A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model

Satellite-based models for estimating concentrations of particulate matter with an aerodynamic diameter less than 2.5 μm (PM<sub>2.5</sub>) have seldom been developed in islands with complex topography over the monsoon area, where the transport of PM<sub>2.5</sub> is influenc...

Full description

Bibliographic Details
Main Authors: Chau-Ren Jung, Wei-Ting Chen, Shoji F. Nakayama
Format: Article
Language:English
Published: MDPI AG 2021-09-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/13/18/3657
_version_ 1827681414809649152
author Chau-Ren Jung
Wei-Ting Chen
Shoji F. Nakayama
author_facet Chau-Ren Jung
Wei-Ting Chen
Shoji F. Nakayama
author_sort Chau-Ren Jung
collection DOAJ
description Satellite-based models for estimating concentrations of particulate matter with an aerodynamic diameter less than 2.5 μm (PM<sub>2.5</sub>) have seldom been developed in islands with complex topography over the monsoon area, where the transport of PM<sub>2.5</sub> is influenced by both the synoptic-scale winds and local-scale circulations compared with the continental regions. We validated Multi-Angle Implementation of Atmospheric Correction (MAIAC) aerosol optical depth (AOD) with ground observations in Japan and developed a 1-km-resolution national-scale model between 2011 and 2016 to estimate daily PM<sub>2.5</sub> concentrations. A two-stage random forest model integrating MAIAC AOD with meteorological variables and land use data was applied to develop the model. The first-stage random forest model was used to impute the missing AOD values. The second-stage random forest model was then utilised to estimate ground PM<sub>2.5</sub> concentrations. Ten-fold cross-validation was performed to evaluate the model performance. There was good consistency between MAIAC AOD and ground truth in Japan (correlation coefficient = 0.82 and 74.62% of data falling within the expected error). For model training, the model showed a training coefficient of determination (<i>R</i><sup>2</sup>) of 0.98 and a root mean square error (RMSE) of 1.22 μg/m<sup>3</sup>. For the 10-fold cross-validation, the cross-validation <i>R</i><sup>2</sup> and RMSE of the model were 0.86 and 3.02 μg/m<sup>3</sup>, respectively. A subsite validation was used to validate the model at the grids overlapping with the AERONET sites, and the model performance was excellent at these sites with a validation <i>R</i><sup>2</sup> (RMSE) of 0.94 (1.78 μg/m<sup>3</sup>). Additionally, the model performance increased as increased AOD coverage. The top-ten important predictors for estimating ground PM<sub>2.5</sub> concentrations were day of the year, temperature, AOD, relative humidity, 10-m-height zonal wind, 10-m-height meridional wind, boundary layer height, precipitation, surface pressure, and population density. MAIAC AOD showed high retrieval accuracy in Japan. The performance of the satellite-based model was excellent, which showed that PM<sub>2.5</sub> estimates derived from the model were reliable and accurate. These estimates can be used to assess both the short-term and long-term effects of PM<sub>2.5</sub> on health outcomes in epidemiological studies.
first_indexed 2024-03-10T07:15:47Z
format Article
id doaj.art-c63d33b780e24b8dadc78b6b8aa10e6e
institution Directory Open Access Journal
issn 2072-4292
language English
last_indexed 2024-03-10T07:15:47Z
publishDate 2021-09-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj.art-c63d33b780e24b8dadc78b6b8aa10e6e2023-11-22T15:06:23ZengMDPI AGRemote Sensing2072-42922021-09-011318365710.3390/rs13183657A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest ModelChau-Ren Jung0Wei-Ting Chen1Shoji F. Nakayama2Japan Environment and Children’s Study Programme Office, National Institute for Environmental Studies, Tsukuba 305-8506, JapanDepartment of Atmospheric Sciences, National Taiwan University, Taipei 106319, TaiwanJapan Environment and Children’s Study Programme Office, National Institute for Environmental Studies, Tsukuba 305-8506, JapanSatellite-based models for estimating concentrations of particulate matter with an aerodynamic