Available online at [Link].

com
Available online at [Link]

ScienceDirect
Available online at [Link]
ScienceDirect
Procedia ScienceDirect
Computer Science 00 (2024) 000–000
Procedia Computer Science 00 (2024) 000–000 [Link]/locate/procedia
Procedia Computer Science 241 (2024) 618–623 [Link]/locate/procedia

The 3rd International Workshop on Edge IA-IoT for Smart Agriculture (SA2IOT)
The 3rd International Workshop
August 5-7, 2024, on Edge
Marshall IA-IoTHuntington,
University, for Smart Agriculture
WV, USA(SA2IOT)
August 5-7, 2024, Marshall University, Huntington, WV, USA
Weather Forecasting Using Artificial Neural Network (ANN): A Review
Weather Forecasting Using Artificial Neural Network (ANN): A Review
OUSSAMA ZEMNAZI*, SANAA EL FILALI, SARA OUAHABI
OUSSAMA ZEMNAZI*, SANAA EL FILALI, SARA OUAHABI
Information Technologies and Modeling Laboratory, University Hassan II - Casablanca, Faculty of Sciences Ben M'Sick, Casablanca, Morocco
Information Technologies and Modeling Laboratory, University Hassan II - Casablanca, Faculty of Sciences Ben M'Sick, Casablanca, Morocco

Abstract
Abstract
Extreme weather occurrences provide issues that necessitate the development of technology capable of accurate analysis and exact prediction
Extreme
in order toweather occurrences
successfully reduceprovide issues Artificial
their effects. that necessitate
neuralthe development
networks (ANNs) of have
technology capable
appeared in theoflast
accurate analysis
few years as aand exact prediction
promising weather
in order to successfully
forecasting reduce their
technology because effects.
of their Artificial
capacity neural intricate
to manage networksand(ANNs) haveweather
nonlinear appeared in the The
factors. last few years
use of as ain
ANNs promising
weather weather
element
forecasting
prediction has technology because of improvements
shown considerable their capacity in to forecasting
manage intricate andand
precision nonlinear weather
accuracy. Factorsfactors. The use
influencing of ANNs
an ANN in weather
model's efficacyelement
include
prediction
input Data hasType shown considerable
and Volume improvements
for Training, HiddeninLayer
forecasting
Neuron precision and accuracy.
Count, network FactorsActivation
architecture, influencingFunctions,
an ANN model's efficacy
and training include
algorithms.
input
In thisData
paper,Typewe and
willVolume
present afor Training,
thorough Hidden
review Layer
of the Neuron Count,
applications network
of Artificial architecture,
Neural NetworksActivation
(ANNs) Functions,
in weather and training specifically
forecasting, algorithms.
In this paper,
focusing we will present
on temperature a thorough
and rainfall reviewover
prediction of the
theapplications of Artificial
past 15 years. To enhance Neural Networks
reader (ANNs) The
comprehension, in weather forecasting,
work of numerous specifically
researchers
focusing on is
in this topic temperature andanalyzed
methodically rainfall prediction
and compared over in
theinpast 15 years.
tabular ToThe
format. enhance
authorsreader
aim comprehension,
to facilitate futureThe work of
research numerous
decisions by researchers
offering an
in this topicreview,
organized is methodically
aiding in analyzed and compared
the determination in in tabular
of suitable format. The
input features, authors
transfer aim to and
functions, facilitate future
training researchfor
algorithms decisions
accuratebytemperature
offering an
organized
and rainfallreview, aiding in the determination of suitable input features, transfer functions, and training algorithms for accurate temperature
predictions.
and rainfall predictions.
© 2024 The Authors. Published by Elsevier B.V.
© 2024
© 2024 The
TheAuthors.
[Link]
Publishedby byElsevier
ElsevierB.V. B.V.
This
This isis an
an open
open access
accessarticle
articleunder
underthe theCCCCBY-NC-ND
BY-NC-NDlicense license([Link]
([Link]
This is an
Peer-review open access article under the CC BY-NC-ND license ([Link]
Peer-review under responsibility of the scientific committeeChairs.
under responsibility of Conference Program of the Conference Program Chair
Peer-review under responsibility of the Conference Program Chairs.
Keywords: Artificial neural networks (ANNs); Extreme weather; temperature and rainfall predictions.
Keywords: Artificial neural networks (ANNs); Extreme weather; temperature and rainfall predictions.

