ISSN (Online): 2455-3662

EPRA International Journal of Multidisciplinary Research (IJMR) - Peer Reviewed Journal

Volume: 9| Issue: 5| May 2023|| Journal DOI: 10.36713/epra2013 || SJIF Impact Factor 2023: 8.224 || ISI Value: 1.188

ENVIRONMENTAL MONITORING WITH MACHINE

LEARNING

Shalu1, Gurjeet Singh2

1Studentof MCA 3rd Sem., Lords University, Chikani, Alwar, Rajasthan-301028
2Dean, Lords School of Computer Applications & IT, Lords University, Chikani, Alwar, Rajasthan-301028 ,

Article DOI: https://doi.org/10.36713/epra13330

DOI No: 10.36713/epra13330

ABSTRACT
The term "environmental monitoring" refers to the practice of keeping tabs on and assessing the state of both natural and built
environments. The purpose of environmental monitoring is to gather information that may be utilized to spot patterns, hazards,
and improvement avenues. Because theycan analyse enormous volumes of data and identify complicated patterns that may not
be clearly detectable using conventional methods, machine learning techniques may be particularly successful for
environmental monitoring. The lack of a reliable method to gather complete data, and the overall lack of data openness, is the
main problem with the status quo. These environmental data are often collected in siloed units, requiring time and money from
environmental protection agencies before they can be made public. In this study, we'll look at how machine learning may be put
to use in environmental surveillance. Two recent cases will be discussed briefly within the framework of our paper before we
wrap things up.

1. INTRODUCTION such as bacterial contamination or harmful algal blooms.

The review article highlights the challenges faced in However, traditional monitoring methods are time-
developing trustworthy monitoring systems for consuming and labor-intensive. The emergence of artificial
environmental applications due to the extreme intelligence (AI), specifically deep learning convolutional
environments sensor platforms are exposed to. neural networks (CNNs), has improved object recognition
Discriminating between genuine data mistakes and and identification from image datasets, narrowing the gap
apparent errors caused by natural phenomena is a complex between human and machine capabilities. By leveraging AI
task, making outlier identification in sensor data technologies, the monitoring process can be expedited,
challenging. Governments worldwide are investing in enhancing efficiency and accuracy in assessing
Early Warning Systems (EWS) to detect and mitigate environmental conditions. Overall, integrating data-driven
potential losses from natural catastrophes. Changes in land approaches and AI technologies into land use planning and
use, driven by urbanization, agriculture, and other factors, environmental monitoring can greatly enhance decision-
have profound impacts on global biodiversity, land making and contribute to sustainable land management
functions, and the carbon cycle. The increasing global practices. [2].
population is expected to further intensify the demand for
land, posing sustainability challenges. Planners and 2. THE DRAWBACK OF TRADITIONAL
policymakers must adopt purposeful land use strategies to ENVIRONMENT
address the growing demand for finite land supplies. The current setup faces two major problems: the inability
Overall, the article highlights the importance of developing to gather comprehensive data and the lack of openness
robust monitoring systems and implementing sustainable around data. Organizations responsible for protecting the
land use strategies to effectively manage environmental environment must invest time and money to collect
challenges and ensure long-term sustainability [1]. environmental data from different sources before making it
public. There is limited time for verifying the data's
In this review article, the importance of data gathering and accuracy and reliability. Furthermore, the lack of
land use mapping for effective land use planning is communication and collaboration among departments
emphasized. Sustainability indicators derived from this renders the collected environmental protection data
data can provide valuable insights for decision-making and useless from a business perspective. However,
problem-solving in land use planning, addressing advancements in big data and AI technologies offer
economic, social, and environmental issues. Environmental solutions. These technologies enable the sharing and
monitoring is crucial to safeguard the environment and comparison of environmental data, promoting openness
human health, particularly in relation to freshwater and transparency. Big data technology can efficiently store
resources. Monitoring microbial assemblages and other information gathered by each department and utilize the
indicators helps identify potential threats to human health, internet to facilitate public involvement and enhance
2023 EPRA IJMR | http://eprajournals.com/ | Journal DOI URL: https://doi.org/10.36713/epra2013------------------------------------------------------------------------- 208
 ISSN (Online): 2455-3662
EPRA International Journal of Multidisciplinary Research (IJMR) - Peer Reviewed Journal
Volume: 9| Issue: 5| May 2023|| Journal DOI: 10.36713/epra2013 || SJIF Impact Factor 2023: 8.224 || ISI Value: 1.188

