TIMBOU-AFRICA

PUBLICATION INTERNATIONAL JOURNAL OF SCIENCE

INTERNATIONAL
JOURNAL NOVEMBER, 2024 RESEARCH AND TECHNOLOGY
EDITIONS. INTERNATIONAL JOURNAL OF:
NOVEMBER, 2024 SCIENCE RESEARCH AND
VOL. 6 NO. TECHNOLOGY
9 E-ISSN VOL. 63026-8095
3026-8796 P-ISSN
EDITIONS.

A
REVIEW OF MACHINE LEARNING
TECHNIQUES APPLICATIONS IN
ENVIRONMENTAL SCIENCE

EGWU L.S.1, ENAYABA, O.F.2, AJIBOYE, A.A.3

DAMOYE T4 OGUNDEJI, I. S5 AND AGBAMS P6.
ABSTRACT 1
Department of Animal and Environmental Biology, Prince
Advances in
Abubakar Audu University, Anyigba, Nigeria. 2Department of
Machine Learning Chemistry, Federal University of Petroleum Resources,
(ML) and Data Effurun, Nigeria. 3Department of Mathematics, Purdue
Science (DS) hold University. 4Department of Computer Science, Babcock
immense University, Nigeria. 5Department of Data Science
potential to Management, Rome Business School, Italy. 6Department of
transform various Agricultural Economics and Extension, Abia State University,
aspects of Uturu, Nigeria.
environmental Corresponding Author: [email protected]
science (ES). DS, a DOI: https://doi.org/10.70382/tijsrat.v06i9.002
broad field
focused on
extracting
insights from data
INTRODUCTION

R
apid advancements in machine learning (Haupt et al.,
using techniques
2022) and deep learning (LeCun et al., 2015) have
like statistical
sparked the scientific community to explore how these
analysis plays a
tools can drive scientific progress and unlocking breakthroughs
crucial role in this that were once considered unattainable. Over decades,
process. Machine environmental science only revolves round how physical and
learning, on the chemical properties alongside with natural resources gears
other hand, interactions of living organisms and the environment. In recent
specializes in time, environmental data are rapidly growing into huge
creating datasets, increasing complexity, resolution, and size. This
algorithms that growth creates interdisciplinary challenges for environmental
scientists, requiring innovative approaches, such as data

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
12
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

enable computers to learn from data and make predictions. Together, these
technologies can deepen our understanding of complex environmental
systems, refine predictive models for climate change, support conservation
efforts, and optimize resource management practices. Such scientific
discovery will enhance ES to make autonomous, real-time decisions by deriving
valuable insights from extensive data. By analysing large datasets, machine
learning algorithms can reveal hidden patterns and insights, empowering
scientists to make data-driven decisions and tackle environmental challenges
more effectively. This article offers a review of the fundamental concepts of
Machine Learning, Deep Learning, and Data Analytics for two groups:
individuals familiar with ML who seek to expand their knowledge, and domain
scientists passionate about integrating these transformative tools into their
research in the environmental science profession.

Keywords: Machine Learning, Environmental Science, Climate Change, Data Science

and Algorithm

processing and big data analysis, to address them effectively. Integrating diverse data
from multiple sources to perform comprehensive analysis and extract meaningful insights
demands a strong foundation in data science. The widespread adoption of data science
techniques has greatly enhanced environmental system management, enabled scenario
modelling and fostered data-driven innovation across industries. Environmental scientists
are increasingly challenged to solve complex interdisciplinary problems through
established and emerging data science methods ((Tharsanee et al., 2020). Data science
enriches environmental science by providing a practical and effective approach to tackle
real-world issues (Karina, et al., 2018).
To gain meaningful insights, data from environmental stressors—collected through
remote sensing satellites, air and water quality sensors, weather and climate
observations, and ground-based sensors that measure the magnitude of earthquakes and
other geological events—must be thoroughly and efficiently analysed (Tharsanee et al.,
2020). Recent advancements in machine learning have empowered scientists and
software engineers to address complex issues in climate variability and weather, fuelling
momentum for national and international workshops (Chantry et al., 2021). A key

