A New Model Using Multiple Feature Clustering and Neural Networks for Forecasting Hourly PM2.5 Concentrations,and Its Applications in China

Particulate matter with an aerodynamic diameter no greater than 2.5 lm(PM2.5)concentration forecasting is desirable for air pollution early warning.This study proposes an improved hybrid model,named multi-feature clustering decomposition(MCD)–echo state network(ESN)–particle swarm optimization(PSO),for multi-step PM2.5 concentration forecasting.The proposed model includes decomposition and optimized forecasting components.In the decomposition component,an MCD method consisting of rough sets attribute reduction(RSAR),k-means clustering(KC),and the empirical wavelet transform(EWT)is proposed for feature selection and data classification.Within the MCD,the RSAR algorithm is adopted to select significant air pollutant variables,which are then clustered by the KC algorithm.The clustered results of the PM2.5 concentration series are decomposed into several sublayers by the EWT algorithm.In the optimized forecasting component,an ESN-based predictor is built for each decomposed sublayer to complete the multi-step forecasting computation.The PSO algorithm is utilized to optimize the initial parameters of the ESN-based predictor.Real PM2.5 concentration data from four cities located in different zones in China are utilized to verify the effectiveness of the proposed model.The experimental results indicate that the proposed forecasting model is suitable for the multi-step high-precision forecasting of PM2.5 concentrations and has better performance than the benchmark models.

一种最优多模式集成方法在我国重污染区域PM2.5浓度预报中的应用

为了提高我国重污染区域PM2.5浓度预报准确率,基于4套国家级以及区域环境气象业务中心发展和维护的空气质量数值预报模式,通过均值集成、权重集成、多元线性回归集成和BP-ANNs集成分别建立集成预报,在实时预报效果评估基础上,建立了最优多模式集成预报。对2015—2016年预报效果进行评估,结果表明:相对于单个空气质量数值预报模式,均值和权重集成对预报偏差的改进幅度有限,但多元线性回归、BP-ANNs和最优集成能较大幅度降低预报偏差;最优集成预报与观测值间的归一化平均偏差(NMB)和均方根误差(RMSE)分别为-10%~10%和10~70μg/m^3,且在更多的站点表现出强相关性,但依然低估了高污染等级的PM2.5浓度。对2018年2月25日—3月4日京津冀地区污染过程进行评估,结果表明:最优集成能较好预报出该过程中PM2.5浓度的变化趋势和量级;在北京、石家庄和郑州3个代表城市中,预报和观测值间的NMB和相关系数(R)分别为-26%^-4%和0.49~0.77;最优集成对轻度污染及中度污染的TS评分为0.39~0.73,重度污染及以上TS评分为0.13~0.30,能为预报员提供客观参考,但对污染峰值的预报能力还需进一步改进。

Overview of Urban PM_(2. 5) Numerical Forecast Models in China

This paper made an overview and introduction of urban PM_(2. 5)numerical forecast models in China,and mainly introduced air quality simulated forecast system of Beijing,Shanghai,and Nanjing. On this basis,it discussed development direction and existing problems of urban PM_(2. 5)forecast models in China. Besides,it revealed significance of numerical models for air quality forecast. In a heavy air pollution of Beijing- Tianjin- Hebei in October 6- 12 th of 2014,the forecast results indicated that pollutants was transported from south to north,so the regional transport exerts great influence on concentration of PM_(2. 5).

江苏省级区域空气质量数值预报模式效果评估

采用中国科学院大气物理研究所开发的嵌套网格空气质量模式系统(NAQPMS),搭建江苏省级区域空气质量数值预报模式系统,并测试了该系统对2013年夏季江苏省PM2.5质量浓度未来24 h预报以及7 d潜势预测的效果。结果表明,该系统成功应用于江苏省的空气质量预报;所有地市的24 h预报效果均在合理范围内(平均分数偏差小于±60%且平均分数误差小于75%);7 d潜势预测效果比24 h预报效果略差,整体能准确把握PM2.5质量浓度的变化趋势。

气象参数对基于BP神经网络的PM_(2.5)日均值预报模型的影响

建立了基于BP神经网络的PM_(2.5)质量浓度预报模型,对广州市5个监测点2012年6月—2013年5月的PM_(2.5)质量浓度日均值进行预报,分析了总体预报误差、不同风速和降雨量下的预报误差,以及天气预报误差对PM_(2.5)质量浓度预报误差的影响。结果表明,BP神经网络模型对5个站点的PM_(2.5)预报结果稳定,平均相对误差为29.71%。在有利于PM_(2.5)扩散的气象条件下预报误差较大,风速较大时与风速较小时预报误差的差异高达15%,而不同降雨量情况下的预报误差较相近。修正天气预报后,各站点的预报误差平均降低了4.67%。这表明可从空气质量数据质量等方面入手改进模型。

