Handbook of HydroInformatics

Front Cover
Handbook of HydroInformatics: Volume II: Advanced Machine Learning Techniques
Copyright
To Late George Edward Pelham Box (British Statistician: 1919-2013)
Contents
Contributors
About the Editors
Preface
Chapter 1: Analyzing spatiotemporal variation of land use and land cover data
	1. Introduction
	2. Data preparation
	3. Visual interpretations
	4. LULC distribution
	5. LULC change detection
	6. Image interpretation
	7. LAI model
	8. Compare the visual interpretation vs image interpretation
	9. Conclusions
	References
Chapter 2: Artificial Intelligence-based model fusion approach in hydroclimatic studies
	1. Introduction
	2. Mathematical concepts
		2.1. Ensemble techniques
			2.1.1. Bagging
			2.1.2. Boosting
			2.1.3. Random forest
			2.1.4. Extremely randomized trees
			2.1.5. Extreme gradient boosting
			2.1.6. Simple linear averaging method
			2.1.7. Linear weighted averaging method
			2.1.8. Bayesian model averaging method
			2.1.9. Stacking (nonlinear ensemble method)
		2.2. Hybrid techniques
			2.2.1. Wavelet-AI methods
			2.2.2. ARIMA-AI methods
			2.2.3. Clustering-based AI methods
			2.2.4. Evolutionary-based AI methods
	3. Some applications
		3.1. Ensemble techniques
		3.2. Hybrid techniques
	4. Conclusions
	References
Chapter 3: Computations of probable maximum precipitation estimates
	1. Introduction
		1.1. Background and importance of PMP estimations
	2. Methodology of PMP estimation
		2.1. Physical method
		2.2. Statistical method
		2.3. Multifractal approach
	3. Statistical PMP estimates: A case-study
		3.1. Hershfield PMP estimates in Malaysia
	4. Conclusions
	References
Chapter 4: Deep learning: Long short-term memory in hydrological time series
	1. Introduction
	2. Model description of long short-term memory (LSTM)
		2.1. Neural network
		2.2. Recurrent neural network
		2.3. LSTM
	3. Training network and backpropagation
		3.1. Feedforward and backward propagation
		3.2. Gradient descent method
		3.3. Backpropagation of RNN
		3.4. Backpropagation through time of RNN
		3.5. Backpropagation through time for LSTM
	4. Variants of LSTM
		4.1. Peephole LSTM
		4.2. Gated recurrent unit
		4.3. Multiplicative LSTM
		4.4. Sequence-to-sequence (seq2seq) LSTM
		4.5. Bidirectional LSTM
	5. Normalization and hyperparameter selection
		5.1. Normalization
		5.2. Estimation of hyperparameters
	6. LSTM applications in hydrometeorological variables
		6.1. LSTM and its variants for prediction
		6.2. Hybrid LSTM
		6.3. Simulation modeling with LSTM
	7. Employed deep learning programs for LSTM
		7.1. Tensorflow and Keras with Python
		7.2. Matlab
	8. Conclusions
	References
Chapter 5: Dimensionality reduction of correlated meteorological variables by Bayesian network-based graphical modeling
	1. Introduction
	2. Study area and data used
		2.1. Study area
		2.2. Data used
	3. Methodology
	4. Results and discussions
		4.1. Directed acyclic graphs obtained from HC and MMHC algorithms
		4.2. Utility of the conditional dependence structure
	5. Conclusions
	References
Chapter 6: The ecohydrological function of the tropical forest rainfall interception: Observation and modeling
	1. Introduction
	2. Canopy water balance: concepts and general aspects of the monitoring
		2.1. Forest canopy water balance
			2.1.1. Rainfall interception
			2.1.2. Throughfall
			2.1.3. Stemflow
		2.2. Water balance in tropical forested watersheds
		2.3. Soil moisture in forest areas
		2.4. Geochemistry in tropical forests
	3. Measurements of the rainfall interception components
		3.1. Standard measurements
		3.2. Ex situ methods for individual trees
		3.3. Forest parameters
	4. Rainfall interception modeling
		4.1. Conceptual models
		4.2. Statistical and machine learning tools for ecohydrological data handling
	5. Conclusions
	References
Chapter 7: Emotional artificial neural network: A new ANN model in hydroinformatics
	1. Introduction
	2. Mathematical concepts of emotional artificial neural network
		2.1. Feed forward neural network (FFNN)
		2.2. Emotional artificial neural network (EANN)
		2.3. Difference between FFNN and EANN
		2.4. Data preprocessing and performance evaluation
		2.5. Dominant inputs selection
	3. Some applications of EANN
	4. Conclusions
	References
Chapter 8: Exploring nature-based adaptation solutions for urban ecohydrology: Definitions, concepts, institutional frame ...
