applied

sciences
Article
Deep Machine Learning-Based Water Level Prediction Model
for Colombo Flood Detention Area
Madhawa Herath 1 , Tharaka Jayathilaka 2 , Yukinobu Hoshino 3 and Upaka Rathnayake 4, *

1 Department of Mechanical Engineering, Faculty of Engineering, Sri Lanka Institute of

Information Technology, Malabe 10115, Sri Lanka
2 Department of Civil Engineering, Faculty of Engineering, Sri Lanka Institute of Information Technology,
Malabe 10115, Sri Lanka
3 School of Systems Engineering, Kochi University of Technology, Tosayamada, Kami 782-8502, Kochi, Japan
4 Department of Civil Engineering and Construction, Faculty of Engineering and Design, Atlantic
Technological University, F91 YW50 Sligo, Ireland
* Correspondence: [email protected] or [email protected]

Abstract: Machine learning has already been proven as a powerful state-of-the-art technique for many
non-linear applications, including environmental changes and climate predictions. Wetlands are
among some of the most challenging and complex ecosystems for water level predictions. Wetland
water level prediction is vital, as wetlands have their own permissible water levels. Exceeding these
water levels can cause flooding and other severe environmental damage. On the other hand, the
biodiversity of the wetlands is threatened by the sudden fluctuation of water levels. Hence, early
prediction of water levels benefits in mitigating most of such environmental damage. However,
monitoring and predicting the water levels in wetlands worldwide have been limited owing to various
constraints. This study presents the first-ever application of deep machine-learning techniques (deep
neural networks) to predict the water level in an urban wetland in Sri Lanka located in its capital.
Moreover, for the first time in water level prediction, it investigates two types of relationships: the
traditional relationship between water levels and environmental factors, including temperature,
humidity, wind speed, and evaporation, and the temporal relationship between daily water levels.
Two types of low load artificial neural networks (ANNs) were developed and employed to analyze
two relationships which are feed forward neural networks (FFNN) and long short-term memory
Citation: Herath, M.; Jayathilaka, T.;
(LSTM) neural networks, to conduct the comparison on an unbiased common ground. The LSTM has
Hoshino, Y.; Rathnayake, U. Deep
outperformed FFNN and confirmed that the temporal relationship is much more robust in predicting
Machine Learning-Based Water Level
Prediction Model for Colombo Flood
wetland water levels than the traditional relationship. Further, the study identified interesting
Detention Area. Appl. Sci. 2023, 13, relationships between prediction accuracy, data volume, ANN type, and degree of information
2194. https://doi.org/10.3390/ extraction embedded in wetland data. The LSTM neural networks (NN) has achieved substantial
app13042194 performance, including R2 of 0.8786, mean squared error (MSE) of 0.0004, and mean absolute error
(MAE) of 0.0155 compared to existing studies.
Academic Editor: Xianpeng Wang

Received: 25 December 2022 Keywords: artificial neural network (ANN); Colombo flood detention area; machine-learning; feed
Revised: 29 January 2023 forward neural network (FFNN); long short-term memory (LSTM); water level prediction; wetlands
Accepted: 7 February 2023
Published: 8 February 2023

1. Introduction
Copyright: © 2023 by the authors.
A wetland is a dynamic system where the land is saturated with water throughout
Licensee MDPI, Basel, Switzerland. the year, either entirely or partially [1]. Wetlands are characterized as transitional zones
This article is an open access article between primarily flooded deep water environments and well-drained hills [2]. They
distributed under the terms and are among the most productive ecosystems in the world [3] and play a broad role in
conditions of the Creative Commons environmental, social, and economic aspects. Wetlands act as the kidneys of the landscape,
Attribution (CC BY) license (https:// filtering and capturing suspended particles, trapping pollutants such as carbon dioxide and
creativecommons.org/licenses/by/ nitrogen, and releasing clean water downstream [4]. They stabilize the global climate, thus
4.0/). acting as carbon sinks [5]. Wetlands accommodate floods, thus reducing flood risk during

Appl. Sci. 2023, 13, 2194. https://doi.org/10.3390/app13042194 https://www.mdpi.com/journal/applsci

Appl. Sci. 2023, 13, 2194 2 of 17

storms [6,7]. Due to the extensive food chain and the abundant biodiversity, wetlands
also have been treated as “nature’s supermarkets” [8]. Pollutants such as nitrate in water
sources are a significant concern, and wetlands are viable sinks for these pollutants [9–11].
Besides, they enhance water quality, control shoreline erosion, rejuvenate the ecosystem,
recharge the groundwater table, facilitate recreation, and aesthetics, etc. [12–14].
Wetlands cover about 6% of the earth’s surface [15]. Due to significant land changes
due to urbanization, infrastructure development, and agriculture, these ecosystems are
under extreme pressure [16]. Wetlands’ vulnerability to climate stress has also been con-
firmed [17]. Therefore, the wetland degradation rate is drastically increasing. According
to the 2021 Global wetland outlook report, “Wetland coverage” has been reduced by 35%
since 1970 [18]. Since it is one of the most significant ecosystems, it is vital to protect
such vulnerable ecosystems. The water level is the crucial parameter that defines the
functionality of a wetland [19]. However, most countries still lack a proper mechanism for
mapping and monitoring wetland water levels. Hence, water level projection will lead to
the protection and proper management of the wetland.
Wetland water levels change in response to variations in water inputs and outputs [20].
Rainfall, surface water inflows, and groundwater inflows are examples of water inputs
to wetlands, whereas evaporation, surface water outflows, and seepage are examples
of water outputs [21]. Water levels link the wetland’s hydrological, meteorological, and
geomorphological characteristics to the surroundings [22]. In addition, wetlands are highly
vulnerable to changes in the atmosphere. As the air temperature increases, rainfall patterns
change and evaporation increases, causing the wetlands to swell [23]. On the other hand,
as the temperature rises, relative humidity falls [24,25]. In addition, high wind speed cause
more evaporation in the water bodies, lowering their water levels [26].
Altunkaynak [27] used neural networks with backpropagation to forecast Lake Van,
Turkey, water levels. They have uncovered that the neural networks give accurate results
even though the relationships between the rainfall and consecutive water levels are highly
dynamic. They have trained the neural network by having three different arrangements
in the input nodes. All three models have produced relatively similar results. One of
the significant drawbacks of this study is that they have yet to consider the rest of the
meteorological parameters other than the rainfall.
Choi et al. [28] emphasized the significance of predicting wetland water levels while
pointing out the difficulties due to the data limitations. They have predicted the water levels
in the Upo wetland, South Korea, using several machine-learning techniques, including
ANN, decision trees (DT), random forest (RF), and support vector machines (SVM). In
addition, researchers have considered the temporal relationship for water level prediction.
However, considering only the previous day’s water levels will not be adequate for better
temporal information extraction through NNs. It is necessary to consider at least 20–30 days
of the previous day’s water levels for accurate temporal relationship learning.
Dadaser-Celik and Cengiz [29] have predicted the monthly average water levels in
Sultan Marshes in Turkey using a multi-layer perceptron-type neural network model. They
have investigated the traditional relationship between climate data (rainfall, temperature,
evapotranspiration) and hydrological data (groundwater levels, spring flow rate, and
the previous month’s water levels). Although they have used several input parameters,
including the last month’s water level, the selected NN type will not support better temporal
relationship learning.
Gopakumar and Takara [30] have carried out a study on the investigation of water
level fluctuations in the Kerala Vembanad Wetland using ANNs. They have predicted
one-day-ahead water levels using rainfall data and the previous day’s water levels. The
models’ results were expressed using several numerical indices, such as the correlation
coefficient and root mean square error. Also, they have found that neural networks only
predict water levels accurately when there is information on wetland storage conditions.
However, this study has not compared both traditional and temporal relationships, which
Appl. Sci. 2023, 13, 2194 3 of 17

