Spatial analysis of saturated hydraulic conductivity in a soil with macropores

Abstract

Saturated hydraulic conductivity (K_S) is an important soil hydraulic parameter for it establishes a limit on the rate of water and solute transmission through soil. However, its determination in the laboratory has been shown to be much influenced by column size. We evaluated the spatial variability of laboratory K_S measurements using three different column sizes: firstly, sixty 5.1 cm long columns of 5 cm diameter were used (type-I), next, thirty 20 cm long and 20 cm diameter columns were considered (type-II), and finally, thirty columns 100 cm long and of 30 cm diameter (type-III) were studied. All columns were collected along a transect in a sandy loam soil with macropores. Estimates of macroporosity at three depths (2.5, 12.5, and 16.5 cm) for twenty-four of the type-II columns were calculated from stained dye patterns obtained during ponded infiltration. The geometric mean of K_S decreased with increasing column size, i.e., from 2.24, 1.68 to 0.56 cm/h for type-I, -II, and -III columns, respectively. The coefficient of variation (CV) based on a log-normal distribution showed a similar trend: 619% for type-I, 217% for type-II, and 105% for type-III. Type-II and type-III columns were large enough to encompass a representative elementary volume (REV). The percentage of dye-staining (macropore cross-sectional area) decreased from 3% at 2.5 cm to 1.7% and 1.6% at 12.5 and 16.5 cm, respectively. Percentage of depth-averaged macropore area was moderately variable with CV = 51%. A geostatistical analysis revealed that a weak spatial structure existed for type-I K_S measurements whereas type-II and type-III columns displayed better spatial correlation with a range of approximately 14 m and 11 m, respectively. Spatial correlation was also observed for depth-averaged macropore area with a range of 12 m. The cross-semivariogram calculated between type-II K_S values and depth-averaged macropore area obtained from the same columns indicated positive spatial cross-correlation for all lags.