Deep spatial-temporal fusion network for fine-grained air pollutant concentration prediction

2021 ◽  
Vol 25 (2) ◽  
pp. 419-438
Author(s):  
Liang Ge ◽  
Kunyan Wu ◽  
Feng Chang ◽  
Aoli Zhou ◽  
Hang Li ◽  
...  
Keyword(s):  
Air Quality ◽  
Environmental Governance ◽  
Tensor Decomposition ◽  
Heterogeneous Data ◽  
Air Pollutant ◽  
Pollutant Concentration ◽  
Air Quality Prediction ◽  
Region Selection ◽  
Similar Region ◽  
Data Completion

Air pollution is a serious environmental problem that has attracted much attention. Predicting air pollutant concentration can provide useful information for urban environmental governance decision-making and residents’ daily health control. However, existing methods fail to model the temporal dependencies or have suffer from a weak ability to capture the spatial correlations of air pollutants. In this paper, we propose a general approach to predict air pollutant concentration, named DSTFN, which consists of a data completion component, a similar region selection component, and a deep spatial-temporal fusion network. The data completion component uses tensor decomposition method to complete the missing data of historical air quality. The similar region selection component uses region metadata to calculate the spatial similarity between regions. The deep spatial-temporal fusion network fuses urban heterogeneous data to capture factors affecting air quality and predict air pollutant concentration. Extensive experiments on a real-world dataset demonstrate that our model achieves the highest performance compared with state-of-the-art models for air quality prediction.

scholarly journals Multi-Group Encoder-Decoder Networks to Fuse Heterogeneous Data for Next-Day Air Quality Prediction

2019 ◽  
Author(s):  
Yawen Zhang ◽  
Qin Lv ◽  
Duanfeng Gao ◽  
Si Shen ◽  
Robert Dick ◽  
...  
Keyword(s):  
Air Quality ◽  
Heterogeneous Data ◽  
Air Pollutant ◽  
Weather Data ◽  
Quality Prediction ◽  
Prevention Measures ◽  
Convolution Operators ◽  
Real World Data ◽  
Air Quality Prediction ◽  
Cross Domain

Accurate next-day air quality prediction is essential to enable warning and prevention measures for cities and individuals to cope with potential air pollution, such as vehicle restriction, factory shutdown, and limiting outdoor activities. The problem is challenging because air quality is affected by a diverse set of complex factors. There has been prior work on short-term (e.g., next 6 hours) prediction, however, there is limited research on modeling local weather influences or fusing heterogeneous data for next-day air quality prediction. This paper tackles this problem through three key contributions: (1) we leverage multi-source data, especially high-frequency grid-based weather data, to model air pollutant dynamics at station-level; (2) we add convolution operators on grid weather data to capture the impacts of various weather parameters on air pollutant variations; and (3) we automatically group (cross-domain) features based on their correlations, and propose multi-group Encoder-Decoder networks (MGED-Net) to effectively fuse multiple feature groups for next-day air quality prediction. The experiments with real-world data demonstrate the improved prediction performance of MGED-Net over state-of-the-art solutions (4.2% to 9.6% improvement in MAE and 9.2% to 16.4% improvement in RMSE).

scholarly journals A DBN-Based Deep Neural Network Model with Multitask Learning for Online Air Quality Prediction

2019 ◽  
Vol 2019 ◽  
pp. 1-9 ◽  
Author(s):  
Jiangeng Li ◽  
Xingyang Shao ◽  
Rihui Sun
Keyword(s):  
Neural Network ◽  
Air Quality ◽  
Network Model ◽  
Neural Network Model ◽  
Prediction Accuracy ◽  
Deep Neural Network ◽  
Multitask Learning ◽  
Air Pollutant ◽  
Quality Prediction ◽  
Air Quality Prediction

