A Diagnostic Analysis of a Long-Term Regional Air Pollutant Transport Model

View More View Less
  • 1 Pacific Northwest Laboratory, Richland, WA 99352
© Get Permissions
Full access

Abstract

Predicted concentrations from the Regional Air Pollutant Transport (RAPT) model are compared with the corresponding observed values of sulfate, and the results used to define strengths and weaknesses in the model formulation.

RAPT was developed to provide long-term (i.e., monthly) average values of pollutants. It has hourly time steps, and incorporates a number of simplifying assumptions on mixing heights, horizontal diffusion and emission averaging. Daily predicted values were analyzed for diagnostic use only, rather than for verification of prediction ability.

The analysis indicates that the model performed reasonably well with regard to short-term temporal predictions of spatially averaged concentrations. Less confidence can be placed in site-specific predictions. Spatial patterns in the analysis highlight the sensitivity of the model's short-term simulations to several features including input data, boundary conditions and the assumption of horizontal dispersion being wholly defined by trajectory variations.

Abstract

Predicted concentrations from the Regional Air Pollutant Transport (RAPT) model are compared with the corresponding observed values of sulfate, and the results used to define strengths and weaknesses in the model formulation.

RAPT was developed to provide long-term (i.e., monthly) average values of pollutants. It has hourly time steps, and incorporates a number of simplifying assumptions on mixing heights, horizontal diffusion and emission averaging. Daily predicted values were analyzed for diagnostic use only, rather than for verification of prediction ability.

The analysis indicates that the model performed reasonably well with regard to short-term temporal predictions of spatially averaged concentrations. Less confidence can be placed in site-specific predictions. Spatial patterns in the analysis highlight the sensitivity of the model's short-term simulations to several features including input data, boundary conditions and the assumption of horizontal dispersion being wholly defined by trajectory variations.

Save