Les méthodes de proximité spatiale

Table des matières

CHAPITRE 1 INTRODUCTION GÉNÉRALE
1.1 Problématique de la thèse
1.2 Organisation de la thèse
CHAPITRE 2 REVUE DE LITTÉRATURE
2.1 Prévision des apports aux sites non-jaugés
2.1.1 Régression linéaire multiple
2.1.2 Proximité spatiale
2.1.3 Similitude physique
2.1.4 Autres méthodes
2.2 Donneurs multiples
2.3 Analyse des approches de régionalisation
2.4Problèmes constatés en régionalisation
2.5 Solutions potentielles
2.5.1Amélioration du calage
2.5.2 Réduction du nombre de paramètres
2.5.3 Modélisation multi-modèle
CHAPITRE 3 ARTICLE 1 : A COMPARISON OF STOCHASTIC OPTIMIZATION ALGORITHMS IN HYDROLOGICAL MODEL CALIBRATION
3.1Introduction
3.2 Optimization algorithms used in the study
3.2.1Adaptive Simulated Annealing (ASA
3.2.2Covariance Matrix Adaptation Evolution Strategy (CMAES
3.2.3 Cuckoo Search (CS)
3.2.4 Dynamically dimensioned search (DDS)
3.2.5Differential Evolution (DE
3.2.6Genetic Algorithm (GA
3.2.7 Harmony Search (HS)
3.2.8Pattern Search (PS
3.2.9 Particle Swarm Optimization (PSO)
3.2.10Shuffled Complex Evolution – University of Arizona (SCE-UA
3.3Models, study area and data
3.3.1 Hydrologic models
3.3.2 Basins
3.4 Benchmarking of the optimization methods
3.5Results
3.5.1Algorithm performance based on ranks
3.5.2Algorithm performance based on convergence speed
3.5.3 Statistical significance tests
3.5.4 Dispersion Metric
3.6 Discussion
3.6.1 On overall performance
3.6.2 On model complexity
3.6.3On the effect of the basin on algorithm performance
3.6.4On convergence speed
3.6.5 On computing power
3.7 Conclusion
3.8Acknowledgements
3.9References
CHAPITRE 4 ARTICLE 2 : A COMPARATIVE ANALYSIS OF 9 MULTI-MODEL AVERAGING APPROACHES IN HYDROLOGICAL CONTINUOUS STREAMFLOW SIMULATION
4.1Introduction
4.2Data, models and multi-model averaging methods
4.2.1Basins, hydrometric and climate data
4.2.2 Hydrological models
4.2.3Multi-model averaging methods
4.2.4Model calibration
4.2.5 Multi-model averaging application
4.3Results
4.3.1Performance of the 15 ensemble members
4.3.2 Performance of the multi-model averaging methods
4.3.3Performance gain quantification
4.3.4Geographical analysis
4.4 Discussion
4.4.1 Individual model performance
4.4.2 Multi-model averaging method analysis
4.4.3 Member contribution in multi-model averaging
4.4.4Geographic analysis
4.5 Conclusion
4.6Acknowledgements
4.7References
CHAPITRE 5 ARTICLE 3 : IMPROVING HYDROLOGICAL MODEL SIMULATIONS USING MULTIPLE GRIDDED CLIMATE DATASETS IN MULTI-MODEL AND MULTI-INPUT AVERAGING FRAMEWORKS
5.1Introduction
5.2Catchments and data
5.3 Models and Methodology
5.3.1 Hydrological models
5.3.2Model parameter calibration process
5.3.3 Model averaging technique
5.3.4Multi-model and multi-input averaging application
5.4Results
5.5 Discussion
5.5.1 Results analysis
5.5.2 Pre-averaging of climate data
5.5.3 Possible further improvements
5.6 Conclusion
5.7Acknowledgements
5.8References
CHAPITRE 6 ARTICLE 4 : CONTINUOUS STREAMFLOW PREDICTION IN UNGAUGED BASINS : THE EFFECTS OF EQUIFINALITY AND PARAMETER SET SELECTION ON UNCERTAINTY IN REGIONALIZATION APPROACHES
6.1Introduction
6.1.1 Equifinality
6.2Scope and aims
6.3 Study area and data
6.3.1 Study area
6.3.2 Meteorological and hydrological datasets
6.4 Methodology
6.4.1 HSAMI hydrological model and calibration
6.4.2 Uncertainty analysis
6.4.3 Generalities common to all regionalization methods
6.4.4Multiple linear regression regionalization method
6.4.5 Physical similarity regionalization method
6.4.6 Spatial proximity regionalization method
