Dataset Information

Correcting Measurement Error in Satellite Aerosol Optical Depth with Machine Learning for Modeling PM2.5 in the Northeastern USA.

ABSTRACT: Satellite-derived estimates of aerosol optical depth (AOD) are key predictors in particulate air pollution models. The multi-step retrieval algorithms that estimate AOD also produce quality control variables but these have not been systematically used to address the measurement error in AOD. We compare three machine-learning methods: random forests, gradient boosting, and extreme gradient boosting (XGBoost) to characterize and correct measurement error in the Multi-Angle Implementation of Atmospheric Correction (MAIAC) 1 × 1 km AOD product for Aqua and Terra satellites across the Northeastern/Mid-Atlantic USA versus collocated measures from 79 ground-based AERONET stations over 14 years. Models included 52 quality control, land use, meteorology, and spatially-derived features. Variable importance measures suggest relative azimuth, AOD uncertainty, and the AOD difference in 30-210 km moving windows are among the most important features for predicting measurement error. XGBoost outperformed the other machine-learning approaches, decreasing the root mean squared error in withheld testing data by 43% and 44% for Aqua and Terra. After correction using XGBoost, the correlation of collocated AOD and daily PM2.5 monitors across the region increased by 10 and 9 percentage points for Aqua and Terra. We demonstrate how machine learning with quality control and spatial features substantially improves satellite-derived AOD products for air pollution modeling.

SUBMITTER: Just AC

PROVIDER: S-EPMC6497138 | biostudies-literature | 2018 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Correcting Measurement Error in Satellite Aerosol Optical Depth with Machine Learning for Modeling PM<sub>2.5</sub> in the Northeastern USA.

Just Allan C AC De Carli Margherita M MM Shtein Alexandra A Dorman Michael M Lyapustin Alexei A Kloog Itai I

Remote sensing 20180522 5

Satellite-derived estimates of aerosol optical depth (AOD) are key predictors in particulate air pollution models. The multi-step retrieval algorithms that estimate AOD also produce quality control variables but these have not been systematically used to address the measurement error in AOD. We compare three machine-learning methods: random forests, gradient boosting, and extreme gradient boosting (XGBoost) to characterize and correct measurement error in the Multi-Angle Implementation of Atmosp ...[more]

PMID: 31057954

Similar Datasets

Project description:Satellite-based PM(2.5) monitoring has the potential to complement ground PM(2.5) monitoring networks, especially for regions with sparsely distributed monitors. Satellite remote sensing provides data on aerosol optical depth (AOD), which reflects particle abundance in the atmospheric column. Thus AOD has been used in statistical models to predict ground-level PM(2.5) concentrations. However, previous studies have shown that AOD may not be a strong predictor of PM(2.5) ground levels. Another shortcoming of remote sensing is the large number of non-retrieval days (i.e., days without satellite data available) due to clouds and snow- and ice-cover. In this paper we propose statistical approaches to overcome these two shortcomings, thereby making satellite imagery a viable method to estimate PM(2.5) concentrations. First, we render AOD a robust predictor of PM(2.5) mass concentration by introducing an AOD daily calibration approach through the use of mixed effects model. Second, we develop models that combine AOD and ground monitoring data to predict PM(2.5) concentrations during non-retrieval days. A key feature of this approach is that we develop these prediction models separately for groups of days defined by the observed amount of spatial heterogeneity in concentrations across the study region. Subsequently, these methodologies were applied to examine the spatial and temporal patterns of daily PM(2.5) concentrations for both retrieval days (i.e., days with satellite data available) and non-retrieval days in the New England region of the United States during the period 2000-2008. Overall, for the years 2000-2008, our statistical models predicted surface PM(2.5) concentrations with reasonably high R(2) (0.83) and low percent mean relative error (3.5%). Also the spatial distribution of the estimated PM(2.5) levels in the study domain clearly exhibited densely populated and high traffic areas. The method we have developed demonstrates that remote sensing can have a tremendous impact on the fields of environmental monitoring and human exposure assessment.

Project description:The countries around the world are dealing with air quality issues for decades due to their mode of production and energy usages. The outbreak of COVID-19 as a pandemic and consequent global economic shutdown, for the first time, provided a base for the real-time experiment of the effect of reduced emissions across the globe in abetting the air pollution issue. The present study dealt with the changes in Aerosol Optical Depth (AOD), a marker of air pollution, because of global economic shutdown due to the coronavirus pandemic. The study considered the countries in south and south-east Asia (SSEA), Europe and the USA for their extended period of lockdown due to coronavirus pandemic. Daily Aerosol Optical Depth (AOD) from Moderate-resolution imaging spectroradiometer (MODIS) and tropospheric column density of NO2 and SO2 from Ozone monitoring instrument (OMI) sensors, including meteorological data such as wind speed (WS) and relative humidity (RH) were analyzed during the pre-lockdown (2017-2019) and lockdown periods (2020). The average AOD, NO2 and SO2 during the lockdown period were statistically compared with their pre-lockdown average using Wilcoxon-signed-paired-rank test. The accuracy of the MODIS-derived AOD, including the changing pattern of AOD due to lockdown was estimated using AERONET data. The weekly anomaly of AOD, NO2 and SO2 was used for analyzing the space-time variation of aerosol load as restrictions were imposed by the concerned countries at the different points of time. Additionally, a random forest-based regression (RF) model was used to examine the effects of meteorological and emission parameters on the spatial variation of AOD. A significant reduction of AOD (-20%) was obtained for majority of the areas in SSEA, Europe and USA during the lockdown period. Yet, the clusters of increased AOD (30-60%) was obtained in the south-east part of SSEA, the western part of Europe and US regions. NO2 reductions were measured up to 20-40%, while SO2 emission increased up to 30% for a majority of areas in these regions. A notable space-time variation was observed in weekly anomaly. We found the evidence of the formation of new particles for causing high AOD under high RH and low WS, aided by the downward vertical wind flow. The RF model showed a distinguishable relative importance of emission and meteorological factors among these regions to account for the spatial variability of AOD. Our findings suggest that the continued lockdown might provide a temporary solution to air pollution; however, to combat persistent air quality issues, it needs switching over to the cleaner mode of production and energy. The findings of this study, thus, advocated for alternative energy policy at the global scale.

Dataset Information

Correcting Measurement Error in Satellite Aerosol Optical Depth with Machine Learning for Modeling PM2.5 in the Northeastern USA.

Publications

Correcting Measurement Error in Satellite Aerosol Optical Depth with Machine Learning for Modeling PM<sub>2.5</sub> in the Northeastern USA.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets