Sensitivity of inverse estimation of 2004 elemental carbon emissions inventory in the United States to the choice of observational networks

Choice of observational networks for inverse re-estimation of elemental carbon (EC) emissions in the United States impacts results. We convert the Thermal Optical Transmittance (TOT) EC measurements to the Thermal Optical Reflectance (TOR) equivalents to make full utilization of available networks in inverse modeling of EC using regional air quality model. Results show that using the Interagency Monitoring of Protected Visual Environments (IMPROVE) network gives significantly lower emissions estimate compared to using the Speciation Trends Network (STN) and other networks or using all available networks together. The re-estimate obtained by using IMPROVE sites alone made overall model performance worse compared to the bottom-up estimate of EC emissions, while both re-estimates, using STN (and others) sites and using all sites together, significantly improved the performance, showing higher robustness. Further analysis suggests that site density with respect to geographical location (downwind of source) impacts the robustness of a network's inverse re-estimate. Citation: Hu, Y., S. L. Napelenok, M. T. Odman, and A. G. Russell (2009), Sensitivity of inverse estimation of 2004 elemental carbon emissions inventory in the United States to the choice of observational networks, Geophys. Res. Lett., 36, L15806, doi:10.1029/2009GL039655.