is one of the major concerns. Moreover, it has considered only rainfall among possible
environmental factors influencing the wetland water level.
Saha et al. [31] studied wetland water depth and area prediction in the Atreyee River
basin using ANN and a non-linear regression model. In this regard, they have used Landsat
satellite images of the river basin during pre- and post-monsoon seasons. There could be
several drawbacks as they have used image processing data. Prediction accuracy would be
limited with a sufficient number of satellite images. Also, images could be noisy because of
the clod forming, dust, etc.
Wetland water level measurements are restricted in most countries, especially Sri
Lanka. This situation is mainly due to unawareness and lack of attention given by the
local authorities. Also, fewer funding allocations for wetland developments have caused
improper wetland management and maintenance. Therefore, many wetlands in Sri Lanka
have faced severe problems, including insufficient water level measurements. Being the
capital of Sri Lanka, wetlands in Colombo city have a significant threat including land
degradation. According to recent research, wetland coverage in Colombo city has been
reduced by 30% [32]. Wetlands act as flood detention basins. As the wetlands’ special
coverage is concentrated in Colombo, frequent flash floods are more prone to occur. The
lack of water level measurements in Colombo wetlands is a major cause of inadequate
flood management in the Colombo region. Also, less attention has been given to water
level prediction in Sri Lankan wetlands. The first attempt to simulate the water levels in the
Colombo flood detention area was taken recently by Jayathilake et al. [33]. However, it was
an initial study to look at the relationship between water levels and climate parameters.
Nevertheless, Jayathilake et al. have found the relationship between wetland water levels
and climate parameters is highly non-linear [33].
Motivated by the importance of wetlands and monitoring their water level along
with the power of deep machine learning techniques (i.e., ANN), this study extends our
previous [33] study to investigate and compare traditional relationship and temporal
relationships on an unbiased common ground for predicting wetland water level for
the first time in this domain using two types of ANNs. In this study, hydrological data
and meteorological data have been considered for wetland water level forecasting. In
addition, artificial neural network models were developed to simulate the water levels at
the Colombo Flood Detention area.
As per the climate parameter prediction literature [34–37], two fundamental types of
existing relationships can be considered for predicting the water level of a wetland. The
first relationship that was investigated is between the water level and several indepen-
dent variables (called the traditional relationship hereafter), including daily rainfall, daily
evaporation, minimum daily temperature, maximum daily temperature, daily relative
humidity during the daytime, daily relative humidity at nighttime, and daily average wind
speed. Second, the temporal relationship between daily water levels was investigated.
Rainfall was considered the primary factor influencing water balance at the spatial and
temporal scales [38]. NNs were selected to model both relationships to maintain a common
platform for comparing the two relationships in this study. A simple feed forward neural
network (FFNN) was chosen to model the traditional relationship and at the same time,
the long short-term memory neural network was used for temporal relationship modeling.
Moreover, the authors need to find the simplest NN architectures that require minimum
computational processing power and memory requirements with substantial prediction
accuracy to run the model on an average computer [39]. It will enable the easy and eco-
nomical use of the prediction model, especially in remote installations in a country like
Sri Lanka.
Primarily, the temporal relationship was never investigated for water level prediction
in the past on a comparative basis. The study further investigates the relative accuracy
of the relationships, advantages vs disadvantages, and limitations in the two types of
predictions. It also examines the available data volume’s impact on prediction accuracy.
Moreover, the information extraction patterns of each NN will be discussed. The selected
Appl. Sci. 2023, 13, 2194 4 of 17

NN architectures are the low-load high, performing NNs selected via detailed model
comparison. This is the first-ever application of deep machine-learning techniques for
predicting wetland water levels in Sri Lanka.
The remainder of the paper is organized as follows: Section 2 briefly discusses two
neural networks considered for water level prediction. Section 3 describes the proposed
approach, data sets, and the related techniques used for comparison. Section 4 presents the
results with a comprehensive discussion, and Section 5 concludes this work.

2. Application of Neural Networks for Water Level Prediction

Over the past years, artificial neural networks (ANN) have been well developed and
employed in many application areas to extract and learn information from various kinds of
data [40]. The ANN was inspired by the working principles of the biological brain and the
neuron system [41]. Types of ANNs are twofold, based on the connection between neurons
as feed forward neural networks and recurrent neural networks (RNN). FFNN does not
have a feedback path from the neuron’s output to its input, but it does in RNN [42]. Usually,
NNs are arranged as interconnected layer architecture [43].
Different types of ANNs have been exercised in various application areas. Selection
of the right ANN type is vital for optimal performance. It is significant to consider how
structural differences between multiple types of ANNs, such as feed forward neural
networks (FFNN), long short-term memory (LSTM) networks, and gated recurrent units
(GRU), affect results. Each type of network has its strengths and weaknesses, and the
choice of the network depends on the specific task and desired outcomes. For example,
FFNN is good at mapping inputs to outputs in a simple, direct way and often functions
for tasks such as image classification and regression. LSTM and GRU networks are better
suited for tasks involving sequential data, where the current output depends on previous
inputs and outputs, as they can maintain long-term dependencies and handle the temporal
dynamics of the data. These networks are often used for language modeling and time series
prediction tasks.
The first type of relationship selected for the investigation is the traditional one, in
which wetland water level as a function of daily rainfall to the catchment, daily evaporation,
minimum daily temperature, maximum daily temperature, daily relative humidity at
daytime, daily relative humidity at nighttime, and daily average wind speed. Since it
is a direct mapping from the aforementioned climate parameters to the water level, the
FFNN was selected. To investigate the other relationship, the temporal relationship that
exists between daily water levels, the LSTM was selected. The daily water level is a time
series signal, and there could exist a temporal relationship between water levels of the
nearby days.

2.1. Feed Forward Neural Network (FFNN)

In FFNN, the input information passes through neurons and synaptic weights from
input layer neurons to output layer neurons. FFNN uses labeled training data to understand
the relationship between the input and output of the given application and accordingly
adjust the hyperparameters of the NN. It follows an optimization algorithm named Back
Propagation during the training process to converge the model. The prediction error (i.e.,
the difference between the NN output and the desired output) must be estimated at each
epoch. The calculated error propagates backward through the NN, and weights and biases
update according to the gradient of error. It uses an error function named Loss Function.
The Loss Function can be used to estimate the loss of the NN model so that the weights and
biases can be updated to reduce the loss in the next iteration (i.e., epoch). The loss function
for the jth neuron at the output layer for nth iteration can be defined as Equation (1).

L ( n ) j = dj − y ( n ) j (1)
θ
 so that the weights and biases can be updated to reduce the loss in the next iteration (i.e.,
epoch). The loss function for the 𝒋th neuron at the output layer for 𝒏th iteration can be
Appl. Sci. 2023, 13, 2194 5 of 17
defined as Equation (1).
𝑳(𝒏)𝒋𝜽 = 𝒅𝒋 − 𝒚(𝒏)𝒋 (1)
where d𝒅j 𝒋 isisthe
where the desired output, y𝒚(𝒏)
desired output, (n)j 𝒋 isisthe
theNN
NN predicted
predicted output,
output, and
and θ𝜽 denotes
denotes the
the
trainable
trainable parameters (i.e., hyperparameters) of the NN. A standard multi-layer FFNN is
parameters (i.e., hyperparameters) of the NN. A standard multi-layer FFNN is
presented
presented in in Figure
Figure1.1.

Figure 1.
Figure 1. A
A Multi-layer
Multi-layer feed
feed forward
forward neural
neural network.
network.