To avoid the adverse effects of severe air pollution on human health, we need accurate real-time air quality prediction. In this paper, for the purpose of improve prediction accuracy of air pollutant concentration, a deep neural network model with multitask learning (MTL-DBN-DNN), pretrained by a deep belief network (DBN), is proposed for forecasting of nonlinear systems and tested on the forecast of air quality time series. MTL-DBN-DNN model can solve several related prediction tasks at the same time by using shared information contained in the training data of different tasks. In the model, DBN is used to learn feature representations. Each unit in the output layer is connected to only a subset of units in the last hidden layer of DBN. Such connection effectively avoids the problem that fully connected networks need to juggle the learning of each task while being trained, so that the trained networks cannot get optimal prediction accuracy for each task. The sliding window is used to take the recent data to dynamically adjust the parameters of the MTL-DBN-DNN model. The MTL-DBN-DNN model is evaluated with a dataset from Microsoft Research. Comparison with multiple baseline models shows that the proposed MTL-DBN-DNN achieve state-of-art performance on air pollutant concentration forecasting.

scholarly journals Numerical Simulation to Air Pollution Emission Control near an Industrial Zone

2017 ◽  
Vol 2017 ◽  
pp. 1-7 ◽  
Author(s):  
Pravitra Oyjinda ◽  
Nopparat Pochai
Keyword(s):  
Air Pollution ◽  
Air Quality ◽  
Industrial Development ◽  
Air Pollutant ◽  
Pollutant Concentration ◽  
Air Pollution Control ◽  
Industrial Zone ◽  
Concentration Data ◽  
Monitoring Point ◽  
Concentration Levels

A rapid industrial development causes several environment pollution problems. One of the main problems is air pollution, which affects human health and the environment. The consideration of an air pollutant has to focus on a polluted source. An industrial factory is an important reason that releases the air pollutant into the atmosphere. Thus a mathematical model, an atmospheric diffusion model, is used to estimate air quality that can be used to describe the sulfur dioxide dispersion. In this research, numerical simulations to air pollution measurement near industrial zone are proposed. The air pollution control strategies are simulated to achieve desired pollutant concentration levels. The monitoring points are installed to detect the air pollution concentration data. The numerical experiment of air pollution consisted of different situations such as normal and controlled emissions. The air pollutant concentration is approximated by using an explicit finite difference technique. The solutions of calculated air pollutant concentration in each controlled and uncontrolled point source at the monitoring points are compared. The air pollutant concentration levels for each monitoring point are controlled to be at or below the national air quality standard near industrial zone index.

Simulation of Air Quality and Pollution Control Countermeasures of Chengdu Urban Agglomeration

2012 ◽  
Vol 610-613 ◽  
pp. 1387-1397 ◽  
Author(s):  
Wen Yong Wang ◽  
Nan Chen ◽  
Xiao Juan Ma
Keyword(s):  
Air Quality ◽  
Sulfur Dioxide ◽  
Pollution Sources ◽  
Air Pollutant ◽  
Urban Agglomeration ◽  
Pollutant Concentration ◽  
Pollutant Emission ◽  
Air Quality Model ◽  
Quality Model ◽  
Allowable Limit

The CMAQ model (Community Multiscale Air Quality model) was used to stimulate the atmospheric environmental quality of Chengdu urban agglomeration. The result shows that air pollutant concentration in some zones of the urban agglomeration is higher than the allowable limit of the national grade II standard. Fortunately, such zones only cover a small area. Zones where the average daily and annual PM10 concentration is higher than the allowable limit only account for 4% of the total area of Chengdu urban agglomeration. Less than 1% of the total area has the concentration of other pollutants higher than the limit. Zones with pollutant concentration higher than the limit are mainly distributed in Chengdu City, Mianyang City, and Meishan City. Pollutants emitted from the cities of Chengdu urban agglomeration shift on to and interact with each other. Therefore, the air pollutant concentration of one city is partially attributable to pollutants emitted from its own pollution sources and a part of or even most of it results from pollutants from other cities. For example, regarding PM10 in air of Deyang City, only 12% comes from its own pollution sources, and 55% comes from pollution sources of Chengdu, and the rest 29% comes from pollution sources of Mianyang. Regarding Sulfur dioxide in air of Chengdu, 59% comes from local pollution sources of Chengdu and 23% comes from pollution sources of Deyang. Other pollutants are also subject to such a rule. As in the urban agglomeration, there are zones where pollutant concentration is higher than the allowable limit, the existing pollution sources must be further controlled by setting reduction target according to the total capacity. The pollutant emission should be reduced by means of eliminating backward productivity, adjusting structure and layout of industries, and controlling pollution sources in depth to effectively improve the regional environmental air quality. At the same time, as pollutants emitted from the cities interact with each other, the 5 cities must sign a joint prevention and control agreement to collaborate in control of sulfur dioxide, nitrogen oxides, smoke and dust, and organic pollutants.