Numerical simulation of an extreme haze pollution event over the North China Plain based on initial and boundary condition ensembles

The North China Plain often su ers heavy haze pollution events in the cold season due to the rapid industrial development and urbanization in recent decades.In the winter of 2015,the megacity cluster of Beijing Tianjin Hebei experienced a seven-day extreme haze pollution episode with peak PM2.5(particulate matter(PM)with an aerodynamic diameter≤2.5μm)concentration of 727μg m 3.Considering the in uence of meteorological conditions on pollu-tant evolution,the e ects of varying initial conditions and lateral boundary conditions(LBCs)of the WRF-Chem model on PM2.5 concentration variation were investigated through ensemble methods.A control run(CTRL)and three groups of ensemble experiments(INDE,BDDE,INBDDE)were carried out based on difierent initial conditions and LBCs derived from ERA5 reanalysis data and its 10 ensemble members.The CTRL run reproduced the meteorological conditions and the overall life cycle of the haze event reasonably well,but failed to capture the intense oscillation of the instantaneous PM2.5 concentration.However,the ensemble forecasting showed a considerable advantage to some extent.Compared with the CTRL run,the root-mean-square error(RMSE)of PM2.5 concentration decreased by 4.33%,6.91%,and 8.44%in INDE,BDDE and INBDDE,respectively,and the RMSE decreases of wind direction(5.19%,8.89%and 9.61%)were the dominant reason for the improvement of PM2.5 concentration in the three ensemble experiments.Based on this case,the ensemble scheme seems an e ective method to improve the prediction skill of wind direction and PM2.5 concentration by using the WRF-Chem model.

A Self-Organizing Memory Neural Network for Aerosol Concentration Prediction

Haze-fog,which is an atmospheric aerosol caused by natural or man-made factors,seriously affects the physical and mental health of human beings.PM2.5(a particulate matter whose diameter is smaller than or equal to 2.5 microns)is the chief culprit causing aerosol.To forecast the condition of PM2.5,this paper adopts the related the meteorological data and air pollutes data to predict the concentration of PM2.5.Since the meteorological data and air pollutes data are typical time series data,it is reasonable to adopt a machine learning method called Single Hidden-Layer Long Short-Term Memory Neural Network(SSHL-LSTMNN)containing memory capability to implement the prediction.However,the number of neurons in the hidden layer is difficult to decide unless manual testing is operated.In order to decide the best structure of the neural network and improve the accuracy of prediction,this paper employs a self-organizing algorithm,which uses Information Processing Capability(IPC)to adjust the number of the hidden neurons automatically during a learning phase.In a word,to predict PM2.5 concentration accurately,this paper proposes the SSHL-LSTMNN to predict PM2.5 concentration.In the experiment,not only the hourly precise prediction but also the daily longer-term prediction is taken into account.At last,the experimental results reflect that SSHL-LSTMNN performs the best.

Air Quality Indices, Sources and Impact on Human Health of PM₁₀and PM_2.5in Alexandria Governorate, Egypt

In this study, PM10 and PM2.5 were measured in seven sites representing different activities (the same sites of EEAA monitoring stations) in addition to eighth site that used as a background. All results were higher than AQLs of EEAA, US/EPA, and EC although PM10 and PM2.5 are considered to be a direct cause of cardiovascular diseases as well as lead to death and it may be a reason for a number of chest diseases in short-term as well as long-term. Results were compared to the Air Quality Forecast system which developed by EEAA and AQI which created by US/EPA was calculated for some PM10 and PM2.5. Probable potential anthropogenic sources for such high concentrations of PM included unpaved roads, indiscriminate demolition and construction work, industrial activities, and solid wastes. This study resulted in a number of suggestions and recommendations include: 1) Implementation of integrated ISO 26000 and ISO 14001, 2) EIMP/EEAA monitoring stations need restructuring plan to cover all areas in Alexandria, 3) EIMP/EEAA must be supported with PM2.5 monitors, 4) PM control systems must be used in all industrial activities to reduce PM pollution from the source, 5) AQL of PM2.5 in the ambient environment must be reduced and it must be included in the working environment parameters, 6) Environmental law must be applied strictly, and 7) Multidisciplinary co-operation especially between environment and public health specialists must be increased.