2.2.
2.2. Long
Long Short-Term
Short-TermMemory
MemoryNeural
NeuralNetworks
Networks(LSTM)
(LSTM)
The
The recurrent neural networks have at least aa single
recurrent neural networks have at least single feedback
feedback connection
connection forfor their
their
neurons,
neurons,allowing
allowingfor forretaining
retainingprevious
previousinformation
information (i.e.,(i.e.,
memory)
memory)within the network
within the network[42].
However, deep RNNs
[42]. However, deep RNNssuffersuffer
from vanishing and exploding
from vanishing and exploding gradient problems
gradient [43]. There
problems [43].
are several
There variants
are several of RNN
variants of existing, including
RNN existing, LSTM,
including LSTM,gatedgated
recurrent unit (GRU),
recurrent and
unit (GRU),
recurrent convolutional neural networks (RCNN) [44]. Among
and recurrent convolutional neural networks (RCNN) [44]. Among them, LSTM has them, LSTM has excellent
resistance to gradient
excellent resistance toproblems [45].
gradient problems [45].
The memory capacity and controllability
The memory capacity and controllability of theofLSTM
the LSTMcell have improved
cell by introduc-
have improved by
ing gates to the LSTM cell. Gates supports NN in dealing with
introducing gates to the LSTM cell. Gates supports NN in dealing with the problem the problem of long-term of
Appl. Sci. 2023, 13, x FOR PEER REVIEW
dependencies. It consists of several types of gates within the cell, including forget 6 gate,
of 18
long-term dependencies. It consists of several types of gates within the cell, including
input
forgetgate,
gate,and output
input gate,gate, as illustrated
and output gate, asinillustrated
Figure 2. in Figure 2.

Figure 2.
Figure 2. An
An illustration
illustration of
of LSTM
LSTM cell.
cell.

The mathematical model of an LSTM cell can be presented in the following Equations
(2)–(7).
𝒇𝒕 = 𝝈 (𝑾𝒇𝒉 𝒉𝒕−𝟏 + 𝑾𝒇𝒙 𝒙𝒕 + 𝒃𝒇 ) (2)

𝒊𝒕 = 𝝈 (𝑾𝒊𝒉 𝒉𝒕−𝟏 + 𝑾𝒊𝒙 𝒙𝒕 + 𝒃𝒊 ) (3)

Appl. Sci. 2023, 13, 2194 6 of 17

The mathematical model of an LSTM cell can be presented in the following Equations (2)–(7).


ft = σ Wfh ht−1 + Wfx xt + bf (2)

it = σ (Wih ht−1 + Wix xt + bi ) (3)

ct = tanh (Wech ht−1 + Wecx xt + bec )
e (4)
ct = ft .ct−1 + it .e
ct (5)
ot = σ (Woh ht−1 + Wox xt + bo ) (6)
ht = ot .tanh(ct ) (7)
where ct denotes the cell state and Wi , Wec , Wo represent the weights while the “operator is
for pointwise multiplication”. During the cell state update, the input gate decides what
information to store in the cell; the output gate decides what information to send out; and
the forget gate decides what information to erase from the cell during the next iteration.
LSTM follows a slightly different algorithm than FFNN, named Back Propagation
Through Time to converge the NN since it deals with temporal information related to
previous timestamps. In addition, it has much better control of deciding what information
to keep and what to remove in the next time stamp. The loss function for LSTM is presented
in Equations (8) and (9).
∑ Lθ
(t)
Lθ = (8)
t

1 T  2
L(θ)(t) = ∑
2T i=1
dj − y(n)j (9)

(t)
where, Lθ is the loss of time stamp t, Lθ is the cumulative loss, and T is the total duration.
Minimizing the computational load in terms of processing power and memory require-
ment for real-world water level prediction model implementation is essential, especially
for rural and remote implementation. Depending on the NN type, hyperparameter com-
binations and the depth of the network will decide the computational load. However,
in-depth comparisons will enable us to find simple NN models that serve powerful pre-
diction performances compared to more complex NN models. In addition to NN models,
genetic programming (GP), and deep GP models are more robust in prediction applications.
However, to have a fair comparison between two types of relationships on common ground
with low computational loading FFNN and LSTM NNs were selected for the study.

3. Materials and Methods

This section presents the mathematical formulation of the problem, the architecture
and hyperparameter selection for the neural networks (NN), input data preparation, input–
output mapping topology, and performance indices used for the analysis. This study
employed two types of NNs, a simple feed forward NN (FFNN) and a long short-term
(LSTM) neural network, for water level prediction purposes.

3.1. Modeling of Feed Forward Neural Network

The FFNNs are used to model the non-linear input-to-output mapping function
between independent variables and the dependent variable, which is the water level of the
wetland, as presented in Equation (10).
 
W L = Fθ RF , Evap , Tmin , Tmax , RHday , RHnight , WS + δ (10)

where WL is the water level of the wetland. RF, Evap, Tmin , Tmax , RHday , RHnight , and
WS are daily rainfall to the catchment (in mm), daily evaporation (in mm), minimum
daily temperature (in ◦ C), maximum daily temperature (in ◦ C), daily relative humidity at
daytime (in %), daily relative humidity at nighttime (in %), and daily average wind speed
Appl. Sci. 2023, 13, 2194 7 of 17

(in km/h). The Fθ is the mapping function, and δ denotes the prediction error. Hence,
the loss function of the FFNN for the given application can be presented as Equation (11),
where θ p represents the trainable parameters of the NN. The architecture of the FFNN used
for the prediction is presented in Table 1.
   2
L p = Fθ RF , Evap , Tmin , Tmax , RHday , RHnight , WS , θ p − W L (11)

Table 1. FNN Network Architecture.

Layer Type Hyper Parameters (Neurons) Activation Function

1st Hidden Layer 128 ReLU
2nd Hidden Layer 64 ReLU
Dropout 0.2 Probability
3rd Hidden Layer 32 ReLU
Dropout 0.2 Probability
4th Hidden Layer 16 ReLU
Dropout 0.2 Probability
5th Hidden Layer 8 ReLU
Output Layer 1 Linear

The proposed FFNN accepts seven independent variable readings as inputs mentioned
above and tries to predict the water level at the output layer. The input variables may have
a varying degree of sensitivity to the water level. Accordingly, the weights and biases of
the NN will be adjusted during the training phase. The NNs are very good at mapping
such non-linear relationships compared to traditional mathematical models.

3.2. Modeling Long Short-Term Neural Network

The water level of a wetland varies continuously throughout the year. Consequently,
the water level of a given day strongly relates to the water levels of previous days. Hence,
it can consider a time series signal (TSS). Therefore, the temporal relationship between
the daily water levels of a wetland can be considered for future water level prediction.
Assuming the temporal information embedded within water level variation during a
consecutive thirty days is strong enough for the future forecast, the last thirty-day water
levels are used as the input for the next day’s water level prediction.
Among the different types of NNs, RNN, LSTM, GRU, and their variants can keep
information related to previous input/output data of the NN within it (i.e., maintaining
memory). Moreover, LSTM presents greater resistance to the exploding and vanishing
gradient, which is one of the critical issues with the RNN family. Considering all facts,
LSTM was selected for TSS-based water level prediction. The loss function of LSTM for the
given application is presented in Equation (12).

T 2
1   
L ( θ )(t) =
2T ∑ θ
F x (t)
− W L (t)
(12)
i =1

(t)
where Lθ is the loss of time step t and x(t) is the input vector to the NN. In this application,
h i
T = 30 and x(t) = W Lt−T −1 , W Lt−(T −1)−1 , . . . , W Lt−1 where W L is the water level. The
architecture of the proposed LSTM network is presented in Table 2.
Appl. Sci. 2023, 13, 2194 8 of 17

Table 2. LSTM Neural Network Architecture.

