Journal of

Marine Science
and Engineering

Review
Ensemble Neural Networks for the Development of Storm
Surge Flood Modeling: A Comprehensive Review
Saeid Khaksari Nezhad *,† , Mohammad Barooni † , Deniz Velioglu Sogut and Robert J. Weaver

Ocean Engineering and Marine Sciences, Florida Institute of Technology, Melbourne, FL 32901, USA;
mbarooni2018@my.fit.edu (M.B.); dvelioglusogut@fit.edu (D.V.S.); rjweaver@fit.edu (R.J.W.)
* Correspondence: skhaksarinez2021@my.fit.edu
† These authors contributed equally to this work.

Abstract: This review paper focuses on the use of ensemble neural networks (ENN) in the develop-
ment of storm surge flood models. Storm surges are a major concern in coastal regions, and accurate
flood modeling is essential for effective disaster management. Neural network (NN) ensembles have
shown great potential in improving the accuracy and reliability of such models. This paper presents
an overview of the latest research on the application of NNs in storm surge flood modeling and
covers the principles and concepts of ENNs, various ensemble architectures, the main challenges
associated with NN ensemble algorithms, and their potential benefits in improving flood forecasting
accuracy. The main part of this paper pertains to the techniques used to combine a mixed set of
predictions from multiple NN models. The combination of these models can lead to improved
accuracy, robustness, and generalization performance compared to using a single model. However,
generating neural network ensembles also requires careful consideration of the trade-offs between
model diversity, model complexity, and computational resources. The ensemble must balance these
factors to achieve the best performance. The insights presented in this review paper are particularly
relevant for researchers and practitioners working in coastal regions where accurate storm surge
flood modeling is critical.

Keywords: deep learning; storm surge prediction; ensemble model; sea level rise
Citation: Nezhad, S.K.; Barooni, M.;
Velioglu Sogut, D.; Weaver, R.J.
Ensemble Neural Networks for the
Development of Storm Surge Flood 1. Introduction
Modeling: A Comprehensive Review.
Rising sea levels increase the risk of coastal flooding depending on the relative rate of
J. Mar. Sci. Eng. 2023, 11, 2154.
mean sea/land level changes [1–3]. The impacts are linked to concurrent near-term trends
https://doi.org/10.3390/
as well as gradual escalation of long-term coastal inundation risk over time [4]. Estuaries
jmse11112154
and coastal areas should adapt to changing climate and implement the necessary mitigation
Academic Editor: Barbara Zanuttigh measures. A complex process such as a storm surge is sensitive to abrupt changes in several
Received: 5 October 2023
storm parameters, such as intensity, surface atmospheric pressure at the center of the storm,
Revised: 27 October 2023
maximum sustained wind speed, size, and forward speed, in addition to the effects driven
Accepted: 7 November 2023
by the characteristics of dynamic coastal settings, such as shoreline geography, estuaries,
Published: 11 November 2023 and bay barriers [5]. The interdependency of these different factors make it notoriously
hard to predict the timing and intensity of the hydrodynamic response (e.g., water levels
and currents) [6–9]. Parametric models conventionally incorporate historical or synthetic
hurricanes using storm size, intensity, and track, allowing for the prediction of storm surge
Copyright: © 2023 by the authors. heights and overland flooding [10,11].
Licensee MDPI, Basel, Switzerland. During a storm surge event (caused by tropical or extratropical cyclones), the potential
This article is an open access article
impacts extend beyond the surge itself and could exacerbate flooding and structural dam-
distributed under the terms and
age. This can be further intensified by the surface gravity waves due to the superimposed
conditions of the Creative Commons
storm tide [12]. Wave driven set-ups can contribute up to 30% of the total increase in water
Attribution (CC BY) license (https://
level (including both typical fluctuations and any additional rise) along the coast [13]. The
creativecommons.org/licenses/by/
combination of elevated water levels along with the destructive power of waves poses
4.0/).

J. Mar. Sci. Eng. 2023, 11, 2154. https://doi.org/10.3390/jmse11112154 https://www.mdpi.com/journal/jmse

J. Mar. Sci. Eng. 2023, 11, 2154 2 of 30

a tremendous danger to densely populated areas adjacent to coastal waters. The U.S.
Atlantic and Gulf Coasts, for example, are expected to experience a sea level rise of, on
average, 0.25–0.30 m in 30 years (2020–2050) [14]. This further increases the vulnerability
of coastal regions to compound flooding (CF), where the interaction of rainfall, rivers,
and ocean storm surges combine and create a cataclysmic force [15]. To overcome these
challenges, physics-based approaches, such as hydrodynamic models, have been used to
estimate hydrological processes and flood hazards/the probability of particular events
that require land–atmosphere–ocean coupling [16]. Although these models explain the
nature of flooding phenomena and show great skill for a wide variety of flood predic-
tion scenarios, they usually deal with the physical dynamics and require various types of
datasets, as the occurrence of floods varies with time and space [17,18]. This requires a large
amount of computation, which makes short-term predictions very challenging. The reader
is kindly referred to [17,19,20] for the comprehensive studies related to the development of
physics-based models, their challenges, and capabilities.
Hydrodynamic modeling has also been extensively used to investigate the spatial
and temporal variability of storm surges. Hydrodynamic models are widely utilized to
describe coastal ocean processes and near-shore circulation and to simulate future scenarios
of possible storm surge flooding [21]. These models are well-developed to account for the
inherent uncertainties associated with sea level rise and storm surges. They also consider
the relative impacts of different meteorological forces in total water levels [22,23]. However,
these models are computationally demanding and time consuming. This limits their ability
to simulate large complex domains or ensembles of events.
Some parametric models, such as the Bayesian model averaging, autoregressive inte-
grated moving average, and peak over threshold methods, are among the most preferred
methods to predict the statistical behavior of storm surge flooding [24,25]. However, these
models are, at times, computationally demanding and typically sophisticated. Furthermore,
generalizing the potential impacts of a storm surge for a particular geographical area to
other areas with different parameters and settings is not a reliable approach [23]. Flood
prediction requires constructing a minimum of a decade of non-tidal residual data from
measurement by sea-level gauges [26]. In small datasets, i.e., those with a lack of large-
sample observational data, even a few outliers will significantly alter the model or affect
the correlation among the predicting variables [27].
Low-fidelity numerical storm surge models such as SLOSH (Sea, Lake, and Over-
land Surges from Hurricane) [28] are used by emergency managers and researchers to
assist in forecasting the hydrodynamic response to a predicted hurricane track, size, and
intensity. These models have significant uncertainty when used for forecasting [29,30].
Coupling ADCIRC (ADvanced CIRCulation model) [31] with WAM (WAve prediction
Model) [32], STWAVE (Steady-State Spectral Wave Model) [33], or SWAN (Simulating
WAves Nearshore) [34] is a widely used method for generating high-resolution storm surge
models of specific regions [35,36]. Considering their additional wave forcing processes,
finer mesh sizes, and smaller time steps, high-fidelity models are computationally more
expensive [37]; thus, the accurate and quick assessment of hurricane-induced flooding has
always been a challenging task.
Surrogate models are another approach to overcome this huge obstacle by simplifying
approximations of more complex, higher-order models [10]. The Surge and Wave Island
Modeling Study (SWIMS) [38] in the USACE, for example, developed a fast surrogate model
by simulating hundreds of hurricanes to predict peak storm surges and hurricane responses
in only a couple of seconds, which is an advantage over high-fidelity coupled simulations.
Considering this issue, in a national-scale effort, the U.S. Army Engineer Research and
Development Center developed a statistical analysis and probabilistic modeling tool named
the StormSim Coastal Hazards Rapid Prediction System (StormSim-CHRPS) [39]. The tool
preserves the accuracy of the high-fidelity hydrodynamic numerical simulation methods,
such as ADCIRC, while significantly reducing computational demands, making it more
convenient for real-time emergency management applications. The intricate input/output
J. Mar. Sci. Eng. 2023, 11, 2154 3 of 30

relationships inherent in high-fidelity numerical models are approximated using a machine

learning method called Gaussian process metamodeling (GPM), enabling the rapid pre-
diction of the peak storm surge and hurricane responses within seconds and for different
hurricane scenarios.
Lee et al. [37] sought to enhance coastal resilience by providing a rapid storm surge
prediction surrogate model called C1PKNet, a combination of a convolutional neural
network model (CNN), principal component analysis, and a k-means clustering method,
which was trained efficiently on a dataset of 1031 high-fidelity storm surge simulations.
The resulting model is capable of predicting peak storm surges from realistic tropical
cyclone track time series. A few studies, such as [40,41], even consider global warming,
earth–moon–sun gravitational attractions, and storm surges to estimate the coastal sea
level at an hourly temporal scale. The model in [40] was developed using an artificial
neural network (ANN) approach called long short-term memory (LSTM) and trained on the
ECMWF (European Center for Medium-Range Weather Forecasts) reanalysis dataset, ERA5
(more information on raw input data generation using ERA5 is available in Section 5.1).
To the best of our knowledge, only a limited number of researchers, such as [37,42–44]
aimed to assess the concept of ANN ensemble learning for storm surge prediction. Braakmann-
Folgmann et al. [43], for example, developed a combined convolutional and recurrent neural
network to analyze both the spatial and the temporal evolution of sea level anomalies in the
northern and central Pacific Ocean. They show how neural network architectures outperform
simple regression to improve predictions for the future sea level. A novel deep learning
architecture was implemented by [44] in contrast to a primitive model called the general ocean
circulation model ensemble or NEMO (Nucleus for European Modelling of the Ocean). Their
aim was to reduce the uncertainty associated with accurate sea level predictions and also to
show the importance of sea level and atmospheric inputs for shorter forecast times. In the
latter study, the ensemble ANN method for sea level forecasting known as HIDRA (HIgh-
performance Deep tidal Residual estimation method using Atmospheric data) implements
variants of temporal convolutional networks (TCN) and LSTM to encode temporal features of
atmospheric and sea-level data. The dataset was trained on a 10-year (2006–2016) time series
of atmospheric surface fields using a single member of the ECMWF atmospheric ensemble.
More recent papers such as [42,43] investigated the capability of different combina-
tions of neural network (NN) models to predict surge levels. The fundamental core of this
research revolves around selecting the best NN architecture for an ensemble approach to
outperform a simple probabilistic model. Tiggeloven et al. [43], for example, combined a
CNN-LSTM (ConvLSTM) model to capture the spatio-temporal dependencies for peak wa-
ter level observations. This research has important implications for the sensitivity analysis
of predictor variables and investigates how uncertainty in the predictions changes with in-
put or architecture complexity. Tropical cyclones can also be parametrically represented via
the joint probabilities method (JPM) [45]. However, the parametric description of complex
systems, such as large-scale, non-frontal, low-pressure tropical cyclones, is intrinsically
difficult to determine. As an alternative approach to these models, data-driven methods
such as multiple linear regression [26,46], decision tree, ANN [40,42,43,47–50], and support
vector machine [51,52] have been widely used for the prediction of storm surge heights.
In most of studies where data-driven surrogate models are trained with physics-based
simulations, such as ADCIRC [37,42,52], a major hurdle is the lack of sufficiently long
datasets for training, validating and testing the surrogate models. As [53] explains, a long
record in a storm surge reconstruction dataset is critical to capture as many storm events as
possible; thus low-probability, high-impact, extreme events could be accounted for.
This review paper is structured as follows. Section 2 highlights the general concept of
neural network ensembles and introduces several challenges and limitations. A theoretical
framework for the geometry of neural networks, transfer learning, and their application
to storm surge prediction models and different ensemble generation methods (i.e., how
to combine the predictions from multiple models) are presented in Section 3. Section 4
discusses the less-debated topic of ensemble pruning and fine-tuning, the next stage after
J. Mar. Sci. Eng. 2023, 11, 2154 4 of 30

ensemble generation. Section 5 introduces data preparation considerations on developing

an ensemble of neural networks, and different sources of datasets commonly used to predict
storm surge levels are presented as well. Section 6 discusses some important factors and
parameters regarding the best model selection and how the performance of the selected
ensemble is evaluated. Finally, in Section 7, a summary is presented.

