Air Pollution Forecasting Research Articles

An Outlier-Robust Point and Interval Forecasting System for Daily PM2.5 Concentration

Air pollution forecasting, particularly of PM2.5 levels, can be used not only to deliver effective warning information to the public but also to provide support for decisions regarding the control and treatment of air pollution problems. However, there are still some challenging issues in air pollution forecasting that urgently need to be solved, such as how to handle and model outliers, improve forecasting stability, and correct forecasting results. In this context, this study proposes an outlier-robust forecasting system to attempt to tackle the abovementioned issues and bridge the gap in the current research. Specifically, the system developed consists of two parts that deal with point and interval forecasting, respectively. For point forecasting, a data preprocessing module is proposed based on outlier handling and data decomposition to mitigate the negative influences of outliers and noise, which can also help the model capture the main characteristics of the original time series. Meanwhile, an outlier-robust forecasting module is designed for better modeling of the preprocessed data. For the model to further improve its accuracy, a nonlinear correction module based on an error ensemble strategy is developed that can provide more accurate forecasting results. Finally, the interval forecasting part of the system is based on a newly proposed artificial intelligence–based distribution evaluation and the results of the point forecasting part to present the range of future changes. Experimental results and analysis utilizing daily PM2.5 concentration from two provincial capital cities in China are discussed to verify the superiority and effectiveness of the system developed, which can be considered an effective technique for point and interval forecasting of daily PM2.5 concentration.

AN AIR QUALITY FORECASTING MODEL BASED ON BIDIRECTIONAL GRU AND LSTM

With the urban and industrial growth, many evolving countries suffer from excessive air pollution. The growing concern about air pollution has been raised by the government and people because it affects individual’s health and sustainable development globally. Several factors influence Air Quality and we must use all of them to interpose and forecast air pollution for the whole city. Thearticle suggested a novel deep learning CBGLSTM model for air quality forecasting based on 1D CNN, Bi-GRU, and Bi-LSTM neural networks. The model has experimented on the PM2.5 dataset taken from UCI Machine Learning Repository. The results demonstrate that compared to other traditional deep learning models, the CBGLSTM model gives better prediction performance and less error.

Forecasting Air Pollution with Sulfur Dioxide Emitted from Burning Desulfurized Diesel Using Artificial Neural Network

Forecasting Air Pollution with Sulfur Dioxide Emitted from Burning Desulfurized Diesel Using Artificial Neural Network

Local design and development of low-cost air sensing and data science methods in sub-Saharan Africa – lessons learnt from Uganda

BACKGROUND AND AIM: Sensing and modelling approaches need to be conceived and designed while taking into account unique local contexts in order to be responsive and effective in addressing the environmental data and evidence gaps for the communities and authorities that need them. We report on experiences of designing, developing, deploying and maintaining a local air quality monitoring, modelling and analysis system in Uganda. METHODS: The AirQo system includes a citywide air quality monitoring network of monitors developed locally to reduce maintenance frequency in dusty environments, leverage solar power, have capability for near-real time monitoring and transmission of air quality data over cellular networks, support for complementary static and mobile (motorcycle) deployment approaches, and application of machine learning approaches for calibration to improve data accuracy. RESULTS:The air quality datasets have enabled design and development of data science products that generate evidence to inform awareness, education and policy interventions including: (1) a digital air quality platform targeted at the city governments in Uganda to visualise trends, forecast and spatial variations of air pollution to the local level. (2) public-facing digital channels such as mobile apps and web tools that provide realtime and historical analysis of air pollution in places of interest, for examples, schools. We also report on experiences from community engagement initiatives where citizens can report observed pollution events to the AirQo platform using social media tools such as WhatsApp. CONCLUSIONS:The lessons from this talk are expected to inform ongoing efforts in other African cities and build local capacity in air quality monitoring monitoring and analysis. KEYWORDS: air quality, monitoring, modelling and analysis

