基于LSTM-GCN的PM2.5浓度预测模型

应用机器学习算法开展空气质量预测已成为当前研究热点之一,空气质量监测数据具有显著的时空特征,即具有时间维度时序特征和空间维度传输演化特征。面向空气质量监测数据,联合LSTM提取的时间特征和GCN提取的空间特征,提出预测PM_2.5浓度的LSTM-GCN组合模型。以北京市35个空气质量监测站2018—2020年监测数据进行仿真实验,并将LSTM-GCN模型与LSTM模型、GCN模型以及时空地理加权回归模型(GTWR)进行对比,结果显示:LSTM-GCN模型相较于LSTM模型均方根误差(RMSE)、平均绝对误差(MAE)分别降低了11.68%、7.34%;相较于GCN模型RMSE、MAE分别降低了40.22%、36.37%;相较于GTWR模型RMSE、MAE分别降低了17.52%、23.69%,表明所提出LSTM-GCN模型在准确率上有所提升。用LSTM-GCN模型预测2021年1—7月PM_2.5浓度,结果显示预测效果较好。     

利用Keras构建神经网络在空气质量预测中的应用

利用python语言搭建了一整套空气质量神经网络预测系统,底层利用Keras设计并建立了基于tensorflow的神经网络模型。选择日平均气压、日平均气温、日平均相对湿度、日降水量、日平均风速、前1日空气质量因子监测数据等因素作为模型输入变量,分别针对广东省所有监测站点和地市的空气质量因子(PM_(2.5)、PM_(10)、NO_2、SO_2、CO、O_3、AQI)进行预测,结果表明,7个因子的地市平均相对误差值为16.15%～27.7%,地市相关系数为0.36～0.77,该模型在城市空气质量预测中具有良好的效果。     

基于自回归GA-BP神经网络的AQI预测

基于徐州市2013年12月—2018年11月的空气质量指数日均值,建立了时间序列自回归输入的GA-BP神经网络模型用于空气质量指数预测。结果表明,所建立的网络模型能够准确预测徐州市空气质量指数的变化趋势,其中夏季预测相对误差18. 23%,仿真均方根误差(RMSE)为14. 59;冬季预测相对误差9. 14%,仿真RMSE为11. 47。     

区域PM2.5时空回归建模与预测

基于区域PM_(2.5)时空建模和预测的需要及PM_(2.5)浓度呈现明显的时空分布趋势的状况,以苏南地区2014年PM_(2.5)日监测数据为实验数据,使用回归克里格对区域PM_(2.5)进行时空建模和估值。利用最小二乘法建立了PM_(2.5)与时空位置的三元二次回归趋势模型,建模点趋势值与实测值间的平均误差接近于0,表明趋势模型拟合效果较好;拟合了样点残差的理论变异函数模型,表明该地区PM_(2.5)的空间和时间相关性范围分别为150 km和4 d;基于该模型,使用时空普通克里格对残差进行时空插值;插值结果与趋势项相加,得到PM_(2.5)回归克里格估值结果;通过对比不考虑趋势的时空普通克里格估值结果,发现考虑时空趋势的时空回归克里格法精度提高了1. 29%。对所提方法进行了创新性分析,并对不足之处进行了讨论。     

粤港珠江三角洲区域空气质量分级技术及信息发布方法研究

根据珠江三角洲空气污染特征.将,SO2、NO2、PM10、O3四项参数作为评价指标,对建立区域空气质量指数(Regional Air Quality Index,RAQI)~型及区域空气质量分级方法,以及开展区域空气质量信息发布的空间表征技术进行了研究,以便直观有效地反映珠江三角洲区域空气质量状况.     