2. Neural Network Ensemble

Ensemble learning refers to techniques that involve combining the predictions of
several base estimators based on classification or regression problems, aiming at improving
predictability. This approach has gained a lot of attention in recent years, and the reported
results regarding sea level rise projections have been satisfactory, such as in [7,44,54].
Ensembles have been reported to achieve higher certified robustness than single machine
learning algorithms, as discussed in Section 2. Therefore, coastal hydrodynamic modeling
techniques have been applied in ensemble with data-driven models such as deep learning
techniques, especially neural networks, to develop ocean circulation and flood simulation
models. This is due to the popularity and application of the finite element methods in
numerical hydrodynamic models and their adequate modeling resolution [55–57]. These
numerical models are conventionally applied to probabilistic coastal ocean forecast systems
such as Surge Guidance System Forecasts (ASGS) or NOAA P-Surge to accommodate
thousands of simulations [58].
Various types of neural networks are helpful to solve regression prediction problems
where the aim is to predict the output of a continuous value such as water levels. Multilayer
perceptrons (MLPs), a classical type of neural network, can reconstruct and validate atmo-
spheric forcing, such as maximum sustained wind speed [59–61]. Convolutional neural
networks (CNNs) have been developed to capture spatial and temporal dependencies for
surge-level observations on a grid-based dataset and could potentially identify and predict
regional and global patterns in storm and climate datasets [62]. They can also extract water
bodies from remote sensing images [63]. Recurrent neural networks (RNNs) could be
helpful in modeling storm behavior and time series of water levels in a sequence prediction
framework [43], which requires a longer training time (not dependent on a fixed input size)
compared to CNNs. Long short-term memory (LSTM), a subtype of RNN, is a successful
model and has been used to capture long-term temporal dependencies of meteorological
forcing [64,65] and to analyze the rapid intensification and occurrences of cyclones [66]. A
diverse set of base learners (individual learners of the ensemble), such as MLPs, CNNs,
and RNNs with appropriate training and tuning, is one empirical way to improve model
performance by generating more complex models [67].
The focus of this paper is to introduce ensemble methods that can predict storm surge
levels using a supervised ANN. Some challenges associated with using ANNs are the
inability to capture peak water levels (due to the complex and nonlinear nature of the
physical processes) [65,68], long-term processes (which are unavailable due to instrument
failures, insufficient data, or sparse observational records), and predictions of storm surges
at ungauged sites [43,69]. However, when utilized appropriately, ANN ensemble models
have the potential to provide better and faster results than finite element hydrodynamic
models. Figure 1 emphasizes the essential need for rapid prediction models, e.g., ENNs,
by presenting a benchmark for the Aransas Wildlife Refuge station in Texas during and
following Hurricane Harvey in 2017 [39]. This descriptive example compares storm surge
predictions from a rapid empirical prediction model against water level observations
from NOAA tide gauges and predictions from operational ADCIRC runs performed at
the U.S. Army Engineer Research and Development Center’s Coastal and Hydraulics
Laboratory (ERDC-CHL). Hurricane Harvey started as a modest tropical storm in August.
However, after re-forming over the Bay of Campeche, it intensified rapidly into a category
4 hurricane. Harvey made its landfall along the central Texas coast and then stalled for
four days, resulting in unprecedented rainfall, exceeding 1520 mm and resulting in a surge
reaching 1.4 m across southeastern Texas [70]. Figure 1 also highlights the rate of change
J. Mar. Sci. Eng. 2023, 11, 2154 5 of 30

and meteorological and oceanographic observations during the hurricane. Forecasts are
typically updated at 6 hour intervals. However, for unusual storm scenarios comparable
to Hurricane Harvey with rapid approach trajectories or extended durations within flood
plains, the expected update intervals can be reduced to 3 h or even shorter.
A thorough and extensive literature review can be found in [1,71], where machine
learning models are compared to traditional physically based models.
Washing ton
St Louis
Louisville
Category 4
Richmond
Category 3
Norfolk
Category 2 Nashville
Knoxville
Category 1 City
O klahoma
Raleig h
NOAA Gauge: 8774230
Memp his Charlotte
Tropical Storm Greenville Aransas Wildlife Refuge (TCOON), TX

Still Water Level (MSL, m)

Tropical Depression 2
Atlanta
Extratropical Storm
Birming ham Observed
Dallas 1.5 Empirical ML
ADCIRC
08/31/17 00Z 1 Landfall
Austin Jacksonville
Houston New O rleans 0.5
San Antonio

O rland o 0
Tamp a

08/25/17 18Z -0.5

08/25 02:24

08/25 08:24

08/25 14:24

08/25 20:24

08/26 02:24

08/26 08:24

08/26 14:24

08/26 20:24

08/26 08:24
M onterrey Miami

XICO
Havana
San Luis Potosi 08/23/17 18Z Date (UTC)
CU BA
Merid a
ara Q uerétaro

M exico City
Port-au-Prince San Juan 1:18,285,727
King ston 0 140 280 560 mi
BELI Z E
0 220 440 880 km
Esri, HERE, Garmin, FAO, NOAA, USGS, EPA

G U A TEM A LA

(a)
100
N
W E Wind Direction Winds
S
75 Gusts

50
Speed (knots)

25

-25
22. Aug 24. Aug 26. Aug 28. Aug 30. Aug

(b)
2.0 1020
Verified
Height in meters (MSL)

Barometric Pressure (mb)

1.5 Predictions 1010

1.0 1000

0.5 990

0.0 980

970
00:00 00:00 00:00 00:00 00:00 22. Aug 24. Aug 26. Aug 28. Aug 30. Aug
8/22 8/24 8/26 8/28 8/30

(c) (d)
Figure 1. Cont.
J. Mar. Sci. Eng. 2023, 11, 2154 6 of 30

95 95

90 90

Degrees (F)
Degrees (F)
85 85

80 80

75 75

70 70
22. Aug 24. Aug 26. Aug 28. Aug 30. Aug 22. Aug 24. Aug 26. Aug 28. Aug 30. Aug

(e) (f)
Figure 1. (a) Best track positions and storm surge predictions from the empirical CHRPS model
compared to water level observations from select NOAA tide gauge and storm surge predictions
from operational ADCIRC simulations performed at CHL [39]. (b) Winds. (c) Hourly heights.
(d) Barometric pressure. (e) Air temperature. (f) Sea surface temperature in Aransas Wildlife Refuge
station, TX, for Hurricane Harvey (August 2017).

3. Theoretical Framework
3.1. Neural Network Architectures
The NN architecture consists of individual members called neurons, which are com-
bined to simulate the biological behavior of the brain to solve real-world problems [37,41].
Neural networks are not an exclusive standardized method; instead, they involve learning
algorithms and architectures that can be applied to a wide range of supervised flood and
storm surge forecasting models. These models use a set of individual independent vari-
ables, such as tidal and meteorological data points, and a real value dependent variable
that represents the phenomenon, such as storm surge levels [42,43,72]. A general scheme
is shown in Figure 2 based on a fully connected MLP representation. In the basic MLP
architecture, the input layer is connected to one or multiple hidden layers and finally to the
output layer to construct a fully connected system. The information is primarily processed
in the forward direction (feed-forward) and is put through a linear transformation using
a weights matrix [47,73]. An activation function defines how the weighted sum of the
input vector is transformed to the neurons of the next layer [47]. The choice of activation
function in both the hidden and output layers significantly influences the performance
of the NN model in learning from the training dataset and predicting storm surge events.
Empirical testing and cross-validation are essential to determine the most appropriate
activation function that can effectively capture non-linear relationships within the data.
Table 1 presents some frequently used activation functions specifically tailored for storm
surge prediction models, as well as the relationship between each activation function and
its corresponding Python library. The elementwise activation function is usually shifted
with a bias to adjust the final output matrix. Different model configurations associated
with learning processes and choices of the right dimensions of the NN structure, including
the number of hidden layers, learning rate, batch size, choice of the activation function and
loss function, etc., are referred to as hyperparameters [74–76]. Table 2 presents a summary
of the major hyperparameters in NN models. These tuning parameters pertain to the
physical components, training/optimization procedures, and regularization effect in a
neural network.
In order to train a MLP feed-forward NN model, a backpropagation NN (BPNN)
is widely used. This algorithm has been identified as one of the simplest and the most
powerful ML prediction tools suitable for flood time series and short-term storm surge
predictions [77–80]. In a BPNN algorithm, the gradient of the loss function (the vector
of the partial derivatives) is calculated through a method called chain rule to adjust each
weight and its contribution to the overall error. Further details of BPNN algorithms can be
found in Appendix A.
J. Mar. Sci. Eng. 2023, 11, 2154 7 of 30

Forward Propagation
Input layer Hidden Layer(s)

x₁ w₁ b
Activation Function
x₂ w₂
Storm Surge
W ∑x�w� Height

Original Dataset x� w�

Updating the Calculating the

Weights and Biases Loss Function

Backpropagation

Figure 2. Flow diagram of transfer learning in NN, including the reuse of a pre-trained model on a
new problem.

Table 1. Frequently used activation functions in ANN storm surge prediction models.

Activation Function Equation Python Library Applications

ReLU (Rectified Linear Unit) f ( x ) = max(0, x ) tensorflow, keras MLP, CNN
Sigmoid f ( x ) = 1+1e−x tensorflow, keras RNN
x e− x
Tanh (Hyperbolic Tangent) f ( x ) = eex −
+e− x
tensorflow, keras RNN
x
Softmax e j tensorflow, keras Classification, normalizing the output
f (xj ) =
∑ k =1 e x k
K

Leaky ReLU f ( x ) = max(αx, x ) tensorflow, keras MLP, CNN

Table 2. Classification of major hyperparameters in NN models .

Physical Components Training/Optimization Procedures Regularization

Number of hidden layers within Degree of regularization
Defining the optimizer algorithm
the network (lambda)
Number of active neurons
Number of hidden Neurons Configuring the learning rate
(dropout rate)
Choice of key activation function Defining the main type of loss function
Choice of evaluation metric for regression problem
Number of training samples (mini-batch)
Setting the random initialization
Number of training cycles (epochs)

3.2. Transfer Learning

In some scenarios, the NN algorithms use different sources of information such as
historical tropical cyclones, topography, meteorological forcing, and other sources to make
a complex network. Training an ensemble of NN models on such a massive volume
of raw data can be computationally expensive [81]. On the other hand, when datasets
are expensive or difficult to collect or data are scarce for a specific problem (such as the
short-term analysis of hurricane tracks) [64,82], obtaining a training dataset to discern a
meaningful pattern could be problematic. Transfer learning, as shown in Figure 3, is a
functional method of tackling these problems through, i.e., building a high performance NN
model while reducing training time [83]. This is performed by obtaining a high-accuracy
and large pre-trained model from a related source and transferring the knowledge from the
trained data to the target domain in a time-saving way [84]. Surge time series data over long
time scales are usually subject to seasonal variability known as seasonality [85–87] (which
J. Mar. Sci. Eng. 2023, 11, 2154 8 of 30

can be simply defined using a Fourier transform and finding the seasonal frequencies).
Removing seasonality from the time series data might happen during data preparation
(which is further discussed in Section 5). Extractions of sparse time series samples from
short-term extreme impacts during dominant seasons could be limited in size, implying that
the insufficient training data are unable to represent the target efficiently [85]. Therefore,
transferring knowledge from a diverse, large-scale, and pre-trained dataset of a time series
of a similar task (with minor adjustments) could be reasonable [88] when a NN model is
adapted to forecast a new time series, thus avoiding the need for additional training [83].

Pre-trained network

Labeled data from

similar tasks

Training Samples
Target Domain

Target Data

Timestamp

Multivariate Timeseries Retraining and �ine-tuning

Figure 3. Flow diagram of transfer learning in NN involving the reuse of a pre-trained model on a
new problem.

3.3. Ensemble Generation Methods

Ensemble neural networks basically consist of [54]: (1) generating multiple base
learners (weak classifiers) and (2) combining the predictions to make a strong learner. The
notion is that various classes of neural networks are created as base learners and then
combined as a strong learner to predict the storm surge [55]. When ensemble members
employ a single-type base learning algorithm but are generated upon a different subset of
training data, they are classified as homogeneous [67,89]. Heterogeneous ensembles, on the
other hand, consist of classifiers (base learners) of different types, such as MLP, CNN, or
LSTM, which are usually trained on the same dataset [67,90]. These ensemble models are
designed such that base learners are generated in sequential or parallel format. The basic
motivation of the former is to create successive learning algorithms over iterations where
predictions of a base learner are corrected and fine-tuned, then provided to the subsequent
base learners. In the latter, the base learners are generated in parallel and independent
from each other. Predictions of the diverse base learners are then combined using ensemble
learning techniques such as bagging and stacking. These methods can potentially reduce
the inference time (the amount of time taken for a forward propagation) and increase the
overall performance [91].
Generating NN ensembles that predict storm surge heights from historical, synthetic,
or predicted hurricanes and/or are able to estimate overland flooding (or surge-induced
J. Mar. Sci. Eng. 2023, 11, 2154 9 of 30

maximum inundation) requires supervised algorithms to learn how to fit the input labeled
data into a continues function [89,91]. This raises the question of how to incorporate
predictions from different models. In this regard, three leading algorithms for combining
weak learners are recognized.
Bootstrap aggregating (bagging): To ensure diversity among base learners, one notion
is to train each learner on a distinct subset of the available training data. An autonomous
training process can be conducted in parallel for each learner through a popular subsam-
pling ensemble method known as bootstrap aggregation, more commonly referred to as
bagging [91,92]. This method uses randomly generated training sets (extracted from the
initial preprocessed dataset) to obtain an ensemble of predictors and subsequently trains an
integrated neural network associated with training sets (Figure 4). Bagging can consider-
ably reduce variance and is an efficient solution to overfitting [92–94] (i.e., it helps with the
generalization of a NN ensemble model to unseen data). Given a series of extreme flood-
ing events in coastal regions with noisy data obtained from the tide stations, particularly
during times when a storm surge coincides with normal high tide, the bootstrap learning
approach could effectively combine uncertainties originating from various measurements.
In a meteorological forecast of the storm’s behavior, for instance, this approach involves
random sampling of the initial training dataset through standard bagging resampling with
replacement, thus resulting in a low-variance ensemble model [95]. In a regression problem,
assuming that the model is trained on the input vector of A = ([ x1 , y1 ], [ x2 , y2 ], . . . , [ xn , yn ]),
to learn the mapping yi = f ( xi ), i = 1, . . . , n, bootstrap aggregation takes the average of
the predictions yi from a collection of bootstrap samples A∗j , j = 1, . . . , m. Each sample
is independent and drawn uniformly among A1∗ , . . . , A∗m with replacement; thus, all the
samples are independent and identically distributed (i.i.d) [92]. The aggregated (bagged)
prediction for each base learner is expressed by

∑m ∗
j =1 A j ( x )
ybs = (1)
m
where A1∗ ( x ), . . . , A∗m ( x ) are the predictions from the i.i.d samples. This method limits
the variance through building different base learners of diverse datasets [96] and helps to
create a more stable and robust overall model. This can be particularly useful in situations
where the data are noisy or where there is high variability in the predicted outcome, such
as in predicting the effects of category 4 and 5 storms. Since ensemble models with low
correlations are preferred in these predictions, the sampling with replacement method
allows more difference in the training dataset and, in turn, results in greater differences
between the predictions of the base learners. It is worth mentioning that the bagging
process, depending on its number of iterations or combination with time series, could be
computationally demanding to fit, as explained in [97]. Figure 5 shows a pseudo-code
for a bagging NN ensemble algorithm; note that this is a simple example, and the actual
implementation of bagging in neural networks may vary depending on each specific case
and library. Additionally, this example does not cover how to handle the overfitting
problem that might occur on these models.
Boosting: This ensemble approach works in a forward stagewise process and learns
the predictions from the previous weak learner by adjusting the weighted data and fitting
the model to an updated training dataset in a sequential order [98] (Figure 6). In the case of
regression, the final output is usually built as the weighted average of a sequence of the
fitted base learners [96,99]. A boosting algorithm reduces the bias owing to the progressive
refinement of the base learner over time [100]. The AdaBoost algorithm, short for Adaptive
Boosting, is one of the most popular boosting algorithms [101]. In this approach, instead of
dividing a training dataset, multiple classifiers are iteratively constructed from the entire
dataset. Using the neural network ensemble model, the subsequent component highlights
the false prediction of the previous step to transform a weak learner into a strong learner. In
other words, training data inaccurately predicted by the former NN become more influential
in the training of the latter NN [92]. This learning approach could be extended to neural
J. Mar. Sci. Eng. 2023, 11, 2154 10 of 30

network ensembles aiming at predicting storm surges or generating a mean estimation of

residual water levels [102]. Figure 7 shows a pseudo-code based on the AdaBoost algorithm
[99]. It is important to note that the actual implementation of boosting in neural networks
may vary depending on the case and library that is implemented. Additionally, there are
other boosting algorithms, such as Gradient Boosting [103] or XGBoost [104], that have
some variations in their pseudo-code.