Layer Type Hyper Parameters Activation Function

LSTM 96 Units
Dropout 0.2 Probability
LSTM 96 Units
Dropout 0.2 Probability
LSTM 96 Units Linear
Dropout 0.2 Probability
LSTM 96 Units
Dropout 0.2 Probability
Dence 1 Neuron

Both networks were implemented using the Python programming language. Pre-
diction performance was measured using three indicators: R2 Score, MSE, and MAE. R2
Score is illustrated in Equation (13). The R2 Score can be determined as a ratio of the total
variation of data points explained by the regression line (sum of squared regression) and
the total variation of data points from the mean (also termed as the sum of squares total or
total sum of squares).
2
∑(ŷi − y)
R2 = 2
(13)
∑ ( yi − y )
where ŷ, and yi represent the predicted value, mean of the values, and actual value,
respectively. Although the R2 score was used for the analysis, the actual vs predicted water
level graphs were also used to demonstrate the prediction performances.
MSE measures the average of the squares of the errors. It can be expressed as in
Equation (14). It is an indication of the difference between measured and observed values.
MSE is always a positive value. A model performs better when the MSE values are lower.
The zero MSE indicates that the model is error-free.
N 2
1 
MSE =
N ∑ Yi, observed − Yi,predicted (14)
i =1

MAE is the magnitude of the difference between the predicted value and the observed
value and the true value of that relevant observation. It is shown in Equation (15).

N
1
MAE =
N ∑ abs(Yi − λ(Xi )) (15)
i =1

where yi is the true target value for the test instance xi , λxi is the predicted target value for
the test instance xi , and n is the number of test instances.
This study used two different NNs (i.e., FFNN and LAST) to investigate the possible
relationships between identified independent variables and water level and the temporal
relationship within daily water levels.
For more accurate results, Hyperparameter tuning is essential for NN-based mod-
els [46]. This study decided to use manual hyperparameter tuning due to the simplicity
of the NN architectures and considering available computational resources. Multiple NN
initialization techniques were tested [47], but there was no considerable contribution from
the initialization method to reduce the convergence of the prediction model. Finally, ran-
dom initialization was selected for both NNs since it was marginally better than the others.
Several optimizers were tested concerning each model. The Adam optimizer [48] was
chosen to compare models due to its unbiased nature. Due to the limited availability of
data points, the data was fed as a single batch for model training. However, all data fields
were normalized before feeding to the NNs for faster model convergence. Considering
the higher possibility of having a non-linear relationship between water level(s) and input
climate parameters, the ReLU activation function was selected for FFNN hidden layers.
Appl. Sci. 2023, 13, 2194 9 of 17

However, both NNs are used with Liner activation at the output for water level prediction.
In addition, the authors trialed several hyperparameter combinations, especially the depth
of the NNs, via multiple NN models for both FFNN and LSTM NNs. The prediction
accuracy (i.e., R2 , MAE, and MSE) and time taken for a single training epoch were ana-
lyzed against the depth of the NN models. After detailed analysis, the proposed two NN
architectures were finalized. The time taken per single training epoch was considered as a
measure of computational load. Both NN models tested for a different number of training
epochs, and 200 epochs was selected considering the optimum accuracy concerning the
epoch number.
Uncertainty quantification (UQ) is essential for any prediction model for more accurate
and reliable estimation [49–51]. However, UQ can be more time-consuming for NN-based
models. It requires substantial amount of data, one of the key-limiting factors for real-
world applications like wetland water level predictions. Compared to NN-based models,
machine-learning techniques like GP and Deep GP can perform UQ readily without the
need for extensive data volume [52].
Proposed models are subjected to both uncertainties, including aleatoric uncertainty
and epistemic uncertainties. Within the study, we tried to reduce the impact of epistemic
uncertainty by manually finetuning the NN hyperparameters and selecting the most
suitable train-test slip after several trials. However, this study had limited data points that
led to poor UQ and model generalization.
An Intel® Core i7 processor with 32GB RAM and a Nvidia® GPU with 2GB memory
were used for all neural network modelling, training, and testing. The Python programming
language was used with Tensorflow® and Keras® (at the backend) for modeling the selected
CNN architectures.
Appl. Sci. 2023, 13, x FOR PEER REVIEW 10 of 18

4. Case Study
The m
1.0–3.0 Colombo
above mean floodsea
detention
level. Thearea is located
Metro in the
Colombo western
Basin coversProvince
105 km2,ofand Srithe
Lanka.
total As
shown in Figure
wetland coverage3, it
is consists
about 20ofkm three
2. wetlands/marshes: Kotte marsh, Kolonnawa marsh,
and HeenThe marsh.
capacityThe urban
of the landscape
Colombo floodsurrounds thesehas
detention basin marshes. Theby
decreased climate
30% as inaColombo
result
isoftropical monsoon.
urbanization [38].The average
Hence, annual
flooding is atemperature rangestofrom
significant threat 26.5 ◦ C
Colombo to 28.5
city. ◦ C, and
Increased
it surface
typically receives
runoff brought2300on
mm byof precipitation
rapid eachdeclining
development, year. Theretention
study area is approximately
areas, and lack of
1.0–3.0 m above
capacities mean sea
in wetlands canlevel. The Metro
be identified Colombo
as the leading Basin
causescovers 105 km
of flooding. 2 , and the total
Consequently,
Colombo
wetland district’sisflood
coverage aboutrisk km2be
20 can . reduced by putting strategies in place.

Figure3.3.Selected
Figure Selectedcatchment
catchment for
for the
the study:
study: Colombo
Colomboflood
flooddetention
detentionbasin.
basin.

Four water level monitoring stations are located within the basin, Kimbulawala
Bridge, Kotte North Canal, Kotte Canal, and Parliament Lake. The data set of each wetland
contains daily water levels and the readings of other independent meteorological
variables. Wetland water level data were obtained from Sri Lanka Land Development
Appl. Sci. 2023, 13, 2194 10 of 17

The capacity of the Colombo flood detention basin has decreased by 30% as a result of
urbanization [38]. Hence, flooding is a significant threat to Colombo city. Increased surface
runoff brought on by rapid development, declining retention areas, and lack of capacities
in wetlands can be identified as the leading causes of flooding. Consequently, Colombo
district’s flood risk can be reduced by putting strategies in place.
Four water level monitoring stations are located within the basin, Kimbulawala Bridge,
Kotte North Canal, Kotte Canal, and Parliament Lake. The data set of each wetland contains
daily water levels and the readings of other independent meteorological variables. Wetland
water level data were obtained from Sri Lanka Land Development Corporation, while
meteorological data were obtained from the Department of Meteorology, Sri Lanka.
The number of data points recorded for each wetland selected for the study differs.
Therefore, considering the number of data points, the train-test data split was decided as
presented in Table 3.

Table 3. Train-Test Data Split.

Total Number of Training Data

Wetland Testing Data Volume
Data Points Volume
Kimbulawala Bridge 1887 1500 387
Kotte Canal 1584 1200 384
Kotte North 457 375 82
Parliament Lake 1126 800 326

The range of each variable is substantially different from the range of the others. For
example, the temperature range between 20 and 35 ◦ C, the RH value; is between 65 and
95%; the evaporation is between 2 and 3 mm; and the water level is between 0.2 and 0.4 m.
This difference in value ranges of each variable will adversely affect the speed of the NN
convergence and the time taken to adjust NN hyperparameters. Hence, each variable was
Appl. Sci. 2023, normalized to set it between the ranges of 0 and 1, the normalized water level
13, x FOR PEER REVIEW 11 of 18variation,
along with the training–testing split presented in Figure 4. In FFNN based prediction
model, the water level is used as output, whereas LSTM-based prediction is used as input
and output both prediction
in themodel, the water
training level is used as output, whereas LSTM-based prediction is
phase.
used as input and output both in the training phase.

(a) (b)

(c) (d)
Figure 4. Train–Test data split of (a) Kimbulawala, (b) Kotte Canal, (c) Kotte North, and (d)
Figure 4. Train–Test dataLake.
Parliament split of (a) Kimbulawala, (b) Kotte Canal, (c) Kotte North, and (d) Parliament Lake.

