approach showed that our new approach has slightly better predictive performance within the measurement range and a more plausible course in the dry range. For our field data, the new approach performed best when only 2 of the water retention parameters were scaled based on the known ρ b . One of them determines the adsorptive water content and the other the shift of the capillary retention function on suction axis. Both parameters can be considered model independent, indicating that the new approach may not need calibration for each capillary retention model within the PDI model system. • We present a new scheme to predict compaction effects on soil hydraulic properties. • Description of soil hydraulic properties from saturation to oven dryness. • Model allows conductivity prediction of compacted soil without conductivity data. • First test with field data showed that only 2 out of 4 parameters needed scaling. • Better performance of new approach than established one.