Investigation of preferential water flow in soil using developed kinematic dispersive wave- van Genuchten model: Study with global optimization analysis

These days the problem of water and soil pollution is one of the factors threatening the sustainability of agricultural production and human’s life and other living things. Also, preferential flow of water and solute is a common phenomenon in the natural saturated and unsaturated soil which generally results in fast contaminant transport and thus greatly increases the risk of groundwater contamination. So mathematical models are widely used in soil physics and hydrology for predicting preferential water flow and contaminants transport through the unsaturated zone. Preferential flow which is the cause of water transport in soil macropores such as underground channels formed by worm activity and root plants growth, is the reason of rapid water and contaminants transport to ground water and its contamination. For process predicting and describing of these types of water flow in soil, in this research the kinematic dispersive wave- van Genuchten model is introduced which is the innovation of this research.
In this research, the experiments were conducted with four different rainfall intensities of 56.97, 107.64, 133.01 and 161.71 mm h-1, which were applied on the surface of a soil column and output water fluxes from the bottom of soil column and mobile water content of whole soil column were recorded. Model coefficients were calculated by minimizing the error function between the observed values and something modeled by equation using particle swarm optimization (PSO) method. To achieve the best results and the minimum amount of error function, several solutions were tried and different values for c1 and c2 which are the learning factors (weights) or acceleration coefficients of optimization algorithm which interfere to make the next algorithm results and control the personal and global best respectively, were tried and chosen and also several equations as the inertia weight, w which used to control the particles/results velocities in the search spaces, were tried.
After applying several amounts for c1 and c2, finally the amount of 1.2 and 2.4 for c1 and c2 respectively, leads to best results and lowest error function. Also for the optimization, after reviewing the results of several different equations, the linear decreasing inertia weight equation which was presented by Xin et al., in 2009 was chosen (31). Based on results, in all rainfall intensities, optimization algorithm could find the best results after 3500 iteration and making frequent generation.
Generally the results have shown that the used algorithm could define the coefficients of kinematic dispersive wave- van Genuchten model in a short time and with reasonable accuracy.