scholarly journals A numerical study of the contribution to the air pollutant in Beijing during CAREBeijing-2006

2011 ◽  
Vol 11 (2) ◽  
pp. 5271-5312 ◽  
Author(s):  
Q. Z. Wu ◽  
Z. F. Wang ◽  
A. Gbaguidi ◽  
C. Gao ◽  
L. N. Li ◽  
...  
Keyword(s):  
Surface Layer ◽  
Air Quality ◽  
Rural Areas ◽  
Air Pollutants ◽  
High Performance ◽  
Numerical Study ◽  
Air Pollutant ◽  
Air Quality Prediction ◽  
Regional Sources ◽  
The Impact

Abstract. An online air pollutant tagged module has been developed in the Nested Air Quality Prediction Model System (NAQPMS) to investigate the impact of local and regional sources on the air pollutants in Beijing during the Campaign of Air Quality Research in Beijing 2006 (CAREBeijing-2006). The NAQPMS model shows high performance in simulating sulfur dioxide (SO2), particulate matter (PM10), nitrogen dioxide (NO2), and ozone (O3) with overall better agreements with the observations at urban sites than rural areas. With the tagged module, the air pollutant contributions from local and regional sources to the surface layer (about 30 m) and the upper layer (about 1.1 km) in Beijing are differentiated and estimated. The air pollutants at the surface layer in Beijing are dominated by the contributions from local sources, accounting for 65% of SO2, 75% of PM10 and nearly 90% of NO2, respectively, comparatively, the upper layer has large source contributions from the surrounding regions (e.g., southern Beijing), accounting for more than 50% of the SO2 and PM10 concentrations. Country scale analysis is also performed and the results suggest that Tianjin is the dominant source of SO2 in Pinggu County, and Langfang, Hebei is the most important regional contributor to PM10 in Beijing. Moreover, the surrounding regions show larger impact on SO2, PM10 and NO2 in the eastern counties of Beijing (e.g., Pinggu, Tongzhou and Daxing) than those in western Beijing, which is likely due to the Beijing's semi-basin topography and the summer monsoon. Our results indicate that the efforts to control the air pollutants in Beijing should focus on controlling both local and regional emissions.

scholarly journals HADOOP-BASED DISTRIBUTED SYSTEM FOR ONLINE PREDICTION OF AIR POLLUTION BASED ON SUPPORT VECTOR MACHINE

2015 ◽  
Vol XL-1-W5 ◽  
pp. 215-219 ◽  
Author(s):  
Z. Ghaemi ◽  
M. Farnaghi ◽  
A. Alimohammadi
Keyword(s):  
Air Pollution ◽  
Support Vector Machine ◽  
Data Analysis ◽  
Air Quality ◽  
Large Scale ◽  
Processing Time ◽  
Air Pollutant ◽  
Pollutant Concentration ◽  
Support Vector ◽  
Prediction Problems