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
13
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

advantage of machine learning algorithms is their ability to identify trends and patterns
in data autonomously. Their predictive accuracy improves as more data becomes
available. Machine learning can also handle multidimensional and diverse data, even in
dynamic or uncertain environments (Loussaief et al., 2016; Ahn et al. 2016 & Brunton et al.
2019).
Although ML and DS has record initial success in Environmental Science, several
challenges still persist. Foremost, many ES researchers are eager to adopt these
techniques but may lack the necessary expertise to apply it correctly, leading to potential
misuse of the technology. Additionally, as data volume and complexity have increased,
more advanced ML applications, such as deep neural networks, are being utilized to
capture complex nonlinear relationships. Lastly, Applicability domain analysis of ML
models is still not commonly practiced by researchers in Environmental and Science
Engineering after model development, except in the case of quantitative structure–
activity relationships (Gadaleta et al., 2016). However, these models are often considered
"black boxes," making model interpretability essential to ensure that the predictions align
with core domain scientific principles. Despite growing attention to model
interpretability, it is still often overlooked in ES researches (Kerckhoffs et al., 2019; Song
et al., 2017; Pak et al., 2020 & Xiao 2018).
According to Hsieh W. (2009), ML, which originates from Artificial Intelligence, has
become a cutting-edge approach in data mining with significant and future potential in
environmental science. These methods are used to process satellite data, predict climate
trends, forecast outcomes, and analyse environmental datasets. To derive meaningful
insights from this data, a modern and effective approach is required, incorporating
techniques such as linear statistical analysis, time series analysis, feedforward neural
networks, nonlinear optimization, generalization learning, classification models,
regression models, principal component analysis, and correlation analysis. Together,
these models provide an optimal framework for tackling a wide range of challenges in
environmental science. This review seeks to outline future directions in ML and ES that
we believe will significantly advance the field, along with brief explanations of various
machine learning techniques and deep learning algorithms for their application in
environmental data analysis.

Fig 1: Graphical illustration of Deep learning

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
14
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

Different neural Network framework (Tharsanee et al., 2020)

Fig 2: A Graphical Method for Machine Learning in Remote Sensing

A deep learning approach for analysing remote sensing data (Zhang et al., 2016)
Fig 3: Graphical Summary of ML Algorithms

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
15
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

Various ML Algorithms (Tharsanee et al., 2020)

Brief Overview of ML Challenges in ES

Machine learning offers significant potential in environmental science, particularly in data
analysis, climate prediction, ecosystem modelling, and resource management. However,
its application faces several challenges;
Data Quality and Availability: Environmental data can be sparse or incomplete, making it
difficult to train accurate ML models. In some cases, data may not be standardized or may
have gaps that require imputation or preprocessing. To effectively collect valid and high-
quality data is the foremost issues with the application of ML in the field of ES. Low-quality
data are not easily suitable for modelling. On the other hand, machine learning demands
a large dataset to create robust predictive models and ensure accurate predictions
(L’heureux et al., 2017). Researchers also contribute to the aforementioned challenges by
often conducting experiments under different conditions, such as variations in water
quality, soil or sediment properties, and catalyst or adsorbent types and loadings, even
when data are available. This leads to discrepancies in the collected data. To develop
predictive models that are truly robust and widely applicable, it is crucial to first build a
large, consistent dataset (Shifa et al., 2021)
Model Interpretability vs Overfitting: Many ML models, particularly deep learning
algorithms, operate as "black boxes," making it challenging to interpret how they arrive

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
16
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

at specific predictions. This lack of transparency can be problematic in environmental

science, where understanding the rationale behind a model's decision is crucial. Model
overfitting is a major issue in large language model development, as it affects
interpretability. Overfitting occurs when a model learns the noise in the training data
instead of the true patterns, leading to poor performance on unseen data and making the
model's decision-making hard to understand. It results in an overly complex model that is
difficult to interpret, due to increased complexity, difficulty with feature selection,
reduced generalizability, and challenges in model explanation. Detecting overfitting is
challenging, but methods such as feature selection can help reduce the risk of overfitting,
data augmentation, cross-validation, regularization, model simplification or choice,
dropout, and early stopping (Kuhn et al., 2019; Bühlmann et al., 2011; Srivastava, et al.,
2014; Yao et al., 2007)
Data Underfitting: Underfitting in ML occurs when a model is too simple to capture the
underlying patterns or relationships in the data. It typically happens when the model has
insufficient complexity, such as having too few features, overly simplistic assumptions, or
inadequate model architecture. As a result, the model fails to learn important trends in
the data, leading to poor performance both on the training set and unseen data.
Key causes of underfitting include:
• Insufficient model complexity: The model lacks the capacity to learn the data’s
patterns.
• Inadequate feature selection: Not enough relevant features are included in the
model.
• Excessive regularization: Too much regularization can prevent the model from
learning sufficiently complex patterns.

