Probability analyses of combining background concentrations with model-predicted concentrations.

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Air Quality Modeling

Background:

• Ambient air quality standards require comparing total concentrations (monitored + modeled) to regulatory limits.
• Current methods often combine background and predicted concentrations inaccurately, leading to overpredictions.
• Existing approaches, like adding 99th percentiles, are mathematically incorrect and do not reflect combined probability distributions.

Purpose of the Study:

• To review current techniques for determining background concentrations and combining them with modeled data.
• To propose a statistically sound method for calculating total pollutant concentrations.
• To ensure compliance modeling aligns with the probabilistic nature of National Ambient Air Quality Standards (NAAQS).

Main Methods:

• Review of existing background concentration determination and combination techniques.
• Development and application of a pairwise combination approach for observed background and predicted concentrations.
• Calculation of joint probabilities from all possible pairwise combinations of daily maximum concentrations.

Main Results:

• The pairwise combination method generates a realistic distribution of total concentrations.
• This approach yields a more accurate 99th percentile total value compared to simple addition of distribution peaks.
• The proposed method avoids the 'double counting' issue inherent in current EPA guidance.

Conclusions:

• Current EPA methods for compliance modeling often overpredict total air pollution impacts.
• Adding the highest values (e.g., 99th percentiles) from separate background and model distributions results in design values exceeding NAAQS probability requirements.
• The pairwise combination approach provides a more accurate and statistically appropriate method for assessing total pollutant concentrations against air quality standards.