understanding of environmental agencies. AI has found image and audio identification, natural language
widespread application in air pollution prediction and early processing, predictive analytics, fraud detection,
warning systems. Nevertheless, for highly variable recommendation systems, andautonomous vehicles.
pollution concentrations, a single machine learning system
may not be sufficient for monitoring. It is necessary to There have been many recent developments in the field of
develop a comprehensive ambient air quality monitoring machine learning that can be useful in monitoring the
system and innovative approaches for making air quality environment effectively [8].
decisions [3]. In the event of extremely variable pollutant Deep Learning, Reinforcement Learning, Generative
concentrations, a single machine learning method is not Adversarial Networks (GANs), Transfer Learning,
sufficient for environmental monitoring. Federated Learning
Therefore, we can that there are several drawbacks of
traditional environment monitoring systems, including: 4. ENVIRONMENT MONITORING
• Limited coverage: Traditional environment monitoring Environmental monitoring refers to the process of
systems are often limited in their coverage area. This collecting and analyzing data on various aspects of the
means that they may not be able to monitor certain areas environment, such as air quality, water quality, soil quality,
thatare of interest [4]. and weather conditions [9]. Environmental monitoring
• Expensive infrastructure: Setting up and maintaining serves the purpose of identifying and quantifying
traditional environment monitoring systems can be environmental changes, assessing the effectiveness of
expensive due to the need for a large amount of environmental policies, and evaluating the impact of
infrastructure, including sensors, data loggers, and human activities on the environment. Different monitoring
communication networks. programs focus on specific aspects, such as air quality or
water quality. Various techniques, including remote
• High power consumption: Traditional environment sensing, satellite imaging, and on-the-ground sampling, are
monitoring systems often require a lot of power to employed for environmentalmonitoring. The data collected
operate, which can be a challenge in remote or off-grid through monitoring programs is utilized by government
areas where power may be limited. agencies, non-governmental organizations, and businesses
• Limited scalability: Traditional environment to make informed decisions regarding environmental
monitoring systems may not be easily scalable, meaning management and conservation. Technological
that it can be difficult to add more sensors or expand the advancements have significantly improved environmental
monitoring area [5]. monitoring, enabling better data collection, analysis, and
• Limited real-time data: Traditional environment interpretation. These advancements include remote
monitoring systems may have a delay in data sensing, geographic information systems (GIS), sensor
transmission and processing, meaning that real-time technology, big data analytics, citizen science, and the
data is not always available. Internet of Things (IoT). Environmental monitoring plays
Data silos: Traditional environment monitoring systems a crucial role in safeguarding natural resources and
often generate large amounts of data,but the data may be promoting sustainable management for future generations.
stored in different locations or in different formats, making [10]. Advancements in environmental monitoring include
it difficult to integrate and analyze. remote sensing, IoT, big data analytics, artificial
intelligence (AI), and mobile apps. Remote sensing enables
3. WHAT EXACTLY IS MACHINE data collection on a large scale, such as tracking
LEARNING? deforestation and monitoring crop growth. IoT devices
Machine learning is a subfield of artificial intelligence that connected to the internet provide real-time data on air and
focuses on the creation of algorithms and statistical models water quality, aiding resource management and pollution
that allow computer systems to learn from data and make control. Big data analytics allows for analysis of large
predictions or judgments based on it. Machine learning datasets identifying trends and patterns. AI algorithms
aims to create systems that can enhance their job analyze data to detect patterns, anomalies, and potential
performance without being explicitly trained to do so [6]. risks. Mobile apps enable crowd sourcing of
There are three kinds of machine learning algorithms: environmental data. Machine learning can be appliedto
supervised learning, unsupervised learning, and environmental monitoring for predictive modeling, image
reinforcement learning. In supervised learning, the recognition, anomaly detection, classification, and time-
algorithm is trained using labeled data, which implies that series analysis. These advancements have improved our
the expected output is displayed alongside the input data understanding of human impacts on the environment and
[7]. The algorithm learns to predict the appropriate output our ability to mitigate and adapt to environmental changes.
based on the input properties. Unsupervised learning [11].
algorithms are trained on unlabeled data and are tasked
with discovering patterns and structure in the data. 5. NEED FOR DETAILED MONITORING OF
Reinforcement learning is a kind of machine learning in THE ENVIRONMENT
which an agent learns how to act in a given The need for detailed monitoring of the environment has
environment to maximize a reward signal. Machine become increasingly important in recent years due to
learning is used in a variety of applications, including various factors such as climate change, pollution,
2023 EPRA IJMR | http://eprajournals.com/ | Journal DOI URL: https://doi.org/10.36713/epra2013------------------------------------------------------------------------- 209
 ISSN (Online): 2455-3662
EPRA International Journal of Multidisciplinary Research (IJMR) - Peer Reviewed Journal
Volume: 9| Issue: 5| May 2023|| Journal DOI: 10.36713/epra2013 || SJIF Impact Factor 2023: 8.224 || ISI Value: 1.188

