See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/353333167

This work is licensed under a Creative Commons Attribution 4.0 International

License A Review on Weather Forecasting using Machine Learning and Deep
Learning Techniques
Research · July 2021
DOI: 10.17148/IARJSET.2021.8537

CITATIONS READS
0 96

1 author:

Jagruti Raut
Viva College
2 PUBLICATIONS 0 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jagruti Raut on 19 July 2021.

The user has requested enhancement of the downloaded file.

 IARJSET ISSN (Online) 2393-8021
ISSN (Print) 2394-1588

International Advanced Research Journal in Science, Engineering and Technology

Vol. 8, Issue 5, May 2021

DOI: 10.17148/IARJSET.2021.8537

A Review on Weather Forecasting using Machine

Learning and Deep Learning Techniques
Jagruti Raut
Assistant Professor, Computer Science, Viva College of Arts, Commerce and Science, Virar, Maharashtra

Abstract: Weather plays a crucial role in everyone’s life. Accurate weather information is necessary to plan our day-to-
day activities. Appropriate weather forecasting is important as it is to protect ourselves as well as our property. The
primary aim of this study is to review different Machine Learning and Deep Learning techniques used in weather
forecasting. Machine Learning is the automated process in which models learn by themselves, identify patterns and
make decision. This review indicates that Machine Learning and Deep Learning techniques can help us to forecast
weather based on input features such as temperature, humidity, Rainfall, Air Pressure and so on.

Keywords: Machine Learning, Deep Learning, Weather Forecasting, ANN, MLP

I. INTRODUCTION

The weather is the reflection state of the atmosphere around us.[1] The weather is a part of natural phenomenon that
maintains equilibrium in the atmosphere. The temperature, humidity, wind speed, Rainfall, Evaporation, Air Pressure,
Vapour Pressure, Sunshine duration, Sea Level, Visibility etc. are some of the elements of the weather.

Weather forecasting is an important research problem due to its effect in our day-to-day life. It is an essential approach
to avoid harmful climatic conditions. Weather prediction plays a significant role in many components in decision
making related to many fields such as agriculture, business, tourism, energy management, human and animal health
etc.[2]

Machine Learning and Deep Learning, the branches of Artificial Intelligence focusing on learning and prediction
provides a practical approach of prediction based on several features. A feature is nothing but the describing properties
of an individual thing. The performance of ML model is measured by performance metric. These models provide sharp
accuracy on deriving features using meteorological dataset.

This objective of this paper is to provide an overview of the different Machine Learning and Deep Learning techniques
that have been used in weather forecasting.

II. LITERATURE REVIEW

[1] had predicted Rainfall using weather parameters like Low Temperature, High Temperature, Humidity and Wind
Speed using Hybrid Machine Learning techniques such as MLP (Multi-Layer Perceptron) based PSO (Particle Swarm
Optimization) and MLP (Multi-Layer Perceptron) based LM(Levenberg-Marquardt) techniques. The MLP based PSO
shown more accuracy with RMSE=0.14 than MLP based LM.

Rainfall and Temperature Prediction was performed by [2] using Rainfall and Temperature dataset. It used ML
Techniques such as Support Vector Regression (SVR) and Artificial Neural Networks (ANN). The results showed that
SVR outperformed the ANN in rainfall prediction.

[3] had predicted Next day weather based on Maximum Temperature, Minimum Temperature, Evaporation, Humidity
and Wind Speed as weather parameters using Linear Regression and Deep Neural Network Regressor. It was found that
the DNN Regressor showed more accuracy than LR.

[4] had predicted Rainfall based on several meteorological parameters using Machine Learning and Deep Learning
techniques such as ARIMA Model, Artificial Neural Network, Support Vector Machine, Multilayer Perceptron (MLP)
Model and Auto-Encoders. It was found that the proposed methodology performed well.

[5] had predicted Temperature, Humidity and Pressure in the Next 24 Hours based on weather parameters such as
Temperature, Humidity and Air Pressure using Long Short-Term Memory (LSTM) Model and Multilayer Perceptron

Copyright to IARJSET IARJSET 226

This work is licensed under a Creative Commons Attribution 4.0 International License
 IARJSET ISSN (Online) 2393-8021
ISSN (Print) 2394-1588

International Advanced Research Journal in Science, Engineering and Technology

Vol. 8, Issue 5, May 2021

DOI: 10.17148/IARJSET.2021.8537

(MLP) Model. It was found that the proposed models were good at prediction, MAE of LSTM = 1.056 and MAE of
MLP = 0.7731.

