A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern Alignment

Accurate air quality forecasting is essential for environmental management and health protection. However, conventional air quality models often exhibit systematic biases and underpredict pollution events due to uncertainties in emissions, meteorology, and atmospheric processes. Addressing these lim...

Full description

Saved in:
Bibliographic Details
Main Authors: Ioannis Stergiou, Nektaria Traka, Dimitrios Melas, Efthimios Tagaris, Rafaella-Eleni P. Sotiropoulou
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Atmosphere
Subjects:
Online Access:https://www.mdpi.com/2073-4433/16/6/739
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849467182573944832
author Ioannis Stergiou
Nektaria Traka
Dimitrios Melas
Efthimios Tagaris
Rafaella-Eleni P. Sotiropoulou
author_facet Ioannis Stergiou
Nektaria Traka
Dimitrios Melas
Efthimios Tagaris
Rafaella-Eleni P. Sotiropoulou
author_sort Ioannis Stergiou
collection DOAJ
description Accurate air quality forecasting is essential for environmental management and health protection. However, conventional air quality models often exhibit systematic biases and underpredict pollution events due to uncertainties in emissions, meteorology, and atmospheric processes. Addressing these limitations, this study introduces a hybrid deep learning model that integrates convolutional neural networks (CNNs) and Long Short-Term Memory (LSTM) for ozone forecast bias correction. The model is trained here, using data from ten stations in Texas, enabling it to capture both spatial and temporal patterns in atmospheric behavior. Performance evaluation shows notable improvements, with a Root Mean Square Error (RMSE) reduction ranging from 34.11% to 71.63%. F1 scores for peak detection improved by up to 37.38%, Dynamic Time Warping (DTW) distance decreased by 72.77%, the Index of Agreement rose up to 90.09%, and the R<sup>2</sup> improved by up to 188.80%. A comparison of four loss functions—Mean Square Error (MSE), Huber, Asymmetric Mean Squared Error (AMSE), and Quantile Loss—revealed that MSE offered balanced performance, Huber Loss achieved the highest reduction in systematic RMSE, and AMSE performed best in peak detection. Additionally, four deep learning architectures were evaluated: baseline CNN-LSTM, a hybrid model with attention mechanisms, a transformer-based model, and an End-to-End framework. The hybrid attention-based model consistently outperformed others across metrics while maintaining lower computational demands.
format Article
id doaj-art-1e33115399a64f3ea25a8949ac8b2bb1
institution Kabale University
issn 2073-4433
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Atmosphere
spelling doaj-art-1e33115399a64f3ea25a8949ac8b2bb12025-08-20T03:32:27ZengMDPI AGAtmosphere2073-44332025-06-0116673910.3390/atmos16060739A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern AlignmentIoannis Stergiou0Nektaria Traka1Dimitrios Melas2Efthimios Tagaris3Rafaella-Eleni P. Sotiropoulou4Air & Waste Management Lab, Polytechnic School, University of Western Macedonia, 50132 Kozani, GreeceAir & Waste Management Lab, Polytechnic School, University of Western Macedonia, 50132 Kozani, GreeceLaboratory of Atmospheric Physics, School of Physics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceAir & Waste Management Lab, Polytechnic School, University of Western Macedonia, 50132 Kozani, GreeceAir & Waste Management Lab, Polytechnic School, University of Western Macedonia, 50132 Kozani, GreeceAccurate air quality forecasting is essential for environmental management and health protection. However, conventional air quality models often exhibit systematic biases and underpredict pollution events due to uncertainties in emissions, meteorology, and atmospheric processes. Addressing these limitations, this study introduces a hybrid deep learning model that integrates convolutional neural networks (CNNs) and Long Short-Term Memory (LSTM) for ozone forecast bias correction. The model is trained here, using data from ten stations in Texas, enabling it to capture both spatial and temporal patterns in atmospheric behavior. Performance evaluation shows notable improvements, with a Root Mean Square Error (RMSE) reduction ranging from 34.11% to 71.63%. F1 scores for peak detection improved by up to 37.38%, Dynamic Time Warping (DTW) distance decreased by 72.77%, the Index of Agreement rose up to 90.09%, and the R<sup>2</sup> improved by up to 188.80%. A comparison of four loss functions—Mean Square Error (MSE), Huber, Asymmetric Mean Squared Error (AMSE), and Quantile Loss—revealed that MSE offered balanced performance, Huber Loss achieved the highest reduction in systematic RMSE, and AMSE performed best in peak detection. Additionally, four deep learning architectures were evaluated: baseline CNN-LSTM, a hybrid model with attention mechanisms, a transformer-based model, and an End-to-End framework. The hybrid attention-based model consistently outperformed others across metrics while maintaining lower computational demands.https://www.mdpi.com/2073-4433/16/6/739deep learningconvolutional neural networks—CNNsLong Short-Term Memory—LSTMCommunity Multiscale Air Quality—CMAQbias correction
spellingShingle Ioannis Stergiou
Nektaria Traka
Dimitrios Melas
Efthimios Tagaris
Rafaella-Eleni P. Sotiropoulou
A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern Alignment
Atmosphere
deep learning
convolutional neural networks—CNNs
Long Short-Term Memory—LSTM
Community Multiscale Air Quality—CMAQ
bias correction
title A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern Alignment
title_full A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern Alignment
title_fullStr A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern Alignment
title_full_unstemmed A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern Alignment
title_short A Deep Learning Method for Improving Community Multiscale Air Quality Forecast: Bias Correction, Event Detection, and Temporal Pattern Alignment
title_sort deep learning method for improving community multiscale air quality forecast bias correction event detection and temporal pattern alignment
topic deep learning
convolutional neural networks—CNNs
Long Short-Term Memory—LSTM
Community Multiscale Air Quality—CMAQ
bias correction
url https://www.mdpi.com/2073-4433/16/6/739
work_keys_str_mv AT ioannisstergiou adeeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT nektariatraka adeeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT dimitriosmelas adeeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT efthimiostagaris adeeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT rafaellaelenipsotiropoulou adeeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT ioannisstergiou deeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT nektariatraka deeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT dimitriosmelas deeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT efthimiostagaris deeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment
AT rafaellaelenipsotiropoulou deeplearningmethodforimprovingcommunitymultiscaleairqualityforecastbiascorrectioneventdetectionandtemporalpatternalignment