diameter less than 2.5 μm (PM<sub>2.5</sub>) have seldom been developed in islands with complex topography over the monsoon area, where the transport of PM<sub>2.5</sub> is influenced by both the synoptic-scale winds and local-scale circulations compared with the continental regions. We validated Multi-Angle Implementation of Atmospheric Correction (MAIAC) aerosol optical depth (AOD) with ground observations in Japan and developed a 1-km-resolution national-scale model between 2011 and 2016 to estimate daily PM<sub>2.5</sub> concentrations. A two-stage random forest model integrating MAIAC AOD with meteorological variables and land use data was applied to develop the model. The first-stage random forest model was used to impute the missing AOD values. The second-stage random forest model was then utilised to estimate ground PM<sub>2.5</sub> concentrations. Ten-fold cross-validation was performed to evaluate the model performance. There was good consistency between MAIAC AOD and ground truth in Japan (correlation coefficient = 0.82 and 74.62% of data falling within the expected error). For model training, the model showed a training coefficient of determination (<i>R</i><sup>2</sup>) of 0.98 and a root mean square error (RMSE) of 1.22 μg/m<sup>3</sup>. For the 10-fold cross-validation, the cross-validation <i>R</i><sup>2</sup> and RMSE of the model were 0.86 and 3.02 μg/m<sup>3</sup>, respectively. A subsite validation was used to validate the model at the grids overlapping with the AERONET sites, and the model performance was excellent at these sites with a validation <i>R</i><sup>2</sup> (RMSE) of 0.94 (1.78 μg/m<sup>3</sup>). Additionally, the model performance increased as increased AOD coverage. The top-ten important predictors for estimating ground PM<sub>2.5</sub> concentrations were day of the year, temperature, AOD, relative humidity, 10-m-height zonal wind, 10-m-height meridional wind, boundary layer height, precipitation, surface pressure, and population density. MAIAC AOD showed high retrieval accuracy in Japan. The performance of the satellite-based model was excellent, which showed that PM<sub>2.5</sub> estimates derived from the model were reliable and accurate. These estimates can be used to assess both the short-term and long-term effects of PM<sub>2.5</sub> on health outcomes in epidemiological studies.https://www.mdpi.com/2072-4292/13/18/3657aerosol optical depthPM<sub>2.5</sub>random forest modelsatellite-based estimation model
spellingShingle Chau-Ren Jung
Wei-Ting Chen
Shoji F. Nakayama
A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model
Remote Sensing
aerosol optical depth
PM<sub>2.5</sub>
random forest model
satellite-based estimation model
title A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model
title_full A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model
title_fullStr A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model
title_full_unstemmed A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model
title_short A National-Scale 1-km Resolution PM<sub>2.5</sub> Estimation Model over Japan Using MAIAC AOD and a Two-Stage Random Forest Model
title_sort national scale 1 km resolution pm sub 2 5 sub estimation model over japan using maiac aod and a two stage random forest model
topic aerosol optical depth
PM<sub>2.5</sub>
random forest model
satellite-based estimation model
url https://www.mdpi.com/2072-4292/13/18/3657
work_keys_str_mv AT chaurenjung anationalscale1kmresolutionpmsub25subestimationmodeloverjapanusingmaiacaodandatwostagerandomforestmodel
AT weitingchen anationalscale1kmresolutionpmsub25subestimationmodeloverjapanusingmaiacaodandatwostagerandomforestmodel
AT shojifnakayama anationalscale1kmresolutionpmsub25subestimationmodeloverjapanusingmaiacaodandatwostagerandomforestmodel
AT chaurenjung nationalscale1kmresolutionpmsub25subestimationmodeloverjapanusingmaiacaodandatwostagerandomforestmodel
AT weitingchen nationalscale1kmresolutionpmsub25subestimationmodeloverjapanusingmaiacaodandatwostagerandomforestmodel
AT shojifnakayama nationalscale1kmresolutionpmsub25subestimationmodeloverjapanusingmaiacaodandatwostagerandomforestmodel