气象因素对大气污染物影响的季节差异分析及预测模型对比——以深圳为例

以深圳为例,通过收集其2014年9月至2017年8月的地面气象观测数据、大气污染物PM_2.5和PM₁₀浓度数据及空气质量指数(AQI),利用SPSS软件分别构建线性回归模型与非线性多层神经网络(MLP)模型,探讨了气象因素对PM₁₀、PM_2.5及AQI影响的季节性差异,并利用构建的两种模型对PM_2.5、PM₁₀和AQI进行预测和对比。结果表明,气温、湿度、风速及风向均对深圳PM_2.5、PM₁₀及AQI有较大影响,且影响因素存在季节差异性;两种模型中,MLP模型在对复杂多变的空气质量预测上更具优越性。研究结果可为深圳空气质量优化以及城市局部和整体规划提供科学依据,并为其他城市空气质量模拟分析提供参考。

The seasonal differential effects of meteorological parameters on atmospheric pollutants and the prediction model comparison:a case study of Shenzhen

In this paper,Shenzhen was selected as research area.The ground surface meteorological observation data,PM2.5 ,PM10 and air quality index(AQI)at Shenzhen from September 2014 to August 2017 were collected to establish the linear regression model and multi-layer perception model(MLP)by using SPSS software.With the help of two established models,the seasonal differential effects of meteorological parameters on atmospheric pollutants were analyzed.Meanwhile,the variation of PM2.5 ,PM10 and AQI was also predicted by these two models,and predicted outcomes were compared.It was concluded that air temperature,humidity,wind speed and wind direction had great influence on PM2.5 ,PM10 and AQI of Shenzhen,and these influences presented significant seasonal differences.Of the two establishing models,MLP was better at solving complex and changeable problems such as air quality prediction.The results could provide scientific basis for Shenzhen air quality improvement and urban planning,and give reference for the air quality improvement of other cities.