Prediction of Particulate Concentration Based on Correlation Analysis and a Bi-GRU Model.

Area of Science:

• Environmental Science
• Data Science
• Public Health

Background:

• Particulate air pollution poses significant health risks, necessitating accurate concentration prediction.
• Understanding spatial-temporal correlations in particle transport is key for effective forecasting.

Purpose of the Study:

• To develop and evaluate a novel model for predicting particulate matter concentrations.
• To leverage spatial-temporal correlations for enhanced air quality forecasting.

Main Methods:

• Utilized Pearson Correlation Coefficients (PCCs) to analyze spatial correlations between monitoring sites.
• Developed a prediction model integrating bi-directional gated recurrent units (Bi-GRUs) with PCCs.
• Tested the model on hourly pollutant concentration data from Beijing air quality monitoring stations.

Main Results:

• The proposed Bi-GRU and PCC model demonstrated superior performance in predicting particulate concentrations within a six-hour forecast window.
• Achieved lower prediction errors compared to three other benchmark models.
• The model requires fewer training samples and enables real-time forecasting.

Conclusions:

• The developed model offers a promising approach for improving the accuracy of short-term fine particle concentration predictions.
• Findings can aid environmental researchers in refining air quality models and policymakers in implementing targeted pollution control strategies.
• Correlation analysis between sites facilitates informed, data-driven environmental policy decisions.