Random Sampling

Bootstrap 1 P1

Bootstrap 2 P2 Storm Surge

Height

Final Prediction
Original Dataset
Bootstrap n P3

Base Classi�iers Training Stage Bagged Classi�iers

Figure 4. A general scheme of the bagging ensemble approach.

Step 1: Initialize the ensemble

ensemble_models = []
Step 2: Build base ensemble models
for i in range(num_ensemble_models):
# Sample dataset with replacement to create a new training set
sample_data = random.sample(original_data, len(original_data))
# Train a model on the sampled dataset
model = train_neural_network(sample_data)
# Add the trained model to the ensemble
ensemble_models.append(model)
Step 3: Make predictions
def ensemble_predict(ensemble_models, input_data):
predictions = []
for model in ensemble_models:
predictions.append(model.predict(input_data))
# Average the predictions to get the final ensemble prediction
ensemble_prediction = np.mean(predictions)
return ensemble_prediction

Figure 5. A simplified pseudo-code of an ensemble learning algorithm for bagging.

J. Mar. Sci. Eng. 2023, 11, 2154 11 of 30

resampling
Calculate the accuracy

Random Sampling
Storm Surge
Height

Original Dataset Create Base Classi�ier Training Stage Final Prediction

Figure 6. A general schematic of the boosting ensemble approach.

Step 1: Initialize the ensemble with a base model

ensemble_models = [base_model]
Step 2: Expand ensemble
for i in range(num_iterations):
# Predictions with current ensemble
ensemble_predictions = ensemble_predict(ensemble_models, data)
# Calculate the error of the ensemble predictions
error = calculate_error(ensemble_predictions, data_labels)
# Train a new neural network to predict the error
new_model = train_neural_network(error)
# Add the new model to the ensemble with a weight
ensemble_models.append((new_model, weight))
Step 3: Make predictions
def ensemble_predict(ensemble_models, input_data):
# Get weighted predictions
predictions = []
for model, weight in ensemble_models:
prediction = weight * model.predict(input_data)
predictions.append(prediction)
# Sum the predictions to get the final ensemble prediction
ensemble_prediction = np.sum(predictions)
return ensemble_prediction

Figure 7. A simplified pseudo-code of an ensemble learning algorithm for boosting.

Assuming that each of n base learners make a prediction yi out of a random sample,
the weighted average of the boosted model would be [105]

∑nj=1 βyi ( x )
ybt = (2)
n
where β is the shrinkage coefficient that controls the rate at which the boosting algorithm
reduces the error. β is similar to the learning rate hyperparameter in NN.
When using synthetic storm data to support the incomplete dataset (or data which
cannot capture an event resulting from instrument failures), it is possible that the generated
dataset could be more biased and less accurate than real-world data, such as in tide
stations [88,106]. Boosting algorithms focus on weak learners to determine which factors
are contributing to false outcomes and treat those factors carefully in testing data, decreasing
the bias error.
J. Mar. Sci. Eng. 2023, 11, 2154 12 of 30

Stacking: Stacked generalization, also known as stacking, is a heterogeneous ensemble

strategy proposed by Wolpert [106] to train a set of diverse weak learners in parallel with
greater predictive accuracy. Base learners (also called level 0/first-level learners) serve as
input to run a combiner or meta-learner (also called the level 1/second-level/super learner)
(Figure 8). Both the precision and diversity of base learners are crucial to the performance
of a stacking ensemble such that various base learners could construct a well-functioning
model with improved results [107].

Model 1 P1

Meta-model Storm Surge

Model 2 P2
Height

Level 1 Final prediction

Original Dataset Meta learner
Model n Pn

Training subsets Level 0 Predictions

base learners

Figure 8. A general scheme of the stacking ensemble approach.

The predictive performance of a stacking ensemble is influenced by the number of

individual base learners [107,108]; however, there are only a few NN combinations available
(as explained in Section 3.3) to investigate the accuracy of combined predictions associated
with different combinations of base learners. Choosing the optimal subset of stacked base
learners is explained in [109–111]. Figure 9 shows a pseudo-code for the stacking ensemble
algorithm. It is to be noted that the provided snippet code is a basic instance, and the actual
implementation of stacking in neural networks might differ according to each specific case
and the implemented methods. Other stacking ensemble algorithms include Blending [112]
and Super Learner [113], which have some variations in this pseudo-code. Let yim = f ( xi )
represent the mapping function applied to the model m with N = 1, . . . , i observations
in the training set N, where predictions from a set of heterogeneous weak learners (sub-
models) m = 1, 2, . . . , M are combined as new training data for the metalearner. The
stacking weights are defined as the minimum value of the Euclidean distance between the
weighted prediction and the target yi [114]
 " # 
N M 2

Wst = arg min ∑ yi − ∑ (m) (m)

W . yi  (3)
i =1 m =1

M  
which leads to the final stacked ensemble prediction yst = ∑ Wst . y(m) .
m =1
Here, the learning method to train the metalearner is based on the most common form
of regression analysis, linear regression. High-fidelity ocean circulation models such as
ADCIRC predict a skewed distribution of the peak storm surge height at the early stages or
with biased subsets of training datasets [42]. Stacking ensembles can help to mitigate the
effects of data bias and improve the overall performance of the model since they take into
account the strengths and weaknesses of sub-models and make robust predictions to the
biases that may be present in any individual subset.
J. Mar. Sci. Eng. 2023, 11, 2154 13 of 30

Step 1: Initialize the ensemble with base models

ensemble_models = [base_model_1, base_model_2, ...]
Step 2: Generate meta-features based on base model predictions
meta_features = []
for model in ensemble_models:
predictions = model.predict(original_data)
meta_features.append(predictions)
meta_features = np.concatenate(meta_features, axis=1)
Step 3: Train a new neural network on the meta-features
meta_model = train_neural_network(meta_features, original_data_labels)
Step 4: Add the meta-model to the ensemble
ensemble_models.append(meta_model)
Step 5: Make predictions
def ensemble_predict(ensemble_models, input_data):
base_predictions = []
for model in ensemble_models[:-1]:
base_predictions.append(model.predict(input_data))
meta_features = np.concatenate(base_predictions, axis=1)
ensemble_prediction = ensemble_models[-1].predict(meta_features)
return ensemble_prediction

Figure 9. A simplified pseudo-code of ensemble learning algorithm for stacking.

An overview of six different studies is outlined in Table 3, summarizing the utilization

of ensemble approaches and evaluation metrics, along with the data collection sources
for each study. A comparative analysis is illustrated in Figure 10 based on a qualitative
reference value (rv) and a representative skill metric (sm) across the different studies
summarized in Table 3.
rv − sm
Relative Score = (4)
rv

10
Study # 1 2 3 4 5 6

8
Rela�ve Score

Run�me Resource U�liza�on Scalability Model Complexity

Figure 10. Qualitative assessment of studies numbered 1 to 6 from Table 2.

J. Mar. Sci. Eng. 2023, 11, 2154 14 of 30

Table 3. Comparative analysis of ensemble approaches, evaluation metrics, and data collection in
different studies (2015–2022).

Study Ensemble Evaluation

Target Goal Methodology Data Collection
Number Approach Metric
Low-probability peak ANN and coupled GBDTR and Synthetic TCs + Historical
RAE, MRAE,
1 [42] storm surge height ADCIRC + SWAN AdaBoost typhoon data in the New
and RMSE
due to TCs simulations Regressor York metropolitan area
Stacking (super-
Hydrodynamic and ensemble) RMSE, US mid-Atlantic and North-
Storm tide and resur-
2 [115] Hydrologic Ensemble based on PRE, east coastline wind and tide
gence
Forecast RMSE and bias and COU data
correction
Hourly surge time se- ANN, CNN, LSTM, Bootstrap RMSE GESLA Version 2 tide station
3 [43]
ries at the global scale and ConvLSTM aggregation and CRPS database
Average
C1PKNet (1D CNN,
of ten trained
Peak storm surges from principal component NACCS synthetic TC surge
4 [37] C1PKNet MSE and CC
TC track time series analysis, and k-means database
model
clustering)
predictions
Storm surge level time series
Real time and accurate CNN and LSTM, RMSE, MAE,
5 [83] – in the southeastern coastal re-
storm surge transfer learning and CC
gion of China
Rapid prediction of ANN and CSTORM- Synthetic storms in the Gulf
6 [116] – RMSE and CC
storm surge time series MS coupled model of Mexico
GBDTR = Gradient Boosted Decision Tree Regressor; RAE = relative absolute error; MRAE = mean relative
absolute error; RMSE = root-mean-square error; MAE = mean absolute error; CC = correlation coefficient;
PRE = peak relative error; COU = coverage of observation uncertainties; CRPS = continuous ranked probability score.

4. Ensemble Pruning and Fine-Tuning

An ensemble model is a systematic process of combining individual diverse base
predictive learners to produce robust and accurate predictions. The concept of an ensemble
model might be potent enough for the default parameters to shine; however, many studies,
such as [117–122], acknowledge that the accuracy could be improved further through
tuning. An intuitive approach is to alter the network’s setup in a process known as pruning.
This is followed by fine-tuning the hyperparameters of the diverse base learners through the
regular process of developing the networks. Pruning entails reducing trivial (or redundant)
parameters from an existing network systematically [123]. In the case that the model has
poor performance after pruning, the hyperparameters are fine-tuned, i.e., the parameters
of each individual model are adjusted, and then the models are retrained to restore the
best possible accuracy [121]. The result is an ensemble of relatively accurate and robust
fine-tuned models with a lower correlation between the independent predictions and
residuals [119]. A general scheme on pruning and fine-tuning steps in a neural network
ensemble is shown in Figure 11.
Pruning: The main idea of pruning networks is to reduce the complexity and energy
required to implement large trained networks and make predictions on new input data in
real time [124]. This could be a crucial stage in predicting storm surge time series [54,55],
such that accurate real-time predictions of storm surge can help emergency management
officials issue evacuation orders, take preemptive measures to protect infrastructures, and
minimize the economic impact of the storm. Typically, the initial network is large and
tends to achieve higher accuracy; generating a smaller network with comparable preci-
sion is preferable. This approach has seen a significant amount of growth over the past
decade [123]. However, a handful of studies, such as [125–127], addressed the process of
ensemble pruning, especially in predicting time series of water surface elevations during
or after storms. One major reason is that some ensemble techniques, such as the Adaboost
J. Mar. Sci. Eng. 2023, 11, 2154 15 of 30

algorithm, inherently mitigate overfitting by independently optimizing input parameters

to reach an optimal value. Once the accuracy of individual base learners slightly surpasses
random guessing, the final model is proven to reduce generalization error, yielding en-
hanced performance as a strong learner [123]. Furthermore, NN ensemble pruning can also
be interpreted as a special type of stacking technique (as introduced in Section 3) in which
a meta-learner is applied to improve the predictive performance of the models [128].

NNE Performance Evaluation

Target Accuracy
No Achieved ? Yes

Training with Adjusted

Learning Rate and
Overfitting Control
Yes
Selective Removal of
Weights and Connections

Re-training No Network Initialization

for Improved with Learned Weights
Accuracy ?

Re-evaluation of Pruned
Network on Validation Set
Creating a Representative
Subset of dataset

Pruning Fine-tuning

Figure 11. General process of pruning and fine-tuning in a neural network ensemble.

