The Emergence of Numerical Air Quality Forecasting Models and Their Application

Abstract

Over the past decade there has evolved an increasing interest by the public in the U.S. and other nations in the day-to-day air quality conditions to which they are exposed. Driven by the increasing awareness of the health aspects of air pollution exposure, especially by subpopulations most sensitive such as children and the elderly, short-term air pollution forecasts are being provided by more and more local authorities. Degradation of visibility in national parks and other pristine areas has also provided motivation for forecasts. Besides issuing alerts and warnings on air quality conditions, some local authorities are relying on air quality forecasts to put in place intermittent short-term management strategies such as free bus/rail fares, additional carpool strategies, burning bans, etc. Efforts to produce these short-term (1-3 day) forecasts, usually of ozone (O3), carbon monoxide (CO), fine particulate matter (PM2.5) and/or visibility rely on techniques ranging from persistence, to simple empirical local “rules-of-thumb”, to various statistical regression or neural network methods, to more complex models of the atmosphere. In a number of cases, nations or regions have launched efforts to use models that span large regional or national areas that can bridge the time and space scales between urban areas containing more dense air quality monitoring networks. This paper discusses the emergence of national/regional numerical air quality forecast (NAQF) model systems based on three dimensional grid (Eulerian) models, driven by mesoscale weather forecast models and source emissions models. We will focus on examples of such systems in the U.S. and Canada, although there are other notable forecast systems in place elsewhere. Recently a review of international air quality forecasting approaches (Cope and Hess, 2005) discussed example systems in Australia, Japan, Europe, and North America. They credit the availability of cost-effective high performance computers, advancements in high-resolution meteorological modeling, and the availability of real-time air quality monitoring data as having spurred the development of more sophisticated air quality model forecast systems. Most of the national efforts have been on regional-scale forecasting using models with horizontal grid sizes of 25-50 km, although model resolution is improving rapidly in many cases. In the review done by Cope and Hess (2005) several Eulerian model systems are highlighted. In Europe such systems are used to produce regional forecasts by Meteo France (Europe/France), University of Cologne (Europe/Germany), Swedish Meteorological and Hydrological Institute (Europe), and Norwegian Institute for Air Research (Northern Hemisphere), among others. The Australian forecast system, one of the first to become operational, employs a nested grid system with 1-5 km grids over the target areas of Sydney and Melbourne, with twice daily forecast cycles.