To address this issue, the following factors must be taken into account:
• Increase the training time
• Boost the model's complexity
• Add more relevant features to the data
• Reduce regularization parameters
• Extend the model's training duration

Complexity of the machine learning process: Machine learning is a complex process that
involves multiple stages, including data collection, preprocessing, model selection,
training, evaluation, and deployment. Each stage requires careful attention to detail and
can introduce challenges that affect the model's performance. The complexity arises from

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
17
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

the need to manage large volumes of data, select appropriate algorithms, fine-tune model
parameters, and ensure generalization to unseen data. Additionally, factors such as
overfitting, underfitting, model interpretability, and computational resources all
contribute to the intricacy of the machine learning process
Bias in machine learning Models: From a technical perspective, bias refers to systematic
errors introduced by the model or the data that lead to unfair, skewed, or inaccurate
predictions. Bias can arise at various stages of the machine learning process, from data
collection to model deployment, and can significantly impact the model's performance
and fairness. Data contamination is a more subtle type of data leakage and can be difficult
to identify without domain expertise. Data omission is a prevalent issue in scientific
communities, as peer-reviewed publications often highlight only the most promising
positive results, leaving out negative results and outliers that are essential for ML model
performance. Additionally, data may be missing due to choices made by scientists, such
as the use of specific reagents, reaction conditions, or sampling plans, or the failure to
collect data that contradicts established theories. These types of anthropogenic biases in
datasets can further degrade ML performance (Charidimou et al., 2019). Bias of
Algorithmic arises when the model’s structure or loss function does not align with the
intended use case. Identifying potential bias in an ML model early is critical for its
successful application, beyond just environmental concerns. Mitigating bias can be
achieved by enhancing the model’s interpretability, allowing for the integration of domain
knowledge to assess its validity. A practical strategy for detecting bias is to use an
ensemble of ML models, comparing their outputs on the same set of problems. This
comparison helps identify inconsistencies in performance and reveals any bias specific to
a particular model (Shifa et al., 2021).
Algorithmic Flaws as Data Expands: As data continues to grow, algorithms may become
outdated in the future. Current models, which are considered the best, may become
inaccurate and will require adjustments. Maintaining algorithms requires constant
monitoring and upkeep. According to the report by Shifa et al. (2021), three main concerns
were highlighted: (1) complete reliance on ML should be avoided, as traditional statistical
tools may be more suitable in certain cases, such as when sample sizes are small; (2) it is
crucial to investigate findings through experimentation or domain expertise, rather than
overestimating the capabilities of ML techniques (3) Always keep in mind that not every
ESE problem can be directly solved using ML tools. Transforming these problems into
ones that can be effectively addressed by ML requires skilful and thoughtful design.
Applications and Implementations of ML

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
18
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

Machine learning has been widely applied across multiple areas of environmental
monitoring and management. These are illustrated below;

