Dataset Information

A Satellite-Based Spatio-Temporal Machine Learning Model to Reconstruct Daily PM_2.5 Concentrations across Great Britain.

ABSTRACT: Epidemiological studies on the health effects of air pollution usually rely on measurements from fixed ground monitors, which provide limited spatio-temporal coverage. Data from satellites, reanalysis, and chemical transport models offer additional information used to reconstruct pollution concentrations at high spatio-temporal resolutions. This study aims to develop a multi-stage satellite-based machine learning model to estimate daily fine particulate matter (PM_2.5) levels across Great Britain between 2008-2018. This high-resolution model consists of random forest (RF) algorithms applied in four stages. Stage-1 augments monitor-PM_2.5 series using co-located PM₁₀ measures. Stage-2 imputes missing satellite aerosol optical depth observations using atmospheric reanalysis models. Stage-3 integrates the output from previous stages with spatial and spatio-temporal variables to build a prediction model for PM_2.5. Stage-4 applies Stage-3 models to estimate daily PM_2.5 concentrations over a 1 km grid. The RF architecture performed well in all stages, with results from Stage-3 showing an average cross-validated R² of 0.767 and minimal bias. The model performed better over the temporal scale when compared to the spatial component, but both presented good accuracy with an R² of 0.795 and 0.658, respectively. These findings indicate that direct satellite observations must be integrated with other satellite-based products and geospatial variables to derive reliable estimates of air pollution exposure. The high spatio-temporal resolution and the relatively high precision allow these estimates (approximately 950 million points) to be used in epidemiological analyses to assess health risks associated with both short- and long-term exposure to PM_2.5.

SUBMITTER: Schneider R

PROVIDER: S-EPMC7116547 | biostudies-literature | 2020 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Satellite-Based Spatio-Temporal Machine Learning Model to Reconstruct Daily PM<sub>2.5</sub> Concentrations across Great Britain.

Schneider Rochelle R Vicedo-Cabrera Ana M AM Sera Francesco F Masselot Pierre P Stafoggia Massimo M de Hoogh Kees K Kloog Itai I Reis Stefan S Vieno Massimo M Gasparrini Antonio A

Remote sensing 20201120 22

Epidemiological studies on the health effects of air pollution usually rely on measurements from fixed ground monitors, which provide limited spatio-temporal coverage. Data from satellites, reanalysis, and chemical transport models offer additional information used to reconstruct pollution concentrations at high spatio-temporal resolutions. This study aims to develop a multi-stage satellite-based machine learning model to estimate daily fine particulate matter (PM<sub>2.5</sub>) levels across Gr ...[more]

PMID: 33408882

Similar Datasets

Project description:Fine particulate matter is considered to be the most significant ambient air pollutant in terms of potential health impacts. Therefore, it is important that regulators are able to accurately assess the exposure of populations to PM10 and PM2.5 across municipal areas. We report on the practicalities of using a laser light scattering portable particulate monitor (Turnkey Instruments DustMate), in combination with a GPS, to map PM10 and PM2.5 concentrations on city-wide scales in Newcastle upon Tyne/Gateshead (UK), during a series of walking surveys. A heated inlet is necessary to remove moisture droplets from the sampled air prior to analysis by the instrument, though this also results in the loss of volatile particulate components, particularly from the PM2.5 fraction. A co-location calibration study was carried out with a reference urban background Tapered Element Oscillating Micro-Balance/Filter Dynamics Measuring System (TEOM-FDMS) system in Newcastle that is part of the UK's Automatic Urban and Rural Network (AURN) of air quality monitoring stations. For PM10, orthogonal regression of the DustMate against TEOM-FDMS data gave a slope and intercept of 1.02 ± 0.06 and -3.7 ± 1.2, respectively (R2 = 0.73), whereas for PM2.5, the respective values were 0.78 ± 0.06 and -0.63 ± 0.55 (R2 = 0.79). These parameters are comparable to literature calibration studies using this technology. There was good agreement between simultaneous samples taken using two DustMate instruments: for PM10, a slope and intercept of 1.05 ± 0.03 and 0.36 ± 0.5, respectively (R2 = 0.73), were obtained, whereas for the PM2.5, the respective values were 0.79 ± 0.01 and 0.19 ± 0.06 (R2 = 0.86). Correction factors based on the slope and intercepts obtained from the calibration exercise were applied to raw data collected from the DustMate. An annually-normalised correction procedure was then used to account for different background particulate concentrations on different sampling days. These corrected PM10 and PM2.5 concentrations and corresponding GPS coordinates were displayed on a base map using Google Fusion Tables and Google Earth Professional. Almost all areas surveyed in Newcastle/Gateshead were well below the EU Air Quality Standards for PM10 and PM2.5.

Project description:Ambient particulate matter of aerodynamic diameter less than 2.5 microns PM2.5) levels in Delhi routinely exceed World Health Organization (WHO) guidelines and Indian National Ambient Air Quality Standards (NAAQS) for acceptable levels of daily exposure. Only a handful of studies have examined the short-term mortality effects of PM in India, with none from Delhi examining the contribution of PM2.5.ObjectivesWe aimed to analyze the association between short-term PM2.5 exposures and daily nonaccidental mortality in Delhi, India.MethodsUsing generalized additive Poisson regression models, we examined the association between daily PM2.5 exposures and nonaccidental mortality between June 2010 and December 2016. Daily exposures to PM2.5 were estimated using an ensemble averaging technique developed by our research group, and mortality data were obtained from the Municipal Corporations of Delhi and the New Delhi Municipal Council.ResultsMedian exposures to PM2.5 were 91.1 µg/m3 (interquartile range = 68.9, 126.2), with minimum and maximum exposures of 21.4 µg/m3 and 276.7 µg/m3, respectively. Total nonaccidental deaths recorded in Delhi during the study period were 700,512. Each 25 µg/m3 increment in exposure was associated with a 0.8% (95% confidence intervals [CI] = 0.3, 1.3%) increase in daily nonaccidental mortality in the study population and a 1.5% (95% CI = 0.8, 2.2%) increase in mortality among those with 60 years of age or over. The exposure-response relationship was nonlinear in nature, with relative risk rising rapidly before tapering off above 125 µg/m3. Meeting WHO guidelines for acceptable levels of exposure over the study period would have likely averted 17,526 (95% CI = 6,837, 25,589) premature deaths, with older and male populations disproportionately affected.DiscussionThis study provides robust evidence of the impact of short-term exposure to PM2.5 on nonaccidental mortality with important considerations for various stakeholders including policymakers and physicians. Most importantly, we find that reducing exposures significantly below current levels would substantially decrease the mortality burden associated with PM2.5.

Dataset Information

A Satellite-Based Spatio-Temporal Machine Learning Model to Reconstruct Daily PM_2.5 Concentrations across Great Britain.

Publications

A Satellite-Based Spatio-Temporal Machine Learning Model to Reconstruct Daily PM<sub>2.5</sub> Concentrations across Great Britain.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets

Dataset Information

A Satellite-Based Spatio-Temporal Machine Learning Model to Reconstruct Daily PM2.5 Concentrations across Great Britain.

Publications

A Satellite-Based Spatio-Temporal Machine Learning Model to Reconstruct Daily PM<sub>2.5</sub> Concentrations across Great Britain.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets

A Satellite-Based Spatio-Temporal Machine Learning Model to Reconstruct Daily PM_2.5 Concentrations across Great Britain.