deforestation, and the depletion of natural resources [12]. the spread of invasive species.
Here are some key reasons why detailed monitoring of
the environment is necessary: The integration of machine learning and AI with
environmental issues has the potential to revolutionize our
5.1 Conservation and Biodiversity: Detailed monitoring approach to addressing and mitigating environmental
allows us to understand the distribution and abundance of challenges. Here are a few key areas where machine
species, their habitats, and any changes occurring within learning and AI can contribute to environmental issues
ecosystems. This information is crucial for effective [14].
conservation efforts and the protection of biodiversity.
5.2 Climate Change: Monitoring environmental 5.7 Environmental Monitoring and Conservation:
parameters such as temperature, precipitation,sea level, Machine learning algorithms can analyze large datasets
and greenhouse gas concentrations helps us track and collected from various sources such as satellite imagery,
understand the impacts of climate change. This sensors, and drones to monitor and assess environmental
information is essential for developing strategies to conditions. This technology can be used to track
mitigate and adapt to climate change and to assess the deforestation, detect changes in land cover, monitor water
effectiveness of such measures. quality, and identify endangered species. By automating
the analysis process, machine learning can provide real-
Pollution Management: Monitoring the quality of air, time insights and help conservation efforts.
water, and soil is essential for identifying pollution sources,
assessing their impact on human health and ecosystems, 5.8 Environmental Decision-Making: AI can assist
and implementing appropriate pollution control measures. policymakers and organizations in making informed
decisions related to environmental management. By
5.3 Natural Resource Management: Detailed monitoring analyzing vast amounts of data and considering various
provides insights into the availability, distribution, and factors, AI algorithms can provide insights, model
usage of natural resources such as forests, water, scenarios, and support evidence-based decision-making
minerals, and fisheries. This information is critical for processes.
sustainable resource management and planning.
6. CONDUCTING EFFECTIVE
5.4 Disaster Management: Monitoring environmental ENVIRONMENT MONITORING
conditions can help in predicting and responding to natural Conducting effective environment monitoring with the
disasters such as hurricanes, floods, wildfires, and help of machine learning can significantly enhance the
droughts. Timely and accurate data enable effective accuracy and efficiency of data analysis [15]. To
emergency planning, response, and recovery efforts [13]. implement environmental monitoring using machine
learning, follow these steps:
5.5 Human Health: Monitoring environmental factors 6.1 Define the Problem: Clearly identify the
such as air quality, water contamination, and exposure to environmental factors you want to monitor and the specific
hazardous substances is vital for understanding the impacts problem you want to solve, such as air quality, water
on human health. pollution, or climatechange.

5.6 Policy Development: Detailed environmental 6.2. Data Collection: Gather relevant data from sources
monitoring provides scientific evidence and data that like sensors, satellites, weather stations, or existing
inform policy development and decision-making databases. Ensure the data is reliable, consistent, and covers
processes. a significant time period.