5. Results and Discussion

The R2 Score, MSE, and MAE were measured for both FFNN and LSTM network-
based models for each wetland water level prediction, and the results are presented in
Table 4. The results confirmed that the LSTM had performed better than FFNN in
Appl. Sci. 2023, 13, 2194 11 of 17

5. Results and Discussion

The R2 Score, MSE, and MAE were measured for both FFNN and LSTM network-based
models for each wetland water level prediction, and the results are presented in Table 4.
The results confirmed that the LSTM had performed better than FFNN in predicting the
water levels of the wetlands in all three performance indices considered for the comparison.

Table 4. The water level prediction performances.

R2 Score MSE MAE

Wetland
FFNN LSTM FFNN LSTM FFNN LSTM
Kimbulawala Bridge 0.1685 0.8118 0.0212 0.0017 0.0931 0.0249
Kotte Canal 0.2076 0.8786 0.0112 0.0004 0.0668 0.0155
Kotte North 0.1478 0.8345 0.0082 0.0011 0.0548 0.0216
Parliament Lake 0.1530 0.6368 0.0173 0.0069 0.0961 0.0371

The performance of both prediction types (i.e., tradition and temporal) for all wetlands
considered for the study is present in Figure 5. It presents similar patterns in both FFNN and
Appl. Sci. 2023, 13, x FOR PEER REVIEW 12 of 18
LSTM networks’ performances against each performance index. Both NNs similarly treated
each wetland without any bias. Hence, it confirmed that there is no specific relationship
between the water level data of a particular wetland and a given NN type (i.e., LSTM or
type (i.e.,
FFNN). LSTM ora detailed
Moreover, FFNN). Moreover,
analysis of aeach
detailed analysis
wetland’s FFNN of performance
each wetland’s FFNN
confirmed
performance
no differenceconfirmed no difference
in the degree in thebetween
of correlation degree of correlation between
independent variablesindependent
selected for
variables selected
investigation withfor
theinvestigation with the
water level among water level among wetlands.
wetlands.

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Kimbulawala Bridge Kotte Canal Kotte North Parliament Lake

R2-FFNN R2-LSTM MSE-FFNN MSE-LSTM MAE-FFNN MAE-LSTM

Figure 5.
Figure Representation of
5. Representation of water
water level
level prediction
prediction performances.
performances.

The performance difference between LSTM- and FFNN-based models for all wetlands
The performance difference between LSTM- and FFNN-based models for all
is substantial. The higher performance of LSTM over FFNN concludes that the temporal
wetlands is substantial. The higher performance of LSTM over FFNN concludes that the
relationship between the daily water level is much stronger than that between the identified
temporal relationship between the daily water level is much stronger than that between
independent variables and water level. However, the number of data points may be
the identified independent variables and water level. However, the number of data points
inadequate to learn the composite relationship between the seven independent variables
may be inadequate
and the water level to in learn
the FFNNthe composite
approach.relationship
The highestbetween
number the seven
of data independent
points used for
variables and the water level in the FFNN approach. The highest
the investigation is 1887 from the Kimbulawala Bridge wetland. As per the thumb number of data points
rule, it
used for theto
is required investigation
have at leastis a1887
fewfrom the Kimbulawala
100,000 data points toBridge wetland.
learn the As per the
relationship, thumb
especially
rule, it is input
for seven required to havesuccessfully.
variables, at least a few 100,000
Hence, data points
increasing to learn
the volume the relationship,
of training data can
especially
improve the forprediction
seven input variables, successfully.
performance of FFNN. Hence, increasing the volume of training
data can
The improve the prediction
NN prediction performance
performances of FFNN.
(R2 ) were further analyzed against the training
The NN prediction performances ( R2) were further analyzed against the training data
data volume of the respective wetland. In order to make a better graphical comparison,
volume of the respective wetland. In order to make a better graphical comparison,
normalized values of each field are calculated as presented in Table 5, and Figure 6
illustrates the performance variation viz. training data volume. The normalization will not
affect the performance comparison since it is only used to analyze the relative shape of
performance pattern variation.
 Appl. Sci. 2023, 13, 2194 12 of 17

normalized values of each field are calculated as presented in Table 5, and Figure 6 illustrates
the performance variation viz. training data volume. The normalization will not affect the
performance comparison since it is only used to analyze the relative shape of performance
pattern variation.

Table 5. Normalized performances and training data volume.

Normalized
Wetland
Training Data R2 Value of R2 Value of Training Data Normalized R2 Normalized R2
Volume FFNN LSTM Value of FFNN Value of LSTM
Volume
Kimbulawala 1500 0.1685 0.7917 0.3870 0.2572 0.2520
Bridge
Appl. Sci.Kotte
2023, Canal 1200
13, x FOR PEER REVIEW 0.2076 0.8786 0.3096 0.3168 0.2796
13 of 18
Kotte North 800 0.126 0.8345 0.2064 0.1923 0.2656
Parliament Lake 375 0.153 0.6368 0.0967 0.2335 0.2026

Kimbulawala Bridge
0.4

0.3

0.2

0.1
Parliament Lake 0 Kotte Canal

Kotte North

Normalized Training Normalized Normalized R2 Sore of

Data Volume R2 Score of FFNN LSTM

Figure 6. Normalized prediction performances (R2 Score) and training data volume analysis.
Figure 6. Normalized prediction performances (R2 Score) and training data volume analysis.
As per the graphical illustration of Figure 6, both FFNN and LSTM prediction models
As per the graphical illustration of Figure 6, both FFNN and LSTM prediction models
followed a similar pattern (only the shape of variation but not performance values) in the
followed a similar
normalized pattern (only
performance. the shape
Hence, of variation
it confirmed both NNsbut not performance
endeavored values)the
to discover in hidden
the
normalized performance. Hence, it confirmed both NNs endeavored
relationships (i.e., two types of relationships) for water level prediction. Moreover, although to discover the
hidden
two relationships (i.e., two types
kinds of relationships of relationships)
considered for the studyfor water level prediction.
(i.e., temporal Moreover,
and traditional) have
although
provided a different level of performance, both have a similar degree of informationand
two kinds of relationships considered for the study (i.e., temporal within
traditional) havedata.
all wetland provided a different
Therefore, both level
NNs of performance,
extracted both of
two types have a similar degree
relationship of
information
information within all wetland data. Therefore, both NNs extracted
from each wetland data in an almost similar ratio. This is confirmed by having normalized two types of
relationship
performance information
patterns with fromtheeachsame wetland
shape.data in anasalmost
However, similar
per Figure 5, itratio.
couldThis is a
not find
confirmed by having normalized performance patterns with the same
strong relationship between the amount of training data and prediction accuracy between shape. However, as
per wetlands.
Figure 5, itHence,
could not find a strong
it confirmed relationship
the degree between the
of relationship amount ofembedded
information training data in the
andgiven
prediction
wetland accuracy
data isbetween
much more wetlands.
importantHence,
thanit the
confirmed
available the degreeofofdata
volume relationship
for accurate
information
water levelembedded
prediction. in the given wetland data is much more important than the
availableSince
volume of data
LSTM-based forprediction
accurate water
providedlevela prediction.
higher performance, LSTM was selected for
Since LSTM-based prediction
further analysis. The comparison between provided a higher performance,
the actual LSTM water
and predicted was selected
levels for
by the
further
LSTM analysis.
network Theforcomparison
each wetland between the actual
is presented and predicted
in Figure 7. Exceptwater for a levels
few days by the
in the
LSTM network Lake
Parliament for each wetland
wetland, for is
allpresented in Figure
other wetlands, the7.LSTM
Except for awas
model few able
daystoinprovide
the
Parliament Lake wetland,
greater prediction accuracy.for all other wetlands,
Although, the LSTM model
in some applications, the R2wasscoreable to to
failed provide
represent
greater prediction
a direct comparison accuracy. Although,
performance in some
of actual vs. applications,
predicted but therepresents
R2 score afailed to
correlation.
represent
However, a direct
in this comparison
application, the performance
comparisonofgraphs actualconfirmed
vs predictedthat Rbut2 was represents a
able to directly
correlation.
compare However, in this application,
actual vs predicted water levelsthe comparison
together graphs
with other confirmedindices
performance that Rthat
2 was MSE
ableand MAE selected
to directly compare foractual
the analysis.
vs predicted water levels together with other performance
indices that MSE and MAE selected for the analysis.
A comparison of actual vs predicted water levels in the graphs for all wetlands
showed that the LSTM model could accurately predict average and extreme values (i.e.,
peaks and valleys). It confirms the greater generalization capability of the LSTM-based
Appl.
Appl. Sci. 2023, 13,
Sci. 2023, 13, 2194
x FOR PEER REVIEW 13 of
14 of 17
18

