Pedotransfer Functions for Estimating Saturated Hydraulic Conductivity of Selected Benchmark Soils in Ghana

Aims: Direct methods of measuring saturated hydraulic conductivity (Ks), either in situ or in the laboratory, are time consuming and very expensive. Several Pedotransfer functions (PTFs) are available for estimating Ks, with each having its own limitations. In this study, the performances of four popular PTFs were evaluated on different soil classes in the semi deciduous zone of Ghana. The PTFs considered herein were Puckett et al. (1985), Campbell and Shiozawa (1994), Dane and Puckett (1994), and Ferrer-Julià et al. (2004). In addition, five local data derived PTFs were used to study the possibility of using local datasets to validate PTF accuracy.

Materials and Methods: A total of 450 undisturbed soil cores were collected from the 0 – 15 cm depth from three benchmark soils, namely, Stagni-Dystric Gleysol (SDG), Plinthi Ferric Acrisol (PFA) and Plinthic Acrisol (PA). The Ks of samples were measured by the falling-head permeameter method in the laboratory. Sand, silt and clay fractions, bulk density, organic matter content, and exchangeable calcium and sodium were measured and used as input parameters for the newly derived PTFs. Accuracy and reliability of the predictions were evaluated by the root mean square error (RMSE), coefficient of correlation (r), index of agreement (d), and the Nash-Sutcliffe efficiency (NSE) between the measured and predicted values from both tested and newly derived PTFs. The relative improvement (RI) of the newly derived PTFs from this study over the existing ones were also evaluated.

Results: The newly derived PTFs in this study had higher prediction accuracy with r, d, RMSE and NSE ranging from 0.80 – 0.99, 0.79 – 0.94, 0.14 – 1.74 and 0.84 – 0.98, respectively, compared with 0.32 – 0.45, 0.27 – 0.50, 4.00 – 4.90 and 0.41 – 0.47 for the tested PTFs. The relative improvement of the newly derived over the tested PTFs ranged from 56.50 – 95.71% in the SDG, 70.73 – 96.89% in the PFA, and 65.37 – 95.81% in the PA. Generally, RI was observed to be highest for Model 1 in the SDG, and Model 4 in both PFA and PA, and lowest for Model 5 in all three soils. It was observed that the inclusion of exchangeable calcium and sodium as predictors increased the predictability of the newly derived PTFs.