Advanced gap-filling techniques in satellite-based PM2.5 exposure models and their applications in air pollution epidemiology Pubblico

Abstract

Satellite-based models relating the satellite-derived parameter aerosol optical depth (AOD) to ground-level PM2.5 have a number of advantages over more traditional approaches to exposure characterization. Namely, observations are available, via satellite, at a daily level over areas that lack traditional ground monitors. However, satellite-based models are also subject to cloud and snow-driven missingness which may bias results in unpredictable ways, particularly when satellite-derived PM2.5 estimates are further used in human health studies. This dissertation focuses on the problem of missing observations in satellite retrievals. In chapter 1 we characterize the scope of the missing data problem in the daily Moderate resolution imaging spectroradiometer (MODIS) Aqua AOD product, finding that observations were only present in the United States on an average of 30% of possible days with the remaining 70% missing as a result of cloud-cover and surface brightness. In chapter 2 we strive to understand drivers, in terms of cloud cover properties and correlated meteorological conditions, behind differences between cloudy and non-cloudy PM2.5 observations. We found that changes in temperature, wind speed, relative humidity, planetary boundary layer height, convective available potential energy, precipitation, cloud effective radius, cloud optical depth, and cloud emissivity were all associated with changes in PM2.5 concentrations and composition at two sites, one in Atlanta, and one in San Francisco. A case study at the San Francisco site confirmed that accounting for cloud-cover and associated meteorological conditions could alter average concentrations and the predicted spatial distribution of daily PM2.5 concentrations. In chapter 3, we aim to develop and compare the ability of different gap-filling methods to eliminate bias resulting from missing AOD observations in human health studies. We find that different gap-filling models produce comparable odds ratios in a study on the relationship between emergency department visits for asthma or wheeze, otitis media, and upper respiratory infection. However, when gap-filling was not used, odds ratios were attenuated towards the null for two of the three possible outcomes. This dissertation highlights the importance of understanding and correcting for missing observations in satellite retrievals when estimating PM2.5 concentrations. 

Table of Contents

Table of contents

 

 

 

Introduction  . . . . . . . . . . .  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1

 

Particulate matter and human mortality . . . . . . . . . . . . . . . . . . .  7

 

  The reanalyses and extensions . . . . . . . . . . . . . . . . . . . . . . . . . .  11

 

  Cohorts galore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

 

The multi-city studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  14

 

PM and hospitalizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  17

 

PM mechanisms of action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

 

Remote sensing of aerosols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

 

Works Cited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  23

 

Chapter 1: Evaluation of Aqua MODIS Collection 6 AOD Parameters for Air Quality Research over the Continental United States . . . . . . . . . . . . . . . . .  31

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  32

Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  35

Satellite and ground datasets . . . . . . . . . . . . . . . . . . . . . .  35

Coverage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Coverage of high confidence retrievals . . . . . . . . . . . . . . 39

Coverage gains with lower-quality retrievals . . . . . . . . . . 40

Accuracy of high confidence retrievals . . . . . . . . . . . . . .  41

Accuracy assessment of lower-confidence retrievals . . .  42

Dependence of retrieval errors on flight geometry and landcover type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  43 

Dependence of retrieval errors on season and weather conditions . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  45

Case study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  47

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Supplementary materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Author contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  52

Conflicts of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  52

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

Chapter 2: The Potential Impact of Satellite-Retrieved Cloud Parameters on Ground-Level PM2.5 Mass and Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  59

Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Study area characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Clouds and 24-hour gravimetric mass . . . . . . . . . . . . . . . .  70

Clouds and speciation of PM2.5 . . . . . . . . . . . . . . . . . . . . . . 72

Application to MAIAC-derived PM2.5 . . . . . . . . . . . . . . . .  73

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

Supplementary materials . . . . . . . . . . .  . . . . . . . . . . . . . . . . . . . . . .  78

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  79

Author contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Conflicts of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  79

Chapter 3: Effect attenuation in the relationship between pediatric ED visits and satellite-based PM2.5 exposure  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  86

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

Methods – Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

Health Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

Exposure model inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

Methods – Exposure models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

No cloud gap-filled satellite model . . . . . . . . . . . . . . . . . . . . . 92

Cloud gap-filled satellite model . . . . . . . . . . . . . . . . . . . . . . . . 92

Methods – Epidemiology models . . . . . . . . . . . . . . . . . . . . . . . . . . . .  93

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

Tables and figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  107

 

About this Dissertation

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Laney Graduate School
Department	Environmental Health Sciences
Degree	Ph.D.
Submission	Dissertation
Language	English
Research Field	Environmental Health Remote Sensing
Parola chiave	AOD Remote sensing Aerosols Air pollution
Committee Chair / Thesis Advisor	Liu, Yang, Emory University
Committee Members	Strickland, Matthew, University of Nevada, Reno Chang, Howard, Emory University Mulholland, James, Georgia Institute of Technology

Ultima modifica

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Advanced gap-filling techniques in satellite-based PM2.5 exposure models and their applications in air pollution epidemiology ()	2018-04-05 14:35:41 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions