Modelling framework for representation of lumped hydrological models

Eleni Mickovska

Master's thesis

Abstract:

Hydrological modeling plays a central role in water resource manage- ment through access to risks and impacts of hydrological phenomena, where the model itself represents a simplified representation of a realworld system. The hydrological models are classified into many different groups depending on different factors. This study will be focusing on the performances of a conceptual lumped rainfall-runoff model. The main objective of this study is to extend and test an existing modular modelling framework, dHRUM which stands for Distributed Hydrological Response Unit Model, used for description of water balance in small and midsize catchments. This modelling framework was initially developed by the Department of Water Resources and Environmental Modelling at the Faculty of Environmental Sciences (Czech University of Life Sciences Prague) and furtherly extended with seven groundwater structures and three soil water structures. The analyzed time-series is the CAMELS (Catchment Attributes and MEteorology for Large-sample Studies) dataset. For measuring the quality of the model’s performance, calibration and validation were performed with the model using the differential split sample test, where calibration was applied for the dry period, and validation was applied for the wet period and vice versa. Three different single objective functions were used: KGE, NSE, MAE. The Differential Evolution algorithm was used as an optimization algorithm. The different structures were compared based on the three different goodness-of-fit criteria: KGE (Kling Gupta Efficiency), NSE (Nash Sutcliffe Efficiency), MAE (Mean Absolute Error) which were results from using the gof (goodness-of-fit) package in R. The results were very successfull when using the KGE objective function, as the validations were more successfull when the calibration was done on the wet period. An analysis of the different groundwater and soil-water structures were made, where ’superior’ and ’non-superior’ structures were chosen.

Keywords

Hydrological modelling, Differential split sample test, modelling framework, rainfall-runoff modelling, KGE, NSE, MAE, CAMELS dataset, DE

Language used: English

Date on which the thesis was submitted / produced: 31. 3. 2022

Thesis defence

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

Czech University of Life Sciences Prague

Faculty of Environmental Sciences

Master programme:
Environmental Modelling