Automatic calibration of groundwater models using global optimization techniques

Abstract

The problem of a groundwater model calibration is posed as a multiextremum (global) optimization problem, rather than the more widely considered single-extremum (local) optimization problem. Several algorithms of randomized search incorporated in the global optimization tool GLOBE are considered (including the canonical genetic algorithm and more recently developed adaptive cluster covering), and applied to the calibration of the groundwater model TRIWACO. The results show the usefulness of global optimization algorithms in the automatic calibration of even complex models having considerable running times. Calage automatique de modèles d'écoulement souterrain utilisant des techniques d'optimisation globales Résumé Par opposition à l'approche classique qui considère le problème du calage des modèles d'écoulement souterrain comme un problème d'optimisation à un extremum (optimisation locale), nous posons ce problème ici en termes d'optimisation à extremums multiples (optimisation globale, OG). L'application de divers algorithmes de recherche aléatoire (recherche aléatoire contrôlée, algorithme génétique canonique et couverture adaptative) au modèle d'écoulement souterrain TRIWACO montre l'intérêt de l'utilisation des algorithmes OG pour le calage de modèles qui, en raison de leur complexité, exigent des temps de calcul considérables.

Parameter Optimization for Simulating Runoff from Highlatitude River Basins Using Land Surface Model and Global Data Sets

“…At the next stage, a new search of the optimum of the objective functions is performed for the reduced parameter space that allows one to reduce the number of realizations. This is especially important for a large set of optimized parameters, because if the feasible parameter space is fixed during optimization, the number of realizations needed to find the optimum with the specified accuracy grows exponentially with an increase in the number of parameters (Solomatine et al, 1999). If it is necessary, further reduction of parameter space may be done and searching the optimum may be continued until there will be no progress in minimization of Ext.…”

“…As it was shown there, when the objective function does not have an analytical expression (as in the present study), application of minimization techniques like a gradient search (Jacobs, 1977) is impossible. In this case, methods of direct search are usually used if Ext is a singleextremum function; otherwise, methods of global optimization are applied (Rosenbrock, 1960;Powell, 1964;Nelder & Mead, 1965;Solomatine et al, 1999;Duan, 2003). Many of them are based on the statistical methods of finding the extremum of Ext (vector of optimized parameters) (Rastrigin, 1968;Gupta et al, 1998;Solomatine et al, 1999).…”

“…In this case, methods of direct search are usually used if Ext is a singleextremum function; otherwise, methods of global optimization are applied (Rosenbrock, 1960;Powell, 1964;Nelder & Mead, 1965;Solomatine et al, 1999;Duan, 2003). Many of them are based on the statistical methods of finding the extremum of Ext (vector of optimized parameters) (Rastrigin, 1968;Gupta et al, 1998;Solomatine et al, 1999). It should be noted, that the method of blind random search in the parameter space with the pseudo-uniform distribution of points is n-times (where n is the total number of parameters) as effective as the method of search on the deterministic grid (Rastrigin, 1968).…”

Optimization of Hydraulic Parameters of Iranshahr Alluvial Aquifer

“…Two methods are commonly used for identification of mode parameters through calibration: trial and error method and automatic parameter estimation (Solomatine et al, 1999;Madsen, 2003). In the trial and error method, parameter values are assigned to the each node of the model and during the calibration these values has been adjusted, until the simulated values (head, discharge,...) are close to the observed ones.…”

Hydroinformatics: computational intelligence and technological developments in water science applications—Editorial

“…Some examples of early papers include the use of artificial neural networks (ANNs) for rainfall-runoff modelling (Minns & Hall, 1996;Dawson & Wilby, 1998;See & Openshaw, 1999;Hu et al, 2001;Rajurkar et al, 2002;Campolo et al, 2003), and the calibration of rainfall-runoff and groundwater models using genetic algorithms (Franchini, 1996) and other optimization methods (Solomatine et al, 1999). Since then the published contributions have continued to grow.…”