Uncertainty Analysis in Rainfall-Runoff Modelling - Application of Machine Learning Techniques: UNESCO-IHE PhD Thesis, 1st Edition (Paperback) book cover

Uncertainty Analysis in Rainfall-Runoff Modelling - Application of Machine Learning Techniques

UNESCO-IHE PhD Thesis, 1st Edition

By Durga Lal Shrestha

CRC Press

224 pages

Purchasing Options:$ = USD
Paperback: 9780415565981
pub: 2010-01-15
$94.95
x
Hardback: 9781138424098
pub: 2017-07-20
$205.00
x


FREE Standard Shipping!

Description

This book describes the use of machine learning techniques to build predictive models of uncertainty with application to hydrological models, focusing mainly on the development and testing of two different models. The first focuses on parameter uncertainty analysis by emulating the results of Monte Carlo simulation of hydrological models using efficient machine learning techniques. The second method aims at modelling uncertainty by building an ensemble of specialized machine learning models on the basis of past hydrological model’s performance. The book then demonstrates the capacity of machine learning techniques for building accurate and efficient predictive models of uncertainty.

Table of Contents

SUMMARY

CHAPTER 1 INTRODUCTION

1.1 Background

1.2 Uncertainty analysis in rainfall-runoff modelling

1.3 Machine learning in uncertainty analysis

1.4 Objective of this study

1.5 Outline of the thesis

CHAPTER 2 UNCERTAINTY ANALYSIS IN RAINFALLRUNOFF MODELLING

2.1 Types of rainfall-runoff models

2.1.1 Data-driven models

2.1.2 Conceptual models

2.1.3 Physically based models

2.1.4 Stochastic models

2.2 Notion of uncertainty

2.3 Classification of uncertainty

2.4 Sources of uncertainty in rainfall-runoff models

2.5 Sources of uncertainty in context of data-driven modelling

2.6 Uncertainty analysis in rainfall-runoff modelling

2.7 Uncertainty representation

2.7.1 Probability theory

2.7.2 Fuzzy set theory

2.7.3 Entropy theory

2.8 Uncertainty analysis methods

2.8.1 Analytical methods

2.8.2 Approximation methods

2.8.3 Simulation and sampling-based methods

2.8.4 Bayesian methods

2.8.5 Methods based on the analysis of model errors

2.8.6 Fuzzy set theory-based methods

CHAPTER 3 MACHINE LEARNING TECHNIQUES

3.1 Introduction

3.2 Machine learning types

3.3 Learning principle and notations

3.4 Application of machine learning in rainfall-runoff modelling

3.5 Artificial neural networks

3.5.1 Learning in ANN

3.5.2 Multi-layer perceptron network

3.6 Model trees

3.7 Instance based learning

3.7.1 Locally weighted regression

3.8 Clustering methods

3.8.1 K-means clustering

3.8.2 Fuzzy C-means clustering

3.8.3 Validity measures

3.9 Selection of input variables

CHAPTER 4 MACHINE LEARNING IN PREDICTION OF PARAMETER UNCERTAINTY: MLUE METHOD

4.1 Introduction

4.2 Monte Carlo techniques for parameter uncertainty analysis

4.3 Problems attached to Monte Carlo based uncertainty analysis methods

4.4 Machine learning emulator in uncertainty analysis

4.5 Methodology

4.5.1 Monte Carlo simulations

4.5.2 Characterisation of uncertainty

4.5.3 Predictive model for emulating MC simulations

4.6 Selection of input variables

4.7 Validation of methodology

4.8 Limitations of method

CHAPTER 5 APPLICATION OF MACHINE LEARNING METHOD TO PREDICT PARAMETER UNCERTAINTY

5.1 Description of the Brue catchment

5.2 Description of rainfall-runoff model

5.3 Experimental setup

5.4 MC simulations and convergence analysis

5.5 Posterior distributions and sensitivity of parameters

5.6 Machine learning techniques in emulating results of MC simulations

5.6.1 Selection of input variables

5.6.2 Modelling prediction intervals

5.6.3 Modelling median and standard deviation

5.6.4 Modelling probability distribution function

5.7 Conclusions

CHAPTER 6 MACHINE LEARNING IN PREDICTION OF RESIDUAL UNCERTAINTY: UNEEC METHOD

6.1 Introduction

6.2 Definition and sources of model errors

6.3 Methodology

6.3.1 Clustering input data

6.3.2 Estimating probability distribution of model errors

6.3.3 Building model for probability distribution of model errors

6.4 Computation of predictive uncertainty of model output

6.5 Selection of input variables

6.6 Validation of UNEEC method

6.7 Limitations and possible extensions of method

CHAPTER 7 APPLICATION OF MACHINE LEARNING METHOD TO PREDICT RESIDUAL UNCERTAINTY

7.1 Application 1: Synthetic data set

7.1.1 Linear regression model of synthetic data set

7.1.2 Prediction interval for linear regression

7.1.3 Experimental setup

7.1.4 Results and discussions

7.1.5 Conclusions

7.2 Application 2: Sieve River catchment

7.2.1 Flow forecasting model of the Sieve River catchment

7.2.2 Experimental setup

7.2.3 Results and discussions

7.2.4 Uncertainty estimation

7.2.5 Conclusions

7.3 Application 3: Brue catchment

7.3.1 Analysis of simulation results

7.3.2 Selection of input variables

7.3.3 Clustering

7.3.4 Uncertainty results and discussions

7.3.5 Comparison of uncertainty results

7.3.6 Conclusions

7.4 Application 4: Bagmati catchment

7.4.1 Description of case study

7.4.2 Calibration of model parameters

7.4.3 Analysis of model residuals

7.4.4 Clustering

7.4.5 Selection of input variables for uncertainty model

7.4.6 Comparison of uncertainty results

7.4.7 Estimation of probability distribution of model errors

7.4.8 Conclusions

7.5 Multiobjective calibration and uncertainty

7.5.1 Preference ordering of the Pareto-optimal points

7.5.2 Results and discussions

7.5.3 Uncertainty assessment

7.6 Conclusions

CHAPTER 8 CONCLUSIONS AND RECOMMENDATIONS

8.1 Rainfall-runoff modelling and uncertainty analysis

8.2 Uncertainty analysis methods

8.3 Machine learning methods for uncertainty analysis

8.3.1 A method for parameter uncertainty analysis

8.3.2 A method for residual uncertainty analysis

8.4 Multiobjective calibration and uncertainty

8.5 Summary of conclusions

8.6 Recommendations

ABBREVIATIONS

NOTATIONS

REFERENCES

SAMENVATTING

ACKNOWLEDGEMENT

ABOUT THE AUTHOR

About the Author

Durga Lal Shrestha is a researcher in the Hydroinformatics and Knowledge Management Department of the UNESCO-IHE Institute for Water Education, Netherlands. He received his Masters degree in hydroinformatics from the UNESCO-IHE Institute for Water Education in 2002. His research interests include hydrological modelling, uncertainty analysis, global and evolutionary optimisation, machine learning techniques and their applications in water based systems.

Subject Categories

BISAC Subject Codes/Headings:
SCI019000
SCIENCE / Earth Sciences / General
SCI031000
SCIENCE / Earth Sciences / Geology
TEC010000
TECHNOLOGY & ENGINEERING / Environmental / General
TEC010030
TECHNOLOGY & ENGINEERING / Environmental / Water Supply