Modelling of hydrological balance in monthly time step using the lumped hydrological model

Bc. Ba Doudou

Master's thesis

Modelling of hydrological balance in monthly time step using the lůumped hydrological model

Abstract:

This study investigated the calibration performance of hydrological mod-els applying series of split-sample to crash-test all potential combinationsof calibration-validation periods under drought type (dry/wet) usinglumped models: BILAN and GR2M. A sub-period focused on the droughtwas systematically selected based on a particular climate characteristic(precipitation, temperature, runoff) and a 7-year moving window. Thisapproach gives a perception into calibrated parameters transferabilityover time under similar or different climate conditions (drought). Theboth lumped models yielded similar results over a set of 6 catchmentsin a main West African river basin located in Senegal: the Gambia riverbasin. The Kling-Glupta Efficiency (KGE) was the objective function toassess models efficiency. A dependency was found then between themodel performance and the extent of input data. Results have shownthat the calibration performance decreases within an extending simula-tion period width. A focus on the impact of drought type on calibrationperformance revealed models simulating better Dry than wet years. Theanalyse on how model performance would be affected when calibrated ina climate condition different to the validation (e.g calibrated in dry(wet)and validated into wet (dry) revealed that calibration over a wetteror dryer condition than the validation and vice-versa may lead to anover(under)estimation of the simulated runoff. The results also indicatea general performance loss due to the transfer of calibrated parametersto independent validation periods of −5 to −25%, on average. Theshift of model parameters in time (validation) may generate significantlevel of errors. The outcome of this study may lead to a master ofthe uncertainty associated with one hydrological model and a betterassessment of runoff in real world application.

Abstract:

This study investigated the calibration performance of hydrological mod-els applying series of split-sample to crash-test all potential combinationsof calibration-validation periods under drought type (dry/wet) usinglumped models: BILAN and GR2M. A sub-period focused on the droughtwas systematically selected based on a particular climate characteristic(precipitation, temperature, runoff) and a 7-year moving window. Thisapproach gives a perception into calibrated parameters transferabilityover time under similar or different climate conditions (drought). Theboth lumped models yielded similar results over a set of 6 catchmentsin a main West African river basin located in Senegal: the Gambia riverbasin. The Kling-Glupta Efficiency (KGE) was the objective function toassess models efficiency. A dependency was found then between themodel performance and the extent of input data. Results have shownthat the calibration performance decreases within an extending simula-tion period width. A focus on the impact of drought type on calibrationperformance revealed models simulating better Dry than wet years. Theanalyse on how model performance would be affected when calibrated ina climate condition different to the validation (e.g calibrated in dry(wet)and validated into wet (dry) revealed that calibration over a wetteror dryer condition than the validation and vice-versa may lead to anover(under)estimation of the simulated runoff. The results also indicatea general performance loss due to the transfer of calibrated parametersto independent validation periods of −5 to −25%, on average. Theshift of model parameters in time (validation) may generate significantlevel of errors. The outcome of this study may lead to a master ofthe uncertainty associated with one hydrological model and a betterassessment of runoff in real world application.

Keywords

rainfall runoff model, lumped hydrological model, Gambia river basin, BILAN model, GR2M model, crach-test, Calibration

Keywords

lumped hydrological model, rainfall runoff modeling, Gambia river basin, BILAN model, GR2M model, crash-test, calibration

Language used: Czech

Date on which the thesis was submitted / produced: 29. 3. 2021

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