摘要
运用统计分析和气象统计预报方法，对南昌市6个环境监测点的污染物观测资料和同期南昌市自动气象观测资料进行了统计，分季节研究了南昌市PM2.5质量浓度的变化规律，对影响大气污染的气象因子进行了综合分析，并分别建立夏季及非夏季PM2.5，质量浓度与气象要素的统计预报模式。结果表明：南昌市PM2.5污染非夏季较夏季更为严重，非夏季有超过一半的时间空气质量超出了国家二级标准。PM2.5质量月均浓度峰值出现在12月和1月，其中1月南昌市平均4d中就有3d超出国家二级标准。PM2.5浓度逐时变化呈现双峰型特征，对于夏季而言，2个峰值基本出现在9：00～11：00和22：00～24：00时段；而非夏季前一个峰值延后至11：00～12：00时段，后一个峰值往往提前至21：00～23：00，这主要与人类活动规律及太阳照射时间有关。建立的非夏季预报模式等级预报准确率较夏季高，对环境气象预报业务有一定实际应用价值。
Abstract： According to the pollutant data observed at six monitoring points in Nanchang and the automatic meteorological observation data of Nanchang in the same period, we used the methods of statistical analysis and meteorological statistical forecast to study the seasonal change law of PM2.5 mass concentration in Nanehang, carried out a comprehensive analysis of the main meteorological factors affecting the air pollution, and then established the statistical forecasting models of PM2.5 mass concentration in summer and non-summer. The results indicated that the average PM2.5 mass concentration in summer in Nanchang was lower than that in non-summer, and the air quality in over half of non-summer exceeded the national standard grade Ⅱ. The peak of monthly mean PM2.5 mass concentration appeared in December and January, and the air quality of Nanchang in 3/4 of January exceeded the national standard grade Ⅱ. The hourly variation in PM2.5 mass concentration was of bimodal distribution, and the two peak values appeared basically at 9 ： 00 - 11 ： 00 or 22 ： 00 - 24 ： 00 in summer and at 11 ： 00 - 12 ： 00 or 21 ： 00 - 23 ： 00 in non- summer, which was related to the law of human activity and sunlight time. The established non-summer forecast model of PM：.s had a higher accuracy than the summer forecast model, and it has a certain practical application value for environmental meteorological forecast.
出处
《江西农业学报》
CAS
2016年第9期95-101,共7页
Acta Agriculturae Jiangxi
基金
公益性行业(气象)可研专项(GYHY201406029
GYHY201306043)
江西省气象服务中心2015年课题"南昌空气污染物(PM2.5)浓度变化特征及预报模型研究"
关键词
PM2.5
南昌
污染特性
预报模式
PM2.5
Nanchang
Pollution characteristics
Forecast model
