Experimental evidence of randomness and nonuniqueness in an unsaturated outflow experiments designed for hydraulic parameter estimation

Abstract

Single transient outflow experiments are commonly conducted for inverse estimation of unsaturated hydraulic parameters. We assess the validity of this procedure through several repeated experiments on the same sample, a medium sand contained in a pressure cell. Outflow was induced by one or multiple step changes in botton boundary suction. We observed that experiments with small initial steps changes were poorly reproducible, even though our setup allowed are production of almost identical initial saturation for each run. Experiments with large step changes were well reproducible, but the outflow response was virtually the same for different steps levels. Neither type of observation was predicted by a theoretical sensitivity analysis of the Richards equation, given the minimal accuracy in our experiments. Inverse estimation yielded incompatible apparent hydraulic parameters for different flow conditions. Out results imply experimental limitations of the inversion procedure.