Fine-grained PM2 Research Articles

MTLPM: a long-term fine-grained PM2.5 prediction method based on spatio-temporal graph neural network.

The concentration of PM2.5 is one of the air quality indicators that the public pays the most attention to. Existing methods for PM2.5 prediction primarily analyze and forecast data from individual monitoring stations, without considering the mutual influence among multiple stations caused by natural environmental factors, e.g., air circulation. Moreover, the existing methods are mostly short-term predictions and perform poorly in long-term forecasting. In this paper, we propose MTLPM, i.e., a spatio-temporal graph neural network model based on an encoder-decoder architecture, which fully exploits the spatial dynamic patterns and long-term dependencies. Firstly, we adopt a message passing mechanism combined with spatial features and complex environmental factors (e.g., temperature, humidity, and wind direction) to update station data, capturing real-time spatial dynamic information. Secondly, we adopt the Multi-head ProbSparse Self-attention to extract temporal features, learning the long-term dependency relationships among the data. Finally, we adopt a generative one-step decoder structure to simultaneously forecast the data for multiple stations over a long period. We conducted experiments on both the project dataset and the publicly available dataset. Compared to existing state-of-the-art methods, MTLPM achieved an average reduction of approximately 1.6 in mean absolute error (MAE) and approximately 0.02 in symmetric mean absolute percentage error (SMAPE) in predicting results. The relevant source code is publicly available on GitHub1.

Sparse Grid Imputation Using Unpaired Imprecise Auxiliary Data: Theory and Application to PM2.5 Estimation

Sparse grid imputation (SGI) is a challenging problem, as its goal is to infer the values of the entire grid from a limited number of cells with values. Traditionally, the problem is solved using regression methods such as KNN and kriging, whereas in the real world, there is often extra information—usually imprecise—that can aid inference and yield better performance. In the SGI problem, in addition to the limited number of fixed grid cells with precise target domain values, there are contextual data and imprecise observations over the whole grid. To solve this problem, we propose a distribution estimation theory for the whole grid and realize the theory via the composition architecture of the Target-Embedding and the Contextual CycleGAN trained with contextual information and imprecise observations. Contextual CycleGAN is structured as two generator–discriminator pairs and uses different types of contextual loss to guide the training. We consider the real-world problem of fine-grained PM2.5 inference with realistic settings: a few (less than 1%) grid cells with precise PM2.5 data and all grid cells with contextual information concerning weather and imprecise observations from satellites and microsensors. The task is to infer reasonable values for all grid cells. As there is no ground truth for empty cells, out-of-sample mean squared error and Jensen–Shannon divergence measurements are used in the empirical study. The results show that Contextual CycleGAN supports the proposed theory and outperforms the methods used for comparison.

A Spatial-Temporal Causal Convolution Network Framework for Accurate and Fine-Grained PM2.5 Concentration Prediction.

Accurate and fine-grained prediction of PM2.5 concentration is of great significance for air quality control and human physical and mental health. Traditional approaches, such as time series, recurrent neural networks (RNNs) or graph convolutional networks (GCNs), cannot effectively integrate spatial–temporal and meteorological factors and manage dynamic edge relationships among scattered monitoring stations. In this paper, a spatial–temporal causal convolution network framework, ST-CCN-PM2.5, is proposed. Both the spatial effects of multi-source air pollutants and meteorological factors are considered via spatial attention mechanism. Time-dependent features in causal convolution networks are extracted by stacked dilated convolution and time attention. All the hyper-parameters in ST-CCN-PM2.5 are tuned by Bayesian optimization. Haikou air monitoring station data are employed with a series of baselines (AR, MA, ARMA, ANN, SVR, GRU, LSTM and ST-GCN). Final results include the following points: (1) For a single station, the RMSE, MAE and R2 values of ST-CCN-PM2.5 decreased by 27.05%, 10.38% and 3.56% on average, respectively. (2) For all stations, ST-CCN-PM2.5 achieve the best performance in win–tie–loss experiments. The numbers of winning stations are 68, 63, and 64 out of 95 stations in RMSE (MSE), MAE, and R2, respectively. In addition, the mean MSE, RMSE and MAE of ST-CCN-PM2.5 are 4.94, 2.17 and 1.31, respectively, and the R2 value is 0.92. (3) Shapley analysis shows wind speed is the most influencing factor in fine-grained PM2.5 concentration prediction. The effects of CO and temperature on PM2.5 prediction are moderately significant. Friedman test under different resampling further confirms the advantage of ST-CCN-PM2.5. The ST-CCN-PM2.5 provides a promising direction for fine-grained PM2.5 prediction.

