Nonseparable Dynamic Nearest Neighbor Gaussian Process Models for Large Spatio-Temporal Data with an Application to Particulate Matter Analysis

Details

• Personal Author:
  Banerjee S ; Datta A ; Finley AO ; Hamm NAS ; Schaap M
• Description:
  Particulate matter (PM) is a class of malicious environmental pollutants known to be detrimental to human health. Regulatory efforts aimed at curbing PM levels in different countries often require high resolution space-time maps that can identify red-flag regions exceeding statutory concentration limits. Continuous spatio-temporal Gaussian Process (GP) models can deliver maps depicting predicted PM levels and quantify predictive uncertainty. However, GP-based approaches are usually thwarted by computational challenges posed by large datasets. We construct a novel class of scalable Dynamic Nearest Neighbor Gaussian Process (DNNGP) models that can provide a sparse approximation to any spatio-temporal GP (e.g., with nonseparable covariance structures). The DNNGP we develop here can be used as a sparsity-inducing prior for spatio-temporal random effects in any Bayesian hierarchical model to deliver full posterior inference. Storage and memory requirements for a DNNGP model are linear in the size of the dataset, thereby delivering massive scalability without sacrificing inferential richness. Extensive numerical studies reveal that the DNNGP provides substantially superior approximations to the underlying process than low-rank approximations. Finally, we use the DNNGP to analyze a massive air quality dataset to substantially improve predictions of PM levels across Europe in conjunction with the LOTOS-EUROS chemistry transport models (CTMs). [Description provided by NIOSH]
• Subjects:
  Air Pollutants, Occupational Air Pollution Database Environmental Exposure Environmental Pollution Geographic Information Systems Occupational Health Particulate Matter Safety
• Keywords:
  Air Pollutants Air Pollution Air Quality Measurement Algorithms Analytical Models Author Keywords: Nonseparable Spatio-temporal Models Bayesian Inference Databases Data Processing Environmental Pollutants Environmental Pollution Geographic Information Systems Geostatistics Information Retrieval Systems Information Systems Markov Chain Monte Carlo Mathematical Models Models Nearest Neighbors Particulates Permissible Concentration Limits Scalable Gaussian Process Statistical Analysis

  More +
• ISSN:
  1932-6157
• Document Type:
  Text
• Funding:
  Grant
• Genre:
  Journal Article
• Place as Subject:
  California ; Maryland ; Michigan ; OSHA Region 3 ; OSHA Region 5 ; OSHA Region 9
• CIO:
  National Institute for Occupational Safety and Health (NIOSH)
• Topic:
  Workplace Safety & Health
• Location:
  North America
• Volume:
  10
• Issue:
  3
• NIOSHTIC Number:
  nn:20051724
• Citation:
  Ann Appl Stat 2016 Sep; 10(3):1286-1316
• Contact Point Address:
  Abhirup Datta, Department of Biostatistics, Johns hopkins University, Baltimore, Maryland 55455, USA
• Email:
  abhidatta@jhu.edu
• Federal Fiscal Year:
  2016
• Performing Organization:
  University of California, Los Angeles
• Peer Reviewed:
  True
• Start Date:
  20130901
• Source Full Name:
  The Annals of Applied Statistics
• End Date:
  20170831
• Collection(s):
  National Institute for Occupational Safety and Health
• Main Document Checksum:
  urn:sha-512:54db6987a2d850e1b9bf7159845fcaa61ff48bc9069cb9406c5fb9a6db249c38c067f151f0a3f4b09709a90241df11cc3f50b7692b53148ffd816a73ba306114
• Download URL:
  https://stacks.cdc.gov/view/cdc/211523/cdc_211523_DS1.pdf
• File Type:
  [PDF - 2.54 MB ]