The major pruning techniques that are applicable to NN ensembles are as follows:
(1) weight decay [129], which involves adding a regularization term to the loss function
that penalizes the complexity of the ensemble; (2) an error-based approach [130], which
involves calculating the prediction errors of each network in the ensemble and removing
the networks with the highest error rates; and (3) neuron pruning [131], which involves
removing the neurons in each network of the ensemble that have the least impact on the
network’s output.
Fine-tuning: Once a pruned ensemble is created, the next common stage is to perform
fine-tuning, where the network is retrained using the pruned architecture, possibly with a
smaller learning rate and fewer training epochs. Fine-tuning can help restore some of the
accuracy lost during pruning and can lead to better generalization performance [132].
Tuning methods cannot be overlooked since less complex but fine-tuned real-time
predictive models could possibly result in accurate predictions of water level and flood
extent [118,119]. which are essential for real-time monitoring and timely warnings of poten-
tial floods. When constructing predictive models, finding a set of optimal hyperparameters
for each individual learner is a challenge. Tuning the base models (learners) individually
and tuning all the models in an ensemble simultaneously are the two fundamental methods
to determine the optimal parameters [67]. In the former approach, the hyperparameter
tuning process for each base model is often carried out as an independent procedure based
on unique sets of hyperparameters. To illustrate, different base models in an ensemble
may use different types of activation functions, optimization algorithms, regularization
techniques, or learning rates. Tuning these hyperparameters separately can help ensure
that each model is individually optimized and contributes to the overall performance of the
ensemble. This conventional approach is described in [133,134]. It is important to note that
the hyperparameter tuning process should also take into account the interactions between
the base models in the ensemble [128,133] (the later approach). The weights assigned to
J. Mar. Sci. Eng. 2023, 11, 2154 16 of 30

each base model have a significant impact on the overall performance of the ensemble, so
these weights may also need to be tuned in conjunction with the hyperparameters of each
individual model. Such a kind of connection is usually more compatible with probabilistic
approaches, such as Bayesian optimization [135]. This method usually involves modeling
the objective function (e.g., accuracy) as a Gaussian process [136], which can be more effi-
cient than other fine-tuning methods, such as grid search [137] and random search [138],in
some cases, as it leverages previous evaluations of the objective function to better guide the
search process [139].

5. Data Preparation
Data preparation in neural network ensembles refers to the process of preprocessing
and organizing raw data before training a group of neural networks together as an en-
semble [140]. The goal of this crucial step is to ensure that the input data are consistent,
relevant, and suitable for use by the ensemble, which can lead to better model performance
and more accurate predictions. A dataset in a traditional ANN can be represented as a
set of input–output pairs, where the input is a vector of features and the output is a scalar
target value [47]. In a regression problem such as water level prediction, a dataset of size N
would be stored as follows:

x11 x12 ... x1j

 
 x21 x22 ··· x2j 
(5)
 
 . .. .. 
 .. . . 
xi1 ··· xij
Each ith row is an observation in the dataset, and each
jth column represents an
individual component of an observation in the dataset xij ∈ R . In contrast to an ANN, the
input xi in convolutional neural networks is a 2D or 3D matrix of pixel values representing
an image with dimensions (height, width, and channels), and a set of convolutional filters
are applied to detect patterns in the image [141]. However, they can also be applied to time
series data by treating the time dimension as a spatial dimension; thus, the input would be
a 1D sequence of data points and a set of 1D filters, which are applied to detect patterns
in the time series [142]. By structuring the time series as a sequence, a CNN can detect
local patterns that correspond to different storm events or meteorological conditions over
shorter time intervals [62].

5.1. Raw Input Data

Datasets are an integral part of ensemble models, and major improvements in the final
prediction highly depend on the availability of high-quality input and training datasets.
There is a diverse assortment of sources and domains that provide data on the oceans and
coasts of the United States. These data can be utilized to improve hurricane prediction
models and create strategies for coping with the impact of climate change on coastal
communities, including rising sea levels [143,144]. With current developments, researchers
can generate various independent records of tropical cyclone datasets from the measured
tide and oceanographic data (Table 4) or take advantage of hindcasting (a retrospective
analysis of past weather conditions) and reanalyzing archives (a more comprehensive
and detailed reconstruction of observations combined with numerical models), such as
high-resolution temperature, pressure, humidity, and wind datasets from a forecast system
(Table 4). Some systems are adept at computing random, short-crested waves in coastal
regions using third-generation wave models, such as WAVEWATCH III, WAM, or SWAN,
or coupling them with other finite-element-based hydrodynamic models [35,36], such as
ADCIRC. Atmospheric and tidal forcing is commonly applied to high-resolution wave
models such as ADCIRC or SWAN [37,52] to simulate the behavior of ocean waves under
different storm conditions and generate synthetic storm datasets that can be used for
assessing flood risk and improving coastal management strategies [11].
J. Mar. Sci. Eng. 2023, 11, 2154 17 of 30

Ensemble NN models have high variability in their input data type and are commonly
considered heterogeneous. While homogeneity could be a desirable property of the in-
put data (in terms of the features and their scales) for neural networks, a heterogeneous
dataset in a regression problem such as storm surge prediction may work better [145]
because it includes a variety of features that capture different aspects of the storm and its
effects on the surge. This helps the neural network learn more robust and diverse features
that can be better generalized to new, unseen data [92,94]. Table 4 presents brief descrip-
tions and features of the ocean datasets that have been extensively used to predict storm
surge levels and flood extents. These datasets address a wide range of features, including:
(1) storm characteristics, such as storm intensity, wind speed and direction, and track;
(2) oceanographic features, such as water temperature, salinity, and currents; (3) meteoro-
logical features, such as air pressure, temperature, and humidity; (4) geographical features,
such as the shape and slope of the coastline, the depth of the ocean floor, islands, and
shoals, and (5) historic storm surge records, including the timing, intensity, and duration of
the surge. Common points and major differences between these datasets are outlined in
Table 5.

Table 4. Description and main features of the most widely used storm and flood datasets. The symbol
3 indicates that the feature is included, while the symbol 7 signifies that the feature is not included.

Dataset Description Features Source

3 Consistent across the entire North
Atlantic Coast region.
A combined set of 1050 syn- 3 Covers storm surge, sea level rise,
North Atlantic Coast thetic tropical and 100 syn- and erosion The U.S. Army
Comprehensive thetic extratropical storms us- 3 Easily accessible Corps of Engineers
Study (NACCS) ing the coupled ADCIRC/ST- 7 Coarse spatial resolution (USACE) [146]
WAVE models 7 Limited temporal scope
7 Relies on certain assumptions and un-
certainties
3 High temporal and spatial resolution
3 Covers a wide range of atmospheric
The latest generation of atmo-
variables Copernicus Climate
spheric reanalysis of the global
ECMWF Re-Analysis 3 Publicly available Change Service (C3S),
climate with detailed informa-
(ERA5) 7 Complex and may require advanced the joint C3S-NOAA
tion on a wide range of atmo-
technical skills project [147,148]
spheric variables.
7 Limited vertical resolution (137 pres-
sure levels)
3 Covers a wide range of extreme sea-
level events
3 Consistent across the entire globe and
Provides 39148 years of sea
different geographic locations University of Hawaii
level data from 1355 station
Global Extreme Sea- 3 Publicly available and the National
records, with information on
Level Analysis Version 2 7 Gaps in the data particularly for re- Oceanic and Atmo-
extreme sea levels, including
(GESLA-2) mote or sparsely populated regions. spheric Administra-
storm surges, tidal cycles, and
7 Relies on certain assumptions and un- tion (NOAA) [2]
rise in sea level.
certainties
7 Limited information on coastal mor-
phology and human activities
J. Mar. Sci. Eng. 2023, 11, 2154 18 of 30

Table 4. Cont.

Dataset Description Features Source

3 Provides high-quality and updated
oceanographic and meteorological data
in real time
3 Global coverage
provides nowcasts (analyses of
3 High spatial and temporal resolution
near-present conditions) and
3 Integration with other models for a National Centers
NOAA Global Real Time forecast guidance on up to
more comprehensive understanding of for Environmental
Ocean Forecasting Sys- eight days of ocean tempera-
storm surge Prediction (NCEP),
tem (RTOFS global ture and salinity, water veloc-
7 Limited data availability for a partic- NOAA [4]
ity, sea surface elevation, sea ice
ular area or time period
coverage, and sea ice thickness.
7 Relies on certain assumptions and un-
certainties
7 Requires significant computational
resources
3 High-quality data
3 High spatial resolution with detailed
National coastal storm hazard information about storm surge patterns
data resource for probabilis- 3 Provides historical data Pacific Coastal and
tic coastal hazard assessment 7 Limited to the coastal areas of the Marine Science Cen-
Coastal Hazards System
(PCHA) results and statistics, United States ter of the United
(CHS)
including measurements of wa- 7 Limited temporal resolution for pre- States Geological
ter level, wind speed, and wave dicting a storm surge during an ongoing Survey (USGS) [149]
height event
7 Needs to be integrated with other
models to make accurate predictions
3 Specifically designed and tested for
storm surge prediction
3 Can be customized to specific geo-
Uses a combination of histori-
graphic areas
cal storm data, topographical
3 Can be integrated with other models,
data, and numerical algorithms National Oceanic
The Sea, Lake and Over- such as atmospheric and wave models
to simulate the impact of a hur- and Atmospheric
land Surges from Hurri- 7 Resource-intensive
ricane on coastal areas and pre- Administration
canes (SLOSH) model 7 Limited data availability (requires in-
dict storm surge heights and (NOAA) [150]
put data, such as atmospheric pressure
flooding potential associated
and wind speed)
with hurricanes.
7 Limited spatial resolution
7 Relies on certain assumptions and un-
certainties
3 Specifically designed and tested for
storm surge prediction
3 Provides historical data National Oceanic and
3 Wide geographic coverage through- Atmospheric Admin-
National Water Level A network of tide gauges that out the United States istration’s (NOAA)
Observation Network can be used for storm surge pre- 7 Limited spatial resolution Center for Opera-
(NWLON) diction. 7 Lack of a comprehensive model for tional Oceanographic
predicting storm surge (needs to be inte- Products and Services
grated with other models) (CO-OPS) [151]
7 Limited data availability in all coastal
regions
J. Mar. Sci. Eng. 2023, 11, 2154 19 of 30

Table 5. General comparison between the datasets in Table 4. The symbol 3 indicates that the feature
is included, while the symbol 7 signifies that the feature is not included.

NACCS ERA5 GESLA2 RTOFS CHS SLOSH NWLON

Spatial reso-
0.25 degrees 0.25 degrees 0.25 degrees 0.08 to 0.25 degrees 0.02 to 0.05 degrees 0.02 to 0.05 degrees 0.08 to 0.33 degrees
lution
Temporal
6h Hourly Monthly Hourly Hourly Hourly Hourly
resolution
Atlantic and Gulf
North Atlantic Coastal areas of the Coastal areas of the
Coverage Global Global Global coasts of the
Coast region United States United States
United States
Open access
(needs license
Availability Open access Open access Open access Limited access Limited access Open access
for real-time
products)
Complexity Highly complex Highly complex Complex Complex Fairly complex Complex Fairly complex
Missing or incom-
Incomplete cov- Limited or no Incomplete cover- Incomplete cover- Missing or incom- Incomplete cover-
plete weather sta-
Possible erage or missing data on certain age or missing data age or missing data plete data for cer- age or missing data
tion data in cer-
data gap data for certain sea levels and for certain time for certain time tain hurricanes or for certain time
tain regions or pe-
time periods time periods periods periods regions periods
riods
Integration
with other 3 7 3 3 3 3 3
models

5.2. Data Preprocessing and Wrangling

Data preprocessing and wrangling are critical steps in any machine learning workflow,
and they often take up a significant amount of time and effort [140,152]. The pre-mentioned
datasets may contain several types of data issues that need to be addressed and prepro-
cessed before NN algorithms can be applied effectively. Some of the most common issues
include missing values, outliers, categorical data (such as storm category, wind direction,
tidal phase, landfall location, and storm direction), correlated and irrelevant features, and
issues related to scaling and normalization [152,153]. The dataset presented in Table 6 dis-
plays a subset of hurricane Harvey’s tracking data (Figure 1) derived from the International
Best Track Archive for Climate Stewardship (IBTrACS) [154,155], which, although com-
prehensive, requires careful data processing to be suitable for ENN. Here, the maximum
sustained wind speed reported from multiple agencies for the current location needs to
be converted to a unified 10 min sustained wind speed. Then, important features must be
extracted and interpolated according to desired time steps. Missing values are handled
using interpolation or imputation techniques, such as mean imputation or predictive mod-
eling. Another dataset can be found in [156], where both recent and historical standard
meteorological and water level information is provided by the National Data Buoy Center
(NDBC). The data can be collected from the stations near an area of interest (Port Aransas,
Texas) combined with the extracted TC tracks and then fed into the ENN model.
As mentioned in Section 3.3, data-driven models are usually agnostic to physical laws
because they rely only on data. However, it is important to note that while data-driven
models do not explicitly incorporate physical laws, they can still be used to make predic-
tions about physical phenomena based on empirical data [55,56]. For example, a NN model
can be trained on data from a time series of gauge data to predict the uncertainty related to
storm surge flooding [55,57], even if the underlying physical laws are not fully understood
or modeled. Therefore, the accuracy and reliability of data are heavily influenced by the
quality of data preprocessing steps, such as cleaning and filtering the data, handling miss-
ing values, normalizing or scaling the data, and feature selection or extraction. Last but
foremost, some important issues related to the data preprocessing stage that can impact the
performance of NN ensemble are as follows:
J. Mar. Sci. Eng. 2023, 11, 2154 20 of 30

Table 6. Sample best-track dataset associated with hurricane Harvey (2017) in the North Atlantic
basin [154,155].

