Predicting storm surges : chaos, computational intelligence, data assimilation, ensembles

Front Cover......Page 1
Dedication......Page 4
Summary......Page 6
Table of Contents......Page 12
1.1 MOTIVATION: NATURAL DISASTERS......Page 26
1.2 MODELING NATURAL PHENOMENA: HYDROINFORMATICS......Page 28
1.3 PREDICTING STORM SURGES......Page 30
1.3.1 Physically-based modeling......Page 31
1.3.2 Data driven modeling: Nonlinear dynamics and chaos theory......Page 32
1.3.3 Main relations between the two modeling paradigms: chaotic modeling......Page 33
1.4 CHAOTIC BEHAVIORS IN OCEAN SURGE AND OTHER AQUATIC PHENOMENA......Page 34
1.5 MAIN OBJECTIVES......Page 35
1.6 THESIS OUTLINE......Page 37
2.1 STUDY AREA: THE NORTH SEA......Page 40
2.2.1 Ocean dynamics......Page 42
2.2.2 Tides and sea level......Page 43
2.3 STORM SURGE CONDITION IN THE NORTH SEA......Page 44
2.3.2 Procedure for issuing warnings and alarms......Page 46
2.4 DATA DESCRIPTION......Page 47
2.5 SUMMARY......Page 48
3.2.1 Ocean waves and its classification......Page 50
3.2.2 Tides......Page 54
3.3 SURGES......Page 56
3.4 SWAN WAVE SPECTRUM MODEL......Page 58
3.5 PHYSCIALLY-BASED STORM SURGE PREDICTION MODEL......Page 60
3.6.1 North West Shelf Operational Oceanographic System (NOOS)......Page 61
3.6.2 KNMI and RIKZ......Page 62
3.6.3 European Centre for Medium-Range Weather Predictions (ECMWF)......Page 67
3.7 LINKING PREDICTIVE CHAOTIC MODEL WITH EUROPEAN OPERATIONAL STORM SURGE MODELS......Page 68
3.8 SUMMARY......Page 70
4.1 INTRODUCTION......Page 72
4.2 ARTIFICIAL NEURAL NETWORKS......Page 75
4.2.1 Mathematical model of artificial neuron......Page 77
4.2.2 Learning methods......Page 78
4.2.3 Multi-layer perceptron and back-propagation algorithm......Page 80
4.2.4 Dynamic neural network......Page 82
4.3 INSTANCE-BASED LEARNING......Page 83
4.3.1 k-nearest neighbors learning......Page 84
4.3.3 Locally weighted regression......Page 85
4.4 HIERARCHICAL MODULAR MODELS......Page 86
4.5 EVOLUTIONARY AND OTHER RANDOMIZED SEARCH ALGORITHMS......Page 89
4.6 SUMMARY......Page 90
5.1 INTRODUCTION......Page 92
5.2.1 Dynamical system......Page 93
5.2.3 Various behaviors of dynamical system......Page 94
5.2.5 Chaos in Iterative Maps......Page 95
5.3.2 Return plot......Page 97
5.4 BIFURCATIONS......Page 98
5.5 NONLINEAR DYNAMICS IN DIFFERENTIAL EQUATIONS......Page 99
5.5.1 Sensitivity to initial conditions......Page 100
5.5.2 Properties of chaos......Page 101
5.6 PHASE SPACE RECONSTRUCTION – METHOD OF TIME DELAY......Page 102
5.7 FINDING APPROPRIATE TIME DELAY......Page 103
5.8.1 Self-similarity: Dimension......Page 104
5.8.3 Cao's method......Page 106
5.8.4 Kolmogorov-Sinai Entropy......Page 107
5.9 ANALYSIS OF STABILITY: LYAPUNOV EXPONENTS......Page 108
5.10 BUILDING CHAOTIC MODEL......Page 110
5.11 RECURRENCE PLOTS......Page 113
5.12 SUMMARY......Page 115
6.1 INTRODUCTION......Page 116
6.3 PHASE SPACE RECONSTRUCTION: FINDING TIME DELAY......Page 118
6.4 CORRELATION DIMENSION......Page 120
6.5 FALSE NEAREST NEIGHBORS......Page 121
6.6 CAO'S EMBEDDING DIMENSION......Page 122
6.7 SPACE-TIME SEPARATION......Page 123
6.9 POINCARÉ SECTIONS......Page 124
6.10 RECURRENCE PLOT......Page 125
6.11 PREDICTIVE CHAOTIC MODEL: GLOBAL AND LOCAL MODELING......Page 128
6.12.1 Univariate predictive chaotic model......Page 129
6.12.2 Multivariate predictive chaotic model......Page 132
6.13 MODEL RESULTS AND DISCUSSION......Page 134
6.14 K-FOLD CROSS VALIDATION......Page 137
6.15 SUMMARY......Page 140
7.1.1 Introduction......Page 142
7.1.2 Problems of dimensionality......Page 143
7.1.4 Reducing the phase space dimension......Page 144
7.1.5 Model results and discussion......Page 145
7.2.1 Introduction......Page 147
7.2.2 Data description......Page 148
7.2.3 Setting up the 1st standard predictive chaotic model......Page 149
7.2.4 Setting up the 2nd model (predictive chaotic model and ANN model......Page 151
7.2.5 Model results and discussion......Page 152
7.3.1 Introduction......Page 153
7.3.4 Cubic spline interpolation......Page 155
7.3.5 Model results and discussion......Page 156
7.4.1 Euclidean distance method......Page 158
7.4.2 The new trajectory based method......Page 159
7.4.3 Model results and discussion......Page 161
7.5 SUMMARY......Page 163
8.1 INTRODUCTION......Page 166
8.2.2 Genetic algorithm (GA)......Page 168
8.2.3 Adaptive cluster covering algorithm (ACCO)......Page 170
8.3 CASE STUDY......Page 171
8.4.1 Main experiment: predictive model for Hoek van Holland......Page 172
8.4.2 Additional experiment: predictive model for the San Juan station......Page 173
8.5 MODEL RESULTS AND DISCUSSION......Page 175
8.6 SUMMARY......Page 178
9.1 INTRODUCTION......Page 180
9.2.2 Learning Algorithm......Page 183
9.3 NARX DATA ASSIMILATION......Page 184
9.5.1 Estimating delay time and embedding dimension......Page 186
9.5.3 Chaotic storm surge models......Page 189
9.5.4 Data assimilation using NARX neural network......Page 190
9.6 SUMMARY......Page 191
10.2 PRINCIPLES OF ENSEMBLE MODEL PREDICTION......Page 194
10.2.1 Information-theoretic model selection......Page 195
10.2.2 Bayesian model averaging......Page 196
10.2.4 Machine learning: modular model......Page 199
10.3 LINEAR PREDICTION COMBINATION......Page 200
10.4.1 Dynamic averaging......Page 201
10.4.2 Dynamic neural networks......Page 202
10.5.2 Local model......Page 203
10.5.4 Dynamic neural networks......Page 205
10.6 SUMMARY......Page 206
11.1 MAIN CONCLUSIONS......Page 208
11.2 LIMITATIONS AND RECOMMENDATIONS......Page 212
REFERENCES......Page 216
ABOUT THE AUTHOR......Page 226
SCIENTIFIC PUBLICATIONS......Page 228
SAMENVATTING......Page 232