The extension of optimization algorithm of Bilan model

Son Hoang Trung

Master's thesis

Abstract:

The Bilan model is a famous Czech hydrological model developed by the T. G. Masaryka Water Research Institute (WRI) in 15 years. Water balance model that can be used to simulate impacts of environmental change on hydrology. The structure of the model is formed by a system of relationships describing the principle of water balance on the land surface, the zone of aeration, including the effect of vegetation cover and groundwater (Kašpárek & Novický, 1997; Vizina et al., 2010). The model has eight free model parameters and uses four optimization algorithms for their calibration using gauged streamflow. The optimization aims at attaining the best fit between the observed and the runoff series, for which five optimizations criteria are available. The Bilan model continues to evolve, with developers focused primarily on increasing the speed of computation and model accuracy, but optimization algorithms are unlikely to change. The model still uses five optimization criteria: "MSE" (mean squared error), "MAE" (mean absolute error), "NS" (Nash-Sutcliffe efficiency), "LNNS" (Nash-Sutcliffe of log-transformed data), "MAPE" (mean absolute percentage error). This thesis discusses a new developmental trend for the Bilan model, which is to build a new optimization function based on the Kling-Gupta coefficient (Gupta H.V., 2009). For testing the accuracy of the new Bilan model, I used the KGE optimization function to simulate the total runoff of 57 catchments across the Czech Republic. The results are then compared to the default optimization functions of the previous Bilan model, and are evaluated using the reference as results of the WRI center provided for each catchment. This paper demonstrates the feasibility of extending the optimization algorithm for the Bilan model using the Kling-Gupta coefficient for calibration.

Keywords

Bilan model, Kling-Gupta coefficient, conceptual hydrological model, hydroGOF

Language used: English

Date on which the thesis was submitted / produced: 18. 4. 2017

Thesis defence

• Supervisor: doc. Ing. Petr Máca, Ph.D.

Czech University of Life Sciences Prague

Faculty of Environmental Sciences

Master programme / field:
Landscape Engineering / Environmental Modelling