[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.

[1] Real-time forecasts of PM 2.5 aerosol mass from seven air quality forecast models (AQFMs) are statistically evaluated against observations collected in the northeastern United States and southeastern Canada from two surface networks and aircraft data during the summer of 2004 International Consortium for Atmospheric Research on Transport and Transformation (ICARTT)/New England Air Quality Study (NEAQS) field campaign. The AIRNOW surface network is used to evaluate PM 2.5 aerosol mass, the U.S. EPA STN network is used for PM 2.5 aerosol composition comparisons, and aerosol size distribution and composition measured from the NOAA P-3 aircraft are also compared. Statistics based on midday 8-hour averages, as well as 24-hour averages are evaluated against the AIRNOW surface network. When the 8-hour average PM 2.5 statistics are compared against equivalent ozone statistics for each model, the analysis shows that PM 2.5 forecasts possess nearly equivalent correlation, less bias, and better skill relative to the corresponding ozone forecasts. An analysis of the diurnal variability shows that most models do not reproduce the observed diurnal cycle at urban and suburban monitor locations, particularly during the nighttime to early morning transition. While observations show median rural PM 2.5 levels similar to urban and suburban values, the models display noticeably smaller rural/urban PM 2.5 ratios. The ensemble PM 2.5 forecast, created by combining six separate forecasts with equal weighting, is also evaluated and shown to yield the best possible forecast in terms of the statistical measures considered. The comparisons of PM 2.5 composition with NOAA P-3 aircraft data reveals two important features: (1) The organic component of PM 2.5 is significantly underpredicted by all the AQFMs and (2) those models that include aqueous phase oxidation of SO 2 to sulfate in clouds overpredict sulfate levels while those AQFMs that do not include this transformation mechanism underpredict sulfate. Errors in PM 2.5 ammonium levels tend to correlate directly with errors in sulfate. Comparisons of PM 2.5 composition with the U.S. EPA STN network for three of the AQFMs show that sulfate biases are consistently lower at the surface than aloft. Recommendations for further research and analysis to help improve PM 2.5 forecasts are also provided.

One year of observations from a network of five 915-MHz boundary-layer radar wind profilers equipped with radio acoustic sounding systems located in California's Central Valley are used to investigate the annual variability of convective boundary-layer depth and its correlation to meteorological parameters and conditions. Results from the analysis show that at four of the sites, the boundary-layer height reaches its maximum in the late-spring months then surprisingly decreases during the summer months, with mean July depths almost identical to those for December. The temporal decrease in boundary-layer depth, as well as its spatial variation, is found to be consistent with the nocturnal low-level lapse rate observed at each site. Multiple forcing mechanisms that could explain the unexpected seasonal behaviour of boundary-layer depth are investigated, including solar radiation, precipitation, boundary-layer mesoscale convergence, low-level cold-air advection, local surface characteristics and irrigation patterns and synoptic-scale subsidence. Variations in solar radiation, precipitation and synoptic-scale subsidence do not explain the shallow summertime convective boundary-layer depths observed. Topographically forced cold-air advection and local land-use characteristics can help explain the shallow CBL depths at the four sites, while topographically forced low-level convergence helps maintain larger CBL depths at the fifth site near the southern end of the valley.