To achieve detailed monitoring, various technologies and 6.3 Data Preprocessing: Clean the collected data by
methods are employed, including satellite remote sensing, removing outliers, handling missing values, and
ground-based sensors, drones, data analytics, and citizen normalizing it for accuracy.
science initiatives. Collaboration between governments,
scientific institutions, environmental organizations, and 6.4 Feature Engineering: Extract meaningful features
communities is crucial for the success of environmental from the data that can provide valuable insights, such as
monitoring efforts. Overall, detailed monitoring of the pollutant levels, weather conditions, and geographical
environment plays a crucial role in understanding the state information.
of our planet, identifying problems, and developing
effective solutions to ensure a sustainable and healthy 6.5Model Selection: Choose an appropriate machine
future. Effective monitoring programs require careful learning model based on your problem and data
planning and implementation to ensure that data is characteristics, such as decision trees, random forests,
accurate, reliable, and meaningful. Regular monitoring can support vector machines (SVM), or neural networks.
help to track changes over time and identify trends that may
require intervention. It can also provide early warning of 6.6. Training the Model: Divide the cleaned data into a
potential environmental threats, such as natural disasters or training set and a test set. Train the model using the training
2023 EPRA IJMR | http://eprajournals.com/ | Journal DOI URL: https://doi.org/10.36713/epra2013------------------------------------------------------------------------- 210
 ISSN (Online): 2455-3662
EPRA International Journal of Multidisciplinary Research (IJMR) - Peer Reviewed Journal
Volume: 9| Issue: 5| May 2023|| Journal DOI: 10.36713/epra2013 || SJIF Impact Factor 2023: 8.224 || ISI Value: 1.188

set to identify trends and connections within the data.

7.2 The Smart Monitoring of Water Pollution (SMWP)
6.7 Model Evaluation: Assess the trained model's System is another advanced technology that uses sensors,
performance using the testing set, using metrics like data analytics, and communication technologies for real-
accuracy, precision, recall, and F1 score. Fine-tune the timemonitoring and management of water pollution. The
model if necessary. system provides accurate and timelyinformation about
the quality and condition of water bodies, enabling
6.8 Deploying the Model: Deploy the model in a real-time prompt actions to mitigate pollution and protect water
or near-real-time environment for continuous resources. [18].
monitoring, integrating it into software applications, 7.3 The SMWP system typically consists of the following
cloud-based platforms, orIoT systems. components
7.4
6.9 Monitoring and Updating: Continuously monitor 7.3.1 Sensor Network: It includes a network of sensors
the model's performance and collect new data for deployed in different water bodies such as rivers, lakes, and
periodic updates. Retrain the model with fresh data as oceans. These sensors measure various parameters related
needed to adapt to changing environmental conditions. to water quality, including temperature, pH level, dissolved
oxygen, turbidity, conductivity, and the presence of
6.10 Interpretation and Action: Analyze the model's contaminants like heavy metals or harmful chemicals [25].
predictions to generate actionable insights. Use the
results to make informed decisions, implement The Smart Monitoring of Water Pollution (SMWP) system
interventions, or raise awareness about environmental involves data collection and transmission through a sensor
issues. network, with data being sent to a central server or cloud-
basedplatform. Advanced algorithms and machine learning
In addition to environmental monitoring, machine learning techniques [19-24] are applied to analyze the collected
can also be applied to climate modeling and prediction, data, identifying patterns, trends, and anomalies in water
energy efficiency and resource management, natural quality. Real-time monitoring capabilities allow
disaster management, waste management and recycling, stakeholders to access data and receive alerts when
and ecosystem restoration and biodiversity conservation pollution levels exceed limits or significant changes
[16]. Collaboration with environmental scientists and occur. User-friendly interfaces and dashboards visualize
stakeholders is crucial for successful implementation. the data for easy interpretation. The system can be
Remember, the effectiveness of environmental monitoring integrated with existing water management infrastructure,
relies not only on machine learning but also on data quality enabling automation of processes and optimization of
and domain expertise. watertreatment and allocation based on real-time pollution
data.
7. RECENT EXAMPLES
Now, before concluding our paper, we'll analyze 2 recent Overall, the Smart Monitoring of Water Pollution (SMWP)
examples where machine learning was adopted into one of system plays a crucial role in enhancing water quality
the aspects of environment monitoring- monitoring and management, aiding in pollution
prevention, and supporting sustainable water resource
7.1 The Urban Flood management practices.
Project is an initiative funded by the European Union (EU)
that aims todevelop and implement innovative technologies 8 CONCLUSION
for monitoring and managing flood risks in urban areas. In conclusion, combining big data and artificial intelligence
The project focuses on utilizing advanced sensor networks, technologies has enormouspotential for addressing urgent
data analytics, and real-time information systems to environmental issues and advancing sustainable
improve flood forecasting, early warning systems, and management and conservation initiatives. The study of vast
emergency response strategies [17]. The Urban Flood and varied environmental datasets is made possible by
project focuses on environment monitoring in urban areas machine learning algorithms, making it easier to spot
to address flood risk management. It utilizes a network of trends, abnormalities, and threats. These algorithms are
sensors to monitor environmental parameters like rainfall essential for environmental modelling and forecasting
intensity, river water levels, and soil moisture. The project because they make it possible to predict future
aims to develop advanced forecasting models and early environmental conditions. Machine learning boosts
warning systems by integrating data from multiple sources. productivity and efficacy in environmental monitoring
Decision support systems are also being developed to assist operations by automating data processing and enabling
emergency response teams and policymakers during flood real-time monitoring. However, to guarantee the
events. The project considers climate change adaptation dependability and accuracy of machine learning models,
and seeks to assess the potential impact of climate significant consideration must be paid to data quality,
change on urban flood risks. Overall, the UrbanFlood model validation, and continual improvement. The ethical
project contributes to mitigating environmental impacts application of machine learning in environmental
and fostering sustainable urbandevelopment. monitoring, as well as transparency, accountability, and
2023 EPRA IJMR | http://eprajournals.com/ | Journal DOI URL: https://doi.org/10.36713/epra2013------------------------------------------------------------------------- 211
 ISSN (Online): 2455-3662
EPRA International Journal of Multidisciplinary Research (IJMR) - Peer Reviewed Journal
Volume: 9| Issue: 5| May 2023|| Journal DOI: 10.36713/epra2013 || SJIF Impact Factor 2023: 8.224 || ISI Value: 1.188