	1. Introduction
	2. Nature-based adaptation solutions (NBaS): Conceptual framework and position
		2.1. NBaS: Concepts and terminology
		2.2. NBaS in the international, regional, and national frameworks
	3. NbaS and ecohydrology
	4. The need for physically-based evidence
		4.1. Potential and limitations of green roofs
			4.1.1. Potentials
			4.1.2. Limitations
		4.2. Green roofs: A means for hybridizing the gray
	5. Conclusions
		5.1. Adapt by learning and learning to adapt
		5.2. Are NBaS the way forward?
	References
Chapter 9: Fuzzy-based large-scale teleconnection modeling of monthly precipitation
	1. Introduction
	2. Materials and methods
		2.1. Teleconnections impact on hydroclimatologic systems
		2.2. Proposed methodology
		2.3. Association rule
		2.4. Fuzzy logic
		2.5. Efficiency criteria
		2.6. Study area and data
	3. Results and discussion
	4. Conclusions
	References
Chapter 10: Hydrologic models classification, calibration, and validation
	1. Introduction
	2. Hydrological modeling for integrated water management
		2.1. Classification of hydrological models
		2.2. Description of some common hydrological model types
			2.2.1. Physical models
			2.2.2. Deterministic models
			2.2.3. Black box and statistical models
			2.2.4. Conceptual models
			2.2.5. Empirical models
		2.3. Rainfall runoff transformation methods in physical models
		2.4. Hydrologic model components
	3. Model calibration and validation
		3.1. Model parameterization
		3.2. Model calibration
			3.2.1. Calibration components
			3.2.2. The objective function
			3.2.3. Nash criteria
		3.3. Optimization methods for model calibration
			3.3.1. Manual calibration
			3.3.2. Automatic calibration
			3.3.3. Genetic algorithms
		3.4. Model validation
		3.5. Model uncertainties
			3.5.1. Data uncertainty
			3.5.2. Uncertainty due to parameters estimation
			3.5.3. Model uncertainty
		3.6. Regionalization methods in rainfall-runoff modeling (case of ungauged basins)
	4. Conclusions
	References
	Further reading
Chapter 11: Identification of soil erosion sites in semiarid zones: Using GIS, remote sensing, and PAP/RAC model
	1. Introduction
	2. Materials and method
		2.1. Study area
		2.2. PAP/RAC model application
			2.2.1. The predictive phase
			2.2.2. The descriptive phase
			2.2.3. The integration phase
		2.3. The accuracy test
	3. Results and discussion
		3.1. Predictive phase
			3.1.1. The slope map
			3.1.2. The lithology map
			3.1.3. The erodibility map
			3.1.4. Land use map
			3.1.5. Vegetation density
			3.1.6. The soil protection map
			3.1.7. The predictive erosion map
		3.2. Descriptive phase
		3.3. Integration phase
		3.4. Global diagnosis of the test accuracy assessment
	4. Conclusions
	References
Chapter 12: Metrics of the water performance engineering modeling
	1. Introduction
	2. Types of hydro-climatological modeling and metrics
		2.1. Point prediction
		2.2. Prediction intervals of modeling
		2.3. Binary classification
		2.4. Input selection methods in modeling
		2.5. Decision making models
		2.6. Hydrographs in hydrological modeling
		2.7. Flow-duration curves
		2.8. Information theory
	3. Some applications of the metrics
		3.1. Application of the point prediction in hydro-climatological modeling
		3.2. Application of the PIs in hydro-climatological modeling
		3.3. Application of the binary classification in hydro-climatological modeling
		3.4. Application of the input selection in hydro-climatological modeling
		3.5. Application of the decision making in hydro-climatological modeling
		3.6. Application of the hydrographs in hydro-climatological modeling
		3.7. Application of the flow-duration curves in hydro-climatological modeling
		3.8. Application of the information theory in hydro-climatological modeling
	4. Agenda for future studies
	References
