WIT Press


Errors In Model Predictions Of NOx Traffic Emissions At Road Level – Impacts Of Input Data Quality

Price

Free (open access)

Paper DOI

10.2495/AIR080271

Volume

116

Pages

15

Page Range

255 - 269

Published

2008

Size

424 kb

Author(s)

R. Smit

Abstract

This study investigates the effects of three important input variables on the prediction accuracy of average speed emission models. These variables are average speed, basic traffic composition (proportion of heavy-duty vehicles) and model choice (COPERT, QGEPA). Sensitivity analysis (conditional NRSA) is used to determine to what extent the possible range in these input variables influences model outcomes (i.e. NOx emissions for road links), and hence accuracy. It is shown that maximum errors can be large (up to a factor of about 3.5). Moreover, they are a function of the level of congestion with errors generally increasing with the level of congestion. Traffic composition is shown to most strongly affect NOx emissions (29-241%), followed by average speed (2- 168%) and model choice (0-177%). The results were similar for arterial roads and freeways. These results can be used to provide direction to the collection of model input data, further model development and model application. The external errors found in this study appear to be of the same order of magnitude as internal errors that have been reported from (partial) road validation studies. This implies that in terms of further improvements of traffic emission modeling, focus should be on both the quality of input data (application) and the quality of the actual emission models (model development). Given the relevance of these results, it would be worthwhile to extend and refine this work by including other air pollutant and greenhouse gas emissions, and to use more complex traffic and emission models. Keywords: accuracy, error, road traffic emission, modeling, sensitivity analysis, NOx, congestion.

Keywords

accuracy, error, road traffic emission, modeling, sensitivity analysis, NOx, congestion.