Despite many years of field measurements there are very few direct observations of the aerodynamic roughness length (z₀) of the sea surface for 10m, averaged wind speeds greater than 25 m/s. Knowledge of z₀ (or equivalently the drag coefficient, C_D10n) under these conditions is needed, for example, for hurricane research and forecasting. Unfortunately it is at these high wind speeds that the available parameterisations diverge significantly. For example, we have recently proposed a formula, based on the height and steepness of the significant waves (Taylor and Yelland, 2001: TY2001 hereafter), which predicts significantly lower roughness at high winds than that which would be predicted by many wave-age based formulas.

This paper will review measurements of surface roughness from the laboratory and the field. We will show that for a range of laboratory studies, the TY2001 formula not only predicts that the roughness should be of similar magnitude to the open ocean, but also predicts detailed characteristics of the observed roughness. The TY2001 formula also predicts the varying roughness characteristics observed in a range of field experiments. This suggests that the mechanisms causing roughness are similar for waves in the laboratory and the field. This conclusion is in marked contrast to wave age based formula which, if tuned to field experiments, significantly over estimate the roughness observed at the extremely short fetches and very young waves studied in the laboratory. Thus proponents of such formulas have argued that laboratory and field waves are fundamentally dissimilar. They show evidence that roughness at very short fetches in the field is higher than observed either in the open ocean or the laboratory.

In attempting to reconcile these opposing views we shall consider the effects of spurious relationships (for example where the Charnock parameter is plotted as a function of wave age). Indeed we will show that, although the TY2001 formula does not contain a wave age term, it does reproduce the apparent wave age dependence of z₀ observed in some field experiments. