Scholarlink

<p>This study first constructs a Geographically Weighted Random Forest (GW-RF) model to analyze the importance scores and spatial distribution of the factors influencing PM<sub>2.5</sub> concentrations across 76 cities in the Yellow River Basin from 2017 to 2022. Subsequently, we developed a Bayesian Iterative Dirichlet Model (BIDM). In this model, the Dirichlet distribution describes the prior probability distribution of the key factors affecting PM<sub>2.5</sub> concentrations. The key influencing factors identified by the GW-RF model then establish a state transition matrix. The columns and vectors of this matrix serve as the observation count vectors for the likelihood function, which derives from the multinomial distribution of these observation count vectors. Finally, Bayesian inference forecasts the probability of occurrence of these key factors in future PM<sub>2.5</sub> concentrations across cities in the Yellow River Basin. The results indicate that: (1) During the study period, the monthly average PM<sub>2.5</sub> concentration in the Yellow River Basin exhibited a distinct “U” shaped curve, with annual average concentrations showing a downward fluctuation. Spatially, a “higher in the east and lower in the west” distribution pattern appeared. (2) The importance scores of factors influencing PM<sub>2.5</sub> concentrations in the cities of the Yellow River Basin reveal that local PM<sub>10</sub> concentration and surrounding urban PM<sub>2.5</sub> concentration have relatively high importance scores, with average values of 0.5602 and 0.4254, respectively. Local CO and NO<sub>2</sub> concentrations also exert some influence on PM<sub>2.5</sub> concentrations, with their average values ranging from 0.2145 to 0.3438. In contrast, meteorological factors have a relatively weak impact on PM<sub>2.5</sub> concentrations. (3) The key factors influencing future PM<sub>2.5</sub> concentrations in the Yellow River Basin are local PM<sub>10</sub> concentration, surrounding urban PM<sub>2.5</sub> concentration, and local CO concentration, with predicted probabilities ranging from 0.3000 to 0.5907, 0.2353 to 0.4783, and 0.0591 to 0.1478, respectively.</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

Prediction of influencing factors of PM2.5 in the Yellow River Basin based on Bayesian iterative Dirichlet model

  • 期刊论文

  • 发表日期：

    2026年4月7日

查看全文
  • Jingya Liu,
  • Weifu Ding,
  • Lifen Chen,
  • Hangzhi Yu

摘要

This study first constructs a Geographically Weighted Random Forest (GW-RF) model to analyze the importance scores and spatial distribution of the factors influencing PM2.5 concentrations across 76 cities in the Yellow River Basin from 2017 to 2022. Subsequently, we developed a Bayesian Iterative Dirichlet Model (BIDM). In this model, the Dirichlet distribution describes the prior probability distribution of the key factors affecting PM2.5 concentrations. The key influencing factors identified by the GW-RF model then establish a state transition matrix. The columns and vectors of this matrix serve as the observation count vectors for the likelihood function, which derives from the multinomial distribution of these observation count vectors. Finally, Bayesian inference forecasts the probability of occurrence of these key factors in future PM2.5 concentrations across cities in the Yellow River Basin. The results indicate that: (1) During the study period, the monthly average PM2.5 concentration in the Yellow River Basin exhibited a distinct “U” shaped curve, with annual average concentrations showing a downward fluctuation. Spatially, a “higher in the east and lower in the west” distribution pattern appeared. (2) The importance scores of factors influencing PM2.5 concentrations in the cities of the Yellow River Basin reveal that local PM10 concentration and surrounding urban PM2.5 concentration have relatively high importance scores, with average values of 0.5602 and 0.4254, respectively. Local CO and NO2 concentrations also exert some influence on PM2.5 concentrations, with their average values ranging from 0.2145 to 0.3438. In contrast, meteorological factors have a relatively weak impact on PM2.5 concentrations. (3) The key factors influencing future PM2.5 concentrations in the Yellow River Basin are local PM10 concentration, surrounding urban PM2.5 concentration, and local CO concentration, with predicted probabilities ranging from 0.3000 to 0.5907, 0.2353 to 0.4783, and 0.0591 to 0.1478, respectively.