Machine-learning regression for forecasting air pollution

BACKGROUND AND AIM: Many scientific researches have already confirmed the link between air pollution and many other respiratory and cardiovascular diseases. Air quality forecasting may play an important role in decision making where clean air is still a serious challenge for public health services. METHODS: This paper examines the ability of supervised machine-learning regression techniques, namely, feed-forward neural network, tree-based ensemble, and generalized linear methods, to anticipate and manage changes in atmospheric pollutant concentrations. In order to infer the performance of the developed models, daily observed data of air quality, Climate Indices and meteorological parameter’s collected during the years 2006-2016, have been used. RESULTS:The analysis stage of the modeling provided clear and intuitive results regarding air quality in Casablanca City. The proposed models achieve comparable results with good generalization performances in forecasting pollutant parameters. CONCLUSIONS:Results from this work have important implications for understanding and forecasting of air quality parameters in Morocco and showed as promising techniques to be applied in other countries. KEYWORDS: Air pollution, Ozone, Particulate matter, machine learning

Data-driven method for the improving forecasts of local weather dynamics

This paper describes the modeling approach for lower atmosphere dynamics in a selected location. The purpose of this model is to provide short-term and long-term forecasts of the weather variables which are used as the input data for the model of the dispersion of radioactive air pollution. The information from this integrated system is important for the implementation of the population safety measures in the case of a nuclear accident with an atmospheric release. We developed a dynamical, probabilistic, and non-parametric model based on Gaussian processes (GPs). GP nonlinear autoregressive model with exogenous inputs and variational training principle was implemented for multi-output training. A Monte Carlo approach to multi-output simulation of the model for long-term forecasts is presented which allows arbitrary prior distributions over function values. The model encompasses all available measurements from the weather stations near the location of interest and combines them with the forecasts from the numerical weather prediction model. The contribution of the developed model is the harvesting of all available information and simultaneously providing interconnected forecasts. The key result of this investigation is the improvement of short-term and long-term weather variable forecasts over those of the numerical weather prediction model. Consequently, we significantly enhance the dispersion forecast of radioactive air pollution for the case study considered. The computationally demanding modeling is accelerated using general-purpose computing on graphics processing units. The proposed method represents a step forward in the assurance of safety in the case of a nuclear accident.

A Novel Framework for Forecasting, Evaluation and Early-Warning for the Influence of PM10 on Public Health

PM2.5 has attracted widespread attention since the public has become aware of it, while attention to PM10 has started to wane. Considering the significance of PM10, this study takes PM10 as the research object and raises a significant question: when will the influence of PM10 on public health end? To answer the abovementioned question, two promising research areas, i.e., air pollution forecasting and health effects analysis, are employed, and a novel hybrid framework is developed in this study, which consists of one effective model and one evaluation model. More specifically, this study first introduces one advanced optimization algorithm and cycle prediction theory into the grey forecasting model to develop an effective model for multistep forecasting of PM10, which can achieve reasonable forecasting of PM10. Then, an evaluation model is designed to evaluate the health effects and economic losses caused by PM10. Considering the significance of providing the future impact of PM10 on public health, we extend our forecasting results to evaluate future changes in health effects and economic losses based on our proposed health economic losses evaluation model. Accordingly, policymakers can adjust current air pollution prevention plans and formulate new plans according to the results of forecasting, evaluation and early-warning. Empirical research shows that the developed framework is applicable in China and may become a promising technique to enrich the current research and meet the requirements of air quality management and haze governance.

Multi-output TCN autoencoder for long-term pollution forecasting for multiple sites

Air pollution is one of the major environmental issues attracting massive attention from researchers and policymakers. Both the developed and developing countries are undergoing a high concentration of pollution levels. Fine particulate matter PM2.5 (particles having a diameter less than 2.5μm) and PM10 (particles having a diameter less than 10μm) can easily penetrate the lungs and respiratory system and causes adverse health issues like heart attacks, cardiovascular diseases, lung function reduction. Real-time pollutant information is of great importance to providing prompt and complete information on air quality. Air pollution forecasting is another significant step of air pollution management, which can help policymakers and citizens make proper decisions to prevent air pollution-related diseases. This research study explores a novel pollutant forecasting model named as Multi-output temporal convolutional network autoencoder (MO-TCNA). The model accumulates each step's predicted values to perform multi-step ahead long-term forecasting for multiple pollutants and multiple sites in a single training model. The MO-TCNA network serves both the PM2.5 and PM10 pollutants forecasting for various locations instead of performing single output and site-specific pollutant forecasting. Consequently, the experimental results show that the MO-TCNA network is time-saving and has better performance than the traditional site-specific forecasting models.