1. Introduction
1. Introduction
Weather significantly influences human life, shaping our daily activities and decisions. The safety of individuals
andWeather
propertysignificantly
from threatsinfluences humanfloods,
like tornadoes, life, shaping our daily
and major storms activities and decisions.
relies heavily The safety
on accurate weatherof forecasting.
individuals
and property from threats like tornadoes, floods, and major storms relies heavily on accurate
Diverse sectors such as Business, Tourism, Sports, Agriculture, Mining, Power, Food Industry, Airport, and Naval weather forecasting.
Diverse
systems sectors such as Business,
are intricately linked to Tourism, Sports, Agriculture,
precise weather [Link], Power, in
For instance, Food Industry, Airport,
agriculture, advancedandweather
Naval
systems are intricately linked to precise weather predictions. For
information empowers farmers to make informed decisions for enhancing crop yields [1].instance, in agriculture, advanced weather
information
For achievingempowers farmers
predictions thereto are
make a informed
number ofdecisions
methods, forranging
enhancingfromcrop yields
naive [1].
methods to those that use more
For
complex techniques such as artificial intelligence (AI), artificial neural networks (ANNs) tobeing
achieving predictions there are a number of methods, ranging from naive methods thoseone
that
of use
the more
most
complex techniques such as artificial intelligence (AI), artificial neural networks (ANNs)
valuable and attractive methods for forecasting tasks. In prediction, ANNs, as opposed to traditional methods being one of the most
in
valuable and attractive
meteorology, are based methods for forecasting
on self-adaptive [Link]
In learn
prediction, ANNs, asand
from examples opposed
capturetofunctional
traditionalrelationships
methods in
meteorology,
between data, are
evenbased
if theon self-adaptive
relationships aremechanisms
unknown or that learntofrom
difficult examples
describe [2]. and capture functional relationships
between data, even if the relationships are unknown or difficult to describe [2].

* Corresponding author. Tel.: +212 649587130.

* Corresponding
E-mail address:author. Tel.: +212 649587130.
[Link]-etu@[Link]
E-mail address: [Link]-etu@[Link]
1877-0509 © 2024 The Authors. Published by Elsevier B.V.
1877-0509
1877-0509
This ©©2024
is an open 2024 The
The
access Authors.
Authors.
article Published
Published
under bybyElsevier
ElsevierB.V.
the CC BY-NC-ND B.V. ([Link]
license
Thisisisananopen
This
Peer-review open
underaccess
access article
article under
under
responsibility of the theBY-NC-ND
the CC CC BY-NC-ND
Conference license
Program license ([Link]
([Link]
Chairs.
Peer-review
Peer-review under
under responsibility
responsibility of the scientific
of the Conference committee of the Conference Program Chair
Program Chairs.
10.1016/[Link].2024.08.090
 OUSSAMA ZEMNAZI et al. / Procedia Computer Science 241 (2024) 618–623 619
2 OUSSAMA ZEMNAZI / Procedia Computer Science 00 (2024)
000–000
The use of Artificial Neural Networks (ANNs) in weather prediction is gaining prominence because of their
capacity to manage intricate and nonlinear weather factors. ANNs have proven large gains in predicting accuracy
when employed to anticipate diverse weather aspects. The performance of an ANN model hinges on several
parameters, including input Data Type and Volume for Training, Hidden Layer Neuron Count, network architecture,
Activation Functions, and training algorithms.
This study offers a detailed analysis of using Artificial Neural Networks (ANNs) for weather prediction,
particularly focusing on temperature and rainfall prediction. The analysis includes a comparison of ANN
performance based on Root Mean Square Error (RMSE) for temperature and rainfall forecasting. The primary
objectives of this study are to address key questions such as: Which optimization algorithms are deemed most
effective for training neural networks? What is the optimal Hidden Layer Neuron Count for superior neural network
performance? Which optimization algorithms are deemed most effective for training neural networks? What
activation function is optimal for a neural network designed for weather forecasting?
The remaining sections of the paper are organized as follows: In Section 2, a review of research literature on the
application of ANN in weather prediction (temperature and rainfall) is presented. Section 3 brings out an analysis of
existing works. The goal of section 4 was to provide a thorough discussion and identification of the input features,
hidden layer neuron count, optimization algorithm, and activation function that are utilized to predict temperature
and precipitation with the least amount of error. Finally, we provide a summary and recommendations for future
study directions in Section 5.

