HYDRUS: Model Use, Calibration, and Validation

Abstract

The HYDRUS numerical models are widely used for simulating water flow and solute transport in variably saturated soils and groundwater. Applications involve a broad range of steady-state or transient water flow, solute transport, and/or heat transfer problems. They include both short-term, one-dimensional laboratory column flow or transport simulations, as well as more complex, long-duration, multi-dimensional field studies. The HYDRUS models can be used for both direct problems when the initial and boundary conditions for all involved processes and corresponding model parameters are known, as well as inverse problems when some of the parameters need to be calibrated or estimated from observed data. The approach to model calibration and validation may vary widely depending upon the complexity of the application. Model calibration and inverse parameter estimation can be carried out using a relatively simple, gradient-based, local optimization approach based on the Marquardt-Levenberg method, which is directly implemented into the HYDRUS codes, or more complex global optimization methods, including genetic algorithms, which need to be run separately from HYDRUS. In this article, we provide a brief overview of the HYDRUS codes, discuss which HYDRUS parameters can be estimated using internally built optimization routines and which type of experimental data can be used for this, and review various calibration approaches that have been used in the literature in combination with the HYDRUS codes.