Environmental and Ecological Statistics

, Volume 18, Issue 1, pp 185–207 | Cite as

A fast Bayesian method for updating and forecasting hourly ozone levels



A Bayesian hierarchical space-time model is proposed by combining information from real-time ambient AIRNow air monitoring data, and output from a computer simulation model known as the Community Multi-scale Air Quality (Eta-CMAQ) forecast model. A model validation analysis shows that the model predicted maps are more accurate than the maps based solely on the Eta-CMAQ forecast data for a 2 week test period. These out-of sample spatial predictions and temporal forecasts also outperform those from regression models with independent Gaussian errors. The method is fully Bayesian and is able to instantly update the map for the current hour (upon receiving monitor data for the current hour) and forecast the map for several hours ahead. In particular, the 8 h average map which is the average of the past 4 h, current hour and 3 h ahead is instantly obtained at the current hour. Based on our validation, the exact Bayesian method is preferable to more complex models in a real-time updating and forecasting environment.


Bayesian inference Eta-CMAQ model Space-time forecasting Hierarchical model Separable models Spatial interpolation 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Cowles MK, Zimmerman DL (2003) A Bayesian space-time analysis of acid deposition data combined from two monitoring networks. J Geophys Res-Atmos. vol 108. doi: 10.1029/2003JD004001
  2. Fuentes M, Raftery A (2005) Model evaluation and spatial interpolation by Bayesian combination of observations with outputs from numerical models. Biometrics 61: 36–45PubMedCrossRefGoogle Scholar
  3. Gelfand AE, Sahu SK (2009) Combining monitoring data and computer model output in assessing environmental exposure. In: O’Hagan A, West M (eds) Handbook of Applied Bayesian Analysis (to appear)Google Scholar
  4. Jun M, Stein ML (2004) Statistical comparison of observed and CMAQ modeled daily sulfate levels. Atmos Environ 38: 4427–4436CrossRefGoogle Scholar
  5. Kang D, Mathur R, Rao ST, Yu S (2008) Bias adjustment techniques for improving ozone air quality forecasts. J Geophys Res 113: D23308. doi: 10.1029/2008JD010151 CrossRefGoogle Scholar
  6. McMillan N, Holland DM, Morara M, Feng J (2008) Combining numerical model output and particulate data using Bayesian space-time modeling. Environmetrics (to appear)Google Scholar
  7. Sahu SK, Yip S, Holland DM (2009) Improved space-time forecasting of next day ozone concentrations in the eastern U.S. Atmos Environ 43:494–501. doi: 10.1016/j.atmosenv.2008.10.028 Google Scholar
  8. Stein M (1999) Interpolation of spatial data: some theory for kriging. Springer, BerlinGoogle Scholar
  9. Zhang H (2004) Inconsistent estimation and asymptotically equal interpolations in model-based geostatistics. J Am Stat Assoc 99: 250–261CrossRefGoogle Scholar
  10. Zimmerman DL, Holland DM (2005) Complementary co-kriging: spatial prediction using data combined from several environmental monitoring networks. Environmetrics 16: 219–234CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2009

Authors and Affiliations

  1. 1.School of MathematicsUniversity of SouthamptonSouthamptonUK
  2. 2.Exeter Climate SystemsUniversity of ExeterExeterUK
  3. 3.U.S. Environmental Protection AgencyNational Exposure Research LaboratoryResearch Triangle ParkUSA

Personalised recommendations