2. Related Works

Rawat and al. (2019) [2] created ANN models to estimate daily rainfall during the monsoon season in the
Parbhani District of Maharashtra, India. The study used 30 years of data (1985-2014). It was discovered that the
models can predict rainfall with appropriate precision.
Nastos and al. (2014) [3] employed ANN to forecast the maximum daily precipitation for the upcoming year. The
data utilized in the analysis are annual maximum daily precipitation totals recorded at the National Observatory of
Athens (NOA) throughout the lengthy period 1891-2009. The results of the research revealed a very excellent
association (R2 = 0.482, IA = 0.817, RMSE = 16.4 mm, and MBE = +5.2 mm) between the anticipated and
observed maximum daily precipitation totals one year ahead.
Hung and al. (2009) [4], used an Artificial Neural Network model to predict rainfall for Bangkok. From 1991 to
2005, historical rainfall data was gathered from 104 locations in the BMA and TMD rain gauge networks. It was
discovered that the ANN model could handle the noisy and unstable data that are common in the case of
meteorological data and was effective at quick calculation.
Amr and al. (2011) [5] created and applied two models in Alexandria, Egypt, to forecast rainfall. These models
are the Multi-Regression MLR model and the Artificial Neural Network ANN model. The data set used in this study
spans the years 1957 through 2009 and consists of daily temperature and rainfall readings. The final results indicate
that, with equal RMSE values (yearly = 32.04, January = 2.83, and December = 10.04), the ANN model may be a
useful tool for local rain forecasting.
Khairudin and al. (2020) [6] used five alternative machine learning models: artificial neural networks (ANN),
support vector regression (SVR), decision trees (DT), random forest algorithms (RFA), and long short-term memory
(LSTM). For this study, the Kuala Krai rainfall station's average weekly rainfall data were used as predictor
variables. According to the model's evaluation results, the LSTM obtained the lowest values for both metrics
(RMSE=8.34 and MAE=4.83).
Sulaiman and al. (2018) [7] used an artificial neural network (ANN) to estimate the monthly precipitation.
Historical rainfall data from 116 rain gauge stations was gathered for this purpose during a 50-year period, from
1965 to 2015. The results show that the ANN model performed better than the ARIMA model, with a 0.01
difference in RMSE
Khalili and al. (2016) [8] employed an artificial neural network to forecast the monthly rainfall at the Iranian
Mashhad synoptic station. They used monthly rainfall data for this synoptic station from 1953 to 2003.
The developed models' effectiveness is confirmed by the performance statistical analysis, which reveals that the
monthly prediction model with the best tuning has a correlation coefficient (R) of 0.93, a root mean square error
(RMSE) of 0.99, and a mean absolute error (MAE) of 6.02 mm.
Nguyen and al. (2021) [9] created a model for the daily and monthly rainfall forecast using artificial intelligence
(AI) that is based on artificial neural networks (ANNs). The Austin KATT station's [Link]
provided the dataset. The period of collection for the daily and monthly rainfall is from February 21, 2013, to July
31, 2017. The outcomes demonstrate that the ANN model is a good algorithm for predicting rainfall on a daily or
monthly basis.
620 OUSSAMA
OUSSAMAZEMNAZI
ZEMNAZI et al. / Procedia
/ Procedia Computer
Computer ScienceScience 241 (2024) 618–623
00 (2024) 3
000–000