Rainfall Prediction was done by [6] using Rainfall measurement attributes including individual months, annual and
combination of 3 consecutive months for 36 sub divisions as the input parameters. The Machine Learning techniques
such as Multiple Linear Regression, Support Vector Regression, Lasso Regression was applied on the dataset. Principal
Component Analysis technique was applied for feature reduction. It was found that Support Vector Regression
outperformed the Multiple Linear Regression and Lasso.

[7] had predicted Rainfall based on weather parameters such as Rain, Relative Humidity, Vapour Pressure, Sunshine
Duration, Cloud Amount and Visibility. Several Machine Learning Techniques were used in this domain. KMeans
clustering and Hierarchical Clustering were used for weather patterns clusters finding. Linear Regression was used to
predict rain based on sky visibility. Multiple Linear Regression was used to predict rain based on cloud amount,
visibility in sky, relative humidity and sun shine duration. Multivariate Multiple Linear Regression was used to predict
rain and cloud visibility based on atmospheric temperature, cloud amount in sky and relative humidity. Among all this,
Multivariate Multiple Linear Regression was performed well in prediction of rain. LR and MLR was also close enough.

[8] had predicted Rainfall based on hourly meteorological data such as Pressure, Temperature, Humidity, Wind Speed,
Wind direction, Sea Level and Rainfall using Deep Learning Techniques Echo state network (ESN) and Deep Echo
state network (DeepESN). The accuracy of predicted rainfall by using the DeepESN was improved compared with
those by using ESN, the BPN and the SVR.

Prediction of Rainfall was done by [9] with the help of weather parameters such as Temperature, Humidity and
Pressure. Random Forest Classification algorithm was used and 87.90% of accuracy was achieved by this technique.

[10] had predicted Maximum and Minimum Temperatures of the Next Day and Mean Temperature of the Next Day
based on weather parameters such as Pressure, Humidity, Rainfall, Temperature, Dust Particles and Light by using
Multiple Linear Regression Model. The 94% accuracy achieved for next day minimum temperature. The 93% accuracy
achieved for next day maximum temperature. The 95% accuracy achieved for next day Mean temperature.

[11] had predicted Rainfall, Humidity, Wind Speed, High Temperature and Low Temperature using weather parameters
such as Wind Speed, Humidity, Temperature and Rainfall. It used several ML Techniques like Support Vector
Regression (SVR), Linear Regression, Bayesian Ridge, Gradient Boosting (GB), Extreme Gradient Boosting
(XGBoost), Category Boosting (CatBoost), Adaptive Boosting (AdaBoost), k-Nearest Neighbors (KNN) and Decision
Tree Regressor (DTR). It was found that, ML-based models are more accurate than conventional methods. However, it
can be seen that DTR and CatBoost methods were almost equivalent but adaptability of DTR was more for nonlinear
data.

Weather Prediction was performed by [12] using ENSO Dataset and Weather Dataset. It used Deep Learning
Techniques like Recurrence Neural Network (RNN), Conditional Restricted Boltzmann Machine (CRBM), and
Convolutional Network (CN) models. The results showed that Recurrent NN using heuristically optimization method
for rainfall prediction based on weather dataset comprises of ENSO variables.

The below table tabulates the various Machine learning and Deep Learning techniques used for weather prediction with
different set of weather parameters.

Copyright to IARJSET IARJSET 227

This work is licensed under a Creative Commons Attribution 4.0 International License
 IARJSET ISSN (Online) 2393-8021
ISSN (Print) 2394-1588

International Advanced Research Journal in Science, Engineering and Technology

Vol. 8, Issue 5, May 2021

DOI: 10.17148/IARJSET.2021.8537

TABLE I

Reference Parameters Technique Result Prediction

s
H.Abdel- Low Temperature, MLP (Multi-Layer MLP based PSO shown more Rainfall
Kader High Temperature, Perceptron) based PSO accuracy (RMSE=0.14) than Prediction
et al.[1] Humidity, And (Particle Swarm MLP based LM.
Wind Speed Optimization),
MLP (Multi-Layer
Perceptron) based
LM(Levenberg-Marquardt)
R.I.Rasel Rainfall dataset Support Vector Regression SVR outperformed the ANN Rainfall and
et al. Temperature (SVR) and Artificial in rainfall prediction Temperature
[2] dataset Neural Networks (ANN) Prediction