Comparative assessment of modeling deep learning networks for modeling ground-level ozone concentrations of pandemic lock-down period

Covid-19 pandemic lock-down has resulted significant differences in air quality levels all over the world. In contrary to decrease seen in primary pollutant species, many of the countries have experienced elevated ground-level ozone levels in this period. Air pollution forecast gains more importance to achieve air quality management and take measures against the risks under such extra-ordinary conditions. Statistical models are indispensable tools for predicting air pollution levels. Considering the complex photochemical reactions involved in tropospheric ozone formation, modeling this pollutant requires efficient non-linear approaches. In this study, deep learning methods were applied to forecast hourly ozone levels during pandemic lock-down for an industrialized region in Turkey. With this aim, different deep learning methods were tested and efficiencies of the models were compared considering the calculated RMSE, MAE, R2 and loss values.

An artificial neural network model for forecasting air pollution

Air pollution has caused a lot of problems to people in terms of health and economy, as well as affecting various floras and faunas. Thus, monitoring air quality levels and forecasting the occurrence of air pollution is important so that preventive measures could be taken. In this study, Artificial Neural Network (ANN) was used to forecast the air pollution index (API) in Kuala Terengganu. This study focused on the prediction of API based on 5 years of data of main pollutants’ daily concentration taken at the air quality monitoring station in Kuala Terengganu. The aim was to develop an Artificial Neural Network model that can predict the API. A Multilayer Perceptron Neural Network (MLP) engine was implemented in the system prototype and developed by using Keras, a deep learning library in Python. The model’s performance was evaluated using the Mean Squared Error (MSE) statistical method and functionality tests were done to ensure the prototype was working correctly. In order to get a good performance model, a hyperparameter tuning process was carried out and the best hyperparameters values were selected. The performance of the model in making predictions was good as the MSE value was 0.0195.

Evaluation of an Air Pollution Forecasting System Based on Micro-Pulse Lidar Cruising Measurements in the South China Sea

In this study, we evaluated the performance of an air pollution forecasting system during a scientific cruise in the South China Sea (SCS) from 9 August to 7 September 2016. The air pollution forecasting system consisted of a Lagrangian transport and dispersion model, the flexible particle dispersion model (FLEXPART), coupled with a high-resolution Weather Research and Forecasting model (WRF). The model system generally reproduced the meteorological variability and reasonably simulated the distribution of aerosols both vertically and horizontally along the cruise path. The forecasting system was further used to study the regional transport of non-local aerosols over the SCS and track its sources during the cruise. The model results showed that Southeast Asia contributed to more than 90% of the non-local aerosols over the northern region of the SCS due to the southwesterly prevailing winds. Specifically, the largest mean contribution was from Vietnam (39.6%), followed by Thailand (25.1%). This study indicates that the model system can be applied to study regional aerosols transport and provide air pollution forecasts in the SCS.

Forecasting Particulate Pollution in an Urban Area: From Copernicus to Sub-Km Scale

Particulate air pollution has aggravated cardiovascular and lung diseases. Accurate and constant air quality forecasting on a local scale facilitates the control of air pollution and the design of effective strategies to limit air pollutant emissions. CAMS provides 4-day-ahead regional (EU) forecasts in a 10 km spatial resolution, adding value to the Copernicus EO and delivering open-access consistent air quality forecasts. In this work, we evaluate the CAMS PM forecasts at a local scale against in-situ measurements, spanning 2 years, obtained from a network of stations located in an urban coastal Mediterranean city in Greece. Moreover, we investigate the potential of modelling techniques to accurately forecast the spatiotemporal pattern of particulate pollution using only open data from CAMS and calibrated low-cost sensors. Specifically, we compare the performance of the Analog Ensemble (AnEn) technique and the Long Short-Term Memory (LSTM) network in forecasting PM2.5 and PM10 concentrations for the next four days, at 6 h increments, at a station level. The results show an underestimation of PM2.5 and PM10 concentrations by a factor of 2 in CAMS forecasts during winter, indicating a misrepresentation of anthropogenic particulate emissions such as wood-burning, while overestimation is evident for the other seasons. Both AnEn and LSTM models provide bias-calibrated forecasts and capture adequately the spatial and temporal variations of the ground-level observations reducing the RMSE of CAMS by roughly 50% for PM2.5 and 60% for PM10. AnEn marginally outperforms the LSTM using annual verification statistics. The most profound difference in the predictive skill of the models occurs in winter, when PM is elevated, where AnEn is significantly more efficient. Moreover, the predictive skill of AnEn degrades more slowly as the forecast interval increases. Both AnEn and LSTM techniques are proven to be reliable tools for air pollution forecasting, and they could be used in other regions with small modifications.

