Quantifying Uncertainty in Concentration Output Generated by Atmospheric and Diffusion Models

One of the main problems that the air-pollution meteorologist faces everyday is how to quantify uncertainty in concentration data obtained from transport and diffusion models. The sources of uncertainty can arise from many model inputs (i.e., weather data) or parameters (i.e., model settings). However, one of the main sources of uncertainty arises from error in the horizontal wind data that is used by models. Since winds are probably the most critical model input in determining source to sampler transport and ultimately the concentration output, it is very important to assess possible errors in these data. This paper will describe a system called CLUES (Concentration Level Uncertainty Ensemble System), which has been developed to provide information about the uncertainty in concentration output generated by atmospheric transport and diffusion models. CLUES use an ensemble approach to obtain the uncertainty information. There are four main steps in the CLUES basic approach. The first is to make use of a statistical mode to generate simulated errors in the wind fields. The next step is to use these simulated errors to perturb the winds and create an ensemble of transport and diffusion output using a Monte Carlo Method. Once the ensemble is created, statistics (standard deviations, confidence limits, and other quantities) are generated and displayed graphically to help the user to assess overall model output uncertainty. Finally, CLUES also provides a way to check for convergence of the ensemble. Besides presenting an overall structure of CLUES, we will also present examples of how the system can be applied to “real” world transport and diffusion models. The CLUES method will be applied to both observed and mesoscale (i.e., RAMS) data.