Chapter 13: Outlier robust extreme learning machine: Predicting river water temperature in the absence of air temperature
	1. Introduction
	2. Study area and data
	3. Materials and methods
		3.1. Outlier robust extreme learning machine(ORELM)
		3.2. Performance assessment of the models
	4. Results and discussion
	5. Conclusions
	References
Chapter 14: Parametric and nonparametric methods for analyzing the trend of extreme events
	1. Introduction
	2. Trend calculation methods
		2.1. Mann-Kendall test (MK)
		2.2. Mann-Kendall test with trend-free prewhitening (MK2)
		2.3. Modified Mann-Kendall tests (MK3)
		2.4. Mann-Kendall test considering LTP (MK4)
			2.4.1. Calculation of Hurst coefficient (H)
			2.4.2. Significance level of H
			2.4.3. Calculation of variance
		2.5. Spearmans rho test
		2.6. Linear regression method
		2.7. Quantile regression method
		2.8. Generalized least squares (GLS) regression with AR errors
		2.9. T-test for difference between means
	3. Case studies
	4. Conclusions
	References
Chapter 15: Voting-based extreme learning machine: Potential of linking soil moisture content to soil temperature
	1. Introduction
	2. Materials and methods
		2.1. Study area and available data
		2.2. Performance metrics of the models
		2.3. Adaptive network based fuzzy inference system (ANFIS)
		2.4. Random vector functional link neural networks (RVFL)
		2.5. Voting-based extreme learning machines (VELM)
	3. Results and discussion
	4. Conclusions
	References
Chapter 16: Prediction of reference crop evapotranspiration: Empirical and machine learning approaches
	1. Introduction
	2. An overview of various empirical methods for reference evapotranspiration estimation
		2.1. FAO-56 Hargreaves method
		2.2. FAO-24 Blaney-Criddle (temperature-based) method
		2.3. FAO-24 radiation method
		2.4. Priestley-Taylor method
		2.5. Turc radiation-based method
		2.6. FAO-56 pan evaporation method
		2.7. FAO-24 Penman combination type
		2.8. FAO-56 Penman-Monteith method
	3. Estimation of ET0 using empirical models
	4. Machine learning techniques used for estimation of evapotranspiration
		4.1. Artificial neural network
		4.2. Genetic algorithm
		4.3. Extreme learning machine (ELM)
		4.4. Support vector machine (SVM)
		4.5. Decision tree
		4.6. Bagging technique
		4.7. Boosting technique
			4.7.1. AdaBoost technique
			4.7.2. Gradient boosting
			4.7.3. XgBoost technique
			4.7.4. Light gradient boosting technique
			4.7.5. Gradient boosting with categorical features support (CatBoost) technique
	5. Modeling of ET0 using machine learning techniques
	6. Deep learning techniques in modeling of ET0
	7. Case study
	8. Conclusions
	References
Chapter 17: Reference evapotranspiration in water requirement: Theory, concepts, and methods of estimation
	1. Introduction
	2. Theory of evapotranspiration
		2.1. Evaporation
		2.2. Transpiration
		2.3. Concepts of evapotranspiration
		2.4. Reference evapotranspiration
		2.5. Reference surface
	3. Methods of calculating the reference evapotranspiration
		3.1. Direct method of estimating reference evapotranspiration
			3.1.1. Drainage lysimeters
			3.1.2. Weighting lysimeters
		3.2. Indirect methods for estimating the reference evapotranspiration
			3.2.1. FAO-Penman-Monteith method
				Location coordinates
				Temperature
				Humidity
				Radiation
				Wind speed
				Latent heat of vaporization
				Humidity coefficient (γ)
				Saturated steam pressure
				Actual steam pressure (ed)
				Steam pressure deficiency (ea-ed)
				Steep pressure curve slope (Delta)
				Extraterrestrial radiation (Ra)
				Number of hours of daylight (N)
				Net radiation (Rn)
				Heat flux into the soil (G)
			3.2.2. Most valid indirect methods of estimating reference evapotranspiration
				Thornthwaite
				Turc
				Priestley-Taylor
				Trajcovic
				Schendel
				Berti et al.
				Ravazzani et al.