SID ISO_TIME NATURE LAT LON WMO_WIND WMO_PRES DIST2LAND LANDFALL

degrees_N degrees_E kts mb km km
2017228N14314 8/25/2017 3:00 TS 25.2924 −94.7578 243 204
2017228N14314 8/25/2017 6:00 TS 25.6 −95.1 90 966 204 170
2017228N14314 8/25/2017 9:00 TS 25.935 −95.4651 160 133
2017228N14314 8/25/2017 12:00 TS 26.3 −95.8 95 949 133 123
2017228N14314 8/25/2017 15:00 TS 26.6999 −96.0652 126 108
2017228N14314 8/25/2017 18:00 TS 27.1 −96.3 105 943 108 67
2017228N14314 8/25/2017 21:00 TS 27.4875 −96.5806 67 34
2017228N14314 8/26/2017 0:00 TS 27.8 −96.8 115 941 34 11
2017228N14314 8/26/2017 3:00 TS 28 −96.9 115 937 11 0
2017228N14314 8/26/2017 6:00 TS 28.2 −97.1 105 948 0 0
2017228N14314 8/26/2017 9:00 TS 28.4534 −97.2205 0 0

Data cleaning: Large amounts of data from various sources, such as weather sen-
sors, tide gauges, and satellite imagery, can be prone to errors, missing data, and outliers,
which can significantly affect the accuracy of the model’s predictions. Therefore, it is
essential to perform data cleaning to remove any errors or inconsistencies in the data before
feeding it into the neural network ensemble model [157]. This process may involve identi-
fying and removing outliers, handling missing data through imputation, and smoothing
noisy signals.
Feature scaling: Neural networks require all features to be on the same scale to
ensure that no feature dominates the others, where feature scaling techniques such as
normalization, standardization, or range scaling can be applied [37]. Choosing the wrong
scaling technique can lead to poor model performance. In storm surge prediction, input
features such as sea level, wind speed, and atmospheric pressure can have very different
scales and ranges. Therefore, it is important to apply feature scaling to ensure that all
features have a similar impact on the model’s predictions.
Feature selection: Ensemble models can have a large number of features, which
can lead to overfitting and poor generalization. The input features may include various
meteorological and oceanographic variables, such as wind speed, air pressure, water
temperature, tidal levels, and ocean currents. However, not all of these features may
be equally important for predicting storm surges. By removing irrelevant or redundant
features, the model can focus on learning the most important patterns in the data, leading to
more accurate predictions [83]. There are various techniques for feature selection (including
filter methods, wrapper methods, and embedded methods) which can be applied before or
during training the NN ensemble model to select the most relevant features.
Data transformation: The goal of data transformation is to convert the input data into
a format that is more suitable for analysis and modeling by the neural network ensemble.
Transforming data to fit a particular distribution can improve the performance of neural
network ensembles and lead to more accurate and robust predictions of storm surges [158].
Some common data transformation techniques include normalization, logarithmic transfor-
mation, PCA transformation, and discretization. However, it is important to choose the
right transformation technique to avoid introducing noise into the data.
Handling class imbalance: This refers to a situation where the distribution of the target
variable is heavily skewed towards one class (base model). In such cases, failing to handle
the class imbalance can lead to biased models with inaccurate predictions that perform
poorly on the minority classes [54]. Various techniques for handling class imbalances
include resampling, synthetic data generation, and cost-sensitive learning.

6. Model Selection and Evaluation

There is no optimal ensemble configuration for predicting peak surge levels under
different scenarios. It is essential to carefully evaluate the performance of different ensem-
ble models and select the one that provides the best trade-off between bias and variance,
accuracy, diversity, stability, generalization, and computational cost [67,91,92,159]. The
J. Mar. Sci. Eng. 2023, 11, 2154 21 of 30

final stage would evaluate and validate the performance of the selected ensemble model
using appropriate evaluation metrics and statistical tests, such as the mean absolute er-
ror (MAE) [21,83,160], root-mean-squared error (RMSE) [106,161], correlation coefficient
(CC) [49,83,106,161], and coefficient of determination (R-squared) [42,43,161–163]. The
following section covers some of the fundamental concepts that are considered when
evaluating a neural network ensemble for storm surge prediction.

6.1. Bias–Variance Tradeoff

The process of designing a NN ensemble, which involves combining multiple mod-
els or algorithms, can be optimized by finding the best balance between bias and vari-
ance [92,164]. Bias refers to the extent to which a model consistently misses the mark
in its predictions, while variance refers to the extent to which a model’s predictions are
sensitive to small perturbations in the training data. A good ensemble should strike a
balance between these two factors in order to minimize the overall prediction error [92]. To
achieve this balance, the optimal choice of weights for each base learner in the ensemble
needs to be determined. The weights are chosen such that they minimize the prediction
error of the ensemble. By doing so, the ensemble becomes more robust to different types
of data and can achieve better overall performance [83]. The bias-variance decomposition
of the mean squared error (MSE) is actually a method for analyzing the behavior of a
stochastic model [92,164,165]. Each individual base learner in the ensemble may have
some degree of stochasticity or variability in its predictions due to factors such as the
initialization of the weights or the selection of the training data. By decomposing the MSE
(between the estimated output variable y and the estimator f ( x )) into its bias and variance
components, it is possible to gain insight into the sources of error in the model [83,165]. For
a given sample dataset x, the error made by the estimator f ( x ) is defined as ε = f ( x) − y;
hence, the MSE of the estimator is defined as the expected value of the squared error,
i.e., MSE( f ( x) ) = E[ε2 ]. For every unseen sample x, the MSE can be decomposed as

E[( f ( x) − y)2 ] = Bias2 ( f ( x) ) + Var ( f ( x) ) + Var (ε) (6)

The last term in Equation (6) contains an irreducible error that is inherent in the
relationship between the input and output and cannot be reduced by any model. This
error arises from the fact that the input may not contain enough information to perfectly
predict the output or that there may be random variations in the data that cannot be
modeled [133,166]. Therefore, an ensemble model cannot reduce irreducible error, but it
can help improve the overall performance of the model by reducing the bias and variance.

6.2. Ensemble Diversity

Ensemble diversity can be particularly important to ensure that the ensemble is able
to accurately capture the complex dynamics of the ocean and the atmosphere that influence
storm surge. By using different training data or model architectures, the ensemble can
better account for different sources of uncertainty in the data and avoid overfitting to any
particular aspect of the data [83,165,166]. As discussed in Section 3.3, there are several tech-
niques that can be used to promote ensemble diversity, including bagging, boosting, and
stacking. One commonly used metric to evaluate ensemble diversity is cross-entropy. Cross-
entropy measures the difference between the predictions of each individual model and the
predictions of the ensemble [164,166,167]. A lower cross-entropy value indicates that the
ensemble is more diverse. Another metric to evaluate ensemble diversity is disagreement,
which measures the degree of disagreement between the predictions of each individual
model [168,169]. A higher disagreement value indicates that the ensemble is more diverse.
Correlation is another metric that can be used to evaluate ensemble diversity [83,106]. It
measures the degree of similarity between the predictions of each individual model. A
lower correlation value indicates that the ensemble is more diverse. When selecting the
final model for a neural network ensemble, a good approach is to choose the model that
achieves good individual performance while contributing to higher ensemble diversity.
J. Mar. Sci. Eng. 2023, 11, 2154 22 of 30

This can be done by evaluating each model’s performance on a validation set and then
evaluating the ensemble’s performance on a separate test set. The final model should
be chosen based on a combination of good individual performance and high ensemble
diversity, as measured by the chosen diversity metric.

6.3. Probabilistic Performance

The predictive ability of probabilistic models can be assessed by probabilistic per-
formance and skill metrics, which can also be used to select the final model in a neural
network ensemble considering ensemble diversity in storm surge prediction [43]. The most
commonly used probabilistic performance metrics are mentioned below. These metrics
can provide a more comprehensive evaluation of the performance of the models in the
ensemble, including their ability to accurately capture the uncertainty in the predictions.
Models that have good individual performance and contribute to higher ensemble diversity
should be chosen.
The Brier skill score (BSS) measures the skill of a forecast by comparing the predictions
with a reference forecast, such as a climatological forecast or a persistence forecast. The
BSS ranges from −∞ to 1, with a score of 1 indicating a perfect forecast and a score of
0 indicating no skill beyond the reference forecast. BSS can be used to evaluate the probabil-
ity of a surge or total water level exceeding a given threshold and thus yields the accuracy
of the system’s probabilistic forecasts [7,170].
The mean square skill score (MSSS) measures the improvement in the mean squared
error (MSE) of the forecast system relative to a reference forecast, such as a climatological
forecast or a persistence forecast. The MSSS ranges from −∞ to 1, with a score of 1
indicating perfect skill and a score of 0 indicating no improvement beyond the reference
forecast. When the system generates a probability distribution for the water level, the MSSS
can measure the improvement in the mean squared error of this distribution over a given
time period compared to the reference forecast [171,172]. The MSSS can be a useful metric
when the focus is on the mean of the forecast distribution rather than the full distribution
itself. However, it does not provide information on the reliability and resolution of the
forecast, which are important for assessing the quality of probabilistic forecasts.
The continuous ranked probability score (CRPS) is used to evaluate the accuracy of
probabilistic forecasts. It measures the distance between the cumulative distribution func-
tion (CDF) of the forecast probability distribution and the CDF of the observed outcomes.
The lower the CRPS, the better the forecast. When the system generates a probability
distribution for the water level, the CRPS can measure the accuracy of this distribution
over a given time period by comparing it to the observed water levels. The CRPS takes into
account both the reliability and sharpness of the forecast probability distribution, which
makes it a more informative metric than the Brier skill score in some cases [148].

7. Summary
The present paper focuses on various approaches that can predict storm surge levels
using ensemble neural networks. The challenges and limitations of accurately predicting
peak water levels, which are often caused by complex interactions between ocean currents,
winds, and atmospheric pressure systems, are also emphasized. Despite the limitations,
supervised neural networks, specifically those utilizing the backpropagation technique,
have proven to be a powerful tool for predicting storm surge levels, particularly for short-
term forecasting. However, the accuracy of BPNN models can be limited by overfitting,
which occurs when the model becomes too complex and fits the training data too closely. To
address the limitations of single BPNN models, ensemble methods that combine multiple
neural network models to improve accuracy and reduce overfitting are preferred. Ensemble
methods involve generating multiple base learners (weak classifiers) and combining their
predictions to create a strong learner. There are three leading meta-algorithms for combining
weak learners: bootstrap aggregating (bagging), boosting, and sitting. Bagging involves
generating multiple training datasets by randomly sampling from the original dataset
J. Mar. Sci. Eng. 2023, 11, 2154 23 of 30

with replacement, then training each base learner on a different dataset. Boosting involves
iteratively training weak classifiers, with each subsequent model focusing on the samples
that were misclassified by the previous model. Stacking involves training a meta-learner
that combines the predictions of multiple base learners. As the networks grow larger,
the importance of pruning and fine-tuning, as well as data preparation and wrangling,
become unquestionable. Data preparation involves preprocessing and organizing raw
data before training a group of neural networks together as an ensemble. The goal of this
crucial step is to ensure that the input data are consistent, relevant, and suitable for use by
the ensemble. The paper highlights different sources of input data type for storm surge
prediction and the need for careful data preprocessing and wrangling to ensure accurate
predictions. However, there is no one-size-fits-all approach for creating an ensemble of
neural networks for predicting storm surge levels. Instead, it is essential to carefully
evaluate the performance of different ensemble models and select the one that provides the
best trade-off between bias and variance, accuracy, diversity, stability, generalization, and
computational cost. Overall, the paper provides valuable insights into the use of ensemble
methods for storm surge flood modeling, which can contribute to better predictions and
preparedness for extreme weather events.
Author Contributions: Conceptualization, S.K.N. and M.B.; methodology, S.K.N. and D.V.S.; investiga-
tion, M.B., S.K.N. and D.V.S.; resources, R.J.W.; data curation, S.K.N.; writing—original draft preparation,
M.B.; writing—review and editing, D.V.S. and R.J.W.; visualization, S.K.N.; supervision, D.V.S.; funding
acquisition, D.V.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding
Institutional Review Board Statement: Not applicable
Informed Consent Statement: Not applicable
Data Availability Statement: Not applicable
Conflicts of Interest: The authors declare no conflict of interest.

Appendix A. Implementation of Backward Propagation of Errors

single layer
input output
w.x z h
x
* + f y
əL/əz əL/əh

W b əL/əW

Figure A1. Simplified BP algorithm in a 1-layer NN with 2D input.

• Defining the sigmoid activation function and its derivative

1 def activation ( x ) :
2 return 1 / (1 + np . exp ( - x ) )
3 def act ivati on_der ivati ve ( x ) :
4 return activation ( x ) * (1 - activation ( x ) )

• Defining the forward propagation function

1 def for ward_propagation (x , weights , biases ) :

2 a = [x]
3 z = []
4 for l in range (1 , len ( weights ) + 1) :
J. Mar. Sci. Eng. 2023, 11, 2154 24 of 30

5 z . append ( np . dot ( weights [ l ] , a [l -1]) + biases [ l ])

6 a . append ( activation ( z [l -1]) )
7 return a , z

• Defining the backward propagation function

1 def backward_propagation (x , y , a , z , weights , biases ,

learn in g_rate ) :
2 L = len ( weights )
3 delta = [ None ] * ( L + 1)
4 gradients = {}

∂L ∂L ∂h ∂z
• Running the error propagation using the chain rule ∂w = ∂h ∂z ∂w , h = f (z) , and Loss
n
1
Function L = n ∑ (h j − y j )
j =1

1 # Compute the output layer delta

2 delta [ L ] = ( a [ L ] - y ) * act ivatio n_der ivativ e ( z [L -1])
3 # Compute deltas for the hidden layers
4 for l in range (L -1 , 0 , -1) :
5 delta [ l ] = np . dot ( weights [ l +1]. T , delta [ l +1]) *
act ivatio n_der ivativ e ( z [l -1])
6 # Compute gradients for weights and biases
7 for l in range (1 , L +1) :
8 gradients [f ’ dW { l } ’] = np . dot ( delta [ l ] , a [l -1]. T )
9 gradients [f ’ db { l } ’] = delta [ l ]

