Assessment of an ensemble of seven real‐time ozone forecasts over eastern North America during the summer of 2004

Abstract

[1] The real-time forecasts of ozone (O 3 ) from seven air quality forecast models (AQFMs) are statistically evaluated against observations collected during July and August of 2004 (53 days) through the Aerometric Information Retrieval Now (AIRNow) network at roughly 340 monitoring stations throughout the eastern United States and southern Canada. One of the first ever real-time ensemble O 3 forecasts, created by combining the seven separate forecasts with equal weighting, is also evaluated in terms of standard statistical measures, threshold statistics, and variance analysis. The ensemble based on the mean of the seven models and the ensemble based on the median are found to have significantly more temporal correlation to the observed daily maximum 1-hour average and maximum 8-hour average O 3 concentrations than any individual model. However, root-mean-square errors (RMSE) and skill scores show that the usefulness of the uncorrected ensembles is limited by positive O 3 biases in all of the AQFMs. The ensembles and AQFM statistical measures are reevaluated using two simple bias correction algorithms for forecasts at each monitor location: subtraction of the mean bias and a multiplicative ratio adjustment, where corrections are based on the full 53 days of available comparisons. The impact the two bias correction techniques have on RMSE, threshold statistics, and temporal variance is presented. For the threshold statistics a preferred bias correction technique is found to be model dependent and related to whether the model overpredicts or underpredicts observed temporal O 3 variance. All statistical measures of the ensemble mean forecast, and particularly the bias-corrected ensemble forecast, are found to be insensitive to the results of any particular model. The higher correlation coefficients, low RMSE, and better threshold statistics for the ensembles compared to any individual model point to their preference as a real-time O 3 forecast.

An evaluation of real‐time air quality forecasts and their urban emissions over eastern Texas during the summer of 2006 Second Texas Air Quality Study field study

“…The region of overlap for all of the models shown in Figure 1a is determined by the STEM-12 km and BAMS-15 km model domain limits. Detailed background information on the models as they relate to both O 3 and PM 2.5 forecasts are provided by McKeen et al [2005] and McKeen et al [2007], respectively. A brief summary of the modeling systems, including the underlying emission inventories used for this study, and updates that they may have undergone between the 2004 and 2006 forecast seasons are given in Appendix A.…”

“…[72] The operational runs were based on CMAQ and interface module configurations described previously [Yu et al, 2007a;Otte et al, 2005;McKeen et al, 2005]. However, to rectify deficiencies in forecast results noted in previous years, several modifications to process modules, coupling between the meteorological and chemical calculations, and input data specification were tested in the developmental CMAQ-NAM configuration as summarized below.…”

Ensemble‐based chemical data assimilation. I: General approach

“…Finally, we discuss covariance inflation obtained through perturbing key model parameters: we call this model-specific inflation. We note that a better approach can be obtained by constructing multi-model ensembles (McKeen et al, 2005).…”

“…multi-model ensemble (McKeen et al, 2005). Another approach to preventing filter divergence is to prevent the ensemble from inbreeding (Houtekamer and Mitchell, 2001) by breaking the filter into two parts each of which acts on the other's input.…”

Predicting Air Quality: Current Status and Future Directions

“…The application of ensemble techniques to air quality forecasts is very recent [35]. As part of a collective, informal model verification project within the ICARTT/NEAQS-2004 study, forecasts of several key meteorological, radiation, and gas-phase atmospheric constituents were gathered in near real time (typically 4-to 10-hour computational delay) from seven CTMs and used to prepare and evaluate forecast skill for predicting surface ozone [80].…”