A plot-scale uncertainty analysis of saturated hydraulic conductivity of a clay soil

Simulating soil hydrological processes at the plot or field scale requires using spatially representative values of the saturated soil hydraulic conductivity, K-s. Sampling campaigns should yield a reliable mean of K-s with a sustainable workload since measuring K-s at many points is challenging. Uncertainty analysis can be used to determine the lowest number of measurements that yield a mean K-s value with a specified accuracy level. Potential and limitations of this analysis were tested in this investigation for different extents of the sampled area and sampling densities. A clay soil was sampled intensively on two plots (plot area = 44 m(2)), two dates and using both small (0.15 m in diameter) and large (0.30 m) rings. With the small rings, intensively sampling an appropriate portion of the total plot area should be enough to establish the number of measurements yielding a certain accuracy level for the entire plot since this level remained nearly constant when the same number of measurements was performed on larger areas. Moreover, for these areas, the spatial resolution of the measurements did not influence appreciably the width of the confidence interval of the mean Ks value. However, working with larger rings was recommended since, in this case, the sampled area did not affect at all normalized confidence levels that, in addition, varied only a little with the number of the considered measurements of K-s. In practice, characterizing the plots required about 20 and 10 measurements with the smaller and the larger rings, respectively. The uncertainty analysis appears promising to plan practically sustainable soil sampling campaigns.

Automated summary

This study simulated soil hydrological processes at the plot or field scale, showing the importance of using spatially representative values of saturated soil hydraulic conductivity, K-s. Uncertainty analysis can help determine the minimum number of measurements needed to achieve a specified accuracy level for the mean K-s value.