(a) (b)

(c) (d)
Figure 7.
Figure 7. Actual
Actual vs.
vs predicted
predicted water
water level
level comparison
comparison for
for LSTM-based
LSTM-based model
model of
of (a) Kimbulawala,
(a) Kimbulawala,
(b) Kotte Canal, (c) Kotte North, and (d) Parliament Lake.
(b) Kotte Canal, (c) Kotte North, and (d) Parliament Lake.

Further,
A the loss
comparison curves
of actual of the LSTM
vs predicted waternetwork during
levels in the thefor
graphs training period
all wetlands were
showed
analyzed and presented in Figure 8. It will provide insights into NN learning
that the LSTM model could accurately predict average and extreme values (i.e., peaks and behavior,
capability,
valleys). It and speed.
confirms Asgreater
the per thegeneralization
loss curves, Kotte Canalof
capability showed the best convergence
the LSTM-based prediction
behavior,
model. which
A few was reflected
variations in theinresults
can be found of the
the graphs, R scorefor
2
especially comparison in Table
the Kotte North 6. It
wetland
concludes
water levelthat the amount
prediction. of embedded
As stated, relationship
the relationship information
between withininthe
water levels data set is
a wetland is
much more
highly substantial
non-linear than the training
and complex. Hence, data volume inprediction
the achieved water level prediction. of LSTM
performance
is outstanding. These results indirectly confirm the robustness of the prediction model
against UQ, especially in extreme values (i.e., peaks and valleys) of water levels.
Further, the loss curves of the LSTM network during the training period were analyzed
and presented in Figure 8. It will provide insights into NN learning behavior, capability,
and speed. As per the loss curves, Kotte Canal showed the best convergence behavior,
which was reflected in the results of the R2 score comparison in Table 6. It concludes that the
amount of embedded relationship information within the data set is much more substantial
than the training data volume in water level prediction.

(a) (b)
 Further, the loss curves of the LSTM network during the training period were
analyzed and presented in Figure 8. It will provide insights into NN learning behavior,
capability, and speed. As per the loss curves, Kotte Canal showed the best convergence
behavior, which was reflected in the results of the R2 score comparison in Table 6. It
Appl. Sci. 2023, 13, 2194 14 of 17
concludes that the amount of embedded relationship information within the data set is
much more substantial than the training data volume in water level prediction.

Appl. Sci. 2023, 13, x FOR PEER REVIEW 15 of 18

(a) (b)

(c) (d)
Figure 8. Loss curves of LSTM-based model for (a) Kimbulawala, (b) Kotte Canal, (c) Kotte North,
Figure 8. Loss curves of LSTM-based model for (a) Kimbulawala, (b) Kotte Canal, (c) Kotte North,
and (d) Parliament Lake.
and (d) Parliament Lake.
Table 6. Comparison of perdition performance of the LSTM model with existing studies.
Table 6. Comparison of perdition performance of the LSTM model with existing studies.
Disaggregation Studies R2 Score MSE MAE
Disaggregation Studies& Abghari, 2016 [53] R2 Score
Vaheddoost, Aksoy, 0.989 MSE - MAE -
Young,
Vaheddoost, Aksoy, Liu, & Hsieh,
& Abghari, 2015 [54]
2016 [53] 0.989 0.985 - - - 1.27
Young, Liu, & Hsieh,
Assem2015 [54]2017 [55]
et al., 0.985 0.853 - - 1.27 0.039
Assem et al., 2017
Daliakopoulos, [55]
Coulibaly, & Tsanis, 2005 [56] 0.853 0.985 - - 0.039 -
Daliakopoulos, Coulibaly, & Tsanis, 2005 [56] 0.985 - -
Current study
Current study 0.8786 0.8786 0.0004 0.0004 0.0155
0.0155

The performance of the proposed model is compared with the performance of the
The studies,
existing performance of the proposed
as presented in Tablemodel is compared
6, although thewith
studythe is
performance of the
for relationship
existing studies, as presented in Table 6, although the study is for relationship
comparison. Table 6 contains two wetland water level predictions: one urban water flow comparison.
Table 6 contains
prediction and onetwogroundwater
wetland water level
level predictions:
prediction one urbanAll
performance. watertheflow
listedprediction and
studies used
one groundwater level prediction performance. All the listed studies
a large volume of training data compared with the current study. In contrast, the proposedused a large volume
of training
LSTM model data
hascompared
achieved with the current
substantial study.
results usingInlimited
contrast, the proposed
training data. LSTM model
has achieved substantial results using limited training data.
One of the critical limitations of the study is the inadequate amount of data for the
OneFor
analysis. of the critical
effective limitations
model of the study
development, is the inadequate
it is required amount
to have at least a fewof100,000
data fordata
the
analysis. For effective model development, it is required to have at
points per wetland. In addition, it was assumed that the independent variable readingsleast a few 100,000 data
points
did notper wetland.
deviate In addition,
considerably it was
during assumed
that day from that the
the independent
record variableHowever,
measurement. readings the
did
not deviate considerably during that day from the record measurement.
study showcases the importance of closely monitoring wetland data in Sri Lanka for However, the study
showcases the importance of closely monitoring wetland data in Sri Lanka for sustainable
sustainable water resources management. It is well understood that traditional data
water resources management. It is well understood that traditional data monitoring process
monitoring process can be cost consuming; however, water level measuring and weather
can be cost consuming; however, water level measuring and weather sensors can be
sensors can be effectively used for long-term data recording without much human
effectively used for long-term data recording without much human intervention.
intervention.

6. Conclusions
Extant literature is limited to considering only the traditional relationship between
environmental parameters and wetland water level for future water level predictions. For
future projections, this study presents a novel concept using a temporal relationship
Appl. Sci. 2023, 13, 2194 15 of 17

6. Conclusions
Extant literature is limited to considering only the traditional relationship between
environmental parameters and wetland water level for future water level predictions.
For future projections, this study presents a novel concept using a temporal relationship
between daily wetland water levels. Moreover, both traditional and temporal relationships
were modeled using two types of ANN, and prediction performance was analyzed in
detail. Several NN architectures were compared within the study, and lightweight, high-
performing NN models for LSTM and FFNN were selected.
The results confirmed that the temporal relationship within daily wetland water levels
is much more robust in predicting future water levels than using the traditional relationship
between environmental parameters and water level. The detailed analysis found that the
degree of embedded relationship information within the wetland data is more critical than
the volume of training data regarding water level prediction. Moreover, the normalized
performance graph comparison presented that although the performance level substantially
differs between FFNN and LSTM models, the prediction accuracy pattern between wetlands
is similar. Hence, two prediction models learned two different relationships from wetland
data in almost the same ratio from each wetland. The actual vs predicted water level graphs
reconfirm the performance measured using the R2 Score, MSE, and MAE. The model can
predict extreme cases like very high and low water levels, which is essential in the said
application area. It also reconfirmed the greater generalization capability of the proposed
LSTM model and its robustness against the uncertainties.
However, ANN models are subjected to considerable deviation in predicting results
out of the training data range. Hence it is required to accommodate a wider range of
data for model training. Manual hyperparameter training is an exhaustive effort, and it is
necessary to have expert knowledge compared to automatic hyperparameter tuning models.
Compared to other techniques like GP that can perform UQ readily, it is comparatively
difficult to accommodate uncertainty quantification for the ANN-based models due to
limited training data and extensive time consumption.
Future studies can investigate and select the most influenceable input parameters
(i.e., variables) for water level predictions than using all seven identified input variables.
Moreover, immediate effect analysis for water level variation from some selected input
variables can be analyzed as an extended study. Finally, the comparative study will be
extended to a comprehensive analysis of soft computing methods, including ANNs, GP,
and other ML techniques for wetland water level prediction.