For the daily rainfall, the testing phase values are 0.8063, 0.2487, and 0.0932 for R, RMSE, and MAE,
respectively. The testing portion's performance metrics for monthly rainfall are 0.8012, 0.0731, and 0.0578 for R,
RMSE, and MAE, respectively.
Khan and al. (2021) [10] provided rainfall predictions for the semi-arid mountainous region of Abha, Saudi
Arabia. The data was taken from the Meteorological Subdivision of Abha in the Saudi Arabian province of Aseer
during the years 1978 and 2016. The study suggests that ANN and MA can predict rainfall in semi-arid mountainous
regions.
MuttalebAlhashimi and al. (2014) [11] constructed rainfall estimating models using ANN, ARIMA, and MLR to
forecast monthly rainfall for the Kirkuk region. The data set utilised in this study comprises monthly observations of
the Kirkuk station's mean temperature, relative humidity, wind speed, and rainfall from 1970 to 2008. The ANN
model with four inputs has a higher R2 (0.91) and a lower RMSE value (27.278) for the testing data set in all
analyses, confirming its suitability as a forecasting tool for monthly rainfall prediction.
Dombaycı and al. (2009) [12] employed an artificial neural network (ANN) model to forecast the daily mean
ambient temperatures in Denizli, southwest Turkey. Temperature readings collected over three years (2003–2005)
by The Turkish State Meteorological Service were utilized as training data while values from 2006 were used as
testing data to train the model. The R2 and RMSE values of this network were 0.9902 and 1.8524, respectively,
indicating that the ANN method is a dependable model for ambient temperature prediction.
Fahimi Nezhad and al. (2019) [13] employed neural network techniques to predict Tehran's highest temperature.
Meteorological data for this study were acquired from the Tehran Mehrabad station between 1951 and 2010. The
findings indicate that a model with three neurons in the input layer, nine neurons in the hidden layer, and a
hyperbolic tangent function in the hidden layer is the most appropriate structure to predict the maximum temperature
in Tehran. The model's correlation coefficient, mean absolute error, and root mean square error are, respectively,
0.104, 0.997, and 0.311.
Rasel and al. (2018) [14] exhibit the effectiveness of data mining and machine learning approaches, specifically
Support Vector Regression (SVR) and Artificial Neural Networks (ANN), for reliable weather prediction. The
Bangladesh Meteorological Department provided a 6-year historical weather record of the temperature and rainfall
in the Chittagong metropolitan, which was used for the experiment. The results of this study show that while ANN
can predict temperature more accurately than SVR with an acceptable error rate deviation, SVR can predict rainfall
better than ANN, producing 0.95% and 0.17% error rates in both types of datasets.
Anjali and al. (2019) [15] suggest a method for predicting temperature using three models: Multiple Linear
Regression (MLR), Artificial Neural Network (ANN), and Support Vector Machine (SVM). They conduct a
comparative analysis using weather data collected from Central Kerala between 2007 and 2015. Compared to ANN
and SVM, MLR is a more accurate model for temperature prediction, as seen by the error metrics and the CC.
Singh and al. (2019) [16] investigated three models for weather forecasting: the Support Vector Machine (SVM),
the Artificial Neural Network (ANN), and the Time Series-based Recurrent Neural Network (RNN). the data was
collected from different airport weather stations in India. This dataset contains data from the previous 12 years 2006-
2018. Time Series-based RNN is shown to be the most effective at predicting weather.
Madhiarasan and al. (2020) [17] provided an ANN-based model for temperature forecasting. The National
Oceanic and Atmospheric Administration (USA) provided the data, which served as the ANN input variables and
goal temperatures, between January 2016 and December 2018. The following performance indexes are achieved by
the proposed ANN model for temperature forecasting applications, which has six inputs, one hidden layer, six
hidden neurons (MRE = 7.2796e−04, R = 1, MAPE = 0.0728, MSE = 0.0059, MAE = 0.0164, RMSE = 0.077.
Finally, Azari and al. (2022) [18] examined the ability of various machine learning approaches (LR, kNN, SVM,
ANN, Random Forest, and AdB) to forecast air temperature in the context of climate change research. The data was
mined from the airport weather station between January 1st, 1980 and September 22nd, 2021. ANN ML approach
with RMSE = 1.1694 was shown to be the most effective method for temperature prediction when dewpoint, relative
humidity, wind speed, sea pressure, and precipitation were used as dependent variables.

3. Analysis

An analysis of research that used ANN to forecast temperature and rainfall is shown in Table 1. The primary
objective is to analyze papers published throughout the last 14 years (2009-2023). The table is split into seven
columns that contain the most pertinent details from the examined papers, including the following: Ref,
Country/Study area, Prediction, Input Data, Dataset, Forecasting Model, Configuration, and Evaluation Criteria.
RMSE was the primary metric utilized in the majority of the papers to compare the algorithm's performances. To
provide the reader a reference of the efficacy of each algorithm, we just put the RMSE value in the table. We
excluded other metrics that were used from the table in order to maintain coherence and prevent confusion.
 OUSSAMA ZEMNAZI et al. / Procedia Computer Science 241 (2024) 618–623 621
4 OUSSAMA ZEMNAZI / Procedia Computer Science 00 (2024)
000–000
Table 1: Analysis of research that used ANN to forecast temperature and rainfall.