				Mahringer
				Businger-Van Bavel
				WMO
				Romanenko modified
				Trabert
				Meyer
				Romanenko
				Hargreaves and Samani
				Dalton
				Kimberly-Penman
				Jensen-Haise
				Szász
				Caprio
				Irmak
				Ritchie
				McGuinness-Bordne
				Baier-Robertson
				Modified Turc
				Valiantzas
				Corrected Makkink method
				FAO-24 Blaney-Criddle
				Radiation FAO
				Makkink
				SCS Blaney-Criddle
				Linacre
			3.2.3. Evaporation pan method
				Cuenca equation
				Snyder equation
				Modified Snyder equation
				Orang equation
			3.2.4. Energy budget models
				Aerodynamic methods
				Energy balance methods
				Combined methods
				Experimental methods
			3.2.5. Case studies
			3.2.6. Statistical evaluation indices
	4. Conclusions
	References
Chapter 18: Extremely randomized trees versus random forest, group method of data handling, and artificial neural network
	1. Introduction
	2. Study area and data
	3. Methodology
		3.1. Multilayer perceptron neural networks (MLPNN)
		3.2. Random forest (RF)
		3.3. Extremely randomized tree (ERT)
		3.4. Group method of data handling (GMDH)
		3.5. Performance assessment of the models
	4. Results and discussion
	5. Conclusions
	References
Chapter 19: Index of resilience and effectiveness of disaster risk management
	1. Introduction
	2. The disaster risk management index, DRMi
	3. Evaluation for Latin America and the Caribbean region
		3.1. Risk identification policy
		3.2. Risk reduction policy
		3.3. Disaster management policy
		3.4. Governance and financial protection policy
		3.5. DRMi results
	4. Conclusions
	References
Chapter 20: Wavelet decomposition based on Gaussian process regression and multiple linear regression: Monthly reservoir  ...
	1. Introduction
	2. Materials and method
		2.1. Study area
		2.2. Gaussian process regression
		2.3. Multi-linear regression
		2.4. Wavelet transform
		2.5. Feature selection method based on correlation (CFS)
		2.6. Proposed methodology
	3. Evaluation criteria
	4. Results and discussion
		4.1. Results of the hybrid models
			4.1.1. The results of WGPR-CFS and WMLR-CFS models
	5. Conclusions
	References
Chapter 21: Sequential Monte-Carlo methods in hydroclimatology
	1. Introduction
	2. Bayes theorem
	3. Basics of sequential Monte-Carlo methods
		3.1. Basic random Monte-Carlo sampling
		3.2. Importance sampling (IS)
		3.3. The algorithm of particle filter
		3.4. Resampling (selection)
			3.4.1. Sequential importance resampling (SIR)
			3.4.2. Residual resampling
			3.4.3. Stochastic universal resampling
		3.5. Tiny numerical example
	4. Particle filters for high-dimensional geoscience applications
		4.1. Proposal density particle filters
			4.1.1. Implicit particle filter
			4.1.2. Equivalent weights particle filter
			4.1.3. Implicit equal weights particle filter
		4.2. Localized particle filters
	5. Adaptive estimation using particle filter
	6. Conclusions
	References
Chapter 22: Smart cities and hydroinformatics
	1. The history of urbanization and smart cities
	2. Definition of a Smart City
		2.1. The technology dimension
		2.2. The human dimension
		2.3. The institutional dimension
	3. Components of a Smart City
		3.1. Smart economy
		3.2. Smart people
		3.3. Smart governance
		3.4. Smart mobility
		3.5. Smart environment
		3.6. Smart living
	4. Smart cities and internet of things
		4.1. Precision
		4.2. Big IoT data
		4.3. Compatibility
		4.4. Investment
		4.5. Security and privacy
	5. Smart City strategies
	6. Smart cities and hydroinformatics
		6.1. Relevant definition of hydroinformatics to smart cities
			6.1.1. Spatial data
			6.1.2. Digital terrain model (DTM)
			6.1.3. Remote sensing data
			6.1.4. Local data
		6.2. The role of data acquisition
		6.3. Data analytics and modeling
		6.4. Applications of hydroinformatics to smart cities
			6.4.1. Water supply and distribution systems
			6.4.2. Wastewater and urban drainage systems
			6.4.3. Water quality management
			6.4.4. Toward integrated water systems
	7. Conclusions
	References
Chapter 23: Support vector regression model optimized with GWO versus GA algorithms: Estimating daily pan-evaporation
	1. Introduction
	2. Study location and data collection
	3. Methodology
		3.1. Gamma test
		3.2. Support vector regression
		3.3. Nature-inspired algorithms
			3.3.1. Grey wolf optimizer
			3.3.2. Genetic algorithm
		3.4. Hybrid SVR models
		3.5. Performance measures
	4. Results and discussion
		4.1. Application of hybrid SVR models
	5. Conclusions
	References
Chapter 24: Univariate, multivariate L-moments and copula functions for drought analysis
	1. Introduction
	2. Materials and methods
		2.1. Univariate L-moments
		2.2. Multivariate L-moments
		2.3. Copula functions
	3. Results
		3.1. Univariate drought index
		3.2. Multivariate drought index
	4. Conclusions
	References
Index
Back Cover