Design of a Spark Big Data Framework for PM2.5 Air Pollution Forecasting.

In recent years, with rapid economic development, air pollution has become extremely serious, causing many negative effects on health, environment and medical costs. PM2.5 is one of the main components of air pollution. Therefore, it is necessary to know the PM2.5 air quality in advance for health. Many studies on air quality are based on the government’s official air quality monitoring stations, which cannot be widely deployed due to high cost constraints. Furthermore, the update frequency of government monitoring stations is once an hour, and it is hard to capture short-term PM2.5 concentration peaks with little warning. Nevertheless, dealing with short-term data with many stations, the volume of data is huge and is calculated, analyzed and predicted in a complex way. This alleviates the high computational requirements of the original predictor, thus making Spark suitable for the considered problem. This study proposes a PM2.5 instant prediction architecture based on the Spark big data framework to handle the huge data from the LASS community. The Spark big data framework proposed in this study is divided into three modules. It collects real time PM2.5 data and performs ensemble learning through three machine learning algorithms (Linear Regression, Random Forest, Gradient Boosting Decision Tree) to predict the PM2.5 concentration value in the next 30 to 180 min with accompanying visualization graph. The experimental results show that our proposed Spark big data ensemble prediction model in next 30-min prediction has the best performance (R2 up to 0.96), and the ensemble model has better performance than any single machine learning model. Taiwan has been suffering from a situation of relatively poor air pollution quality for a long time. Air pollutant monitoring data from LASS community can provide a wide broader monitoring, however the data is large and difficult to integrate or analyze. The proposed Spark big data framework system can provide short-term PM2.5 forecasts and help the decision-maker to take proper action immediately.

An unequal adjacent grey forecasting air pollution urban model

In grey theory, the principle of difference information holds that difference is information. To mine the difference information between data, the unequal accumulation is proposed in this paper. The unequal accumulation reduces the loss of difference information and improves the forecasting performance of the grey model. Based on the unequal accumulation, the unequal adjacent discrete multivariable grey model is proposed. According to the comparison with other models, the proposed model has good forecast performance and algorithm efficiency. To further prove the practical significance of the model, it is used to predict the public concern about air pollution in three cities of China. The results show that the public concern about air pollution in cities with different air pollution conditions is quite different. Environmental protection departments could pay attention to the public concern about air pollution to better control air pollution.

Spatiotemporal causal convolutional network for forecasting hourly PM2.5 concentrations in Beijing, China

Air pollution in Northeastern Asia is a serious environmental problem, especially in China where PM2.5 levels are quite high. Accurate PM2.5 predictions are significant to environmental management and human health. Recently, deep learning has received increasing attention from relevant researchers. In this work, a spatiotemporal causal convolutional neural network (ST-CausalConvNet) for short-term PM2.5 prediction is proposed. The distinguishing characteristics of the proposed model is that the convolutions in the model architecture are causal, where an output at a certain time step is convolved only with elements from the same or earlier time steps in the previous layer. Accordingly, no information leakage is induced from the future to the past in this model. The spatial dependence between multiple monitoring stations was also considered in the model. Spatiotemporal correlation analysis was performed to select relevant information from monitoring stations that have a high relationship with the target station. The information from the target and related stations were then employed as the inputs and fed into the model. A case study from May 1, 2014 to April 30, 2015 in Beijing, China was conducted. The next hour PM2.5 concentration was predicted by the proposed model by using historical air quality and meteorological data from 36 monitoring stations. Experimental results show that the trends of the predicted PM2.5 concentrations and the observed values were consistent. The proposed method achieved a better prediction performance than the other three comparative models, namely artificial neural network (ANN), gated recurrent unit (GRU), and long short-term memory (LSTM). Furthermore, the effects of the important parameters and the model transferability were also conducted. We conclude that the proposed ST-CausalConvNet is a potential effective model for air pollution forecasting.

