Abstract:Increased concern about the ability to accurately model and predict emissions from motor vehicles prompted this research. The ability of the mathematical algorithms contained in version 4 of the CALINE line source dispersion model (CALINE4) developed by Caltrans to accurately predict carbon monoxide (CO) emissions from a fleet of motor vehicles is assessed. The CALINE4 model contains algorithms that predict CO emissions from discrete modal events of idle, cruise, acceleration, and deceleration. A BASIC computer program is used to assess and compare the performance of the CALINE4 algorithms with those incorporated in version 7F of the EMFAC model (EMFAC7F), which is used and developed by the California Air Resources Board. The statistical assessment includes comparisons of mean prediction bias, Theil's U-Statistic, and the linear correlation coefficient. The analyses demonstrate that the currently used CALINE4 algorithms perform similarly to those contained in EMFAC7F, but when modified to use individual emission rates (instead of fleet average emission rates), the CALINE4 algorithms generally outperform the EMFAC7F algorithms. For short- to medium-term microscale model improvements, it is recommended that the CALINE4 model be revised to (a) incorporate individual emission rates into its emission estimation algorithms, (b) update its statistically derived model coefficients, and (c) update the modal activity algorithms to cover all modeling scenarios. For long-term modeling improvements, it is recommended that a more robust modal model be estimated based on second-by-second data and additional causal variables, and true vehicle simulation models be used to estimate vehicle activity.