Inherent uncertainties in atmospheric models: Weather and air pollution

Abstract

It is well known that there are reducible and irreducible uncertainties in both uncoupled and coupled meteorology-atmospheric chemistry models. Reducible (i.e., structural and parametric) uncertainties are attributable to our incomplete or inadequate understanding of the relevant atmospheric processes (e.g. chemical mechanism, PBL evolution, modeling domain, grid resolution, cloud treatment) and errors in model input data (e.g., emissions, boundary conditions). Inherent or irreducible uncertainties stem from our inability to properly characterize the atmosphere with appropriate initial conditions. When the initial state of the atmosphere is unknown, its future state cannot be predicted with great accuracy. There is an emerging need to properly assess these types of modeling uncertainties in order to improve the prediction accuracy of modeling systems. This work focuses on the assessment of inherent uncertainties in atmospheric and air quality modeling systems by estimating the impacts of various options for initial conditions on weather parameters and their consequent effect on atmospheric pollutant concentrations. Support for the modeling efforts is given by data collected from surface measurement networks for the meteorological and air quality parameters. We focus on the changes in atmospheric variables that strongly affect the fate and transport of air pollutants like ozone and aerosols.