References
1. Heberger, M.; Cooley, H.; Herrera, P.; Gleick, P.H.; Moore, E. Potential impacts of increased coastal flooding in California due to
sea-level rise. Clim. Chang. 2011, 109, 229–249. [CrossRef]
2. Woodruff, J.D.; Irish, J.L.; Camargo, S.J. Coastal flooding by tropical cyclones and sea-level rise. Nature 2013, 504, 44–52.
[CrossRef] [PubMed]
3. Barooni, M.; Nezhad, S.K.; Ali, N.A.; Ashuri, T.; Sogut, D.V. Numerical study of ice-induced loads and dynamic response analysis
for floating offshore wind turbines. Mar. Struct. 2022, 86, 103300. [CrossRef]
4. Cahoon, D.R.; Hensel, P.F.; Spencer, T.; Reed, D.J.; McKee, K.L.; Saintilan, N. Coastal wetland vulnerability to relative sea-level
rise: Wetland elevation trends and process controls. In Wetlands and Natural Resource Management; Springer: Berlin/Heidelberg,
Germany, 2006, pp. 271–292.
5. Dube, S.; Jain, I.; Rao, A.; Murty, T. Storm surge modelling for the Bay of Bengal and Arabian Sea. Nat. Hazards 2009, 51, 3–27.
[CrossRef]
6. Hashemi, M.R.; Spaulding, M.L.; Shaw, A.; Farhadi, H.; Lewis, M. An efficient artificial intelligence model for prediction of
tropical storm surge. Nat. Hazards 2016, 82, 471–491. [CrossRef]
7. Flowerdew, J.; Horsburgh, K.; Wilson, C.; Mylne, K. Development and evaluation of an ensemble forecasting system for coastal
storm surges. Q. J. R. Meteorol. Soc. 2010, 136, 1444–1456. [CrossRef]
8. Lynett, P.J.; Gately, K.; Wilson, R.; Montoya, L.; Arcas, D.; Aytore, B.; Bai, Y.; Bricker, J.D.; Castro, M.J.; Cheung, K.F.; et al.
Inter-model analysis of tsunami-induced coastal currents. Ocean. Model. 2017, 114, 14–32. [CrossRef]
9. Arabi, M.G.; Sogut, D.V.; Khosronejad, A.; Yalciner, A.C.; Farhadzadeh, A. A numerical and experimental study of local
hydrodynamics due to interactions between a solitary wave and an impervious structure. Coast. Eng. 2019, 147, 43–62. [CrossRef]
10. Al Kajbaf, A.; Bensi, M. Application of surrogate models in estimation of storm surge: A comparative assessment. Appl. Soft
Comput. 2020, 91, 106184. [CrossRef]
11. Qiao, C.; Myers, A.T.; Arwade, S.R. Validation and uncertainty quantification of metocean models for assessing hurricane risk.
Wind. Energy 2020, 23, 220–234. [CrossRef]
12. Arns, A.; Dangendorf, S.; Jensen, J.; Talke, S.; Bender, J.; Pattiaratchi, C. Sea-level rise induced amplification of coastal protection
design heights. Sci. Rep. 2017, 7, 40171. [CrossRef]
13. Weaver, R.J.; Slinn, D.N. Effect of wave forcing on storm surge. In Coastal Engineering 2004: (In 4 Volumes); World Scientific:
Singapore, 2005; pp. 1532–1538.
14. Sweet, W.V.; Kopp, R.E.; Weaver, C.P.; Obeysekera, J.; Horton, R.M.; Thieler, E.R.; Zervas, C. In Global and Regional Sea Level Rise
Scenarios for the United States; Technical Report; National Oceanic and Atmospheric Administration: Washington, DC, USA, 2017.
J. Mar. Sci. Eng. 2023, 11, 2154 25 of 30

15. Liu, Z.; Cheng, L.; Hao, Z.; Li, J.; Thorstensen, A.; Gao, H. A framework for exploring joint effects of conditional factors on
compound floods. Water Resour. Res. 2018, 54, 2681–2696. [CrossRef]
16. Xi, D.; Lin, N. Understanding uncertainties in tropical cyclone rainfall hazard modeling using synthetic storms. J. Hydrometeorol.
2022, 23, 925–946. [CrossRef]
17. Dtissibe, F.Y.; Ari, A.A.A.; Titouna, C.; Thiare, O.; Gueroui, A.M. Flood forecasting based on an artificial neural network scheme.
Nat. Hazards 2020, 104, 1211–1237. [CrossRef]
18. Velioglu, D. Advanced Two-and Three-Dimensional Tsunami Models: Benchmarking and Validation. Doctoral Dissertation,
Middle East Technical University, Ankara, Turkey, 2017.
19. Chen, Y.; Li, J.; Xu, H. Improving flood forecasting capability of physically based distributed hydrological models by parameter
optimization. Hydrol. Earth Syst. Sci. 2016, 20, 375–392. [CrossRef]
20. Agudelo-Otálora, L.M.; Moscoso-Barrera, W.D.; Paipa-Galeano, L.A.; Mesa-Sciarrotta, C. Comparación de modelos físicos y de
inteligencia artificial para predicción de niveles de inundación. Tecnol. Cienc. Agua 2018, 9, 209–235. [CrossRef]
21. Zhang, Z.; Liang, J.; Zhou, Y.; Huang, Z.; Jiang, J.; Liu, J.; Yang, L. A multi-strategy-mode waterlogging-prediction framework for
urban flood depth. Nat. Hazards Earth Syst. Sci. 2022, 22, 4139–4165. [CrossRef]
22. Oddo, P.C.; Lee, B.S.; Garner, G.G.; Srikrishnan, V.; Reed, P.M.; Forest, C.E.; Keller, K. Deep uncertainties in sea-level rise and
storm surge projections: Implications for coastal flood risk management. Risk Anal. 2020, 40, 153–168. [CrossRef]
23. Ju, Y.; Lindbergh, S.; He, Y.; Radke, J.D. Climate-related uncertainties in urban exposure to sea level rise and storm surge flooding:
A multi-temporal and multi-scenario analysis. Cities 2019, 92, 230–246. [CrossRef]
24. Makris, C.V.; Tolika, K.; Baltikas, V.N.; Velikou, K.; Krestenitis, Y.N. The impact of climate change on the storm surges of the
Mediterranean Sea: Coastal sea level responses to deep depression atmospheric systems. Ocean. Model. 2023, 181, 102149.
[CrossRef]
25. Camargo, S.J.; Barnston, A.G.; Zebiak, S.E. A statistical assessment of tropical cyclone activity in atmospheric general circulation
models. Tellus A Dyn. Meteorol. Oceanogr. 2005, 57, 589–604. [CrossRef]
26. Tadesse, M.; Wahl, T.; Cid, A. Data-driven modeling of global storm surges. Front. Mar. Sci. 2020, 7, 260. [CrossRef]
27. Bevacqua, E.; Maraun, D.; Vousdoukas, M.; Voukouvalas, E.; Vrac, M.; Mentaschi, L.; Widmann, M. Higher probability of
compound flooding from precipitation and storm surge in Europe under anthropogenic climate change. Sci. Adv. 2019,
5, eaaw5531. [CrossRef]
28. Jelesnianski, C.P. Numerical computations of storm surges without bottom stress. Mon. Weather. Rev. 1966, 94, 379–394. [CrossRef]
29. Kim, Y.H. Assessment of coastal inundation due to storm surge under future sea-level rise conditions. J. Coast. Res. 2020,
95, 845–849. [CrossRef]
30. Seo, J.; Ku, H.; Cho, K.; Maeng, J.H.; Lee, H. Application of SLOSH in estimation of Typhoon-induced Storm Surges in the Coastal
Region of South Korea. J. Coast. Res. 2018, 551–555. . [CrossRef]
31. Dietrich, J.C.; Tanaka, S.; Westerink, J.J.; Dawson, C.N.; Luettich, R.; Zijlema, M.; Holthuijsen, L.H.; Smith, J.; Westerink, L.;
Westerink, H. Performance of the unstructured-mesh, SWAN+ ADCIRC model in computing hurricane waves and surge. J. Sci.
Comput. 2012, 52, 468–497. [CrossRef]
32. De Las Heras, M.; Burgers, G.; Janssen, P. Wave data assimilation in the WAM wave model. J. Mar. Syst. 1995, 6, 77–85. [CrossRef]
33. Bender, C.; Smith, J.M.; Kennedy, A.; Jensen, R. STWAVE simulation of Hurricane Ike: Model results and comparison to data.
Coast. Eng. 2013, 73, 58–70. [CrossRef]
34. Booij, N.; Holthuijsen, L.; Ris, R. The “SWAN” wave model for shallow water. Coast. Eng. 1996, 668–676.
. [CrossRef]
35. Reffitt, M.; Orescanin, M.M.; Massey, C.; Raubenheimer, B.; Jensen, R.E.; Elgar, S. Modeling storm surge in a small tidal two-inlet
system. J. Waterw. Port, Coastal, Ocean. Eng. 2020, 146, 04020043. [CrossRef]
36. Ramos Valle, A.N.; Curchitser, E.N.; Bruyere, C.L.; Fossell, K.R. Simulating storm surge impacts with a coupled atmosphere-
inundation model with varying meteorological forcing. J. Mar. Sci. Eng. 2018, 6, 35. [CrossRef]
37. Lee, J.W.; Irish, J.L.; Bensi, M.T.; Marcy, D.C. Rapid prediction of peak storm surge from tropical cyclone track time series using
machine learning. Coast. Eng. 2021, 170, 104024. [CrossRef]
38. Smith, J.M.; Westerink, J.J.; Kennedy, A.B.; Taflanidis, A.A.; Cheung, K.F.; Smith, T.D. SWIMS Hawaii hurricane wave, surge, and
runup inundation fast forecasting tool. In Proceeings of the Solutions to Coastal Disasters Conference, Anchorage, AK, USA,
25–29 June 2011; pp. 89–98.
39. Torres, M.J.; Nadal-Caraballo, N.C.; Ramos-Santiago, E.; Campbell, M.O.; Gonzalez, V.M.; Melby, J.A.; Taflanidis, A.A. StormSim-
CHRPS: Coastal Hazards Rapid Prediction System. J. Coast. Res. 2020, 95, 1320–1325. [CrossRef]
40. Ishida, K.; Tsujimoto, G.; Ercan, A.; Tu, T.; Kiyama, M.; Amagasaki, M. Hourly-scale coastal sea level modeling in a changing
climate using long short-term memory neural network. Sci. Total Environ. 2020, 720, 137613. [CrossRef] [PubMed]
41. Tebaldi, C.; Ranasinghe, R.; Vousdoukas, M.; Rasmussen, D.; Vega-Westhoff, B.; Kirezci, E.; Kopp, R.E.; Sriver, R.; Mentaschi, L.
Extreme sea levels at different global warming levels. Nat. Clim. Chang. 2021, 11, 746–751. [CrossRef]
42. Ayyad, M.; Hajj, M.R.; Marsooli, R. Machine learning-based assessment of storm surge in the New York metropolitan area. Sci.
Rep. 2022, 12, 19215. [CrossRef]
43. Tiggeloven, T.; Couasnon, A.; van Straaten, C.; Muis, S.; Ward, P.J. Exploring deep learning capabilities for surge predictions in
coastal areas. Sci. Rep. 2021, 11, 17224. [CrossRef] [PubMed]
J. Mar. Sci. Eng. 2023, 11, 2154 26 of 30

