Derivation of Pedotransfer Functions for Estimating Penetration Resistance, Aggregate Stability and Parameters of van Genuchten Moisture Curve Model in Fandoglou Forest Lands of Ardabil

Direct measurement of soil hydraulic and mechanical properties is time consuming, costly and sometimes unreliable due to soil heterogeneity and experimental errors. On the other hand, these properties can be estimated from readily available soil properties using pedotransfer functions (PTFs). This research describes regression based PTFs to predict dependent soil properties including penetration resistance (PR), mean weight diameter of aggregates (MWD), some points of soil water retention curve (SWRC), its slope at the inflection point (Sp), the van Genuchten a and n parameters from independent soil properties including sand, silt, clay, geometric mean (dg) and standard deviation (δg) of the soil particles diameter, relative bulk (Db-rel) and particle density (Dp), total porosity (n), water repellency (WR), initial water content (Pm), organic carbon (OC) and equivalent calcium carbonate (CaCO3). Totally, 75 soil samples were taken from Fandoglou forest lands of Ardabil, Iran. Then they were randomly divided in two sets for training (60 data) and testing (15 data) the derived models. The best PTF for estimating the saturated water content (qs) was obtained with input variables of OC, Db-rel and CaCO3 according to the coefficient of agreement (d), mean error (ME), root mean square error (RMSE) and bias-variance trade-off (BVTO) statistics. Also, the best PTFs for estimating the PR were obtained with input variables of Pm, Db rel, dg and CaCO3 and for MWD with input variables of clay to silt ratio, OC and Db-rel. The results showed that in the studied forest soil, 79 percent of the physical quality index (Sp) variations were determined by Db-rel and CaCO3 variables.