Online Program Home
My Program

Abstract Details

Activity Number: 359
Type: Contributed
Date/Time: Tuesday, August 2, 2016 : 10:30 AM to 12:20 PM
Sponsor: Section on Statistics and the Environment
Abstract #319529 View Presentation
Title: Modeling Nonstationary and Anisotropic Geostatistical Data Processes
Author(s): Claude Hill* and David M. Thompson
Companies: University of Oklahoma Health Sciences Center and University of Oklahoma Health Sciences Center
Keywords: Geostatistics ; Nonstationarity ; Anisotropy ; Pollution ; Kriging ; NO2
Abstract:

Introduction: Quantification of air quality is a fundamental issue for identification of exposure to pollutants. NOx can combine with other pollutants to form nitrogen dioxide (NO2), ozone (O3) and fine particulate matter (PM2.5) which are associate with the increased occurrence of respiratory illnesses and increased mortality rates. Methods to analyze nonstationary and anisotropic geostatistical processes are used to model NO2.

Methods: Data from the Environmental Protection Agency's pollution monitoring stations in the United States are used. Daily totals of NO2 are collected at 445 locations between January 1, 2014 and June 30, 2015. A random sample of 10% of the sites are removed from the data. We create four models based on a combination of nonstationarity and anisotropic assumptions. We then use geostatistical kriging methods to predict the NO2 levels at these sites. A comparison of the four models are performed using mean square prediction error (MSPE).

Results: Results will be presented according to the spatial assumptions used for each model. The predicted values at the locations will be compared to the observed levels. The model with the lowest mean squar


Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

 
 
Copyright © American Statistical Association