Comparison of conventional and geostatistical procedures for assessing hydraulic conductivity for drainage design

A comparative assessment of the conventional and geostatistical techniques of hydraulic conductivity estimates was made to know the likely improvement in the estimated value and to address the issue of intensity of investigations. A 10 ha plot in an alluvial soil was selected for which hydraulic conductivity measurements were available at 81 points. Three data sets representing 81, 42 and 21 observation points were used in the analysis. Arithmetic, geometric and the median values besides Russell's mean and weighted average have been calculated in case of conventional methods. The average value of krigged hydraulic conductivity has been estimated using geostatistical techniques. For the series studied, the weighted average from the conventional and geostatistical methods gave almost identical results. Compared with arithmetic mean, the maximum difference has been less than 10%. The results revealed that 21–41 samples may suffice to assess a reasonable value at 80% confidence with an error of 10%. However, weighted mean assessed through geostatistical or conventional methods proved that only 21 samples would yield a good estimate that matches well with the data series. Within the arithmetic mean the error with 21 points was less than 10%. It was concluded that the geostatistical techniques although theoretically sound, do not provide any major advantage over conventional techniques for assessing the hydraulic conductivity for the subsurface drainage design. Rather, it adds to the complexity of the design procedure. Within the conventional techniques, weighted average and Russell's mean may be preferred. Arithmetic mean, geometric mean or the median values may be used once it is possible to clearly identify the distribution of the data series.