Scholarlink

<p>Exploring the intercity relationship of PM<sub>2.5</sub> helps in understanding the mechanisms of the pollution transport and enhancing the forecasting capabilities of data-driven models. In this study, we have developed a framework to identify the potential source region (PSR) for a target city based on the Granger causality (GC) test. Initially, we filter out the low-frequency component of the PM<sub>2.5</sub> series and conduct the GC test on the high-frequency component. Subsequently, only cities with statistical significance from the GC test, and that have at least one backward trajectory of the target city passing through them, are retained as the PSR cities. Based on the results from 309 cities, we observe that the patterns of the PSR cities exhibit a clear seasonal change, reflecting the monsoon direction. The influence direction is primarily north–south and the region with strong intercity influence is located in the North China Plain in fall, expanding to South China in winter. Forecasting experiments using the XGBoost model indicate that, for the high-frequency component of PM<sub>2.5</sub>, incorporating PSR information enhances the <i>R</i><sup>2</sup> value by 0 to 0.2 for most cities. Furthermore, in addition to the PM<sub>2.5</sub> level in the PSR, the meteorological parameters in the PSR, particularly the meridional wind speed, are also valuable for the forecasting model.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Framework for Mining Intercity Relationship of PM2.5 Based on Granger Causality Test: Application to Cities in China

  • 期刊论文

  • 发表日期：

    2025年4月14日

查看全文
  • Ran-Ran He,
  • Dong-Qin Li,
  • Sun Zhang

摘要

Exploring the intercity relationship of PM2.5 helps in understanding the mechanisms of the pollution transport and enhancing the forecasting capabilities of data-driven models. In this study, we have developed a framework to identify the potential source region (PSR) for a target city based on the Granger causality (GC) test. Initially, we filter out the low-frequency component of the PM2.5 series and conduct the GC test on the high-frequency component. Subsequently, only cities with statistical significance from the GC test, and that have at least one backward trajectory of the target city passing through them, are retained as the PSR cities. Based on the results from 309 cities, we observe that the patterns of the PSR cities exhibit a clear seasonal change, reflecting the monsoon direction. The influence direction is primarily north–south and the region with strong intercity influence is located in the North China Plain in fall, expanding to South China in winter. Forecasting experiments using the XGBoost model indicate that, for the high-frequency component of PM2.5, incorporating PSR information enhances the R2 value by 0 to 0.2 for most cities. Furthermore, in addition to the PM2.5 level in the PSR, the meteorological parameters in the PSR, particularly the meridional wind speed, are also valuable for the forecasting model.