Author Contributions: Conceptualization, U.R.; methodology, M.H.; software, M.H.; validation,

M.H.; formal analysis, M.H.; resources, U.R.; data curation, M.H.; writing—original draft preparation,
M.H. and T.J.; writing—review and editing, Y.H. and U.R.; supervision, U.R.; project administration,
U.R. 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: The meteorological and water level data used in this research work
can be requested by corresponding authors only for research purposes.
Acknowledgments: The authors acknowledge the support that they have received from the Sri Lanka
Institute of Information Technology (SLIIT), Sri Lanka, on data purchase.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Hammer, D.A.; Bastian, R.K. Wetlands ecosystems: Natural water purifiers? In Constructed Wetlands for Wastewater Treatment;
CRC Press: Boca Raton, FL, USA, 2020; pp. 5–19.
Appl. Sci. 2023, 13, 2194 16 of 17

2. Pashaei, M.; Kamangir, H.; Starek, M.J.; Tissot, P. Review and evaluation of deep learning architectures for efficient land cover
mapping with UAS Hyper-spatial imagery: A case study over a wetland. Remote Sens. 2020, 12, 959. [CrossRef]
3. Mitsch, W.J.; Gosselink, J.G. Wetlands, 5th ed.; Wiley: Hoboken, NJ, USA, 2015.
4. Mitsch, W.J.; Bernal, B.; Hernandez, M.E. Ecosystem services of wetlands. Int. J. Biodivers. Sci. Ecosyst. Serv. Manag. 2015, 11, 1–4.
[CrossRef]
5. Were, D.; Kansiime, F.; Fetahi, T.; Cooper, A.; Jjuuko, C. Carbon Sequestration by Wetlands: A Critical Review of Enhancement
Measures for Climate Change Mitigation. Earth Syst. Environ. 2019, 3, 327–340. [CrossRef]
6. Janssen, R.; Goosen, H.; Verhoeven, M.L.; Verhoeven, J.T.; Omtzigt, A.Q.A.; Maltby, E. Decision support for integrated wetland
management. Environ. Model. Softw. 2005, 20, 215–229.
7. Goodarzi, D.; Mohammadian, A.; Pearson, J.; Abolfathi, S. Numerical modelling of hydraulic efficiency and pollution transport in
Waste Stabilization Ponds. Ecol. Eng. 2022, 182, 106702. [CrossRef]
8. Wang, W.; Sun, M.; Li, Y.; Zhao, S.; Zhang, Z.; Luan, X. Multi-level comprehensive assessment of Constructed Wetland Ecosystem
Health: A case study of cuihu wetland in Beijing, China. Sustainability 2022, 14, 13439. [CrossRef]
9. Noori, R.; Farahani, F.; Jun, C.; Aradpour, S.; Bateni, S.M.; Ghazban, F.; Hosseinzadeh, M.; Maghrebi, M.; Naseh, M.R.V.;
Abolfathi, S. A non-threshold model to estimate carcinogenic risk of nitrate-nitrite in drinking water. J. Clean. Prod. 2022, 363,
132432. [CrossRef]
10. Dehestaniathar, S.; Nesari, S.; Borzooei, S.; Abolfathi, S. Application of natural biodegradable fiber as biofilm medium and carbon
source in denitrifying ammonium oxidation (DEAMOX) process for nitrogen removal from wastewater. J. Taiwan Inst. Chem. Eng.
2021, 119, 108–114. [CrossRef]
11. Yousaf, A.; Khalid, N.; Aqeel, M.; Noman, A.; Naeem, N.; Sarfraz, W.; Ejaz, U.; Qaiser, Z.; Khalid, A. Nitrogen Dynamics in
Wetland Systems and Its Impact on Biodiversity. Nitrogen 2021, 2, 196–217. [CrossRef]
12. Savage, R.; Baker, V. The Importance of Headwater Wetlands and Water Quality in North Carolina; NC Department of Environment
and Natural Resources, NC Division of Water Quality: Raleigh, NC, USA, 2007.
13. Taillardat, P.; Thompson, B.; Garneau, M.; Trottier, K.; Friess, D. Climate change mitigation potential of wetlands and the
cost-effectiveness of their restoration. Interface Focus 2020, 10, 20190129. [CrossRef] [PubMed]
14. Stein, E.D.; Fetscher, A.E.; Clark, R.P.; Wiskind, A.; Grenier, J.L.; Sutula, M.; Collins, J.N.; Grosso, C. Validation of a wetland rapid
assessment method: Use of EPA’sEPA’s level 1-2-3 framework for method testing and refinement. Wetlands 2009, 29, 648–665.
15. Junk, W.J.; An, S.; Finlayson, C.M.; Gopal, B.; Květ, J.; Mitchell, S.A.; Mitsch, W.J.; Robarts, R.D. Current state of knowledge
regarding the world’s wetlands and their future under Global Climate Change: A synthesis. Aquat. Sci. 2012, 75, 151–167.
[CrossRef]
16. Lefebvre, G.; Redmond, L.; Germain, C.; Palazzi, E.; Terzago, S.; Willm, L.; Poulin, B. Predicting the vulnerability of seasonally-
flooded wetlands to climate change across the Mediterranean Basin. Sci. Total Environ. 2019, 692, 546–555. [CrossRef] [PubMed]
17. Malekmohammadi, B.; Uvo, C.B.; Moghadam, N.T.; Noori, R.; Abolfathi, S. Environmental Risk Assessment of Wetland
Ecosystems Using Bayesian Belief Networks. Hydrology 2023, 10, 16. [CrossRef]
18. Convention on Wetlands. Global Wetland Outlook: Special Edition 2021; Secretariat of the Convention on Wetlands: Gland, Switzerland, 2021.
19. Junk, W.J.; Piedade, M.T.F.; Schöngart, J.; Cohn-Haft, M.; Adeney, J.M.; Wittmann, F. A classification of major naturally-occurring
Amazonian lowland wetlands. Wetlands 2011, 31, 623–640.
20. Bradley, C. Simulation of the annual water table dynamics of a floodplain wetland, Narborough Bog, UK. J. Hydrol. 2002, 261,
150–172.
21. Acreman, M.C.; Miller, F. Hydrological impact assessment of wetlands. In Proceedings of the ISGWAS Conference on Groundwater
Sustainability, Alicante, Spain, 24–27 January 2006; pp. 225–255.
22. Cózar, A.; García, C.M.; Gálvez, J.A.; Loiselle, S.A.; Bracchini, L.; Cognetta, A. Remote sensing imagery analysis of the lacustrine
system of Ibera wetland (Argentina). Ecol. Model. 2005, 186, 29–41. [CrossRef]
23. Helfer, F.; Lemckert, C.; Zhang, H. Impacts of climate change on temperature and evaporation from a large reservoir in Australia.
J. Hydrol. 2012, 475, 365–378. [CrossRef]
24. Trenberth, K. Changes in precipitation with climate change. Clim. Res. 2011, 47, 123–138. [CrossRef]
25. Lacombe, G.; Chinnasamy, P.; Nicol, A. Review of Climate Change Science, Knowledge and Impacts on Water Resources in South Asia;
Background Paper 1; International Water Management Institute (IWMI): Colombo, Sri Lanka, 2019. [CrossRef]
26. Davarzani, H.; Smits, K.; Tolene, R.M.; Illangasekare, T. Study of the effect of wind speed on evaporation from soil through
integrated modeling of the atmospheric boundary layer and shallow subsurface. Water Resour. Res. 2014, 50, 661–680. [CrossRef]
27. Altunkaynak, A. Forecasting Surface Water Level Fluctuations of Lake Van by Artificial Neural Networks. Water Resour. Manag.
2006, 21, 399–408. [CrossRef]
28. Choi, C.; Kim, J.; Han, H.; Han, D.; Kim, H.S. Development of Water Level Prediction Models Using Machine Learning in
Wetlands: A Case Study of Upo Wetland in South Korea. Water 2020, 12, 93. [CrossRef]
29. Dadaser-Celik, F.; Cengiz, E. A neural network model for simulation of water levels at the Sultan Marshes wetland in Turkey.
Wetl. Ecol. Manag. 2013, 21, 297–306. [CrossRef]
30. Gopakumar, R.; Takara, K. Water level forecasting in the Vembanad water system using artificial neural networks. In Proceedings
of the 2009 International Conference on Advances in Recent Technologies in Communication and Computing, Kerala, India,
27–28 October 2009; IEEE: Piscataway, NJ, USA, 2009; pp. 948–953.
Appl. Sci. 2023, 13, 2194 17 of 17

