EVALUATION OF POPULATION BASED EVOLUTIONARY OPTIMIZATIONALGORITHMS IN THE CONCEPTUAL HYDROLOGICAL MODEL CALIBRATION

Abstract

The conceptual hydrological models are generally deterministic and lumped, and these models are based on water budget equations and provide the identification of hydrological cycle elements by means of different parameters. The usage of these models in water resources engineering is rather crucial. These models are used in a variety of areas such as explaining rainfall-runoff relationships of the basin, simulating flows for basins without observation, and analyzing the possible effects of climate change on streamflows. The capability of conceptual models in representing rainfall-runoff relationship of a basin depends on accurate estimation of flow at the outlet of the basin. Hence, it is realized by calibrating of parameters related to hydrological model. This process transforms into an optimization problem based upon determining parameters that minimize the errors between model flows and observed flows. As the number of hydrological model parameters increases, it becomes more difficult to perform the optimization process with manual methods. For this reason, the usage of global optimization algorithms can increase the reliability of the predicted parameters. In the study, some population based evolutionary algorithms were chosen from different optimization techniques and they were assessed in the calibration of a five-parameter hydrological model termed dynamic water balance model. The utilized algorithms mimic some nature inspired principles such as swarm intelligence and the survival of the best within the numerical algorithms. In the study, convergence performances of genetic algorithm (GA), particle swarm optimization (PSO), differential evolution algorithm (DEA), invasive weed algorithm (IWA), and artificial bee colony (ABC) algorithm were compared and it was questioned which algorithms could be preferred in the hydrological model calibration.