Applying Deep Learning and Machine Learning Algorithms to Estimate PM Concentration Using Satellite Data and Meteorological Data

Air pollution, particularly fine particulate matter (PM<inline-formula><tex-math notation="LaTeX">_{2.5}</tex-math></inline-formula>), poses significant health risks and environmental challenges worldwide. Therefore, it is essential to monitor air pollution to effec...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE journal of selected topics in applied earth observations and remote sensing S. 1 - 18
Hauptverfasser: Thapa, Ishwor, Devkota, Bidur, Lamichhane, Badri Raj, Devkota, Bhawana Poudel, Dhakal, Raju, Horanont, Teerayut
Format: Journal Article
Sprache:Englisch
Veröffentlicht: IEEE 11.10.2025
Schlagworte:
Accuracy
Air pollution
Atmospheric modeling
Deep Learning
Google Earth Engine
Machine Learning
MODIS
Monitoring
PM<inline-formula xmlns:ali="http://www.niso.org/schemas/ali/1.0/" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"> <tex-math notation="LaTeX"> _{2.5}</tex-math> </inline-formula>
Predictive models
Public healthcare
Remote sensing
Satellites
Sentinel-5P
Support vector machines
ISSN:1939-1404, 2151-1535
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Air pollution, particularly fine particulate matter (PM<inline-formula><tex-math notation="LaTeX">_{2.5}</tex-math></inline-formula>), poses significant health risks and environmental challenges worldwide. Therefore, it is essential to monitor air pollution to effectively act against it. In this study, PM<inline-formula><tex-math notation="LaTeX">_{2.5}</tex-math></inline-formula> levels were estimated using meteorological data and Sentinel-5P air pollution data through machine learning algorithms. The meteorological data utilized included air temperature, relative humidity (RH), wind speed (WS), and Sentinel-5P data. Three Air Quality Monitoring (AQM) stations in Kathmandu, Nepal, were selected as the study area. The effectiveness of several machine learning methods, such as K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Random Forest (RF), ensemble methods, and hybrid methods, were evaluated. Both RF and XGBoost consistently outperformed SVM and KNN regarding PM<inline-formula><tex-math notation="LaTeX">_{2.5}</tex-math></inline-formula> estimation accuracy. Among all the methods studied, XGBoost achieved the highest R 2 ; value of 0.8284 and the lowest RMSE value of 11.0024 using only the Sentinel-5P dataset. The addition of meteorological data further improved the model's performance. After including meteorological data with the Sentinel-5P data, the stacking ensemble demonstrated a maximum R 2 ; score of 0.8324 and a minimum RMSE score of 10.8747. Hence, this study demonstrated that utilizing advanced technologies such as machine learning (ML), deep learning (DL), and novel datasets obtained from satellites can accurately estimate PM<inline-formula><tex-math notation="LaTeX">_{2.5}</tex-math></inline-formula> levels. This approach can significantly aid in monitoring and controlling air pollution by providing precise and timely information on air quality. These findings have significant implications for stakeholders such as policymakers and urban planners, as integrating noble technologies and datasets like machine learning with satellite and meteorological data can lead to more effective air quality management strategies. By availing low-cost solutions for accurate and timely air pollution estimates, this approach can support informed decision-making, reduce public exposure to pollutants, and improve general public health.
ISSN:1939-1404
2151-1535
DOI:10.1109/JSTARS.2025.3620408