Towards greener smart cities and road traffic forecasting using air pollution data

Road traffic flow forecasting is among the most important use case associated with smart cities. Traffic forecasting allows drivers to select the fastest route towards their target destinations. A prerequisite for traffic flow management is accurate traffic forecasting. In this study, we introduce a framework for traffic forecasting that uses data on air pollution. The reason to select that data is air pollution rates are often associated with traffic congestion, and there is intensive research that exists to forecast air pollution by road traffic. To the best of our knowledge, any effort to enhance road traffic prediction through air quality and ensemble regression model techniques is not yet been proposed. In this research, our contribution is twofold. Firstly, we have performed a comparative analysis of 7 different regression models to find out which model gives better accuracy. Secondly, we propose a framework using regression models in which the first regression model's result is boosted using boosting ensemble method and is passed to the next regression model which shows that the proposed framework gives more satisfying results than the above 7 regression models. The experimental findings show the effectiveness of the proposed framework which decreases the error rate by 2.47 %.

An improved pollution forecasting model with meteorological impact using multiple imputation and fine-tuning approach

Air pollution forecasting is a significant step for air quality pollution management to mitigate pollution’s negative impact on the environment and people’s health. The data-driven forecasting model can help a better understanding of environmental air quality. The existing data-driven forecasting models usually ignore missing values, the correlations between the pollutant and meteorological factors and fail to perform temporal modeling effectively, affecting prediction accuracy. In response to these issues, we present a deep learning-based Convolutional LSTM–SDAE (CLS) model to forecast the particulate matter level, revealing the correlation between particulate matter and meteorological factors. In the proposed architecture, the k nearest neighbor (KNN) imputation technique is employed to recover the air quality dataset’s missing values. The Convolutional Long Short Term Memory (CNN–LSTM) unit identifies the vast dataset’s hidden features and performs pollutants’ temporal modeling. In addition, Bidirectional Gatted Recurrent Unit (BIGRU) is implemented as both encoder and decoder in Sparse Denoising Autoencoder, which reconstructs the CNN–LSTM model’s output in the dynamic fine-tuning layer to get robust prediction results. The experimental results in Talcher, India, and Beijing, China indicate that the model can improve forecasting accuracy and outperforms the other state of art and baseline models.

ADAPTATION OF THE WEB-SERVICE OF AIR POLLUTION FORECASTING FOR OPERATION WITHIN CLOUD COMPUTING PLATFORM OF THE UKRAINIAN NATIONAL GRID INFRASTRUCTURE