44. Žust, L.; Fettich, A.; Kristan, M.; Ličer, M. HIDRA 1.0: Deep-learning-based ensemble sea level forecasting in the northern
Adriatic. Geosci. Model Dev. 2021, 14, 2057–2074. [CrossRef]
45. Ho, F.P.; Myers, V.A. Joint probability method of tide frequency analysis applied to Apalachicola Bay and St. George Sound, Florida; U.S.
Department of Commerce, National Oceanic and Atmospheric Administration, National Weather Service, Office of Hydrology:
Washington, DC, USA, 1975; Volume 18.
46. Feng, J.; Li, D.; Li, Y.; Liu, Q.; Wang, A. Storm surge variation along the coast of the Bohai Sea. Sci. Rep. 2018, 8, 11309. [CrossRef]
47. Ramos-Valle, A.N.; Curchitser, E.N.; Bruyère, C.L.; McOwen, S. Implementation of an artificial neural network for storm surge
forecasting. J. Geophys. Res. Atmos. 2021, 126, e2020JD033266. [CrossRef]
48. Igarashi, Y.; Tajima, Y. Application of recurrent neural network for prediction of the time-varying storm surge. Coast. Eng. J. 2021,
63, 68–82. [CrossRef]
49. Kim, S.W.; Lee, A.; Mun, J. A surrogate modeling for storm surge prediction using an artificial neural network. J. Coast. Res. 2018,
pp. 866–870. [CrossRef]
50. Royston, S.; Lawry, J.; Horsburgh, K. A linguistic decision tree approach to predicting storm surge. Fuzzy Sets Syst. 2013,
215, 90–111. [CrossRef]
51. Bezuglov, A.; Blanton, B.; Santiago, R. Multi-output artificial neural network for storm surge prediction in north carolina. arXiv
2016, arXiv:1609.07378.
52. Bass, B.; Bedient, P. Surrogate modeling of joint flood risk across coastal watersheds. J. Hydrol. 2018, 558, 159–173. [CrossRef]
53. Tadesse, M.G.; Wahl, T. A database of global storm surge reconstructions. Sci. Data 2021, 8, 125. [CrossRef]
54. Palmer, M.; Domingues, C.; Slangen, A.; Dias, F.B. An ensemble approach to quantify global mean sea-level rise over the 20th
century from tide gauge reconstructions. Environ. Res. Lett. 2021, 16, 044043. [CrossRef]
55. Bruneau, N.; Polton, J.; Williams, J.; Holt, J. Estimation of global coastal sea level extremes using neural networks. Environ. Res.
Lett. 2020, 15, 074030. [CrossRef]
56. Chen, R.; Zhang, W.; Wang, X. Machine learning in tropical cyclone forecast modeling: A review. Atmosphere 2020, 11, 676.
[CrossRef]
57. De Oliveira, M.M.; Ebecken, N.F.F.; De Oliveira, J.L.F.; de Azevedo Santos, I. Neural network model to predict a storm surge. J.
Appl. Meteorol. Climatol. 2009, 48, 143–155. [CrossRef]
58. Taylor, A.A.; Glahn, B. Probabilistic guidance for hurricane storm surge. In Proceedings of the 19th Conference on Probability
and Statistics, New Orleans, LA, USA, 21–24 January 2008; Volume 74.
59. Feng, X.; Ma, G.; Su, S.F.; Huang, C.; Boswell, M.K.; Xue, P. A multi-layer perceptron approach for accelerated wave forecasting in
Lake Michigan. Ocean. Eng. 2020, 211, 107526. [CrossRef]
60. Deo, R.C.; Ghorbani, M.A.; Samadianfard, S.; Maraseni, T.; Bilgili, M.; Biazar, M. Multi-layer perceptron hybrid model integrated
with the firefly optimizer algorithm for windspeed prediction of target site using a limited set of neighboring reference station
data. Renew. Energy 2018, 116, 309–323. [CrossRef]
61. Kulkarni, P.A.; Dhoble, A.S.; Padole, P.M. Deep neural network-based wind speed forecasting and fatigue analysis of a large
composite wind turbine blade. Proc. Inst. Mech. Eng. Part C J. Mech. Eng. Sci. 2019, 233, 2794–2812. [CrossRef]
62. Chattopadhyay, A.; Hassanzadeh, P.; Pasha, S. Predicting clustered weather patterns: A test case for applications of convolutional
neural networks to spatio-temporal climate data. Sci. Rep. 2020, 10, 1317. [CrossRef] [PubMed]
63. Luo, Y.; Feng, A.; Li, H.; Li, D.; Wu, X.; Liao, J.; Zhang, C.; Zheng, X.; Pu, H. New deep learning method for efficient extraction of
small water from remote sensing images. PLoS ONE 2022, 17, e0272317. [CrossRef]
64. Hunt, K.M.; Matthews, G.R.; Pappenberger, F.; Prudhomme, C. Using a long short-term memory (LSTM) neural network to boost
river streamflow forecasts over the western United States. Hydrol. Earth Syst. Sci. 2022, 26, 5449–5472. [CrossRef]
65. Zilong, T.; Yubing, S.; Xiaowei, D. Spatial-temporal wave height forecast using deep learning and public reanalysis dataset. Appl.
Energy 2022, 326, 120027. [CrossRef]
66. Varalakshmi, P.; Vasumathi, N.; Venkatesan, R. Tropical Cyclone prediction based on multi-model fusion across Indian coastal
region. Prog. Oceanogr. 2021, 193, 102557. [CrossRef]
67. Sagi, O.; Rokach, L. Ensemble learning: A survey. Wiley Interdiscip. Rev. Data Min. Knowl. Discov. 2018, 8, e1249. [CrossRef]
68. Young, C.C.; Liu, W.C.; Hsieh, W.L. Predicting the water level fluctuation in an alpine lake using physically based, artificial
neural network, and time series forecasting models. Math. Probl. Eng. 2015, 2015, 708204. [CrossRef]
69. Kim, S.; Matsumi, Y.; Pan, S.; Mase, H. A real-time forecast model using artificial neural network for after-runner storm surges on
the Tottori coast, Japan. Ocean. Eng. 2016, 122, 44–53. [CrossRef]
70. Blake, E.S.; Zelinsky, D.A. National Hurricane Center Tropical Cyclone Report; Hurricane Harvey. National Hurricane Center,
National Oceanographic and Atmospheric Association: Miami, FL, USA, 2017.
71. Qin, Y.; Su, C.; Chu, D.; Zhang, J.; Song, J. A Review of Application of Machine Learning in Storm Surge Problems. J. Mar. Sci.
Eng. 2023, 11, 1729. [CrossRef]
72. Yu, Y.; Zhang, H.; Singh, V.P. Forward prediction of runoff data in data-scarce basins with an improved ensemble empirical mode
decomposition (EEMD) model. Water 2018, 10, 388. [CrossRef]
73. Le, X.H.; Ho, H.V.; Lee, G.; Jung, S. Application of long short-term memory (LSTM) neural network for flood forecasting. Water
2019, 11, 1387. [CrossRef]
J. Mar. Sci. Eng. 2023, 11, 2154 27 of 30

74. Liao, L.; Li, H.; Shang, W.; Ma, L. An empirical study of the impact of hyperparameter tuning and model optimization on the
performance properties of deep neural networks. ACM Trans. Softw. Eng. Methodol. (TOSEM) 2022, 31, 1–40. [CrossRef]
75. Victoria, A.H.; Maragatham, G. Automatic tuning of hyperparameters using Bayesian optimization. Evol. Syst. 2021, 12, 217–223.
[CrossRef]
76. Yu, T.; Zhu, H. Hyper-parameter optimization: A review of algorithms and applications. arXiv 2020, arXiv:2003.05689.
77. Hu, C.; Wu, Q.; Li, H.; Jian, S.; Li, N.; Lou, Z. Deep learning with a long short-term memory networks approach for rainfall-runoff
simulation. Water 2018, 10, 1543. [CrossRef]
78. Zhang, X.q.; Jiang, S.q. Study on the application of BP neural network optimized based on various optimization algorithms in
storm surge prediction. Proc. Inst. Mech. Eng. Part M J. Eng. Marit. Environ. 2022, 236, 539–552. [CrossRef]
79. Lee, T.L. Back-propagation neural network for the prediction of the short-term storm surge in Taichung harbor, Taiwan. Eng.
Appl. Artif. Intell. 2008, 21, 63–72. [CrossRef]
80. Tsai, C.; You, C.; Chen, C. Storm-surge prediction at the Tanshui estuary: Development model for maximum storm surges. Nat.
Hazards Earth Syst. Sci 2013, 1, 7333–7356.
81. Alzubaidi, L.; Zhang, J.; Humaidi, A.J.; Al-Dujaili, A.; Duan, Y.; Al-Shamma, O.; Santamaría, J.; Fadhel, M.A.; Al-Amidie, M.;
Farhan, L. Review of deep learning: Concepts, CNN architectures, challenges, applications, future directions. J. Big Data 2021,
8, 1–74. [CrossRef] [PubMed]
82. Giffard-Roisin, S.; Yang, M.; Charpiat, G.; Kumler Bonfanti, C.; Kégl, B.; Monteleoni, C. Tropical cyclone track forecasting using
fused deep learning from aligned reanalysis data. Front. Big Data 2020, 3, 1.. [CrossRef]
83. Wang, T.; Liu, T.; Lu, Y. A hybrid multi-step storm surge forecasting model using multiple feature selection, deep learning neural
network and transfer learning. Soft Comput. 2023, 27, 935–952. [CrossRef]
84. Weiss, K.; Khoshgoftaar, T.M.; Wang, D. A survey of transfer learning. J. Big Data 2016, 3, 1–40. [CrossRef]
85. Wu, W.; Westra, S.; Leonard, M. A basis function approach for exploring the seasonal and spatial features of storm surge events.
Geophys. Res. Lett. 2017, 44, 7356–7365. [CrossRef]
86. Wolf, J.; Flather, R. Modelling waves and surges during the 1953 storm. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 2005,
363, 1359–1375. [CrossRef] [PubMed]
87. Feng, J.; von Storch, H.; Jiang, W.; Weisse, R. Assessing changes in extreme sea levels along the coast of C hina. J. Geophys. Res.
Ocean. 2015, 120, 8039–8051. [CrossRef]
88. Bloemendaal, N.; Haigh, I.D.; de Moel, H.; Muis, S.; Haarsma, R.J.; Aerts, J.C. Generation of a global synthetic tropical cyclone
hazard dataset using STORM. Sci. Data 2020, 7, 40. [CrossRef]
89. Adhikari, R.; Agrawal, R. A homogeneous ensemble of artificial neural networks for time series forecasting. arXiv 2013,
arXiv:1302.6210.
90. Guan, H.; Mokadam, L.K.; Shen, X.; Lim, S.H.; Patton, R. Fleet: Flexible efficient ensemble training for heterogeneous deep neural
networks. Proc. Mach. Learn. Syst. 2020, 2, 247–261.
91. Zhou, Z.H.; Zhou, Z.H. Ensemble Learning; Springer: Berlin/Heidelberg, Germany, 2021.
92. Zhou, Z.H.; Wu, J.; Tang, W. Ensembling neural networks: Many could be better than all. Artif. Intell. 2002, 137, 239–263.
[CrossRef]
93. Ghojogh, B.; Crowley, M. The theory behind overfitting, cross validation, regularization, bagging, and boosting: Tutorial. arXiv
2019, arXiv:1905.12787.
94. Brodeur, Z.P.; Herman, J.D.; Steinschneider, S. Bootstrap aggregation and cross-validation methods to reduce overfitting in
reservoir control policy search. Water Resour. Res. 2020, 56, e2020WR027184. [CrossRef]
95. Altman, N.; Krzywinski, M. Ensemble methods: Bagging and random forests. Nat. Methods 2017, 14, 933–935. [CrossRef]
96. Dietterich, T.G. Ensemble methods in machine learning. In Proceedings of the Multiple Classifier Systems: First International
Workshop, MCS 2000, Cagliari, Italy, 21–23 June 2000; pp. 1–15.
97. Cassales, G.; Gomes, H.; Bifet, A.; Pfahringer, B.; Senger, H. Improving the performance of bagging ensembles for data streams
through mini-batching. Inf. Sci. 2021, 580, 260–282. [CrossRef]
98. Maxwell, A.E.; Warner, T.A.; Fang, F. Implementation of machine-learning classification in remote sensing: An applied review.
Int. J. Remote Sens. 2018, 39, 2784–2817. [CrossRef]
99. Zounemat-Kermani, M.; Batelaan, O.; Fadaee, M.; Hinkelmann, R. Ensemble machine learning paradigms in hydrology: A review.
J. Hydrol. 2021, 598, 126266. [CrossRef]
100. Elith, J.; Leathwick, J.R.; Hastie, T. A working guide to boosted regression trees. J. Anim. Ecol. 2008, 77, 802–813. [CrossRef]
101. Freund, Y.; Schapire, R.E. A decision-theoretic generalization of on-line learning and an application to boosting. J. Comput. Syst.
Sci. 1997, 55, 119–139. [CrossRef]
102. Lawry, J.; He, H. Linguistic decision trees for fusing tidal surge forecasting models. In Combining Soft Computing and Statistical
Methods in Data Analysis; Springer: Berlin/Heidelberg, Germany, 2010; pp. 403–410.
103. Bentéjac, C.; Csörgő, A.; Martínez-Muñoz, G. A comparative analysis of gradient boosting algorithms. Artif. Intell. Rev. 2021,
54, 1937–1967. [CrossRef]
104. Chen, T.; Guestrin, C. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM Sigkdd International Conference
on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016, pp. 785–794.
J. Mar. Sci. Eng. 2023, 11, 2154 28 of 30

