Soil water retention function variability based on soil structure and moisture dynamics at field conditions affected by compaction in an Andosol

The variability of soil hydraulic properties across different spatial and temporal scales leads to heterogeneous sub-surface water flows, affecting the accuracy of predicting soil water distribution and solute transport. Since soils such as Andosols can describe extreme physical behaviours and water rarely is in hydraulic equilibrium in the porous media, it is still challenging to derive hydraulic functions that realistically represent the influence of soil structure dynamics under different land uses, as well as to predict the occurrence of non-uniform flows at field conditions. This work aimed to describe the spatiotemporal variability of unimodal and dual-porosity (bi-modal) soil water retention (SWR) functions using high-resolution field observations in structured soil affected by compaction. We focused on the influence of water-filled pores volumes at wet and dry conditions, wetting/drying cycles, and soil structure dynamics using three depths (10, 20, and 60 cm). The land use was a diverse grassland sown in September 2019, including three compaction levels (0.65, 0.75, and 0.85 g cm^-3, named Control, T1, and T2, respectively) in an Andosol of Southern Chile. A two-year 10-min-resolution dataset (June 2020 to June 2022) of soil moisture content (48 sensors) and matric potential (18 sensors) collected by 6 monitoring stations was analysed by i) separating wet and dry periods dynamics based on soil moisture states, ii) determining wetting and drying cycles using time derivatives of soil moisture content, and iii) fitting and comparing the parameters of unimodal and dual-porosity formulations of the Mualen-van Genuchten numerical solution. Separating soil moisture observations in wet and dry conditions, as well as in wetting and drying cycles, resulted in different SWR curves starting from contrasting water-filled pores volumes. This dynamic-based hydrological parameterisation resulted in a range of high goodness of fit (mean R² of 0.89 ± 0.07 and 0.94 ± 0.06 for unimodal and dual-porosity van Genuchten models) while deriving SWR functions. However, the dual-porosity formulation better represented complex curvatures in SWR curves towards the soil surface in wet conditions, which would increase our capacity to describe near-saturation macropore dynamics at high resolution. Thus field observations allowed the representation of expected spatial variability between soil depths due to different physical properties and compaction influence. While at the same time, high-resolution time series were used to describe significant different SWR curves for wet and dry conditions when soil structure is affected by compaction, mainly influencing α and n parameters in unimodal formulations, and n₁, α₂, and n₂ in dual-porosity formulations.