Evaluation of pedotransfer functions to predict saturated hydraulic conductivity of Ukrainian soils

Abstract

Literature overview. The parameterization of hydrological models requires knowledge of the soil filtration properties. Generally, soil profiles are characterized by properties such as sand, silt and clay content, bulk density, organic carbon fraction or humus content, and no data on filtration properties are available. Ukrainian soil database, created in Geoecophysics of soil laboratory of National Scientific Center “Institute for Soil Science and Agrochemistry Researched named after O.N. Sokolovsky” (Laktionova et al., 2012), among other properties has extensive data on texture and bulk density for more than 2000 profiles, less on organic carbon content, and almost no data on saturated hydraulic conductivity (K_sat). The most probable ranges of K_sat for most types of Ukrainian soils are given in the Atlas of natural conditions and natural resources of the Ukrainian SSR (“Pochvenno-meliorativnoye rayonirovaniye. Masshtab 1:4000000,” 1978), however, the data doesn’t present K_sat for different textures inside one soil type. To fill this gap, the best solution is the applying of pedotransfer function (PTF). 

The purpose of this work is to synthesize the most realistic K_sat of the main soil groups of Ukraine, corresponding to a scale map of 1:2 500 000 (Krupskiy, 1977), as well as their genetic horizons, on the basis of calculated and experimental values available in the literature.

Material and methods. Ten PTFs used in the study are based on regression equations (Cosby et al., 1984; Saxton & Rawls, 2006; Weynants et al., 2009; Wösten et al., 1999), decision tree (Tóth et al., 2015), or neural network (Zhang & Schaap, 2017). K_sat was estimated for 942 horizons of 171 profiles which represented all 40 soil groups (corresponding to the legend of 1:2 500 000 map) of Ukraine according to Dokuchaev classification. 

Results. Wösten and Rosetta3 PTFs are determined as the most relevant by comparing the calculated K_sat values with the available data of the bottom (horizons A₂, B, C) and top (A₀, A₁) soil layers of Ukraine. In particular, they are relevant for widespread soils such as Soddy podzolic soils (WRB – Eutric podzoluvisols), dark gray podzolized soils (Phaeozems Albic), chernozems podzolized (Chernozems Chernic), chernozems southern (Chernozems Calcic), meadow-chernozemic soils (Phaeozems Haplic), dark chestnut and chestnut soils (Kastanozems Haplic and Kastanozems Luvic), meadow soils (Umbrisols Gleic, Fluvisols Dystrict, Fluvisols Eutryc, Leptosols Umbric), mountain soils (Cambisols), and top layer of Chernozems ordinary (Chernozems Chernic). Unfortunately, all ten PTFs underestimate 2-4 times K_sat of bottom layer of ordinary and typical chernozems (Chernozems Chernic) and overestimate 2-5 times for relatively impermeable horizons (< 2 mm/h).

Conclusions. Based on the calculated and experimental values, the map of K_sat of the top and bottom soil layers was obtained. Sandy soils, common in Polissia, have the highest filtration rate. K_sat of loam and clay soils of forest-steppe and steppe can differ between different types by an order. The highest K_sat have soils with high structural properties (Chernozems Luvic, Chernozems Chernic). The lowest K_sat (0.2-3 mm/h) have Phaeozems Sodic, Solonetz, Solonchaks, Planosols Albic, and bottom layer of soddy manly gley (Arenosols Protic/ Haplic) and loamy soddy podzolic soils (Albeluvisols Umbric). The estimated values should be considered as the most probable because K_sat depends on landscape location of soil profile, tillage operations, and soil temperate. The results are acceptable to use in hydrological calculations and modeling.