Premium
Robust Estimation of the Generalized Solute Transfer Function Parameters
Author(s) -
Javaux M.,
Vanclooster M.
Publication year - 2003
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj2003.8100
Subject(s) - reflectometry , transfer function , soil water , estimation theory , range (aeronautics) , log normal distribution , convection , dispersion (optics) , biological system , mathematics , soil science , time domain , mechanics , materials science , statistics , environmental science , physics , computer science , engineering , biology , electrical engineering , optics , composite material , computer vision
A method is presented to mathematically characterize a wide range of solute transport processes in soils from time‐domain‐reflectometry (TDR)‐based breakthrough curve (BTC) measurements. To do this, we combined the flexible generalized transfer function model (GTF) with the definition of the time normalized resident concentrations C rt * The GTF is a four‐parameter flexible transfer function able to describe both the convection‐dispersive (CD) and the stochastic‐convective (SC) process of dispersion in a soil. In addition, it allows other dispersion processes typical for heterogeneous soils to be modeled. To obtain robust estimations of the GTF model parameters, closed‐form expressions of the GTF time normalized resident concentrations C rt * and temporal moments of C rt * were defined. These expressions allow the transport parameters to be estimated from TDR‐based BTCs without relying on problematic probe calibrations and mass recovery assumptions. Since GTF is a four‐parameter model, the robustness of the parameterization procedure was tested by fitting the C rt * of the GTF model to numerically generated, error‐contaminated BTCs. Using the least‐square optimization technique, robust estimates of the GTF parameters could be obtained from TDR observations at two different soil depths. Application of the proposed method was also tested for a real undisturbed sandy subsoil and compared with the convective lognormal transfer (CLT) and CD models. For this case, it was shown that the GTF model improved greatly the goodness of prediction of the BTCs. The proposed method offers a new powerful tool for analyzing transport processes of nonreactive solutes in soil.