Modeling the space/time distribution of particulate matter in Thailand and optimizing its monitoring network


The space/time distribution of PM10 in Thailand is modeled using the Bayesian maximum entropy (BME) method of modern spatiotemporal geostatistics. Three kinds of BME spatiotemporal maps over Thailand are sought on the most polluted day for each year of a 6-year period from 1998 to 2003. These three maps are (1) the map of the BME estimate of daily PM10, (2) the map of the associated BME prediction error, and (3) the BME non-attainment map showing areas where the BME estimate does not attain a 68% probability of meeting the ambient standard for PM10. These detailed space/time PM10 maps provide invaluable information for decision-makers in air quality management. Knowing accurately the spatiotemporal distribution of PM10 is necessary to develop and evaluate strategies used to abate PM10 levels. The space/time BME estimate of PM10 on the worst day of the year offers a general picture as to where daily PM10 levels are not in compliance with the air-quality standard. Delineating these areas leads to the BME non-attainment maps, which are useful in identifying unhealthy zones, where sensitive population such as asthmatic children, seniors, or those with cardiopulmonary disease should be advised to avoid outdoor activities. The results of the space/time BME analysis of PM10 are further extended to assess whether the current monitoring network is adequate. The current distribution of monitoring stations can be evaluated by combining the available demographic information with the BME estimation error maps. Administrative districts with large population size and high BME normalized estimation error are suggested as the target for adding new monitoring stations. © 2007 Elsevier Ltd. All rights reserved.


Puangthongthub, S. Wangwongwatana, S. Kamens, R. M. Serre, M. L.


Atmospheric Environment, 2007


Bayesian maximum entropy, Monitoring network, Particulate matter, PM10, Space-time analysis, Thailan