A suggested method for dispersion model evaluation.

Area of Science:

• Atmospheric Science
• Environmental Modeling
• Air Quality Assessment

Background:

• Operational atmospheric dispersion models are often evaluated on short-term concentration maxima.
• This approach is insufficient for a valid model evaluation.
• A valid procedure should focus on ensemble-average patterns in hourly concentration values.

Purpose of the Study:

• To present a valid model evaluation procedure for atmospheric dispersion models.
• To assess models' ability to replicate ensemble-average patterns.
• To provide statistically significant comparisons of model performance.

Main Methods:

• Analyzing observations to create average patterns for comparison.
• Accounting for uncertainties in observational data.
• Comparing AERMOD and ISCST3 simulation results with EPRI Kincaid experiment tracer data.
• Utilizing BOOT software with the ASTMD 6589 resampling procedure for objective assessment.

Main Results:

• Demonstrated a procedure to evaluate total mass at the receptor level (crosswind integrated concentration).
• Assessed the accuracy of lateral dispersion (mass spreading).
• Quantified uncertainty in transport characterization.
• Showcased objective assessment of model performance differences.

Conclusions:

• Air quality models predict average concentration patterns, not stochastic short-term maxima.
• The presented procedure facilitates proper estimation of ensemble average concentrations.
• It identifies model bias and provides statistical tests for skill differences between models.