collaboration betweendomain experts and data scientists, Natural disasters and extreme events in agriculture:
are crucial. In general, machine learning has the power to Impacts and mitigation, 1-22.
completely transform environmental management by 14. Ghazal, T. M., Hasan, M. K., Alshurideh, M. T., Alzoubi, H.
delivering data-driven insights and predictive capabilities M., Ahmad, M., Akbar, S. S.,... & Akour, I. A. (2021). IoT
for smart cities: Machine learning approaches in smart
for well-informed decision-making and proactive healthcare—A review. Future Internet, 13(8), 218.
environmental preservation. 15. Oztekin, A., Delen, D., Turkyilmaz, A., & Zaim, S. (2013).
A machine learning-basedusability evaluation method for
9 REFERENCE eLearning systems. Decision Support Systems, 56, 63-
1. Pyayt, A. L., Mokhov, I. I., Lang, B., Krzhizhanovskaya, V. 16. Krishnan, S. R., Nallakaruppan, M. K., Chengoden, R.,
V., & Meijer, R. J. (2011). Machine learning methods Koppu, S., Iyapparaja, M.,Sadhasivam, J., & Sethuraman,
for environmental monitoring and flood protection. S. (2022). Smart Water Resource Management Using
International Journal of Computer and Information 17. Artificial Intelligence—A Review. Sustainability, 14(20),
Engineering, 5(6), 549- 554. 13384.
2. Roy Choudhury, A. K. (2014). Environmental impacts 18. Genovese, E. (2006). A methodological approach to land
of the textile industry and itsassessment through life use-based flood damage assessment in urban areas:
cycle assessment. Roadmap to sustainable textiles Prague case study. Technical EUR Reports, EUR, 22497.
andclothing: environmental and social aspects of textiles 19. Pérez, J., Basaguren, A., Descals, E., Larranaga, A.,
and clothing supply chain, 1-39. & Pozo, J. (2013). Leaf-litterprocessing in headwater
3. Zhang, Z., Geiger, J., Pohjalainen, J., Mousa, A. E. D., Jin, streams of northern Iberian Peninsula: moderate levels
W., & Schuller, B. (2018). Deep learning for ofeutrophication do not explain breakdown rates.
environmentally robust speech recognition: An overview of Hydrobiologia, 718, 41-57.
recent developments. ACM Transactions on Intelligent 20. Singh, G., & Nager, P. (2012). A case Study on Nutek India
Systems and Technology (TIST), 9(5), 1-28. Limited Regarding Deep Falling in Share
4. Leduc, G. (2008). Road traffic data: Collection methods Price. Researchers World--Journal of Arts, Science &
and applications. WorkingPapers on Energy, Transport Commerce, 3(2), 3.
and Climate Change, 1(55), 1-55. 21. Singh, Dr. Gurjeet, An Analysis of Outliers for Fraud
5. Alamri, A., Ansari, W. S., Hassan, M. M., Hossain, M. S., Detection in Indian Stock Market (January 2, 2022). Nagar,
Alelaiwi, A., & Hossain, M. A.(2013). A P., & Singh, G. (2012). An Analysis of Outliers for Fraud
Survey on sensor-cloud: architecture, applications, and Detection in Indian Stock Market, Researchers World :