ISpray

Despite regulations and policies to improve city-level air quality in the long run, there lack precise control measures to protect critical urban spots from heavy air pollution. In this work, we propose iSpray, the first-of-its-kind data analytics engine for fine-grained PM2.5 and PM10 control at key urban areas via cost-effective water spraying. iSpray combines domain knowledge with machine learning to profile and model how water spraying affects PM25 and PM10 concentrations in time and space. It also utilizes predictions of pollution propagation paths to schedule a minimal number of sprayers to keep the pollution concentrations at key spots under control. In-field evaluations show that compared with scheduling based on real-time pollution concentrations, iSpray reduces the total sprayer switch-on time by 32%, equivalent to 1, 782 m3 water and 18, 262 kWh electricity in our deployment, while decreasing the days of poor air quality at key spots by up to 16%.

Retrieval of Fine-Grained PM2.5 Spatiotemporal Resolution Based on Multiple Machine Learning Models

Due to the country’s rapid economic growth, the problem of air pollution in China is becoming increasingly serious. In order to achieve a win-win situation for the environment and urban development, the government has issued many policies to strengthen environmental protection. PM2.5 is the primary particulate matter in air pollution, so an accurate estimation of PM2.5 distribution is of great significance. Although previous studies have attempted to retrieve PM2.5 using geostatistical or aerosol remote sensing retrieval methods, the current rough resolution and accuracy remain as limitations of such methods. This paper proposes a fine-grained spatiotemporal PM2.5 retrieval method that comprehensively considers various datasets, such as Landsat 8 satellite images, ground monitoring station data, and socio-economic data, to explore the applicability of different machine learning algorithms in PM2.5 retrieval. Six typical algorithms were used to train the multi-dimensional elements in a series of experiments. The characteristics of retrieval accuracy in different scenarios were clarified mainly according to the validation index, R2. The random forest algorithm was shown to have the best numerical and PM2.5-based air-quality-category accuracy, with a cross-validated R2 of 0.86 and a category retrieval accuracy of 0.83, while both maintained excellent retrieval accuracy and achieved a high spatiotemporal resolution. Based on this retrieval model, we evaluated the PM2.5 distribution characteristics and hourly variation in the sample area, as well as the functions of different input variables in the model. The PM2.5 retrieval method proposed in this paper provides a new model for fine-grained PM2.5 concentration estimation to determine the distribution laws of air pollutants and thereby specify more effective measures to realize the high-quality development of the city.

MCS-RF: mobile crowdsensing–based air quality estimation with random forest

It is a great challenge to offer a fine-grained and accurate PM2.5 monitoring service in urban areas as required facilities are very expensive and huge. Since PM2.5 has a significant scattering effect on visible light, large-scale user-contributed image data collected by the mobile crowdsensing bring a new opportunity for understanding the urban PM2.5. In this article, we propose a fine-grained PM2.5 estimation method based on random forest with data announced by meteorological departments and collected from smartphone users without any PM2.5 measurement devices. We design and implement a platform to collect data in the real world including the image provided by users. By combining online learning and offline learning, the method based on random forest performs well in terms of time complexity and accuracy. We compare our method with two kinds of baselines: subsets of the whole data sets and six classical models (such as logistic, naive Bayes). Six kinds of evaluation indexes (precision, recall, true-positive rate, false-positive rate, F-measure, and receiver operating characteristic curve area) are used in the evaluation. The experimental results show that our method achieves high accuracy (precision: 0.875, recall: 0.872) on PM2.5 estimation, which outperforms the other methods.

