Hybrid models for Hydrological Forecasting: integration of data-driven and conceptual modelling techniques : UNESCO-IHE PhD Thesis

ISBN-13: 9780415565974 / Angielski / Miękka / 2009 / 228 str.

Perez Corzo

Opis

This text discusses possibilities and different architectures for integrating hydrological knowledge and conceptual models with data-driven models for use in hydrological flow forecasting. Models resulting from such integration are referred to as hybrid models.

Spis treści

Summary 1 Introduction1.1 Background1.2 Flood management and forecasting1.2.1 Flood management measures1.2.2 Operational flow forecasting1.3 Hydrological models1.3.1 Classification1.3.2 HBV process-based model1.4 Data-driven models1.5 Objectives of the research1.6 Terminology1.7 Outline 2 Framework for hybrid modeling2.1 Introduction2.2 General considerations and assumptions2.3 Hybrid modelling framework2.3.1 Classification of hybrid models2.3.2 Relationships between model classes2.4 Committee machines and modular models2.5 Measuring model performance2.6 Discussion and conclusions 3 Optimal modularization of data-driven models3.1 Introduction3.2 Methodology of modular modelling3.3 Modularization using clustering (MM1)3.4 Modularization using sub-process identification (MM2)3.5 Modularization using time-based partitioning (MM3)3.6 Modularization using spatial-based partitioning3.7 Optimal combination of modularization schemes3.8 Conclusions 4 Building data-driven hydrological models: data issues4.1 Introduction4.2 Case study (Ourthe river basin - Belgium)4.3 Procedure of data-driven modelling4.4 Preparing data and building a model4.5 The problem of input variables selection4.5.1 Inputs selection based on correlation analysis4.5.2 Selection based on Average Mutual Information (AMI)4.6 Influence of data partitioning4.7 Influence of ANN weight initialization4.7.1 Models not using past discharges as inputs (RR)4.7.2 Models using past discharges as inputs (RRQ)4.8 Various measures of model error4.9 Comparing the various types of models4.10 Discussion and conclusions 5 Time and process based modularization5.1 Introduction5.2 Catchment descriptions5.3 Input variable selection5.4 Comparison to benchmark models5.5 Modelling process5.6 Results and discussion5.7 Conclusions 6 Spatial-based hybrid modelling6.1 Introduction6.2 HBV-M model for Meuse river basin6.2.1 Characterisation of the Meuse River basin6.2.2 Data validation6.3 Methodology6.3.1 HBV-M model setup6.3.2 Scheme 1: Sub-basin model replacement6.3.3 Scheme 2: Integration of sub-basin models6.4 Application of Scheme 16.4.1 Inputs selection and data preparation for DDMs6.4.2 Data-driven sub-basin models6.4.3 Analysis of HBV-S simulation errors6.4.4 Replacements of sub-basin models by ANNs6.5 Application of Scheme 26.6 Discussion6.6.1 Scheme 16.6.2 Scheme 26.7 Conclusions 7 Hybrid parallel and sequential models7.1 Introduction7.2 Metodology and models setup7.2.1 Meuse river basin data and HBV model7.2.2 ANN model setup7.3 Data assimilation (error correction)7.4 Committee and ensemble models7.5 Forecasting scenario7.6 Results and discussion7.6.1 Single forecast results7.6.2 Results on multi step forecast7.7 Conclusions 8 Downscaling with modular models8.1 Introduction8.2 Fuzzy committee8.3 Case study: Beles River Basin, Ethiopia8.4 Beles River Basin8.5 Methodology8.5.1 ANN model setup8.5.2 Committee and modular models8.5.3 Fuzzy committee machine8.6 Results8.7 Conclusions 9 Conclusions and Recommendations9.1 Hybrid modelling9.2 Modular modelling9.3 Downscaling with modular models9.4 Parallel and serial modelling architectures9.5 Data-driven modelling9.6 Conclusion in brief BibliographyA State-Space to input-output transformationA.1 State-space and input-output modelsB Data-driven ModelsB.1 Arti¯cial Neural Networks (Multi-layer perceptron)B.2 Model Trees (M5P)B.3 Support Vector MachinesC Hourly forecast models in the MeuseC.1 MethodologyC.2 Neural network model (ANN)C.3 ResultsList of FiguresList of TablesList of acronymsSamenvattingAcknowledgementsAbout the author

O autorze

Gerald Corzo received his degree of civil engineering from the Escuela Colombian de Ingenieria, Faculty of Engineering (Bogotá, Colombia). He joined the Universidad Francisco de Paula Santander (UFPS), as a lecturer in Mathematics and Statistics (2000-2003). After graduating with the highest degree in his course he moved to Bogotá and served as lecturer in different universities. In October 2003 he moved to Delft in the Netherlands, there he joined the Hydroinformatics Master program at the UNESCO-IHE Institute for Water Education. He was awarded Master of Science in May 2005. His work explored the use of rule-based modelling and Committee of data-driven models in hydrological forecasting. In his thesis he showed a modular scheme approaching the incorporation of hydrological knowledge in a committee model. He continued with his work in the Phd, where he extended the analysis of artificial intelligence methods applied to hydrological phenomena.