Comparison of several interpolators for smoothing hydraulic conductivity data in South West Iran
1993
Hosseini, E. | Gallichand, J. | Caron, J.
Hydraulic conductivity is an important factor in the design of subsurface drainage systems. The high spatial variability of hydraulic conductivity prevents the identification of homogeneous zones to which uniform drain spacing can be assigned. This study was conducted to determine the suitability of several interpolators for smoothing hydraulic conductivity data. Interpolation techniques used were ordinary and universal kriging, trend surface, moving average, and distance weighted moving average. The precision of interpolation was measured by cross-validation using the mean absolute error (MAE) between interpolated and observed values. The data set consisted of 383 auger hole hydraulic conductivity measurements from a 16,000 ha area in South West Iran. The semi-variogram was characterized by a range of influence of 4.5 km and a nugget effect corresponding to 67% of the sill. Values of MAE ranged from 1.373 to 2.159 m d-1 depending on the method and the number of neighbors used in the interpolation. The best methods were ordinary kriging, universal kriging with a linear drift, and squared distance weighted moving average, for which the MAE values were 1.373, 1.383, and 1.392 m d-1, respectively. Although the MAE values of these three methods were similar, the interpolated surfaces were very different. Because of its precision of interpolation and the smoothness of the interpolated surface, ordinary kriging was the most appropriate method for smoothing hydraulic conductivity data in this area.
显示更多 [+] 显示较少 [-]