The critical impact of air pollution on human health and environment in one hand and the complexity of pollutant concentration behavior in the other hand lead the scientists to look for advance techniques for monitoring and predicting the urban air quality. Additionally, recent developments in data measurement techniques have led to collection of various types of data about air quality. Such data is extremely voluminous and to be useful it must be processed at high velocity. Due to the complexity of big data analysis especially for dynamic applications, online forecasting of pollutant concentration trends within a reasonable processing time is still an open problem. The purpose of this paper is to present an online forecasting approach based on Support Vector Machine (SVM) to predict the air quality one day in advance. In order to overcome the computational requirements for large-scale data analysis, distributed computing based on the Hadoop platform has been employed to leverage the processing power of multiple processing units. The MapReduce programming model is adopted for massive parallel processing in this study. Based on the online algorithm and Hadoop framework, an online forecasting system is designed to predict the air pollution of Tehran for the next 24 hours. The results have been assessed on the basis of Processing Time and Efficiency. Quite accurate predictions of air pollutant indicator levels within an acceptable processing time prove that the presented approach is very suitable to tackle large scale air pollution prediction problems.

scholarly journals Evaluation of Relationship Between Air Pollutant Concentration and Meteorological Elements in Winter Months

2014 ◽  
Vol 22 (1) ◽  
pp. 25-32 ◽  
Author(s):  
Andrzej Żyromski ◽  
Małgorzata Biniak-Pieróg ◽  
Ewa Burszta-Adamiak ◽  
Zenon Zamiar
Keyword(s):  
Air Quality ◽  
Wind Speed ◽  
Relative Humidity ◽  
Air Temperature ◽  
Winter Season ◽  
Air Pollutant ◽  
Pollutant Concentration ◽  
Pollution Monitoring ◽  
Air Pollution Monitoring ◽  
Meteorological Elements

Abstract The paper presents the evaluation of the relation between meteorological elements and air pollutants’ concentrations. The analysis includes daily concentrations of pollutants and variation of meteorological elements such as wind speed, air temperature and relative humidity, precipitation and total radiation at four monitoring stations located in the province of Lower Silesia in individual months of the winter half-year (November–April, according to hydrological year classification) of 2005–2009. Data on air quality and meteorological elements came from the results of research conducted in the automatic net of air pollution monitoring conducted in the range of the State Environment Monitoring. The effect of meteorological elements on analysed pollutant concentration was determined using the correlation and regression analysis at significance level α < 0.05. The occurrence of maximum concentration of NO, NO2, NOX, SO2 and PM10 occurred in the coldest months during winter season (January, February and December) confirmed the strong influence of “low emission” on air quality. Among the meteorological factors assessed wind speed was most often selected component in step wise regression procedure, then air temperature, less air relative humidity and solar radiation. In the case of a larger number of variables describing the pollution in the atmosphere, in all analyzed winter seasons the most common set of meteorological elements were wind speed and air temperature.

scholarly journals Research on the air quality prediction model of Wuhai mining area based on deep learning

2021 ◽  
Vol 300 ◽  
pp. 02005
Author(s):  
Jinghua Wang ◽  
Jin Cheng ◽  
Fang Liu ◽  
Lei Yan ◽  
Taijie Tang
Keyword(s):  
Neural Network ◽  
Air Quality ◽  
Large Scale ◽  
Prediction Models ◽  
Mining Area ◽  
Air Pollutant ◽  
Quality Prediction ◽  
Weather Factors ◽  
Air Quality Prediction ◽  
Deep Recurrent Neural Network

With the large-scale and high-intensity mining of coal resources in the Wuhai mining area, the destruction of soil and erosion of rocks has intensified, causing a large amount of surface soil spalling from the mine body and serious damage to the surface vegetation, which has had a serious impact on the quality of the environment in and around the mine. This paper focuses on the corresponding early warning research on air quality in the mining area of Wuhai, and constructs Deep Recurrent Neural Network (DRNN) and Deep Long Short Time Memory Neural Network (DLSTM) air quality prediction models based on the filtered weather factors. The simulation results are also compared and find that the prediction results of DLSTM are better than those of DRNN, with a prediction accuracy of 92.85%. The model is able to accurately predict the values and trends of various air pollutant concentrations in the mining area of Wuhai.

Sign in / Sign up

Export Citation Format

Share Document