ML application into Environmental Impact Assessments

Integrating Machine Learning (ML) with Environmental Impact Assessments (EIAs) raises
various ethical considerations that must be carefully addressed to ensure the fair and
responsible use of technology in environmental science. A significant ethical concern is
algorithmic bias, which can arise when ML models reinforce or amplify existing
environmental inequalities. To address this issue, it is essential to use interpretable
models like decision trees and to develop interpretability techniques that are
independent of the specific model type. Embedding ethical considerations into the core
design of machine learning applications lays the groundwork for responsible and
sustainable technology use that fosters environmental stewardship (Yilmaz et al., 2019 &
Obulesu et al., 2024)
ML in Agriculture: Machine learning (ML) has transformed agriculture by providing data-
driven insights and automation that improve productivity, efficiency, and sustainability.
Machine learning is transforming agriculture by providing tools for forecasting crop
yields, optimizing farming conditions using historical data and future trends, and enabling
accurate predictions and strategic planning (Sharma et al., 2020). These approaches not
only promote sustainable agricultural production but also improve resource efficiency,
such as water and fertilizer usage, supporting environmental conservation efforts (Benos
et al., 2021). For example, machine learning algorithms have been applied to refine
fertilizer usage, achieving a balance between enhancing crop yields and reducing
environmental harm (Peng, et al., 2023). Furthermore, machine learning plays a vital role
in assessing soil health by analysing key soil properties to guide crop rotation and soil
management practices, ensuring soil fertility and supporting healthy crop production
(Mohamed et al., 2023).
The effective implementation of these techniques has ushered in a new era of precision
agriculture, greatly enhancing crop management and boosting yields. Research efforts
such as (Maheswari et al., 2023 & Kuradusenge et al., 2023) emphasize the precision of
machine learning models in predicting crop yields, enabling farmers and stakeholders to
make well-informed decisions. The application of machine learning in disease
management, as illustrated by (Prem et al., 2018 & Sai et al., 2023), highlights the
effectiveness of convolutional neural networks (CNNs) in early detection of plant
diseases, helping to minimize losses and decrease reliance on chemical pesticides.
Advancements in soil analysis using machine learning, as highlighted in (Prem et al., 2018;

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
19
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

Sai et al., 2023 & Sirsat et al., 2018), have facilitated more accurate evaluations of soil
health, enhanced nutrient management and promoting soil conservation. The emergence
of smart agricultural tools, ranging from plant classification to soil erosion modelling,
underscores the transformative impact of these technologies (Sai et al., 2023 & Elavarasan
et al., 2018). These examples emphasize the broad use of these technologies in
agriculture, spanning from disease detection and yield prediction to automating
harvesting and orchard navigation. By utilizing advanced algorithms, the agricultural
sector can greatly boost productivity, sustainability, and its ability to adapt to climate
change challenges

Fig 4: An illustration demonstrating the use of machine learning techniques in agricultural

research, emphasizing the application of traditional ML and deep learning methods, while
acknowledging the absence of AutoML implementations. (Dania et al., 2024)

Table 1 (Dania et al., 2024). Provide an overview of the applications of ML and Deep
Learning techniques in agriculture
ML Technique Agricultural Application
Decision Tree Prediction of Crop Yields, Disease Identification, Soil
Analysis
Random Forest Prediction of Crop Yields, Disease Identification, Soil
Analysis

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
20
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

Extreme Gradient Boosting Crop Yield Estimation, Soil Health Assessment

Naive Bayes Crop Yield Estimation, Soil Health Assessment
K-Nearest Neighbors Crop Yield Prediction, Disease Diagnosis
Ensemble Traditional Crop Yield Prediction
ML Models
Multi-Linear Regressor Crop Yield Prediction
RNN Crop Yield Prediction
LSTM Crop Yield Prediction
Support Vector Regression Crop Yield Prediction
CNN Crop Yield Prediction, Disease Detection
GNN Crop Yield Prediction
U-Net Crop Yield Prediction
ANN Crop Yield Prediction, Disease Detection
DBSCAN Crop Yield Prediction
Support Vector Machine Crop Yield Prediction, Disease Detection, Smart
Farming
Vision Transformers Disease Detection
VGG-RNN Hybrid Soil Assessment
MLP Soil Assessment

Application of ML Techniques in Wildfire science and Management

ML, a subset of Artificial intelligence has various methods that are used for wildfire
science and management. These approaches are grouped into three: supervised learning,
unsupervised learning, or agent-based learning which are briefly elucidate below.
Supervised Learning: in this method, the model is trained on labelled data. This means the
input data is paired with corresponding correct output labels. The algorithm learns the
relationship between the input and output during training, with the goal of making
accurate predictions or classifications when new, unseen data is provided. Supervised
learning methods are typically used for tasks like regression (predicting continuous
values) and classification (predicting discrete categories). Examples of supervised
learning algorithms include linear regression, decision trees, and support vector machines
(SVM)
Unsupervised Learning: Unsupervised learning is a type of machine learning where the
model is trained on data without labelled responses. In this approach, the algorithm tries
to identify patterns and structures in the data on its own, such as grouping similar data
points (clustering) or reducing the dimensionality of the data (dimensionality

