An analysis of error propagation in AERMOD lateral dispersion using Round Hill II and Uttenweiller experiments in reduced averaging times

Abstract

Dispersion modelling was proved by researchers that most part of the models, including the regulatory models recommended by the Environmental Protection Agency of the United States (AERMOD and CALPUFF), do not have the ability to predict under complex situations. This article presents a novel evaluation of the propagation of errors in lateral dispersion coefficient of AERMOD with emphasis on estimate of average times under 10 min. The sources of uncertainty evaluated were parameterizations of lateral dispersion ([Formula: see text]), standard deviation of lateral wind speed ([Formula: see text]) and processing of obstacle effect. The model's performance was tested in two field tracer experiments: Round Hill II and Uttenweiller. The results show that error propagation from the estimate of [Formula: see text] directly affects the determination of [Formula: see text], especially in Round Hill II experiment conditions. After average times are reduced, errors arise in the parameterization of [Formula: see text], even after observation assimilations of [Formula: see text], exposing errors on Lagrangian Time Scale parameterization. The assessment of the model in the presence of obstacles shows that the implementation of a plume rise model enhancement algorithm can improve the performance of the AERMOD model. However, these improvements are small when the obstacles have a complex geometry, such as Uttenweiller.

Modeling the air pollutant concentration near a cement plant co-processing wastes

“…To validate the reliability of a model, many researchers use statistical indicators to compare the discrepancy between the observations and predictions. 38 The commonly used simple metrics employed to quantify the difference between the modeled and observed concentrations is the fraction of predictions within a factor of two of observations (FAC2). The model is considered acceptable when the value of FAC2 is in the range of 0.5 to 2: 0.5 ≤ FAC2 = C m / C o ≤ 2 where C o and C m are the observed and modeled concentrations, respectively.…”