基于深度学习的空气质量预报方法新进展

空气质量预报与人们的日常生活密切相关,其基本思想是分析历史空气质量数据,发现其内在的时空相关性,结合未来气象信息以及污染源排放量,对未来的空气质量进行预测。目前,环境管理和社会公众服务对空气质量预报提出了长时间、多维度、高精度的预测要求,一些新型的空气质量预测方法仍处于起步探索阶段。近年来,随着人工智能的普及与推广(特别是云计算与大数据的发展),深度学习这项基于传统人工神经网络的技术被国内外研究者所重视。笔者对现有典型的空气质量预报方法进行了阐述,包括数值预测模型方法、统计预报模型方法、基于机器学习模型的预测方法等,并重点介绍了该领域最新进展:基于深度学习模型的预测方法,并在此基础上进行了总结与展望。     

基于OPAQ的城市空气质量预报系统研究

空气质量预测在国内的关注度日益提高,传统的空气质量预测系统通常运用数值化学传输模型,利用物理方程来计算污染物的扩散、沉降和化学反应。而化学传输模型的预测准确性很大程度上需要依赖详细的污染源排放信息和气象模型的输出结果。基于统计模型的OPAQ空气质量预报业务系统,采用人工神经网络算法,可预测各污染物的日均值或日最大值。并对北京空气质量预报的结果进行了评价,OPAQ空气质量预报业务系统对空气质量预测的准确性较高,能够利用较低的计算资源得到较为准确的预测结果。与数值预报相比,OPAQ空气质量预报业务系统不需要大量的基础数据作为输入,可弥补数值预报的不足,并成为数值预报的有力补充。     

基于复杂网络及其模体的京津冀及周边城市空气污染空间关联与城市协同治理

为了探究京津冀及周边31个城市空气污染的空间关联与季节演化情况,于2015年1月1日至2021年12月31日选取该区域的空气质量指数(AQI)日均值作为样本,首先对31个城市AQI的时间变化特征和空间分布特征进行了分析;然后计算不同城市AQI的皮尔逊相关系数,结合引力模型,构建了该区域空气污染空间关联网络;最后对网络的整体特征与季节演化情况进行了分析。结果表明,该区域空气污染空间关联网络密度较高,关联比较紧密,度值与中心性较高的城市对空气污染有更高的传输贡献,石家庄、邢台及邯郸的中心性最高,对其他城市空气污染传输的控制能力最强;网络共分为3个凝集子群,石家庄、邢台及邯郸位于子群发生关联的中心位置;对4个季节空间关联网络图进行对比分析,春季网络密度更高,不同城市之间AQI的关联更为紧密;在模体A的四季关联网络中,石家庄、邢台和邯郸为主要传播城市;在模体B的四季关联网络中,邯郸和北京为枢纽城市,这些城市在空气污染的传播过程中起着重要作用。提出,加强石家庄、邢台、邯郸、北京这4个城市与其他城市空气污染协同治理力度,完善监督管理机制,促进环保产业发展;从行政、法律、经济等多方面入手,为解决京津冀大气污染问题提供充分保障;结合京津冀地区的季节气候条件,加强冬季煤改电、煤改气等管理力度,实行区域或集中采暖供热,减少燃料消耗和烟尘排放。     

基于复合LSTM模型的PM2.5浓度预测

在对淄博市19个空气质量监测站点监测数据进行分析后，提出了一种基于机器学习的复合模型——灰色关联度分析(GRA)-改进的完备总体经验模态分解(ICEEMD)-长短期记忆网络(LSTM)模型。通过分析淄博市2019年大气污染物和气象数据，选用LSTM模型预测PM_2.5浓度。由于传统单一模块机器学习模型具有训练时间较长和预测精度较低的问题，提出了复合LSTM模型。该模型由3部分组成：GRA,用于PM_2.5浓度影响因素变量筛选；ICEEMD,用于PM_2.5分解、分量筛选和原始大气污染物及气象数据处理；LSTM,用于PM_2.5浓度预测。预测结果表明：淄博市中部丘陵地带PM_2.5浓度高于南部山区和北部平原，东部高于西部；淄博市逐月PM_2.5浓度呈“U”形分布，1月最高，8月最低；淄博市PM_2.5浓度受PM₁₀和CO影响较大，受湿度和温度影响较小。对比单一LSTM模型和GRA-LSTM模型，GRA-ICEEMD-LSTM模型...     

