On hourly prediction of PM2.5 using spatial–temporal graph convolutional network

Globally, air pollution has garnered significant attention in the past few years. Accurate air pollution like PM 2.5 hourly prediction is an essential instrument for air protection because it can be utilized in developing air pollution prevention and control techniques and provides an early warning...

Full description

Saved in:
Bibliographic Details
Published in:Earth science informatics Vol. 18; no. 2; p. 402
Main Authors: Ren, Zhenxing, Ji, Xinxin
Format: Journal Article
Language:English
Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.06.2025
Springer Nature B.V
Subjects:
Air pollution
Artificial neural networks
Decomposition
Early warning systems
Earth and Environmental Science
Earth Sciences
Earth System Sciences
Forecasting
Forecasting models
Information Systems Applications (incl.Internet)
Meteorological data
Ontology
Particulate matter
Permutations
Pollution prevention
Prediction models
Search algorithms
Simulation and Modeling
Space Exploration and Astronautics
Space Sciences (including Extraterrestrial Physics
Beijing China
China
Spatial–temporal graph convolutional network
Attention mechanism
Hourly prediction of air pollutants
Empirical mode decomposition
ISSN:1865-0473, 1865-0481
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Globally, air pollution has garnered significant attention in the past few years. Accurate air pollution like PM 2.5 hourly prediction is an essential instrument for air protection because it can be utilized in developing air pollution prevention and control techniques and provides an early warning system for dangerous air pollutants. Conventional forecasting models are limited due to the extraction of spatial and spatiotemporal features between variables, emphasizing more on the temporal characteristics of the variables. Considering the spatiotemporal correlation of air pollution, we propose a novel hourly prediction model ICEEMDAN-SSA-VMD-ABSTGCN. Initially, the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) approach is utilized for the decomposition of observed PM 2.5 concentrations. Variational mode decomposition (VMD) is then leveraged to further decompose the selected signals with the highest multi-scale permutation entropy (MPE) after its parameters have been optimized using the sparrow search algorithm (SSA). Next, an attention-based spatial–temporal graph convolutional network (ABSTGCN) was built by merging several effect factors, such as air pollution factors, and meteorological data. This methodology can be used for concentration forecasting of PM 2.5 in the future.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1865-0473
1865-0481
DOI:10.1007/s12145-025-01890-1