approaches. International Journal of Distributed Sensor Journal of Arts, Science and Commerce, Vol.-III, Issue 4(4),
Networks, 9(2), 917923. 10-15., Available at
6. Sil, R., Roy, A., Bhushan, B., & Mazumdar, A. K. (2019, SSRN: https://ssrn.com/abstract=3998394
October). Artificial intelligence and machine learning based 22. Nagar, P., & Issar, G. S. DETECTION OF OUTLIERS IN
legal application: the state-of-the-art and future research STOCK MARKET USING REGRESSION ANALYSIS.
trends. In 2019 International Conference on Computing, 23. Singh, G. (2022). Machine learning models in stock market
Communication, and Intelligent Systems (ICCCIS) (pp. 57- prediction. arXiv preprint arXiv:2202.09359.
62). IEEE. 24. Singh, G., Kalra, N., Yadav, N., Sharma, A., & Saini, M.
7. Alloghani, M., Al-Jumeily, D., Mustafina, J., Hussain, A., & (2022). SMART AGRICULTURE: A REVIEW. Siberian
Aljaaf, A. J. (2020). Asystematic review on supervised and Journal of Life Sciences and Agriculture, 14(6), 423-454.
unsupervised machine learning algorithms for datascience. 25. Singh, G. (2023). Earthquake Prediction using
Supervised and unsupervised learning for data science, 3- Convolutional Neural Network. Disaster Advances, 16(6),
21. 45-52. https://doi.org/10.25303/1606da45052
8. Singh, A., Sharma, A., Ahmed, A., Sundramoorthy, A. K., 26. Mastrangeli, M., Aydogmus, H., Dostanic, M., Motreuil-
Furukawa, H., Arya, S., & Khosla, A. (2021). Recent Ragot, P., Revyn, N., De Wagenaar, B., ... & Sarro, P. M.
advances in electrochemical biosensors: (2021, June). Microelectromechanical Organs-on-Chip. In
Applications,challenges, and future scope. Biosensors, 2021 21st International Conference on Solid-State
11(9), 336. Sensors, Actuators and Microsystems (Transducers) (pp.
9. Artiola, J., Pepper, I. L., & Brusseau, M. L. (Eds.). (2004). 102-107). IEEE.
Environmental monitoring and characterization. Academic
Press.
10. Asad, M., Anwar, M. I., Abbas, A., Younas, A., Hussain, S.,
Gao, R., ... & Khan, S. (2022). AIE based luminescent
porous materials as cutting-edge tool for environmental
monitoring: State of the art advances and perspectives.
Coordination Chemistry Reviews, 463, 214539.
11. Ghannam, R. B., & Techtmann, S. M. (2021). Machine
learning applications in microbial ecology, human
microbiome studies, and environmental monitoring.
Computational and Structural Biotechnology Journal, 19,
1092-1107.
12. De Sherbinin, A., VanWey, L. K., McSweeney, K.,
Aggarwal, R., Barbieri, A., Henry, S.,... & Walker, R.
(2008). Rural household demographics, livelihoods and the
environment. Global environmental change, 18(1), 38-53.
13. Sivakumar, M. V. (2005). Impacts of natural disasters
in agriculture, rangeland and forestry: an overview.
2023 EPRA IJMR | http://eprajournals.com/ | Journal DOI URL: https://doi.org/10.36713/epra2013------------------------------------------------------------------------- 212