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
21
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

reduction). In unsupervised learning, the canonical tasks are dimensionality reduction

and clustering, with relationships or patterns being extracted from the data without any
guidance as to the “correct” answer.
Agent Based Learning: In this approach, an "agent" is a computational entity that makes
decisions based on its observations of the environment and learns from the
consequences of its actions. The agent's goal is typically to optimize its performance or
achieve specific objectives, often through trial and error. This method utilizes incomplete
information about the target variables. Major problem with this approach is
reinforcement in which critical parts of the environment can only be observed
interactively through trial and error (Sutton R. et al., 1998)

Fig 5: diagram depicting the primary types of machine learning, data types, and modeling
tasks, highlighting their associations with widely used algorithms and applications in wildfire
science and management. Algorithms in bold indicate core ML methods, whereas non-bolded
algorithms are generally not classified as core ML (Piyush et al., 2020).

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
22
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

ML application on Water Bodies: The rapid growth of artificial intelligence and the
increasing volume of data on aquatic environments have made machine learning a vital
tool for data analysis, classification, and prediction. Machine learning is a powerful tool
increasingly utilized by environmental science researchers to tackle challenges in water
treatment and management systems. Its applications span across water resource
allocation, pollutant source tracking, real-time monitoring, prediction, pollutant
concentration estimation, and the optimization of water treatment technologies
(Mengyuan et al., 2022).

Fig. 1. Machine learning is extensively applied in water systems. WWTP: wastewater

treatment plant (Mengyuan et al., 2022).

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
23
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

Fig: Machine learning algorithms applied across various water treatment and management
systems. support vector machine, random forest, artificial neural network; SOM: self-
organizing map, decision tree, principal component analysis; XGBoost: extreme gradient
boosting, dissolved oxygen, micropollutant (Mengyuan et al., 2022)

Fig Pictorial representation of four ML applications in Environmental Science and

Engineering (ESE) (Shifa et al., 2021)

Methodology
We adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-
Analyses) guidelines outlined by Liberati et al. (2009) to identify relevant articles and

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
24
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

conduct the review. Our search strategy involved using Google to access the "Web of
Science" and "ScienceDirect" databases, employing various combinations of relevant
keywords such as machine learning, environmental science and climate change.

Authors contribution
This study was conducted through collaboration among all the authors

Declaration of Competing Interests

The authors have declared no existing competing interests

Concluding remark
This paper has highlighted the efficacy of machine learning (ML) and its transformative
role for environmental professionals in environmental science, showcasing how it
continues to grow as a solution for addressing environmental issues. The importance of
ML in this field cannot be overemphasized due to its futuristic potential. Our findings
emphasize the widespread application of this critical technology, spanning traditional ML
techniques to advanced approaches. Additionally, the incorporation of Automated
Machine Learning (AutoML) offers significant, yet largely untapped, potential.
Despite its promise, the adoption of ML in environmental science faces challenges.
Addressing these limitations will require greater collaboration among data scientists,
environmental researchers, and policymakers to enhance model transparency and
usability. By embracing these technologies, the environmental science community can
contribute to more sustainable solutions for pressing global challenges.
Researchers eager to utilize these powerful tools should first master the fundamentals of
their application to avoid discrepancies in findings. Finally, while ML is invaluable, it should
not be solely relied upon. Traditional methods and experimental approaches must be
integrated alongside ML to ensure accuracy and reliability in environmental research.