B.S. Maximum Linear Regression, Deep The DNN Regressor shown Next day weather
Panda et Temperature, Neural Network Regressor more accuracy than LR.
al. Minimum
[3] Temperature,
Evaporation,
Humidity and
Wind Speed

C.Z. Meteorological ARIMA Model, Artificial The proposed methodology Rainfall

Basha et Data Neural Network, Support had outperformed. Prediction
al. Vector Machine,
[4] Multilayer Perceptron
(MLP) Model, Auto-
Encoders
Z.Q. Temperature, Long Short-Term Proposed models were good Temperature,
Huang Humidity, And Air Memory (LSTM) Model at prediction. humidity,
et al. Pressure and The Multilayer MAE (LSTM) = 1.056 and Pressure in the
[5] Perceptron (MLP) Model MAE (MLP) = 0.7731 Next 24 Hours

M. Rainfall Multiple Linear Regression SVR outperformed the MLR Rainfall

Mohamm measurement Support Vector Regression and Lasso. Prediction
ed attributes including Lasso Regression
et al. individual months,
[6] annual and Principal Component
combination of 3 Analysis- for feature
consecutive reduction
months for 36 sub
divisions.

Copyright to IARJSET IARJSET 228

This work is licensed under a Creative Commons Attribution 4.0 International License
 IARJSET ISSN (Online) 2393-8021
ISSN (Print) 2394-1588

International Advanced Research Journal in Science, Engineering and Technology

Vol. 8, Issue 5, May 2021

DOI: 10.17148/IARJSET.2021.8537

A.M. Rain, Relative Kmeans clustering and Multivariate Multiple Linear Rainfall
Suresha[7 Humidity, Vapour Hierarchical Clustering - Regression was performed Prediction
] Pressure, Sunshine weather patterns clusters well in prediction of rain.
Duration, Cloud finding,
Amount, Visibility LR and MLR was also close
Linear Regression – to enough.
predict rain based on sky
visibility

Multiple Linear Regression

– to predict rain based on
cloud amount, visibility in
sky, relative humidity and
sun shine duration.

Multivariate Multiple
Linear Regression – to
predict rain and cloud
visibility based on
atmospheric temperature,
cloud amount in sky and
relative humidity.

M.H.Yen Hourly Echo state network (ESN) The accuracy of predicted Rainfall
et al. meteorological and Deep Echo state rainfall by using the Prediction
[8] data (Pressure, network (DeepESN) DeepESN was
Temperature, improved compared with
Humidity, Wind those by using ESN, the BPN
Speed, Wind and the SVR
direction, Sea
Level, Rainfall)
N. Singh Temperature, Random Forest Accuracy of 87.90% was Rainfall
et al. Humidity and Classification achieved. Prediction
[9] Pressure

A. Pressure, Multiple Linear Regression 94% accuracy achieved for Maximum and
Parashar Humidity, Rainfall, Model next day minimum Minimum
[10] Temperature, Dust temperature. Temperatures of
Particles And the Next Day
Light 93% accuracy achieved for Mean
next day maximum Temperature of
temperature. the Next Day

95% accuracy achieved for

next day Mean temperature.

A. Wind Speed, Support Vector ML-based Rainfall

Mahabub Humidity, Regression (SVR), Linear models are more accurate Prediction
[11] Temperature And Regression, Bayesian than conventional methods. Humidity
Rainfall Ridge, Gradient Boosting Prediction
(GB), Extreme Gradient It can be said that DTR and Wind Speed
Boosting (XGBoost), CatBoost methods were Prediction
Category almost equivalent but High
Boosting (CatBoost), adaptability of DTR was Temperature
Adaptive oosting more for nonlinear data. prediction
(AdaBoost), k-Nearest Low Temperature
Neighbors (KNN) and prediction
Decision Tree Regressor

Copyright to IARJSET IARJSET 229

This work is licensed under a Creative Commons Attribution 4.0 International License
 IARJSET ISSN (Online) 2393-8021
ISSN (Print) 2394-1588

International Advanced Research Journal in Science, Engineering and Technology

Vol. 8, Issue 5, May 2021

DOI: 10.17148/IARJSET.2021.8537

(DTR)

A. G. Wind, Oscillation Recurrence Neural Recurrent NN using Weather

Salman Index, Sea Surface Network (RNN), heuristically optimization Prediction
et al. Temperature and Conditional Restricted method for
[12] Outgoing Long Boltzmann Machine rainfall prediction based on
Wave Radiation (CRBM), and weather dataset comprises of
(ENSO Dataset) Convolutional Network ENSO variables
(CN) models.
Mean
Temperature, Max
Temperature,
Minimum
Temperature,
Precipitation
Temperature,
Relative
Humidity, Mean
Sea Level
Pressure, Mean
Station
Pressure,
Visibility, Average
Win, Maximum
Wind,
Wind and Rainfall
(Weather Dataset)
III.CONCLUSION
Weather Forecasting is a challenging task but very important research problem. Because it is related to our day-to-day
life. Machine Learning and Deep Learning techniques can help us to forecast weather based on several input features.
Also IoT techniques can significantly combine with Machine Learning and Deep Learning to produce better result.
Accuracy in prediction is strongly dependent on the time period and location of weather station. The results showed
that hybrid ML techniques and Deep Learning strategies can achieve better accuracy.

REFERENCES
[1] H. Abdel-Kader, M. A.-E. Salam, and ..., “Hybrid Machine Learning Model for Rainfall Forecasting,” J. Intell. …, vol. 1, no. 1, pp. 5–12,
2021, doi: 10.5281/zenodo.3376685.
[2] R. I. Rasel, N. Sultana, and P. Meesad, “An application of data mining and machine learning for weather forecasting,” Adv. Intell. Syst.
Comput., vol. 566, pp. 169–178, 2018, doi: 10.1007/978-3-319-60663-7_16.
[3] B. S. Panda, P. V Lasyasri, D. Maneesha, P. Goutham, K. Suresh, and C. Pranavsankar, “A N o v e l A p p r o a c h f o r W e a t h e r F o r e
c a s t i n g u s i n g M a c h i n e L e a r n i n g T e c h n i q u e s,” pp. 25–30.
[4] C. Z. Basha, N. Bhavana, P. Bhavya, and V. Sowmya, “Rainfall Prediction using Machine Learning Deep Learning Techniques,” Proc. Int.
Conf. Electron. Sustain. Commun. Syst. ICESC 2020, no. Icesc, pp. 92–97, 2020, doi: 10.1109/ICESC48915.2020.9155896.
[5] Z. Q. Huang, Y. C. Chen, and C. Y. Wen, “Real-time weather monitoring and prediction using city buses and machine learning,” Sensors
(Switzerland), vol. 20, no. 18, pp. 1–21, 2020, doi: 10.3390/s20185173.
[6] M. Mohammed, R. Kolapalli, N. Golla, and S. S. Maturi, “Prediction of rainfall using machine learning techniques,” Int. J. Sci. Technol.
Res., vol. 9, no. 1, pp. 3236–3240, 2020, doi: 10.36227/techrxiv.14398304.
[7] A. M. Suresha, “Machine Learning for mining weather patterns and weather forecasting,” no. October, 2020, doi:
10.13140/RG.2.2.22296.42243.
[8] M. H. Yen, D. W. Liu, Y. C. Hsin, C. E. Lin, and C. C. Chen, “Application of the deep learning for the prediction of rainfall in Southern
Taiwan,” Sci. Rep., vol. 9, no. 1, pp. 1–9, 2019, doi: 10.1038/s41598-019-49242-6.
[9] N. Singh, S. Chaturvedi, and S. Akhter, “Weather Forecasting Using Machine Learning Algorithm,” 2019 Int. Conf. Signal Process.
Commun. ICSC 2019, pp. 171–174, 2019, doi: 10.1109/ICSC45622.2019.8938211.
[10] A. Parashar, “IoT based automated weather report generation and prediction using machine learning,” 2019 2nd Int. Conf. Intell. Commun.
Comput. Tech. ICCT 2019, pp. 339–344, 2019, doi: 10.1109/ICCT46177.2019.8968782.
[11] A. Mahabub and A. S. Bin Habib, “An Overview of Weather Forecasting for Bangladesh Using Machine Learning Techniques,” pp. 1–36,
2019.
[12] A. G. Salman, B. Kanigoro, and Y. Heryadi, “Weather forecasting using deep learning techniques,” ICACSIS 2015 - 2015 Int. Conf. Adv.
Comput. Sci. Inf. Syst. Proc., pp. 281–285, 2016, doi: 10.1109/ICACSIS.2015.7415154.

Copyright to IARJSET IARJSET 230

This work is licensed under a Creative Commons Attribution 4.0 International License
View publication stats