Evaluation of the COPERT III emission model with on-road optical remote sensing measurements

At present, the COPERT III model is the most commonly used model in Europe for official national inventories of emissions from road traffic. In this study, the COPERT III model (version 2.1) was evaluated by utilizing a dataset available from on-road optical remote sensing emission measurements on a large number of vehicles at three different sites in Gothenburg, Sweden, in 2001 and 2002. The remote sensing dataset contained fuel-specific emissions (grams of pollutant emitted per liter of fuel burnt) of carbon monoxide (CO), nitrogen oxide (NO) and hydrocarbons (HC) as well as speed and acceleration data for individual vehicles. For gasoline passenger cars, a total of approximately 20 000 records with valid CO and HC remote sensor readings, and 16 000 records with valid NO readings were available for the COPERT III evaluation. For diesel passenger cars and heavy-duty vehicles the remote sensing dataset contained 1100 and 650 records with valid NO readings, respectively. Average fuel-specific emission factors derived from the remote sensing measurements were compared with corresponding emission factors derived from COPERT III calculations for urban, hot stabilized conditions and average speed around 45 km h−1. The results show a good agreement between the two methods for gasoline passenger cars’ NOx emissions for all COPERT III subsectors (i.e. cylinder volume classes) and technology classes (e.g. EURO 1, 2, 3). In the case of CO emissions, the agreement was less favorable, with the model overpredicting emissions for all but one of the technology classes. For gasoline passenger cars’ HC emissions the model-to-measurement agreement was reasonably good, although with a tendency for the model to overpredict the emissions. There was also a relatively good agreement for NOx emission factors for diesel passenger cars. Finally, NOx emission factors for heavy-duty vehicles according to the COPERT III model were systematically lower than those from the remote sensing measurements, and in particular the reduction between EURO 2 and EURO 3 tended to be overestimated by the model. The study has demonstrated the potential and usefulness of on-road optical remote sensing for emission model evaluation purposes.