Ref Country/ Prediction Input Data Dataset Forecasting Configuration Evaluation

Study area Model Criteria
[2] India, Daily rainfall Temperature, relative 1985–2014 9-8-8-1, Activation Best result:
Maharashtra for monsoon humidity, wind speed, Function = RMSE = 7.26
season vapor pressure, Rainfall Hyperbolic tangent
function
[3] Greece, Maximum daily Annual daily maximum 1891–2009 – RMSE= 16.4
Athens precipitation precipitation
[4] Thailand, Rainfall Rainfall intensity, 1991–2005 9-22-11-1, RMSE= 0.71
Bangkok Relative humidity, wet Activation Function
bulb temperature, = Hyperbolic
pressure, Cloudiness, tangent
Average rainfall
intensity, Rainfall
intensity at surrounding
stations 1, 2, and 3
[5] Egypt, Rainfall on a Daily measurements of 1957–2009 2-10-7-6-1, 2-8-5-4- Yearly: RMSE=
Alexandria yearly and rainfall, maximum and 1, 2-8-9-5-1, 32.04
monthly basis minimum temperature Activation Function January:
= tan sigmoid and RMSE= 2.83
ANN Log sigmoid December:
RMSE= 10.04
[6] Malaysia, Rainfall Average weekly rainfall 1970–2019 Two hidden layers, RMSE= 9.10
Kuala Krai values Activation Function
= ReLu and linear
function
[7] Malaysia, Heavy Monthly and past 1965–2015 12-10-10-1, Best result:
Pahang precipitation on monthly rainfall data Activation Function Model C
a monthly basis = sigmoid function RMSE= 0.06

[8] Iran, Monthly rainfall Average rainfall, 1953–2003 7-4-1, Activation Best result:
Mashhad’s previous year's rainfall Function = sigmoid M741
for the predicted month, function RMSE= 0.96
and the last five bi-
monthly rainfall
[9] United States, Daily and Temperature, humidity, 2013–2017 18-16-8-4-1, Daily rain
Texas monthly rainfall dew point, pressure, Algorithm = (RMSE=
wind speed, and conjugate gradient, 0.2487)
visibility Activation Function Monthly rain
= Hyperbolic (RMSE=0.073)
tangent function
Saudi Arabia, Rainfall Total annual rainfall 1978–2016 Algorithm = Best result:
[10] Abha Levenberg– RMSE = 1.339
Marquardt
Iraq, Kirkuk Rainfall Month rainfall, average 1970–2008 4-8-1, Activation RMSE = 27.278
[11] air temperature, wind Function = sigmoid
speed, and relative function
humidity

Turkey, Daily mean Month of the year, day 2003–2005 3-6-1, Algorithm = Best result:
[12] Denizli. ambient of the month, average Levenberg– RMSE= 1.9655
temperatures temperature of the Marquardt,
previous day. Activation Function
= Hyperbolic
tangent sigmoid
function and purelin
function

Iran, Tehran Maximum (Tmax), (Rtempt), (n), 1951–2010 3-9-1, Activation Best result:
[13] temperature (U2), (RH mean), (T Function = Model 4
mean) hyperbolic tangent RMSE= 0.104

Bangladesh, (Temperature, Rainfall and temperature 6-year Maximum 3 hidden Best result:
[14] Chittagong rainfall) layer networks, rainfall
Activation Function RMSE= 18.43,
= Leaky ReLU temperature
RMSE= 3.31
622 OUSSAMA ZEMNAZI et al. / Procedia Computer Science 241 (2024) 618–623 5
OUSSAMA ZEMNAZI / Procedia Computer Science 00 (2024)
000–000

India, Kerala Temperature Humidity, pressure, wind 2007–2015 5-2-2-2-1 RMSE= 2.0100
[15] speed, wind direction, ANN
temperature
India Temperature Temperature, pressure 2006 -2018 RMSE= 3.1
[16] relative humidity, mean
wind direction, cloud
cover, horizontal –
visibility, and dew point
temperature.
United States Temperature Temperature, wind 2016 -2018 6-6-1, Activation Best result:
[17] speed, solar irradiance, Function = RMSE = 0.0770
relative humidity, dew Hyperbolic tangent
point, precipitation sigmoid function in
the hidden and
output layers.
[18] United States, Temperature Dewpoint, wind speed, 1980 -2021 Algorithm = RMSE = 1.1694
Memphis relative humidity, sea Levenberg–
pressure, and Marquardt
precipitation

4. Discussion
Extensive research has been conducted in the field of weather prediction, particularly in rainfall and temperature
prediction. with significant improvements in the performance of the models. The table above displays the best-
reported error values, specifically the RMSE on several models. Nevertheless, the outcomes are presented in graphs
for analytical purposes.

