AN EXAMINATION OF THE GUELPH PERMEAMETER FOR MEASURING THE SOILS HYDRAULIC-PROPERTIES

The Guelph permeameter method (GPM) is a constant-head well permeamete r or shallow well pump-in technique for measuring soil hydraulic prope rties in the unsaturated zone, i.e., the field-saturated hydraulic con ductivity, K(fs), and the matric flux potential, phi(m). Unfortunately it is not uncommon to obtain negative K(fs)- and phi(m)-values with t he GPM, especially on heterogeneous soils. The objective of this study was therefore to thoroughly examine the GPM for different soil types over a range of initial soil water content conditions and to evaluate the existence of negative K(fs)- and phi(m)-values. Two permeameters o f different diameter were constructed to test the GPM. Three sets of 1 6 wells were prepared at sites of three selected soil types, viz., a N orfolk sandy loam, a Lucedale loam, and a Troup loamy sand, each set h aving a different well diameter ranging from 36 to 86 mm. Steady state flow in each well was measured at 4 water levels ranging from 50 to 1 25 mm. For the same wells the measurements were repeated for relativel y wet and relatively dry initial soil water content conditions. For ye t another site, K(fs)-values were compared with the saturated hydrauli c conductivity values (K(s)) obtained by the constant-head method appl ied to undisturbed soil cores. The in-situ measurements and the undist urbed cores were taken at a depth of 0.34 m at 60 locations on a 50-m x 100-m experimental field. Approximately 40% of the results obtained with the GPM were negative. This was attributed to random heterogeneit ies and systematic soil textural changes with depth. Fewer negative re sults were obtained for the coarser soil types. Positive values also s eemed to have been affected as indicated by extremely high values of t he exponential factor in the hydraulic conductivity-pressure head rela tion. A regression analysis on the K(fs)- and K(s)-values hardly showe d any correlation between the results of the GPM and the constant-head method. A co-kriging procedure, using clay content and water content data as covariates, to obtain K(s)-values for exactly the same locatio ns as where the K(fs)-values were obtained, gave a slightly higher reg ression coefficient, but it was still concluded that no significant co rrelation existed between the two methods.