This article considers the application of functional data analysis methods to modelling particulate matter emission profiles from dynamometer experiments. In particular the functional convolution model is introduced as an extension of the distributed lag model to functional (smooth and continuous) observations