چکیده
|
Abstract. Unsaturated soil hydraulic conductivity K is a fundamental transfer property of soil but its measurement is
costly, difficult, and time-consuming due to its large variations with water content (q) or matric potential (h). Recently,
C. Doussan and S. Ruy proposed a method/model using measurements of the electrical conductivity of soil core samples
to predict K(h). This method requires the measurement or the setting of a range of matric potentials h in the core
samples—a possible lengthy process requiring specialised devices. To avoid h estimation, we propose to simplify that
method by introducing the particle-size distribution (PSD) of the soil as a proxy for soil pore diameters and matric
potentials, with the Arya and Paris (AP) model. Tests of this simplified model (SM) with laboratory data on a broad range
of soils and using the AP model with available, previously defined parameters showed that the accuracy was lower for
the SM than for the original model (DR) in predicting K (RMSE of logK =1.10 for SM v. 0.30 for DR; K in ms–1).
However, accuracy was increased for SM when considering coarse- and medium-textured soils only (RMSE of
logK = 0.61 for SMv. 0.26 for DR). Further tests with 51 soils from the UNSODA database and our own measurements,
with estimated electrical properties, confirmed good agreement of the SM for coarse–medium-textured soils (<35–40%
clay). For these textures, the SM also performed well compared with the van Genuchten–Mualem model. Error analysis
of SM results and fitting of the AP parameter showed that most of the error for fine-textured soils came from poorer
adequacy of the AP model’s previously defined parameters for defining the water retention curve, whereas this was much
less so for coarse-textured soils. The SM, using readily accessible soil data, could be a relatively straightforward way to
estimate, in situ or in the laboratory, K(h) for coarse–medium-textured soils. This requires, however, a prior check of the
predictive
|