Air Quality Forecasting
The role of air pollution forecasts is growing as an AQM tool. Most forecasts produce 1–3-day O3 forecasts that predict exceedances of specific concentration thresholds. These forecasts can be used as a warning system for individuals sensitive to poor air quality and as an AQM tool. The techniques used to make these forecasts usually fall into one of four general types (Dye et al. 2000; NOAA 2001):
Phenomenological forecasts depend most heavily on the skill of the forecaster, who subjectively processes both air quality and meteorological information and, on the basis of past experience, formulates a prediction of pollution levels in the future. Despite its subjective nature, this approach can useful when used in conjunction with one or more of the objective methods discussed below, which have their own limitations.
Climatological methods rely on the association observed between increased pollution levels and specific meteorological conditions to predict future pollution episodes. The predictions can be based on a variety of approaches ranging from simple assumptions of persistence (for example, pollution levels remaining high the day after high levels occur) to more complex weather typing schemes (for example, identifying recurring weather patterns that are accompanied by high pollution levels).
Statistical methods rely on more quantitative relationships between pollution concentrations and meteorological parameters (for example, temperature and humidity) derived from past observations and weather forecasts. The statistical relationships can be as simple as linear regressions to more complex neural networks. These statistical relationships are then used in conjunction with weather forecasts to predict future pollution levels.
Mechanistic methods use chemical transport models (CTMs) in a so-called four-dimensional data assimilation mode to predict future pollutant concentrations. This approach typically makes use of data from monitoring networks to specify current air pollutant concentrations and meteorological fields and detailed meteorological forecasts from the National Weather Service to calculate pollutant concentrations in the future (McHenry et al. 2000; Chang and Cardelino 2002).
Although air pollution forecasting for secondary pollutants, such as O3, is a relatively new scientific endeavor, and the accuracy and predictive skill of the methods described above have yet to be evaluated comprehensively (NRC 2001a), such forecasting has begun to be used in an operational mode.
In some instances, predictions of future air pollution episodes are released to the public as an advisory. Individuals, especially those most sensitive to air pollution, can, if they choose, use the information from these advisories to alter their behavior in a way that would minimize their exposure to the pollution. For example, in AIRNow, probably the largest national air pollution forecasting effort, EPA provides the public with same-day and next-day O3 forecasts as well as PM2.5 forecasts for many cities. The AIRNow program developed an infrastructure to collect daily air quality forecasts from state and local agencies for over 250 cities and provide real-time and forecasted air quality data to the public through media outlets (Wayland et al. 2002). EPA compiles air quality forecasts and posts them to the AIRNow website for public utilization (www.epa.gov/airnow).