Reference
Dania T., Xiuquan W., & Morteza M (2024). A Review of Machine Learning Techniques in Agroclimatic Studies. 14(3),
481; https://doi.org/10.3390/agriculture14030481
Mengyuan Zhu, Jiawei Wang, Xiao Yang, Yu Zhang, Linyu Zhang, Hongqiang Ren, Bing Wu, Lin Ye (2022). A review of the
application of machine learning in water quality evaluation. Eco-Environment & Health; 107–116
Charidimou, A.; Zonneveld, H. I.; Shams, S.; Kantarci, K.; Shoamanesh, A.; Hilal, S.; Yates, P. A.; Boulouis, G.; Na, H. K.; Pasi,
M.; et al. APOE and cortical superficial siderosis in CAA: Metaanalysis and potential mechanisms. Neurology 2019, 93
(4), e358− e371
Obulesu Varikunta, Mr. A. Sarveswara Reddy2, Mr. K. Sathish (2024). Incorporating Machine Learning into Environmental
Impact Assessments for Sustainable Development. https://www.researchgate.net/publication/380464590
Sharma, R.; Kamble, S.S.; Gunasekaran, A.; Kumar, V.; Kumar, A. A systematic literature review on machine learning
applications for sustainable agriculture supply chain performance. Comput. Oper. Res. 2020, 119, 104926. [Google
Scholar] [CrossRef]
Benos, L.; Tagarakis, A.C.; Dolias, G.; Berruto, R.; Kateris, D.; Bochtis, D. Machine Learning in Agriculture: A Comprehensive
Updated Review. Sensors 2021, 21, 3758. [Google Scholar] [CrossRef] [PubMed]
Peng, W.; Karimi Sadaghiani, O. A review on the applications of machine learning and deep learning in agriculture section
for the production of crop biomass raw materials. Energy Sources Part A Recover. Util. Environ. Eff. 2023, 45, 9178–9201.
[Google Scholar] [CrossRef]

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
25
 NOVEMBER, 2024 EDITIONS. INTERNATIONAL JOURNAL OF:
SCIENCE RESEARCH AND TECHNOLOGY VOL. 6