Fig 1: Comparison of ANN performance based on RMSE for temperature forecasting. Fig 2: Comparison of ANN performance based on RMSE for rainfall forecasting.

Fig. 1 depicts the error metrics in terms of RMSE for several temperature forecasting models discussed in the
literature. The models provided by Fahimi Nezhad et al. (2019) [13] and Madhiarasan, M., Tipaldi, M., & Siano, P.
(2020) [17] had lower RMSE values, indicating higher temperature prediction accuracy. Fig. 2 illustrates the
prediction accuracy of rainfall forecasting models based on RMSE. It can be seen that the models suggested by
Hung, Nguyen Q., et al. (2009) [4], Sulaiman, J., & Wahab, S. H. (2017) [7], Khalili, Najmeh, et al. (2016) [8],
Khan, Roohul Abad, et al. (2021) [10] revealed the lowest error values compared to other models, demonstrating
their superior accuracy in forecasting rainfall.
Based on the evaluation results, a comparison may be drawn between the works given in the literature. Indeed,
several factors influence Artificial Neural Network (ANN) performance. Key among them are input features, data
format, dataset size, optimization algorithms, activation functions, and configuration parameters such as the number
of hidden neurons and layers. Since there is no method for selecting the optimal number of hidden neurons and
layers to avoid network underfitting and overfitting, they were primarily discovered through trial and error.
For temperature forecasting, it can be observed that temperature, Humidity, and wind speed values turn out to be
the common parameters in most research. In addition, the Levenberg-Marquardt optimization technique is frequently
used and outperforms others due to its high learning rate and low prediction errors. the most used combination of
activation functions reported is the Hyperbolic tangent sigmoid function.
In the context of predicting rainfall, Hyperbolic Tangent or the Sigmoid Function are the most implemented
functions. In addition, the Levenberg-Marquardt and the conjugate gradient are more suitable for building the
rainfall forecast model with its small RMSE values.

5. Conclusion
According to the literature review, applying the Artificial Neural Network (ANN) approach for weather
prediction involves four steps: selecting input variables, determining model architecture and training algorithms,
selecting activation functions, and testing the model's accuracy. Additionally, the review demonstrates that the
efficacy of the ANN model is dependent on a variety of factors, including the type and quantity of input data
 OUSSAMA ZEMNAZI et al. / Procedia Computer Science 241 (2024) 618–623 623
6 OUSSAMA ZEMNAZI / Procedia Computer Science 00 (2024)
000–000
employed for training, the Hidden Layer Neuron Count, the overall network architecture, and the choice of training
algorithms. In particular, determining the number of neurons in the hidden layer is crucial, and several strategies are
used with the primary goal of minimizing prediction errors. It is emphasized that utilizing too many neurons in the
hidden layer can lead to overfitting.
In the majority of the reviewed studies, the Levenberg–Marquardt optimization algorithm has been employed for
prediction, owing to its effectiveness in predicting weather variables. Additionally, it has been demonstrated that the
Hyperbolic Tangent or the Sigmoid Function are more commonly used in temperature and rainfall prediction
because they provide better training performance for ANN. With these backgrounds, it shows that, ANN is a
powerful tool in predicting temperature and rainfall of a specific place, provided input parameters of the model are
well chosen. Moreover, in the future, we plan to employ Artificial Neural Networks (ANNs) for the prediction of
rainfall and temperature, with a focus on using the best performing training algorithm and the optimal activation
function.

