郑州市大气污染时空分布特征分析及预测模型建立

Spatiotemporal distribution characteristics analysis and prediction model establishment of air pollution in Zhengzhou city

基于郑州市中心城区空气质量监测数据,结合环境空气质量指数(AQI),采用反距离权重插值法(IDW)对郑州市的空气质量进行评价和时空分析。选取气压、气温、湿度、风速、风向、降水量6项常规污染物监测项,以及前一日AQI值总计13项作为预测因子,应用BP和RBF神经网络方法对郑州市未来短期空气质量时空分布进行预测。结果表明:(1)2016年和2017年郑州市空气质量主要以优良和轻度污染为主;(2)PM 10 和PM 2.5 的年平均浓度分别超过我国环境空气质量标准中二级标准年平均浓度限值103.79%、142.03%;(3)管城区污染频率较高,多呈现持续性的污染天气状况,冬季时的极端污染天气多集中于惠济区;(4)BP、RBF神经网络预测模型可以较好地预测郑州市大气污染的时空分布状况。本研究结果对后续研究治理郑州市空气污染具有一定意义。

Based on the monitoring data of air quality on the city of Zhengzhou downtown, the paper evaluates the quality and analyzes the spatiotemporal distribution of the air by Ambient Air Quality Index (AQI) and inverse distance weight interpolation method. The paper forecasts the spatiotemporal distribution of the air quality of Zhengzhou City in the future short time using BP and RBF neural network methods by taking 13 items as predictors, including six conventional pollutant detection index (air pressure, temperature, humidity, wind speed, wind direction and precipitation), as well as the previous day’s AQI value .The studies show that:(1) The air quality of Zhengzhou City is mainly dominated by excellent and light pollution in 2016 and 2017;(2) The annual average concentration of PM 10 and PM 2.5 exceeds the limits of annual average concentration of China s secondary ambient air quality standards by 103.79% and 142.03%;(3) With continuous contaminated weather, the pollution frequency of Guancheng District is relatively high;and the extreme pollution weather in winter is concentrated in the Huiji District;(4) It’s perfect to predict the spatiotemporal distribution of pollution of Zhengzhou downtown using BP and RBF neural network prediction models. This is significant for the follow-up research on air pollution control.