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

Abstract

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.

Synergistic or Antagonistic Health Effects of Long- and Short-Term Exposure to Ambient NO2 and PM2.5: A Review

“…The authors of the IARC monograph [ 53 ] underline that PM exposure from different sources features mutagenic and carcinogenic effects in people. Unfortunately, the number of air monitoring ground stations is limited, and the spatial distribution is discontinuous, and thus to obtain a fine-grained spatiotemporal distribution of PM 2.5 , a retrieval model can be used [ 54 ]. In an estimation of future global mortality from changes in air pollution, Silva et al [ 55 ] predicted 55,600 (−34,300 to 164,000) deaths in 2030 and 215,000 (−76,100 to 595,000) in 2100 due to PM 2.5 worldwide (countering by 16% the global decrease in PM 2.5 -related mortality.…”

An Attempt to Determine the Relationship between Air Pollution and the Real Estate Market in 2010–2020 in Gdańsk Using GLM and GRM Statistical Models

“…Some researchers use spatial econometric models and hedonic price models to understand the effect of air pollution on housing prices, considering their spatial autocorrelation. The most commonly-used spatial hedonic models are the spatial lag model (SLM) [33,34], spatial error model (SEM), spatial Durbin model (SDM) [35], geographically weighted regression (GWR) [36], quantile regression models (QRM) [37], bootstrap autoregressive distributed lag (BARDL) [38], and the fine-grained PM 2.5 data retrieval model that includes high-resolution satellite imagery data [39].…”

Regional Representativeness Analysis of Ground-Monitoring PM2.5 Concentration Based on Satellite Remote Sensing Imagery and Machine Learning Techniques

“…In recent years, some scholars have started to try to use machine learning methods to explore the correlation between PM 2.5 concentrations and the relevant bands of multispectral satellite remote sensing data and use them as the basis for inversion studies of PM 2.5 concentrations. For example, Zhang et al (2019) constructed a PM x estimation model based on the back-propagation neural network commonly used in machine learning with the reflectance of different bands in Landsat 8 remote sensing images as the main basis and selected the Beijing area for testing, and they obtained more satisfactory estimation results [22]; Ma et al (2022) tried to use multiple linear regression, K-nearest neighbor, support vector regression, decision tree, random forest, back-propagation neural network, and other different machine learning methods for PM 2.5 estimation of Landsat 8 remote sensing image data, and in the test with Hangzhou as the target area, the random forest method achieved the best estimation results [23].…”