105. Drucker, H. Improving regressors using boosting techniques. In Proceedings of the Fourteenth International Conference on
Machine Learning (ICML 1997), Nashville, TN, USA, 8–12 July 1997; Volume 97, pp. 107–115.
106. Muis, S.; Apecechea, M.I.; Dullaart, J.; de Lima Rego, J.; Madsen, K.S.; Su, J.; Yan, K.; Verlaan, M. A high-resolution global dataset
of extreme sea levels, tides, and storm surges, including future projections. Front. Mar. Sci. 2020, 7, 263. [CrossRef]
107. Sesmero, M.P.; Ledezma, A.I.; Sanchis, A. Generating ensembles of heterogeneous classifiers using stacked generalization. Wiley
Interdiscip. Rev. Data Min. Knowl. Discov. 2015, 5, 21–34. [CrossRef]
108. Barton, M.; Lennox, B. Model stacking to improve prediction and variable importance robustness for soft sensor development.
Digit. Chem. Eng. 2022, 3, 100034. [CrossRef]
109. Džeroski, S.; Ženko, B. Is combining classifiers with stacking better than selecting the best one? Mach. Learn. 2004, 54, 255–273.
[CrossRef]
110. Breiman, L. Stacked regressions. Mach. Learn. 1996, 24, 49–64. [CrossRef]
111. Zucco, C. Multiple Learners Combination: Stacking. In Encyclopedia of Bioinformatics and Computational Biology; Ranganathan, S.,
Gribskov, M., Nakai, K., Schönbach, C., Eds.; Academic Press: Oxford, UK, 2019; pp. 536–538. [CrossRef]
112. Sill, J.; Takács, G.; Mackey, L.; Lin, D. Feature-weighted linear stacking. arXiv 2009, arXiv:0911.0460.
113. Young, S.; Abdou, T.; Bener, A. Deep super learner: A deep ensemble for classification problems. In Proceedings of the Advances
in Artificial Intelligence: 31st Canadian Conference on Artificial Intelligence, Canadian AI 2018, Toronto, ON, Canada, 8–11 May
2018; pp. 84–95.
114. Wolpert, D.H. Stacked generalization. Neural Netw. 1992, 5, 241–259. [CrossRef]
115. Ayyad, M.; Orton, P.M.; El Safty, H.; Chen, Z.; Hajj, M.R. Ensemble forecast for storm tide and resurgence from Tropical Cyclone
Isaias. Weather. Clim. Extrem. 2022, 38, 100504. [CrossRef]
116. Kim, S.W.; Melby, J.A.; Nadal-Caraballo, N.C.; Ratcliff, J. A time-dependent surrogate model for storm surge prediction based on
an artificial neural network using high-fidelity synthetic hurricane modeling. Nat. Hazards 2015, 76, 565–585. [CrossRef]
117. Guo, T. Hurricane Damage Prediction based on Convolutional Neural Network Models. In Proceedings of the 2021 2nd
International Conference on Artificial Intelligence and Computer Engineering (ICAICE), Hangzhou, China, 5–7 November 2021;
pp. 298–302.
118. Gebrehiwot, A.; Hashemi-Beni, L.; Thompson, G.; Kordjamshidi, P.; Langan, T.E. Deep convolutional neural network for flood
extent mapping using unmanned aerial vehicles data. Sensors 2019, 19, 1486. [CrossRef]
119. Accarino, G.; Chiarelli, M.; Fiore, S.; Federico, I.; Causio, S.; Coppini, G.; Aloisio, G. A multi-model architecture based on Long
Short-Term Memory neural networks for multi-step sea level forecasting. Future Gener. Comput. Syst. 2021, 124, 1–9. [CrossRef]
120. Kaur, S.; Gupta, S.; Singh, S.; Koundal, D.; Zaguia, A. Convolutional neural network based hurricane damage detection using
satellite images. Soft Comput. 2022, 26, 7831–7845. [CrossRef]
121. Korzh, O.; Joaristi, M.; Serra, E. Convolutional neural network ensemble fine-tuning for extended transfer learning. In Proceedings
of the Big Data–BigData 2018: 7th International Congress, Held as Part of the Services Conference Federation, SCF 2018, Seattle,
WA, USA, 25–30 June 2018; pp. 110–123.
122. Becherer, N.; Pecarina, J.; Nykl, S.; Hopkinson, K. Improving optimization of convolutional neural networks through parameter
fine-tuning. Neural Comput. Appl. 2019, 31, 3469–3479. [CrossRef]
123. Blalock, D.; Gonzalez Ortiz, J.J.; Frankle, J.; Guttag, J. What is the state of neural network pruning? Proc. Mach. Learn. Syst. 2020,
2, 129–146.
124. Araghinejad, S.; Azmi, M.; Kholghi, M. Application of artificial neural network ensembles in probabilistic hydrological forecasting.
J. Hydrol. 2011, 407, 94–104. [CrossRef]
125. Chen, W.; Hong, H.; Li, S.; Shahabi, H.; Wang, Y.; Wang, X.; Ahmad, B.B. Flood susceptibility modelling using novel hybrid
approach of reduced-error pruning trees with bagging and random subspace ensembles. J. Hydrol. 2019, 575, 864–873. [CrossRef]
126. Du, L.; Gao, R.; Suganthan, P.N.; Wang, D.Z. Bayesian optimization based dynamic ensemble for time series forecasting. Inf. Sci.
2022, 591, 155–175. [CrossRef]
127. Pham, B.T.; Jaafari, A.; Nguyen-Thoi, T.; Van Phong, T.; Nguyen, H.D.; Satyam, N.; Masroor, M.; Rehman, S.; Sajjad, H.; Sahana,
M.; et al. Ensemble machine learning models based on Reduced Error Pruning Tree for prediction of rainfall-induced landslides.
Int. J. Digit. Earth 2021, 14, 575–596. [CrossRef]
128. Rooney, N.; Patterson, D.; Nugent, C. Reduced ensemble size stacking [ensemble learning]. In Proceedings of the 16th IEEE
International Conference on Tools with Artificial Intelligence, Boca Raton, FL, USA, 15–17 November 2004; pp. 266–271.
129. Naftaly, U.; Intrator, N.; Horn, D. Optimal ensemble averaging of neural networks. Netw. Comput. Neural Syst. 1997, 8, 283.
[CrossRef]
130. Huang, W.; Hong, H.; Bian, K.; Zhou, X.; Song, G.; Xie, K. Improving deep neural network ensembles using reconstruction error.
In Proceedings of the 2015 International joint conference on neural networks (IJCNN), Killarney, Ireland, 12–17 July 2015, pp. 1–7.
131. Zeng, X.; Yeung, D.S. Hidden neuron pruning of multilayer perceptrons using a quantified sensitivity measure. Neurocomputing
2006, 69, 825–837. [CrossRef]
132. Smith, C.; Jin, Y. Evolutionary multi-objective generation of recurrent neural network ensembles for time series prediction.
Neurocomputing 2014, 143, 302–311. [CrossRef]
133. Shahhosseini, M.; Hu, G.; Pham, H. Optimizing ensemble weights and hyperparameters of machine learning models for
regression problems. Mach. Learn. Appl. 2022, 7, 100251. [CrossRef]
J. Mar. Sci. Eng. 2023, 11, 2154 29 of 30

134. Palaniswamy, S.K.; Venkatesan, R. Hyperparameters tuning of ensemble model for software effort estimation. J. Ambient. Intell.
Humaniz. Comput. 2021, 12, 6579–6589. [CrossRef]
135. Snoek, J.; Larochelle, H.; Adams, R.P. Practical bayesian optimization of machine learning algorithms. Adv. Neural Inf. Process.
Syst. Curran Associates, Inc.: Red Hook, NY, USA, 2012, Volume 25.
136. Wu, J.; Chen, X.Y.; Zhang, H.; Xiong, L.D.; Lei, H.; Deng, S.H. Hyperparameter optimization for machine learning models based
on Bayesian optimization. J. Electron. Sci. Technol. 2019, 17, 26–40.
137. Priyadarshini, I.; Cotton, C. A novel LSTM–CNN–grid search-based deep neural network for sentiment analysis. J. Supercomput.
2021, 77, 13911–13932. [CrossRef] [PubMed]
138. Huang, G.B.; Chen, L. Enhanced random search based incremental extreme learning machine. Neurocomputing 2008, 71, 3460–3468.
[CrossRef]
139. Agnihotri, A.; Batra, N. Exploring bayesian optimization. Distill 2020, 5, e26. [CrossRef]
140. Zhou, J.; Peng, T.; Zhang, C.; Sun, N. Data pre-analysis and ensemble of various artificial neural networks for monthly streamflow
forecasting. Water 2018, 10, 628. [CrossRef]
141. Aloysius, N.; Geetha, M. A review on deep convolutional neural networks. In Proceedings of the 2017 International Conference
on Communication and Signal Processing (ICCSP), Melmaruvathur, India, 6–8 April 2017; pp. 0588–0592.
142. Kiranyaz, S.; Avci, O.; Abdeljaber, O.; Ince, T.; Gabbouj, M.; Inman, D.J. 1D convolutional neural networks and applications: A
survey. Mech. Syst. Signal Process. 2021, 151, 107398. [CrossRef]
143. Trice, A.; Robbins, C.; Philip, N.; Rumsey, M. Challenges and Opportunities for Ocean Data to Advance Conservation and Management;
Ocean Conservancy: Washington, DC, USA, 2021.
144. Velioglu Sogut, D.; Yalciner, A.C. Performance comparison of NAMI DANCE and FLOW-3D® models in tsunami propagation,
inundation and currents using NTHMP benchmark problems. Pure Appl. Geophys. 2019, 176, 3115–3153. [CrossRef]
145. Costa, W.; Idier, D.; Rohmer, J.; Menendez, M.; Camus, P. Statistical prediction of extreme storm surges based on a fully supervised
weather-type downscaling model. J. Mar. Sci. Eng. 2020, 8, 1028. [CrossRef]
146. Cialone, M.A.; Massey, T.C.; Anderson, M.E.; Grzegorzewski, A.S.; Jensen, R.E.; Cialone, A.; Mark, D.J.; Pevey, K.C.; Gunkel, B.L.;
McAlpin, T.O.; et al. North Atlantic Coast Comprehensive Study (NACCS) Coastal Storm Model Simulations: Waves and Water Levels;
US Army Engineer Research and Development Center, Coastal and Hydraulics Laboratory: Vicksburg, MS, USA , 2015.
147. Yang, C.; Leonelli, F.E.; Marullo, S.; Artale, V.; Beggs, H.; Nardelli, B.B.; Chin, T.M.; De Toma, V.; Good, S.; Huang, B.; et al.
Sea surface temperature intercomparison in the framework of the Copernicus Climate Change Service (C3S). J. Clim. 2021,
34, 5257–5283. [CrossRef]
148. Hersbach, H. Decomposition of the continuous ranked probability score for ensemble prediction systems. Weather. Forecast. 2000,
15, 559–570. [CrossRef]
149. Wallendorf, L.; Cox, D.T. Coastal Structures and Solutions to Coastal Disasters 2015: Tsunamis; American Society of Civil Engineers:
Reston, VA, USA, 2017.
150. Conver, A.; Sepanik, J.; Louangsaysongkham, B.; Miller, S. Sea, Lake, and Overland Surges from Hurricanes (SLOSH) Basin
Development Handbook v2.0; NOAA/NWS/Meteorological Development Laboratory: Silver Springs, MD, USA, 2008.
151. Miller, A.; Luscher, A. NOAA’s national water level observation network (NWLON). J. Oper. Oceanogr. 2019, 12, S57–S66.
[CrossRef]
152. Raschka, S. Python Machine Learning; Packt Publishing Ltd.: Birmingham, UK, 2015.
153. Yang, H. Data preprocessing. In Data Mining: Concepts and Techniques ; Pennsylvania State University, CiteSeerX, USA 2018.
154. Knapp, K.R.; Kruk, M.C.; Levinson, D.H.; Diamond, H.J.; Neumann, C.J. The International Best Track Archive for Climate
Stewardship (IBTrACS). Bull. Am. Meteorol. Soc. 2010, 91, 363–376. [CrossRef]
155. Knapp, K.R.; Diamond, H.J.; Kossin, J.P.; Kruk, M.C.; Schreck, C.J. In International Best Track Archive for Climate Stewardship
(IBTrACS) Project; Version 4; NOAA National Centers for Environmental Information: Asheville, NC, USA, 2018; [CrossRef]
156. NOAA National Data Buoy Center. In Meteorological and Oceanographic Data Collected from the National Data Buoy Center Coastal-
Marine Automated Network (C-MAN) and Moored (Weather) Buoys; NOAA National Centers for Environmental Information, Dataset:
Port Aransas, TX, USA, 1971.
157. Adebisi, N.; Balogun, A.L.; Min, T.H.; Tella, A. Advances in estimating Sea Level Rise: A review of tide gauge, satellite altimetry
and spatial data science approaches. Ocean. Coast. Manag. 2021, 208, 105632. [CrossRef]
158. Kyprioti, A.P.; Taflanidis, A.A.; Plumlee, M.; Asher, T.G.; Spiller, E.; Luettich, R.A.; Blanton, B.; Kijewski-Correa, T.L.; Kennedy, A.;
Schmied, L. Improvements in storm surge surrogate modeling for synthetic storm parameterization, node condition classification
and implementation to small size databases. Nat. Hazards 2021, 109, 1349–1386. [CrossRef]
159. Queipo, N.V.; Nava, E. A gradient boosting approach with diversity promoting measures for the ensemble of surrogates in
engineering. Struct. Multidiscip. Optim. 2019, 60, 1289–1311. [CrossRef]
160. Freeman, J.; Velic, M.; Colberg, F.; Greenslade, D.; Divakaran, P.; Kepert, J. Development of a tropical storm surge prediction
system for Australia. J. Mar. Syst. 2020, 206, 103317. [CrossRef]
161. Beuzen, T.; Goldstein, E.B.; Splinter, K.D. Ensemble models from machine learning: An example of wave runup and coastal dune
erosion. Nat. Hazards Earth Syst. Sci. 2019, 19, 2295–2309. [CrossRef]
162. Goodarzi, L.; Banihabib, M.E.; Roozbahani, A. A decision-making model for flood warning system based on ensemble forecasts.
J. Hydrol. 2019, 573, 207–219. [CrossRef]
J. Mar. Sci. Eng. 2023, 11, 2154 30 of 30

163. Chang, L.C.; Amin, M.Z.M.; Yang, S.N.; Chang, F.J. Building ANN-based regional multi-step-ahead flood inundation forecast
models. Water 2018, 10, 1283. [CrossRef]
164. Neal, B.; Mittal, S.; Baratin, A.; Tantia, V.; Scicluna, M.; Lacoste-Julien, S.; Mitliagkas, I. A modern take on the bias-variance
tradeoff in neural networks. arXiv 2018, arXiv:1810.08591.
165. Ganaie, M.A.; Hu, M.; Malik, A.; Tanveer, M.; Suganthan, P. Ensemble deep learning: A review. Eng. Appl. Artif. Intell. 2022,
115, 105151. [CrossRef]
166. James, G.; Witten, D.; Hastie, T.; Tibshirani, R. An Introduction to Statistical Learning; Springer: Berlin/Heidelberg, Germany,
2013;m Volume 112.
167. Ortega, L.A.; Cabañas, R.; Masegosa, A. Diversity and generalization in neural network ensembles. In Proceedings of the
International Conference on Artificial Intelligence and Statistics, Valencia, Spain, 28–30 March 2022; pp. 11720–11743.
168. Tsymbal, A.; Pechenizkiy, M.; Cunningham, P. Diversity in search strategies for ensemble feature selection. Inf. Fusion 2005,
6, 83–98. [CrossRef]
169. Dutta, H. Measuring Diversity in Regression Ensembles. In Proceedings of the ICAI, Las Vegas, NV, USA, 13–16 July 2009;
Volume 9, p. 17.
170. Horsburgh, K.; Flowerdew, J. Real-Time Coastal Flood Forecasting. In Applied Uncertainty Analysis for Flood Risk Management;
World Scientific Publishing Co. Pte. Ltd.: London, UK, 2014; pp. 538–562.
171. Murphy, A.H. Skill scores based on the mean square error and their relationships to the correlation coefficient. Mon. Weather. Rev.
1988, 116, 2417–2424. [CrossRef]
172. Tonani, M.; Pinardi, N.; Fratianni, C.; Pistoia, J.; Dobricic, S.; Pensieri, S.; De Alfonso, M.; Nittis, K. Mediterranean Forecasting
System: Forecast and analysis assessment through skill scores. Ocean. Sci. 2009, 5, 649–660. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.