Introduction. Air pollution modeling is a powerful tool that allows developing scientifically justified solutions to reduce the risks posed by atmospheric emissions of pollutants. Problem Statement. Cloud computing infrastructures provide new opportunities for web-based air pollution forecasting systems. However the implementation of these capabilities requires changes in the architecture of the existing systems. Purpose. The purpose is to adapt the web service of forecasting the atmospheric pollution in Ukraine to operate in the cloud computing platform of the Ukrainian National Grid infrastructure. Materials and Methods. The web client – web server – cloud computing architecture was used. The calculation of the model is performed in the cloud infrastructure, while the client and server parts operate on separate computers. Results. With the developed service the forecast of air pollution is possible for every point at the territory of Ukraine for more than thirty substances, including chlorine, ammonia, hydrogen sulfide and others. The forecast is performed using the data of the WRF-Ukraine numerical weather prediction system and visualized through a web interface. The capabilities of the developed system were demonstrated by the example of simulation of air pollution in part of Kyiv affected by the releases from the Energia incineration plant during pollution episode in September, 2019. The total releases of toluene gas from incineration plant and from the fire on spontaneous waste landfill, which is located a few km from Kyiv, were estimated and analyzed. For the considered period the fire could bring considerable additional amounts of pollutants to the studied region. The confidence interval for the maximum airborne concentration for the considered period is estimated from 0.7 to 2.1 mg·m-3 which is higher than the permissible value (0.6 mg· m-3). Conclusions. The presented system could be used by institutions responsible for response to environmental accidents. Keywords: air pollution, atmospheric dispersion, web-systems, cloud computing. Introduction. Air pollution modeling is a powerful tool that allows developing scientifically justified solutions to reduce the risks posed by atmospheric emissions of pollutants. Problem Statement. Cloud computing infrastructures provide new opportunities for web-based air pollution forecasting systems. However the implementation of these capabilities requires changes in the architecture of the existing systems. Purpose. The purpose is to adapt the web service of forecasting the atmospheric pollution in Ukraine to operate in the cloud computing platform of the Ukrainian National Grid infrastructure. Materials and Methods. The web client – web server – cloud computing architecture was used. The calculation of the model is performed in the cloud infrastructure, while the client and server parts operate on separate computers. Results. With the developed service the forecast of air pollution is possible for every point at the territory of Ukraine for more than thirty substances, including chlorine, ammonia, hydrogen sulfide and others. The forecast is performed using the data of the WRF-Ukraine numerical weather prediction system and visualized through a web interface. The capabilities of the developed system were demonstrated by the example of simulation of air pollution in part of Kyiv affected by the releases from the Energia incineration plant during pollution episode in September, 2019. The total releases of toluene gas from incineration plant and from the fire on spontaneous waste landfill, which is located a few km from Kyiv, were estimated and analyzed. For the considered period the fire could bring considerable additional amounts of pollutants to the studied region. The confidence interval for the maximum airborne concentration for the considered period is estimated from 0.7 to 2.1 mg·m-3 which is higher than the permissible value (0.6 mg· m-3). Conclusions. The presented system could be used by institutions responsible for response to environmental accidents.

Artificial intelligence accuracy assessment in NO2 concentration forecasting of metropolises air

Air quality has been the main concern worldwide and Nitrous oxide (NO2) is one of the pollutants that have a significant effect on human health and environment. This study was conducted to compare the regression analysis and neural network model for predicting NO2 pollutants in the air of Tehran metropolis. Data has been collected during a year in the urban area of Tehran and was analyzed using multi-linear regression (MLR) and multilayer perceptron (MLP) neural networks. Meteorological parameters, urban traffic data, urban green space information, and time parameters are applied as input to forecast the daily concentration of NO2 in the air. The results demonstrate that artificial neural network modeling (R2 = 0.89, RMSE = 0.32) results in more accurate predictions than MLR analysis (R2 = 0.81, RMSE = 13.151). According to the result of sensitivity analysis of the model, the value of park area, the average of green space area and one-day time delay are the crucial parameters influencing NO2 concentration of air. Artificial neural network models could be a powerful, effective and suitable tool for analysis and modeling complex and non-linear relation of environmental variables such as ability in forecasting air pollution. Green spaces establishment has a significant role in NO2 reduction even more than traffic volume.

A wavelet-based random forest approach for indoor BTEX spatiotemporal modeling and health risk assessment.

This study reports on BTEX concentrations in one of the largest parking garages in Iran with a peak traffic flow reaching up to ~9300 vehicles in the last few days of the Nowruz holidays. Samples were obtained on different days of the week at three main locations in the Zaer Parking Garage. A novel wavelet-based random forest model (WRF) was trained to estimate BTEX concentrations by decomposing temperature, day of the week, sampling location, and relative humidity data with a maximal overlap discrete wavelet transform (MODWT) function and subsequently inputted into the WRF model. The results suggested that the WRF model can reasonably estimate BTEX trends and variations based on high R2 values of 0.96, 0.95, and 0.98 for training, validation, and test data subsets, respectively. The carcinogenic (LTCR) and non-carcinogenic health risk (HI) assessment results indicated a definite carcinogenic risk of benzene (LTCR = 2.22 × 10-4) and high non-carcinogenic risk (HI = 4.51) of BTEX emissions. The results of this study point to the importance of BTEX accumulation in poorly ventilated areas and the utility of machine learning in forecasting air pollution in diverse airsheds such as parking garages.