6.4.7Regression-augmented approach
6.5Results
6.5.1 Regression-based approach
6.5.2Physical similarity approach
6.5.3 Spatial proximity
6.5.4 Inverse distance weighting
6.5.5 Regression-augmented
6.5.6Inter-method comparison
6.5.7Success Rate vs. Nash-Sutcliffe Efficiency
6.5.8Hydrograph analysis
6.6 Discussion
6.6.1 Number of donor catchments
6.6.2 Regionalization methods analysis
6.6.3 Comparison with other studies
6.6.4 Parameter set selection uncertainty
6.6.5 Type I errors in hypothesis testing
6.7Conclusions
6.8Acknowledgments
6.9References
CHAPITRE 7 ARTICLE 5 : MULTI-MODEL AVERAGING FOR CONTINUOUS STREAMFLOW PREDICTION IN UNGAUGED BASINS
7.1Introduction
7.1.1 Multi-model averaging
7.1.2 Multi-model averaging in regionalization
7.1.3 Averaging methods description
7.2Models, study area and data
7.2.1 Hydrological models
7.2.2 Study area
7.2.3 Meteorological and hydrological datasets
7.3 Methodology
7.3.1Model calibration
7.3.2 Donor basin selection scheme
7.3.3Model averaging strategies
7.3.4Multi-donor averaging
7.4Results
7.4.1 Initial model calibration and weighting method evaluation
7.4.2Regionalization under the multi-model averaging framework
7.4.3 Weights distribution
7.5 Analysis and discussion
7.5.1 Overview of model averaging methods performances
7.5.2 Multi-model averaging in regionalization
7.5.3Model robustness
7.5.4Multi-donor aspect
7.6Conclusions
7.7Acknowledgments
7.8References
CHAPITRE 8 ARTICLE 6 : ANALYSIS OF CONTINUOUS STREAMFLOW REGIONALIZATION METHODS USING A REGIONAL CLIMATE MODEL ENVIRONMENT FRAMEWORK
8.1Introduction
8.2Data and Methodology
8.2.1 Description of the virtual-world setting
8.2.2 Meteorological data
8.2.3 Virtual-world setting hydrometric data
8.2.4 Catchment descriptors
8.2.5 HSAMI hydrological model
8.2.6Model Calibration
8.2.7 Regionalization methods
8.2.8Methodology
8.3Results
8.4Analysis
8.4.1Real world and CRCM environment
8.4.2 Analysis of the methods performance
8.4.3 Evaluation metrics and donor quality analysis
8.4.4Predicting probability of success
8.5 Conclusion
8.6Acknowledgements
8.7References
CHAPITRE 9 ARTICLE 7 : PARAMETER DIMENSIONALITY REDUCTION OF A CONCEPTUAL MODEL FOR STREAMFLOW PREDICTION IN UNGAUGED BASINS
9.1Introduction
9.2Scope and aims
9.3 Study area and data
9.3.1 Meteorological and hydrological datasets
9.4 Methodology
9.4.1 Hydrological models
9.4.2Model calibration
9.4.3 Sobol’ Global sensitivity analysis
9.4.4Sequential model parameter fixing and recalibration
9.4.5 Regionalization methods application
9.5Results
9.5.1 Model calibration performance
9.5.2Regionalization application results
9.5.3 Robustness evaluation
9.6 Discussion
9.6.1 Verification of the main hypothesis
9.6.2 Sobol’ sensitivity analysis
9.6.3 Parameter fixing
9.6.4 Regionalization performance
9.7Conclusions
9.8 Acknowledgments
9.9 References
CHAPITRE 10 DISCUSSION GÉNÉRALE
10.1Analyse de l’équifinalité en régionalisation
10.2Caractéristiques physiques des bassins versants et paramètres des modèles
10.3Comparaisons entre le monde réel et le monde virtuel
10.4 Analyse des appoches multi-modèle
10.4.1Approches multi-modèle en régionalisation
CONCLUSION ET CONTRIBUTIONS
RECOMMANDATIONS
ANNEXE I ARTICLE EN COLLABORATION 1 : POTENTIAL OF GRIDDED DATA AS INPUTS TO HYDROLOGICAL MODELING
ANNEXE II ARTICLE EN COLLABORATION 2 : REDUCING THE
PARAMETRIC DIMENSIONALITY FOR RAINFALL-RUNOFF
MODELS : A BENCHMARK FOR SENSITIVITY ANALYSIS METHODS
ANNEXE III LISTE DES CONTRIBUTIONS SCIENTIFIQUES
LISTE DE RÉFÉRENCES BIBLIOGRAPHIQUES