# Methods for Estimating the Effect of Air Pollution on Asthma under a Changing Climate Open Access

## Alhanti, Brooke (2016)

Permanent URL: https://etd.library.emory.edu/concern/etds/6h440t06t?locale=en
Published

## Abstract

Climate models are complex mathematical representations of global and regional climate, which vary over space and time. Estimates of local meteorological variables provided by climate models are biased. More accurate estimates of future meteorological conditions are essential to investigating health impacts of climate change. Here we consider the association between ambient air pollution and respiratory disease under a changing climate. We use regional climate and air quality model outputs and develop calibration methods that aim to produce more accurate projections.

We first assess the association between ambient ozone (O3) and fine particulate matter (PM2.5) and asthma in the metro Atlanta area by age under recent meteorological conditions (1993-2009). We then propose a quantile-mapping approach where the quantile values of climate model outputs and historical data are regressed against each other with integrated piecewise splines to create calibrated projections.

We utilize copulas to develop a bivariate quantile calibration method that simultaneously calibrates the marginal distribution of each variable while capturing dependence between variables. This method estimates the bias between monotonic increasing quantile functions for climate models and monitoring data and applies this estimated bias to future climate projections. The Gumbel and Frank copulas are used to estimate the dependence between daily average temperature and solar radiation. Inference is conducted under a Bayesian framework to account for all sources of uncertainty in the projection. A cross-validation study is performed using historical data to evaluate the proposed approach. We apply our method to projections from four different climate models in Atlanta and find evidence for higher mean temperature and lower mean solar radiation in 2041-2070 compared to 1991-2000.

Finally, we consider spatial dependence in climate model outputs over a contiguous grid. We develop a spatial calibration method that calibrates climate model outputs over multiple locations and projects calibrated values on a gridded area. This spatial extension allows calibrated projections at locations that do not have a monitor. We project future levels of two air pollutants: O3 and PM2.5 on a spatial field covering northern Alabama and Georgia. We subsequently use these projected pollutant concentrations to project future asthma cases in Atlanta and Birmingham.

Contents

1 Introduction 4 2 Aim 1: Air Pollution and Respiratory Disease 10 2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3 Aim 2: Simultaneously calibrate climate model outputs with multiple variables 22 3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.2 Quantile Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.3 I-Splines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.3.1 I-Spline Examples . . . . . . . . . . . . . . . . . . . . . . . . 26 3.4 Bayesian Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 3.5 Univariate Calibration Method . . . . . . . . . . . . . . . . . . . . 33 3.6 Application of Univariate Method . . . . . . . . . . . . . . . . . . . 35 3.7 Bivariate Calibration Method . . . . . . . . . . . . . . . . . . . . . 44 3.7.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 3.7.2 Motivating Example . . . . . . . . . . . . . . . . . . . . . . 45 3.7.3 Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 3.7.4 General Modeling Framework . . . . . . . . . . . . . . . . . 49 3.7.5 Model Specication . . . . . . . . . . . . . . . . . . . . . . . 51 3.7.6 Parameter Estimation . . . . . . . . . . . . . . . . . . . . . 52 3.7.7 Cross-Validation Study . . . . . . . . . . . . . . . . . . . . . 53 3.7.8 Application to Climate Projections . . . . . . . . . . . . . . 55 3.7.9 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 4 Aim 3: Projection of Future Ambient Air Pollution and Asthma Emergency Department Visits 63 4.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 4.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 4.3 Data Descriptives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.4 Parameter Estimation . . . . . . . . . . . . . . . . . . . . . . . . . 68 4.4.1 Statistical Model . . . . . . . . . . . . . . . . . . . . . . . . 69 4.4.2 Prediction of New Location . . . . . . . . . . . . . . . . . . 74 4.5 Health Impact Projections . . . . . . . . . . . . . . . . . . . . . . . 75 4.5.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 4.5.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 4.5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 4.5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 5 Concluding Remarks 89