Mohamed, S.A.; Metwaly, M.M.; Metwalli, M.R.; AbdelRahman, M.A.E.; Badreldin, N. Integrating Active and Passive Remote
Sensing Data for Mapping Soil Salinity Using Machine Learning and Feature Selection Approaches in Arid
Regions. Remote Sens. 2023, 15, 1751. [Google Scholar] [CrossRef]
Maheswari, M.U.; Ramani, R. A Comparative Study of Agricultural Crop Yield Prediction Using Machine Learning Techniques.
In Proceedings of the 2023 9th International Conference on Advanced Computing and Communication Systems
(ICACCS), Coimbatore, India, 17–18 March 2023; IEEE: Piscataway, NJ, USA, 2023; Volume 1, pp. 1428–1433. [Google
Scholar]
Kuradusenge, M.; Hitimana, E.; Hanyurwimfura, D.; Rukundo, P.; Mtonga, K.; Mukasine, A.; Uwitonze, C.; Ngabonziza, J.;
Uwamahoro, A. Crop Yield Prediction Using Machine Learning Models: Case of Irish Potato and
Maize. Agriculture 2023, 13, 225. [Google Scholar] [CrossRef]
F. Yilmaz, “Performance and Environmental Impact Assessment of a geothermal-assisted combined plant for multi-
generation products,” Sustainable Energy Technologies and Assessments, vol. 46, p. 101291, 2021.
doi:10.1016/j.seta.2021.10129
Haupt, S. E., and Coauthors, 2022: The history and practice of AI in the environmental sciences. Bull. Amer. Meteor. Soc.,
103, E1351–E1370, https://doi.org/10.1175/BAMS-D-20-0234.1
LeCun, Y., Y. Bengio, and G. Hinton, 2015: Deep learning. Nature, 521, 436–444, https://doi.org/10.1038/nature1
Shifa Zhong, Kai Zhang, Majid Bagheri, Joel G. Burken, April Gu, Baikun Li, Xingmao Ma, Babetta L. Marrone, Zhiyong Jason
Ren, Joshua Schrier, Wei Shi, Haoyue Tan, Tianbao Wang, Xu Wang, Bryan M. Wong, Xusheng Xiao, Xiong Yu, Jun-Jie
Zhu, and Huichun Zhang (2021). Machine Learning: New Ideas and Tools in Environmental Science and Engineering,
https://doi.org/10.1021/acs.est.1c01339
Karina Gibert, Jeffery S.Horsburgh, Ioannis N Athanasiadis and Geoff Holmes 2018 Environmental Data Science
L’heureux, A.; Grolinger, K.; Elyamany, H. F.; Capretz, M. A. Machine learning with big data: Challenges and approaches. IEEE
Access 2017, 5, 7776−7797
Gadaleta, D.; Mangiatordi, G. F.; Catto, M.; Carotti, A.; Nicolotti, O. Applicability domain for QSAR models: where theory
meets reality. International Journal of Quantitative Structure-Property Relationships (IJQSPR) 2016, 1 (1), 45−63.
Tharsanee R.M., Soundariya R.S., Vishnupriya B. (2020). Machine Learning and Data Analytics for Environmental Science: A
Review, Prospects and Challenges, doi:10.1088/1757-899X/955/1/012107
Loussaief, S.; Abdelkrim, A. In Machine Learning Framework for Image Classification, 2016 7th International Conference on
Sciences of Electronics, Technologies of Information and Telecommunications (SETIT), 2016; IEEE, 2016; pp 58−61.
Ahn, W.-Y.; Ramesh, D.; Moeller, F. G.; Vassileva, J. Utility of machine-learning approaches to identify behavioral markers
for substance use disorders: impulsivity dimensions as predictors of current cocaine dependence. Frontiers in
Psychiatry 2016, DOI: 10.3389/fpsyt.2016.00034.
Brunton, S. L.; Kutz, J. N. Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control;
Cambridge University Press, 2019
Chantry, M., H. Christensen, P. Dueben, and T. Palmer, 2021: Opportunities and challenges for machine learning in weather
and climate modelling: Hard, medium and soft AI. Philos. Trans. Roy. Soc., A379, 20200083,
https://doi.org/10.1098/rsta.2020.0083
L. Zhang, L. Zhang and B. Du 2016 IEEE Geoscience and Remote Sensing Magazine 4(2) pp. 22-40.
Kerckhoffs, J.; Hoek, G.; Portengen, L. t.; Brunekreef, B.; Vermeulen, R. C. Performance of prediction algorithms for
modeling outdoor air pollution spatial surfaces. Environ. Sci. Technol. 2019, 53 (3), 1413−1421.
Pak, U.; Ma, J.; Ryu, U.; Ryom, K.; Juhyok, U.; Pak, K.; Pak, C. Deep learning-based PM2.5 prediction considering the
spatiotemporal correlations: A case study of Beijing, China. Sci. Total Environ. 2020, 699, 133561.
Song, R.; Keller, A. A.; Suh, S. Rapid Life-Cycle Impact Screening Using Artificial Neural Networks. Environ. Sci. Technol. 2017,
51 (18), 10777−10785.
Kuhn, M.; Johnson, K. Applied Predictive Modeling; Springer, 2013; Vol. 26. (141) Shorten, C.; Khoshgoftaar, T. M. A survey
on image data augmentation for deep learning. Journal of Big Data 2019, 6 (1), 60.
Bühlmann, P.; Van De Geer, S. Statistics for High-Dimensional Data: Methods, Theory and Applications; Springer Science &
Business Media, 2011. (143) Rexstad, E.; Innis, G. S. Model simplification three applications. Ecol. Modell. 1985, 27 (1−2),
1−13.
Srivastava, N.; Hinton, G.; Krizhevsky, A.; Sutskever, I.; Salakhutdinov, R. Dropout: A simple way to prevent neural networks
from overfitting. Journal of Machine Learning Research 2014, 15 (1), 1929−1958.
Yao, Y.; Rosasco, L.; Caponnetto, A. On early stopping in gradient descent learning. Constructive Approximation 2007, 26
(2), 289−315.
Sutton, R.S., and Barto, A.G. 1998. Introduction to reinforcement learning. Vol. 135. MIT Press, Cambridge, Mass., U.S.A.
Piyush Jain Sean C.P. Coogan, Sriram Ganapathi Subramanian, Mark Crowley, Steve Taylor, and Mike D. Flannigan (2020). A
review of machine learning applications in wildfire science and management. https://doi.org/10.1139/er-2020-0019
Xiao, Q.; Chang, H. H.; Geng, G.; Liu, Y. An ensemble machine-learning model to predict historical PM2. 5 concentrations in
China from satellite data. Environ. Sci. Technol. 2018, 52 (22), 13260−13269.

TIJSRAT
E-ISSN 3026-8796
P-ISSN 3026-8095
26