基于改进U-Net网络与联合损失函数的海南自然保护区高分辨率遥感变化检测模型

海南自然保护区遥感监测对森林资源监测、生态保护及热带亚热带地表研究具有重要意义。基于深度学习方法,针对海南省自然保护区大范围变化检测问题,对U-Net网络结构进行了改进,在每一个卷积层后加入标准化层,以跳线连接的形式将原有卷积模块改进为优化模块,同时在编码器底端添加金字塔池化模块以更好提取全局信息,形成了改进U-Net网络模型。模型训练采用基于交叉熵损失函数和广义骰子损失函数构建的联合损失函数,配合多种优化策略实现端到端的地物变化信息提取。该模型应用于公开数据集和研究构建的海南自然保护区数据集的变化检测任务,总体精度分别为97.21%(Kappa系数0.88)和95.12%(Kappa系数0.90),相比原始U-Net效果提升显著。     

支撑京津冀区域大气污染联防联控的大气监测体系构建

京津冀区域已成为全国大气污染最严重的地区之一,并且呈现出明显的区域性污染特征,加强区域间的环境合作,实施区域联防联控是解决京津冀区域大气污染问题的有效手段。对照《生态环境监测网络建设方案》的要求,目前京津冀区域大气监测体系还存在着监测网络不健全、监测项目覆盖不全面、监测新技术应用不足、质控体系不完善、信息产品供给与公众需求有差距等问题,与京津冀区域大气污染联防联控的要求不相适应。为全面提升京津冀区域大气监测体系对区域联防联控的技术支撑能力,亟需进一步完善京津冀区域的监测网络,增设传输研究、污染监控等特殊监测点位;逐步拓展监测项目,开展颗粒物化学组分和O_3前体物监测;加快遥感监测等立体监测技术在大气监测中的应用,全面分析污染物时空分布特征与传输规律;加强空气质量预报预警能力建设,为重污染天气应对提供技术支持;规范监测事权上收后的全过程质控体系,建立颗粒物标准方法比对和O_3量值传递质控机制;构建环境监测大数据平台,加强数据分析应用与信息公开。     

上海市工业区大气污染预警监测信息平台研究与设计

根据上海市工业区大气污染排放特征,在构建大气特征污染监控网络的基础上,搭建了工业区大气污染预警监测信息平台,并研究了适用于大气特征污染自动站的通信传输规范,确保数据联网和应用。平台设置了大气质量监控、大气数据管理、污染源一厂一档、大气质量分析评价、污染源溯源分析、应急监测数据管理、智能移动终端查询等7个子系统,可为上海市工业区大气污染预警和管控提供实时监测数据。     

珠江三角洲地区空气自动监控网络数据空间模拟适用方法

珠江三角洲地区已构建起区域性空气质量自动监测网络,由于初步建立的自动监控网络监测站点对整个区域而言极其有限,如以单离散数据形式发布则不能反映出区域空气污染的连续变化特征和全貌.对有限点位监测结果进行空间模拟研究才能满足反映区域性污染状况的需求.通过运用、比对、评价不同空间插值模型于自动监控网络离散站点的模拟应用,提出了珠江三角洲自动监控网络数据空间模拟适用方法.     

新标准下环境空气质量监控网络完善实践 ——以珠三角地区佛山市为例

介绍珠三角及佛山市新标准监测网络结构,从拓宽网络有效覆盖范围、提升网络对污染研究的支撑能力、以及巩固数据实时发布工作基础等多角度分析了网络完善需求,提出了下一阶段网络建设的重点措施建议,并介绍了佛山市目前计划进行的网络完善工作。     

广州市天地空一体化环境空气质量监测网络的建设

简述了原有广州市环境空气自动监测网络情况,通过对环境空气自动监测网络的集成优化,增设站点、扩展监测项目和引入新技术,建立了一个"金字塔形"的集天地空一体化的环境空气监测网络,弥补了原有监测网络在反映环境空气质量总体水平、污染来源解析、预报预警所需的环境质量信息、对环境空气质量的评估等方面的不足,该网络已成功应用于广州亚运会的空气质量保障。     

Site location optimization of regional air quality monitoring network in China: methodology and case study

Regional air quality monitoring networks (RAQMN) are urgently needed in China due to increasing regional air pollution in city clusters, arising from rapid economic development in recent decades. This paper proposes a methodological framework for site location optimization in designing a RAQMN adapting to air quality management practice in China. The framework utilizes synthetic assessment concentrations developed from simulated data from a regional air quality model in order to simplify the optimal process and to reduce costs. On the basis of analyzing various constraints such as cost and budget, terrain conditions, administrative district, population density and spatial coverage, the framework takes the maximum approximate degree as an optimization objective to achieve site location optimization of a RAQMN. An expert judgment approach was incorporated into the framework to help adjust initial optimization results in order to make the network more practical and representative. A case study was used to demonstrate the application of the framework, indicating that it is feasible to conduct site optimization for a RAQMN design in China. The effects of different combinations of primary and secondary pollutants on site location optimization were investigated. It is suggested that the network design considering both primary and secondary pollutants could better represent regional pollution characteristics and more extensively reflect temporal and spatial variations of regional air quality. The work shown in this study can be used as a reference to guide site location optimization of a RAQMN design in China or other regions of the world.     

