Predictability and error dynamics of street-canyon flows

Predictability is a subject of great importance in atmospheric modelling, yet there has been little research on urban boundary-layer flows. In the present work the predictability of street-canyon flow is examined numerically via large-eddy simulation. The nature of the error growth depends on the aspect ratio, but there is little evidence of an inverse error cascade. The error growth in physical space is rather inhomogeneous and shows clear differences between the shear layer and canyon. It is shown that linear theory yields a good estimate of the error growth rate and of the spatial structure of the error field. Implications for modelling will be discussed.