Dataset Information

Robust prediction of hourly PM_2.5 from meteorological data using LightGBM.

ABSTRACT: Retrieving historical fine particulate matter (PM_2.5) data is key for evaluating the long-term impacts of PM_2.5 on the environment, human health and climate change. Satellite-based aerosol optical depth has been used to estimate PM_2.5, but estimations have largely been undermined by massive missing values, low sampling frequency and weak predictive capability. Here, using a novel feature engineering approach to incorporate spatial effects from meteorological data, we developed a robust LightGBM model that predicts PM_2.5 at an unprecedented predictive capacity on hourly (R²= 0.75), daily (R²= 0.84), monthly (R²= 0.88) and annual (R²= 0.87) timescales. By taking advantage of spatial features, our model can also construct hourly gridded networks of PM_2.5. This capability would be further enhanced if meteorological observations from regional stations were incorporated. Our results show that this model has great potential in reconstructing historical PM_2.5 datasets and real-time gridded networks at high spatial-temporal resolutions. The resulting datasets can be assimilated into models to produce long-term re-analysis that incorporates interactions between aerosols and physical processes.

SUBMITTER: Zhong J

PROVIDER: S-EPMC8566180 | biostudies-literature | 2021 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Robust prediction of hourly PM2.5 from meteorological data using LightGBM.

Zhong Junting J Zhang Xiaoye X Gui Ke K Wang Yaqiang Y Che Huizheng H Shen Xiaojing X Zhang Lei L Zhang Yangmei Y Sun Junying J Zhang Wenjie W

National science review 20210105 10

Retrieving historical fine particulate matter (PM2.5) data is key for evaluating the long-term impacts of PM2.5 on the environment, human health and climate change. Satellite-based aerosol optical depth has been used to estimate PM2.5, but estimations have largely been undermined by massive missing values, low sampling frequency and weak predictive capability. Here, using a novel feature engineering approach to incorporate spatial effects from meteorological data ...[more]

PMID: 34858602

Similar Datasets

Project description:With the continuous promotion of urbanization in China, air pollution problems have become increasingly prominent in recent years. Various factors, such as emissions, meteorology, and physical and chemical reactions, jointly affect the severity of PM2.5 pollution to a large extent. This study selected five meteorological variables (planetary boundary layer height (PBLH), wind speed (WS), temperature(T), water vapor mixing ratio(Q), and precipitation (PCP)) for perturbation, and 21 different scenarios were set up. In this study, the effects of changes in a single meteorological variable on the pollutants produced in the area were represented by subtracting the baseline scenario (i.e., without perturbation of meteorological variables) simulated in January 2017 separately from each post-disturbance scenario. The results showed that Handan (HD) has the highest annual mean PM2.5 concentration of 85.75 μg/m3 in 2017, while all cities in study area exceeded the secondary concentration limit of urban atmospheric particulate matter. The correlation coefficient (R) between the simulation values of models and the actual monitoring values ranges from 0.41 to 0.74, indicating good model performance and acceptable simulation errors. PBLH (±10%-±20%), WS(±10%-±20%), and PCP(±10%-±20%) all showed a single adverse effect among the five meteorological variables, meaning that a reduction in these three factors led to an increase in PM2.5 concentrations. However, T (±1 K-±1.5 K) and Q (±10%-±20%) could indicate a positive impact under certain conditions. From the sensitivity calculations of single meteorological variables, it is clear that WS, PBLH, and PCP show a highly linear trend in all cities at the 0.01 level of significance. The hypothesis that T changes linearly in 10 cities in the study area is valid, while for Q, the hypothesis that Q changes linearly only occurs in Shijiazhuang and Baoding. When different meteorological variables are disturbed, there are significant spatial differences in the main affected areas of PM2.5 concentrations. By discussing the impact of meteorological variable disturbance on air quality in critically polluted cities in China, this study identified the meteorological variables that can substantially affect PM2.5 concentration. The more complex T and Q should be considered when formulating relevant emission measures.

Project description:Epidemiologic studies have found associations between fine particulate matter (PM2.5) exposure and adverse health effects using exposure models that incorporate monitoring data and other relevant information. Here, we use nine PM2.5 concentration models (i.e., exposure models) that span a wide range of methods to investigate i) PM2.5 concentrations in 2011, ii) potential changes in PM2.5 concentrations between 2011 and 2028 due to on-the-books regulations, and iii) PM2.5 exposure for the U.S. population and four racial/ethnic groups. The exposure models included two geophysical chemical transport models (CTMs), two interpolation methods, a satellite-derived aerosol optical depth-based method, a Bayesian statistical regression model, and three data-rich machine learning methods. We focused on annual predictions that were regridded to 12-km resolution over the conterminous U.S., but also considered 1-km predictions in sensitivity analyses. The exposure models predicted broadly consistent PM2.5 concentrations, with relatively high concentrations on average over the eastern U.S. and greater variability in the western U.S. However, differences in national concentration distributions (median standard deviation: 1.00 μg m-3) and spatial distributions over urban areas were evident. Further exploration of these differences and their implications for specific applications would be valuable. PM2.5 concentrations were estimated to decrease by about 1 μg m-3 on average due to modeled emission changes between 2011 and 2028, with decreases of more than 3 μg m-3 in areas with relatively high 2011 concentrations that were projected to experience relatively large emission reductions. Agreement among models was closer for population-weighted than uniformly weighted averages across the domain. About 50% of the population was estimated to experience PM2.5 concentrations less than 10 μg m-3 in 2011 and PM2.5 improvements of about 2 μg m-3 due to modeled emission changes between 2011 and 2028. Two inequality metrics were used to characterize differences in exposure among the four racial/ethnic groups. The metrics generally yielded consistent information and suggest that the modeled emission reductions between 2011 and 2028 would reduce absolute exposure inequality on average.

Dataset Information

Robust prediction of hourly PM_2.5 from meteorological data using LightGBM.

Publications

Robust prediction of hourly PM<sub>2.5</sub> from meteorological data using LightGBM.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets