A modified simulated annealing algorithm for estimating solute transport parameters in streams from tracer experiment data

It is difficult to estimate solute transport parameters in streams empirically and numerous approaches have been investigated in the past. In the current study, we explored the use of the simulated annealing (SA) algorithm for estimating solute transport parameters in streams from tracer experiment data. For a simple one-dimensional dispersion test case, the standard SA algorithm was very slow to converge. To remedy the problem of slow convergence of the annealing optimization, we proposed three strategies to modify the standard SA algorithm and improve the converging speed. The proposed three strategies are: (1) imposing parameter space constraints; (2) adding a valve value for inner loop break; and (3) including an inner loop memory function. We conducted a numerical experiment to test and demonstrate the effectiveness of the modified SA algorithm for estimating three major solute transport parameters: longitudinal dispersion coefficient, cross-sectional averaged flow velocity, and tracer mass loading. We then discussed the advantages and limitations of the proposed algorithm.