Scientists from the Earth and Environment department at Boston University were able to integrate large temporal dataset provided by the High Performance Monitoring System into a model, which estimates on-road emissions in space and time.
Their model is unique because it does not rely on proxy variables such as population size and road density. This allows them to limit the size of the errors and therefore to conduct a full cross-section/time series panel regression of population density and vehicle emissions at local scale.
The team believes that the approach will be particularly useful to urban planners, because statistically emissions are strongly correlated with households, population density, diversity, and others. On the other hand, scientists are now able to examine the interrelation of emissions with population and income.
The findings were verified with on-road emissions data from the Emissions Database for Global Atmospheric Research, as well as with estimates derived from the state fuel consumption statistics. The findings indicate that density is positively correlated with emissions only up to 2000 persons per square kilometer. When the density increases to 4000, the correlation becomes negative, indicating that the higher the density the less on-road emissions. The explanation is that in smaller towns, the increase in population means more drivers on the road.
The new method has the potential to help the scientists find a solution to the on-road CO2 problem. Lucy Hutyra, assistant professor of earth and environment and study co-author, indicates that using emissions inventories with temporal and spatial characteristics, is the key to successful estimations. The lack of proxy values decreased the size of the error and allowed emphasizing on the shape of the response surface between density and on-road emissions.
In addition, the scientists outline the importance of using non-linear predictors when constructing emission inventories.