LaSVM-based big data learning system for dynamic prediction of air pollution in Tehran

Due to critical impacts of air pollution, prediction and monitoring of air quality in urban areas are important tasks. However, because of the dynamic nature and high spatio-temporal variability, prediction of the air pollutant concentrations is a complex spatio-temporal problem. Distribution of pollutant concentration is influenced by various factors such as the historical pollution data and weather conditions. Conventional methods such as the support vector machine (SVM) or artificial neural networks (ANN) show some deficiencies when huge amount of streaming data have to be analyzed for urban air pollution prediction. In order to overcome the limitations of the conventional methods and improve the performance of urban air pollution prediction in Tehran, a spatio-temporal system is designed using a LaSVM-based online algorithm. Pollutant concentration and meteorological data along with geographical parameters are continually fed to the developed online forecasting system. Performance of the system is evaluated by comparing the prediction results of the Air Quality Index (AQI) with those of a traditional SVM algorithm. Results show an outstanding increase of speed by the online algorithm while preserving the accuracy of the SVM classifier. Comparison of the hourly predictions for next coming 24 h, with those of the measured pollution data in Tehran pollution monitoring stations shows an overall accuracy of 0.71, root mean square error of 0.54 and coefficient of determination of 0.81. These results are indicators of the practical usefulness of the online algorithm for real-time spatial and temporal prediction of the urban air quality.     

Forecasting Air Quality Parameters Using Hybrid Neural Network Modelling

Urban air pollution has emerged as an acute problem in recent years because of its detrimental effects on health and living conditions. The research presented here aims at attaining a better understanding of phenomena associated with atmospheric pollution, and in particular with aerosol particles. The specific goal was to develop a form of air quality modelling which can forecast urban air quality for the next day using airborne pollutant, meteorological and timing variables.Hourly airborne pollutant and meteorological averages collected during the years 1995–1997 were analysed in order to identify air quality episodes having typical and the most probable combinations of air pollutant and meteorological variables. This modelling was done using the Self-Organising Map (SOM) algorithm, Sammon's mapping and fuzzy distance metrics. The clusters of data that were found were characterised by statistics. Several overlapping Multi-Layer Perceptron (MLP) models were then applied to the clustered data, each of which represented one pollution episode. The actual levels for individual pollutants could then be calculated using a combination of the MLP models which were appropriate in that situation.The analysis phase of the modelling gave clear and intuitive results regarding air quality in the area where the data had been collected. The resulting forecast showed that the modelling of gaseous pollutants is more reliable than that of the particles.     

Air quality assessment of Estarreja, an urban industrialized area, in a coastal region of Portugal

Despite the increasing concern given to air quality in urban and industrial areas in recent years, particular emphasis on regulation, control, and reduction of air pollutant emissions is still necessary to fully characterize the chain emissions–air quality–exposure–dose–health effects, for specific sources. The Estarreja region was selected as a case study because it has one of the largest chemical industrial complexes in Portugal that has been recently expanded, together with a growing urban area with an interesting location in the Portuguese coastland and crossed by important road traffic and rail national networks. This work presents the first air quality assessment for the region concerning pollutant emissions and meteorological and air quality monitoring data analysis, over the period 2000–2009. This assessment also includes a detailed investigation and characterization of past air pollution episodes for the most problematic pollutants: ozone and PM10. The contribution of different emission sources and meteorological conditions to these episodes is investigated. The stagnant meteorological conditions associated with local emissions, namely industrial activity and road traffic, are the major contributors to the air quality degradation over the study region. A set of measures to improve air quality—regarding ozone and PM10 levels—is proposed as an air quality management strategy for the study region.