References
[1] Jaseena, K. U., & Kovoor, B. C. (2022). Deterministic weather forecasting models based on intelligent predictors: A survey. Journal of
King Saud University-Computer and Information Sciences, 34(6), 3393-341.
[2] Rawat, A., Kumar, P., & Deoli, V. (2019). Daily Monsoon Rainfall Prediction using Artificial Neural Network (ANN) for Parbhani District
of Maharashtra, India. International Journal of Current Microbiology and Applied Sciences, 8(12), 1949-1963.
[3] Nastos, P. T., Paliatsos, A. G., Koukouletsos, K. V., Larissi, I. K., & Moustris, K. P. (2014). Artificial neural networks modeling for
forecasting the maximum daily total precipitation at Athens, Greece. Atmospheric Research, 144, 141-150.
[4] Hung, N. Q., Babel, M. S., Weesakul, S., & Tripathi, N. K. (2009). An artificial neural network model for rainfall forecasting in Bangkok,
Thailand. Hydrology and Earth System Sciences, 13(8), 1413-1425.
[5] Amr, H., El-Shafie, A., El Mazoghi, H., Shehata, A., & Taha, M. R. (2011). Artificial neural network technique for rainfall forecasting
applied to Alexandria, Egypt. Int. J. Phys. Sci, 6(6), 1306-1316.
[6] Khairudin, N. B. M., Mustapha, N. B., Aris, T. N. B. M., & Zolkepli, M. B. (2020, September). Comparison of machine learning models
for rainfall forecasting. In 2020 international conference on computer science and its application in agriculture (ICOSICA) (pp. 1-5). IEEE.
[7] Sulaiman, J., & Wahab, S. H. (2018). Heavy rainfall forecasting model using artificial neural network for flood prone area. In IT
Convergence and Security 2017: Volume 1 (pp. 68-76). Springer Singapore.
[8] Khalili, N., Khodashenas, S. R., Davary, K., Baygi, M. M., & Karimaldini, F. (2016). Prediction of rainfall using artificial neural networks
for synoptic station of Mashhad: a case study. Arabian Journal of Geosciences, 9, 1-9.
[9] Nguyen, H. N., Nguyen, T. A., Ly, H. B., Tran, V. Q., Nguyen, L. K., Nguyen, M. V., & Ngo, C. T. (2021). Prediction of daily and
monthly rainfall using a backpropagation neural network. Journal of Applied Science and Engineering, 24(3), 367-379.
[10] Khan, R. A., El Morabet, R., Mallick, J., Azam, M., Vambol, V., Vambol, S., & Sydorenko, V. (2021). Rainfall Prediction using Artificial
Neural Network in Semi-Arid mountainous region, Saudi Arabia. Ecological Questions, 32(4), 127-133.
[11] MuttalebAlhashimi, S. A. (2014). Prediction of monthly rainfall in Kirkuk using artificial neural network and time series models. Journal
of engineering and development, 18(1), 1-12.
[12] Dombaycı, Ö. A., & Gölcü, M. (2009). Daily means ambient temperature prediction using artificial neural network method: A case study
of Turkey. Renewable Energy, 34(4), 1158-1161.
[13] Fahimi Nezhad, E., Fallah Ghalhari, G., & Bayatani, F. (2019). Forecasting maximum seasonal temperature using artificial neural
networks “Tehran case study”. Asia-Pacific Journal of Atmospheric Sciences, 55, 145-153.
[14] Rasel, R. I., Sultana, N., & Meesad, P. (2018). An application of data mining and machine learning for weather forecasting. In Recent
Advances in Information and Communication Technology 2017: Proceedings of the 13th International Conference on Computing and
Information Technology (IC2IT) (pp. 169-178). Springer International Publishing.
[15] Anjali, T., Chandini, K., Anoop, K., & Lajish, V. L. (2019, July). Temperature prediction using machine learning approaches. In 2019 2nd
International Conference on intelligent computing, instrumentation and control technologies (ICICICT) (Vol. 1, pp. 1264-1268). IEEE.
[16] Singh, S., Kaushik, M., Gupta, A., & Malviya, A. K. (2019, March). Weather forecasting using machine learning techniques. In
Proceedings of 2nd international conference on advanced computing and software engineering (ICACSE).
[17] Madhiarasan, M., Tipaldi, M., & Siano, P. (2020). Analysis of artificial neural network performance based on influencing factors for
temperature forecasting applications. Journal of High-Speed Networks, 26(3), 209-223.
[18] Azari, B., Hassan, K., Pierce, J., & Ebrahimi, S. (2022). Evaluation of machine learning methods application in temperature prediction.
Environ Eng, 8(1),