tailieunhanh - A Bayesian kriged Kalman model for short-term forecasting of air pollution levels

The presentation "Biomass pollution basics" addresses the basics of biomass burning and introduces participants to the concept of incomplete combustion, the wide range of pollutants emitted from wood fires and stoves and typical pollutant concentrations. Two pollutants are of primary interest for both health effects and IAP monitoring: particulate matter (PM) and carbon monoxide (CO). Smaller particles ( and PM1) are likely to be most harmful, as they penetrate deep into the human lung. Larger particles are more likely to get 'filtered' by the upper respiratory tract. Considering available technologies and the relative cost and ease of monitoring, it is recommended that organizations focus on measuring levels of . While the. | Appl. Statist. 2005 54 Part 1 pp. 223-244 A Bayesian kriged Kalman model for short-term forecasting of air pollution levels Sujit K. Sahu University of Southampton UK and Kanti V. Mardia a UniveesityofLeeds UK Received April 2003. Final revision January 2004 A_ Kda Ul. a Summary. Short-term forecasts of air pollution levels in big cities are now reported in news-papersandother media outlets. Studiesindicatethateven short-term exposure to high levels of an air pollutant called atmospheric particulate matter can lead to long-term health effects. Data are typically observed at fixed monitoring stations throughout a study region of interest at different time points. Statistical spatiotemporal models are appropriate for modelling these S tíỊ. We consider short-term forecasting of these spatiotemporal processes by using a Bayesian kriged Kalman filtering model. The spatial prediction surface of the model is built by using the well-knownmethodefkriging for optimum spatial prediction and the temporal effects are analysed by using the models underlying the Kalman filtering method. The full Bayesian model isimplemented by using Markov chain Monte Carlo techniques which enable US to obtain the optimal Bayesian forecasts in time and space. A new cross-validation method based on the Mehalanobio distance hgtween the forecasts and observed data is also developed to assess the forecasting performance of the model implemented. Keywordm Benyingenergy Gibbs sampler Kalman filter Kriging Markov chain Monte Carlo methods Spatial temporal modelling state space model 1 IntsoducSion In recern yeyrs there bas rym a trommdour i y b in she sBBbstícd modds and techniques to analyse spatbsSemporal dath budhas dibpoilutioc bath. b ciBo icdii y. iair thild crine in many other ssntextm the primary rntereds in ynybhSbdg. duCm daSd dre to onth dnbprediot tmw sr oy some .