31. Saha, T.; Pal, S.; Sarkar, R. Prediction of wetland area and depth using linear regression model and artificial neural network based
cellular automata. Ecol. Inform. 2021, 62, 101272. [CrossRef]
32. Mcudp, G. Final Metro Colombo Wetland Management Strategy. Consultancy Services for the Preparation of Management Strategy for
Wetlands and Carrying out an Assessment of Water Quality in the Inland Waterways and Lakes within Metro Colombo Area; MCUDP,
Government of Sri Lanka: Colombo, Sri Lanka, 2016.
33. Jayathilake, T.; Sarukkalige, R.; Hoshino, Y.; Rathnayake, U. Wetland Water Level Prediction Using Artificial Neural Networks—A
Case Study in the Colombo Flood Detention Area, Sri Lanka. Climate 2023, 11, 1. [CrossRef]
34. Amaratunga, V.; Wickramasinghe, L.; Perera, A.; Jayasinghe, J.; Rathnayake, U. Artificial Neural Network to Estimate the Paddy
Yield Prediction Using Climatic Data. Math. Probl. Eng. 2020, 2020, 8627824. [CrossRef]
35. Rhif, M.; Ben Abbes, A.; Farah, I.R.; Martínez, B.; Sang, Y. Wavelet Transform Application for/in Non-Stationary Time-Series
Analysis: A Review. Appl. Sci. 2019, 9, 1345. [CrossRef]
36. Wang, F.; Zhen, Z.; Wang, B.; Mi, Z. Comparative Study on KNN and SVM Based Weather Classification Models for Day Ahead
Short Term Solar PV Power Forecasting. Appl. Sci. 2018, 8, 28. [CrossRef]
37. Franklin, J. Mapping Species Distributions: Spatial Inference and Prediction; Cambridge University Press: Cambridge, UK, 2010.
38. Liu, Y.; Xu, J.; Yuan, Z.; Sha, Z. Analysis of the influence of the water balance process on the change of landscape patterns in the
upper reaches of the Yangtze River. Water Policy 2022, 24, 261–281. [CrossRef]
39. Khan, S.; Yairi, T. A review on the application of deep learning in system health management. Mech. Syst. Signal Process. 2018,
107, 241–265. [CrossRef]
40. Sazli, M.H. A Brief Review of Feedforward Neural Networks. Commun. Fac. Sci. Univ. Ank. Ser. A2-A3 2006, 50, 11–17.
41. Staudemeyer, R.C.; Morris, E.R. Understanding LSTM—A tutorial into long short-term memory recurrent neural networks. arXiv
2019, arXiv:1909.09586.
42. Sharkawy, A.N. Principle of neural network and its main types. J. Adv. Appl. Comput. Math. 2020, 7, 8–19. [CrossRef]
43. Lipton, Z.C.; Kale, D.C.; Elkan, C.; Wetzel, R. Learning to diagnose with LSTM recurrent neural networks. arXiv 2015,
arXiv:1511.03677.
44. Yu, Y.; Si, X.; Hu, C.; Zhang, J. A review of recurrent neural networks: LSTM cells and network architectures. Neural Comput.
2019, 31, 1235–1270.
45. Liang, M.; Hu, X. Recurrent convolutional neural network for object recognition. In Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition, Boston, MA, USA, 7–12 June 2015; pp. 3367–3375.
46. Sajjad, U.; Hussain, I.; Hamid, K.; Ali, H.M.; Wang, C.C.; Yan, W.M. Liquid-to-vapor phase change heat transfer evaluation and
parameter sensitivity analysis of nanoporous surface coatings. Int. J. Heat Mass Transf. 2022, 194, 123088.
47. Zhang, Z. Improved adam optimizer for deep neural networks. In Proceedings of the 2018 IEEE/ACM 26th International
Symposium on Quality of Service (IWQoS), Banff, AB, Canada, 4–6 June 2018; IEEE: Piscataway, NJ, USA, 2018; pp. 1–2.
48. Narkhede, M.V.; Bartakke, P.P.; Sutaone, M.S. A review on weight initialization strategies for neural networks. Artif. Intell. Rev.
2022, 55, 291–322. [CrossRef]
49. Noori, R.; Ghiasi, B.; Salehi, S.; Esmaeili Bidhendi, M.; Raeisi, A.; Partani, S.; Meysami, R.; Mahdian, M.; Hosseinzadeh, M.;
Abolfathi, S. An Efficient Data Driven-Based Model for Prediction of the Total Sediment Load in Rivers. Hydrology 2022, 9, 36.
[CrossRef]
50. Ghiasi, B.; Noori, R.; Sheikhian, H.; Zeynolabedin, A.; Sun, Y.; Jun, C.; Hamouda, M.; Bateni, S.M.; Abolfathi, S. Uncertainty
quantification of granular computing-neural network model for prediction of pollutant longitudinal dispersion coefficient in
aquatic streams. Sci. Rep. 2022, 12, 4610. [CrossRef]
51. Abdar, M.; Pourpanah, F.; Hussain, S.; Rezazadegan, D.; Liu, L.; Ghavamzadeh, M.; Fieguth, P.; Cao, X.; Khosravi, A.;
Acharya, U.R.; et al. A review of uncertainty quantification in Deep learning: Techniques, applications and challenges. Inf.
Fusion 2021, 76, 243–297. [CrossRef]
52. Donnelly, J.; Abolfathi, S.; Pearson, J.; Chatrabgoun, O.; Daneshkhah, A. Gaussian process emulation of spatio-temporal outputs
of a 2D inland flood model. Water Res. 2022, 225, 119100. [CrossRef] [PubMed]
53. Vaheddoost, B.; Aksoy, H.; Abghari, H. Prediction of water level using monthly lagged data in Lake Urmia, Iran. Water Resour.
Manag. 2016, 30, 4951–4967. [CrossRef]
54. Young, C.; Liu, W.; Hsieh, W. 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, 1–11. [CrossRef]
55. Assem, H.; Ghariba, S.; Makrai, G.; Johnston, P.; Gill, L.; Pilla, F. Urban water flow and water level prediction based on Deep
Learning. In Proceedings of the Machine Learning and Knowledge Discovery in Databases, Skopje, Macedonia, 18–22 September
2017; pp. 317–329. [CrossRef]
56. Daliakopoulos, I.N.; Coulibaly, P.; Tsanis, I.K. Groundwater level forecasting using artificial neural networks. J. Hydrol. 2005, 309,
229–240. [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.