Comparison of saturated hydraulic conductivity estimated by surface NMR and empirical equations

Abstract

This study used eight empirical equations such as Kozeny-Carman (K-C), Terzaghi, Hazen, Slitcher, Kruger, Sauerbrei, NAVFACDM7, and Zamarin to estimate saturated hydraulic conductivity from soil particle size distribution and surface NMR porosity. The estimated empirical hydraulic conductivities were compared with the saturated surface nuclear magnetic resonance (surface NMR) hydraulic conductivity. It was found that most of the empirical equations showed poor correlation with the surface NMR hydraulic conductivity except Kozeny-Carman (KC), Hazen, and Zamarin equations. The value of the coefficient (β) of each equation is modified in such a way as to obtain a linear equation with the unit slope to improve the performance and accuracy of empirical equations. The modified Kozeny-Carman equation showed the maximum correlation R2=0.904, Pearson's r=0.951, and RMSE =6.36 in comparison with other empirical equations. The NAVFACDM7 and the Terzaghi equations showed the highest underestimated hydraulic conductivity data (64%) and overestimated hydraulic conductivity data (38%); however, the Zamarin equation showed the highest overlapped hydraulic conductivity data (44%), respectively, with the 1:1 line.