# Decomposing drivers of air pollutant emissions in China: A hybrid LMDI and Geographically Weighted Regression approach

## TL;DR

This paper introduces a new method combining LMDI and GWR to analyze and reduce air pollution in China, showing how factors like energy intensity impact emissions.

## Contribution

The novel three-stage framework integrates LMDI decomposition with GWR for spatially adaptive emission reduction strategies.

## Key findings

- Energy intensity in Tangshan reduces emissions by −14.834 million tons annually.
- SO2-PM2.5 synergy in Beijing Tianjin Hebei achieves 297,000 tons annual emission reduction.
- GWR model outperforms others with 12.78% lower Akaike information criterion.

## Abstract

Air pollution control is an urgent problem in the field of environment, and it is crucial to accurately identify emission driving factors and collaborative emission reduction paths. In order to construct and analyze the driving mechanism of atmospheric pollutant emissions and explore the potential for regional collaborative emission reduction, an innovative three-stage progressive analysis framework was developed by combining Logarithmic Mean Divisia Index (LMDI) decomposition and Geographically Weighted Regression (GWR), which includes factor decomposition, spatial modeling, and collaborative optimization. Through empirical analysis, it was found that the energy intensity effect in Tangshan city reduces emissions by an average of −14.834 million tons per year, becoming the core driving force. The synergistic emission reduction ratio of SO2-PM2.5 in the Beijing Tianjin Hebei region reached 1: 0.38, with an average annual emission reduction of 297000 tons and a regional synergy index of 0.85 (p < 0.01), significantly better than other pollutant combinations. The adjusted R2 of the GWR model reached 0.86, the residual Moran’s I index was 0.07, and the proportion of significant variables reached 75%, which is 15.28% higher than other models. In addition, the Akaike information criterion corrected by the GWR model was reduced by an average of 12.78% compared to other models. The results indicated that the synergistic effect of multi factor decomposition and spatial heterogeneity analysis could significantly enhance the regional adaptability of emission reduction strategies, providing scientific support for cross regional collaborative governance.