Pixel-wise depth based intelligent station for inferring fine-grained PM2.5

Air pollution seriously affects people’s lives, among which PM2.5 is especially harmful for humans health. Although many countries have established air quality monitoring stations (AQMS) to monitor air pollution, the costs of constructing and maintaining for AQMS are extremely expensive and the density of AQMS is very low. Taking advantage of the artificial intelligence (AI) and the internet of things (IoT), we established image based intelligent station (IBIS) to monitor air pollution. To study the relationship between PM2.5 concentration and images information, an IoT based smart systems has been established which has collected photos for consecutive 16 months. Furthermore, we utilize deep learning algorithm to acquire pixel-wise depth information. By combining Bayesian estimation with pixel-wise depth information, knowledge learned from well developed IBIS can be transferred to other newly built IBIS. The performance of the proposed method has been evaluated by real dataset. The results show that, compared with three baselines, our proposed algorithm can reduce up to 35% prediction error in average.

Third-Eye

Air pollution has raised people's public health concerns in major cities, especially for Particulate Matter under 2.5μm (PM2.5) due to its significant impact on human respiratory and circulation systems. In this paper, we present the design, implementation, and evaluation of a mobile application, Third-Eye, that can turn mobile phones into high-quality PM2.5 monitors, thereby enabling a crowdsensing way for fine-grained PM2.5 monitoring in the city. We explore two ways, crowdsensing and web crawling, to efficiently build large-scale datasets of the outdoor images taken by mobile phone, weather data, and air-pollution data. Then, we leverage two deep learning models, Convolutional Neural Network (CNN) for images and Long Short Term Memory (LSTM) network for weather and air-pollution data, to build an end-to-end framework for training PM2.5 inference models. Our App has been downloaded more than 2,000 times and runs more than 1 year. The real user data based evaluation shows that Third-Eye achieves 17.38 μg/m3 average error and 81.55% classification accuracy, which outperforms 5 state-of-the-art methods, including three scattered interpolations and two image based estimation methods. The results also demonstrate how Third-Eye offers substantial enhancements over typical portable PM2.5 monitors by simultaneously improving accessibility, portability, and accuracy.

An estimated method of urban [formula omitted] concentration distribution for a mobile sensing system

The frequent hazy weather has brought enormous threaten to human health, especially in developing countries. Modeling a fine-grained distribution of PM2.5 (Particulate matter with diameters less than 2.5μm) concentrations is of great importance for industrial and environmental applications. The contradiction between the need of fine-grained distribution and the approach of coarse-grained collection has to be addressed immediately.In this paper, we focus on solving the problem of inferring PM2.5 concentration of unobserved areas based on data samples collected by mobile sensors. We propose a Probabilistic Concentration Estimation Method (PCEM) for a regional fine-grained PM2.5 distribution considering the nature of particle motion. It simulates particles transport in an open air-flow field referring the concept of random walk. In the meantime, quadratic programming and heuristic function are also constructed to optimize the algorithm accuracy and robustness.We employ several mobile sensors to collect original data randomly in a region of HangZhou City for certain period of time and utilize the proposed PCEM to generate the real time fine-grained PM2.5 concentration distribution. The results can demonstrate up to 100 times higher resolution level of the PM2.5 concentration distribution than the traditional approaches based on monitoring sites. The degree of correlation between estimated PM2.5 concentration and real measured data is up to 0.9735. The average calculation error of PCEM can be reduced about 41.0% compared with widely used artificial network (ANN). Furthermore, PCEM can easily adapt to other PM distribution inference with different built-in sensors, which could help with a deeper understanding of an informed air quality monitoring system in the future.