Comparison of three measurement methods of saturated hydraulic condutivity

Abstract. After pointing out the importance of the saturated hydraulic conductivity (k_s) measurements and the difficulties and uncertainties that are present, and after recalling salient aspects of three well-known measurement methods of this parameter (i.e. constant-head tension infiltrometer (TI) method, constant-head pressure infiltrometer (PI) method and soil core (SC) estimates method), the results of an investigation on data which were obtained during a measurement campaign on an area of 800 m², on a sandy loam hillslope, located in Southern Italy, were carried out again here. Three sets of values of k_s, obtained with these measurement methods, were analyzed statistically, verifying that the log-normal distribution describes these better than the normal one; moreover, the more significant statistical parameters of each set were compared (average value , amplitude A, coefficient of variation CV and standard deviation SD), individualizing the more significant differences. The greatest value of hydraulic conductivity was found with method (PI), while the smallest with (SC) and the intermediate with (TI); these differences were translated into macroporosity and into the influence of the single measurement method.

Moreover, referring to the possible factors affecting the results, the importance can be noted of the structure, the texture and the soil events, in terms of utilization, which can affect the measure of k_s leading often to very different values even for similar soils, but with a different history, independently of the coincidence of the measurement points and they can be determining to explain the differences affecting the results obtained in analogous investigations by other researchers.

Having confirmed that generalization is not possible, the need was emphasized to adopt the necessary devices relating to the specific measurement method, case by case, and to carefully explain the obtained results, in the light of the peculiarities and the limits of each situation.

Finally, the results of similar statistical analysis carried out on a greater number of k_s values, measured through the (TI) and (PI) methods are shown in this paper, with some statistical considerations on the increasing of the measurements number.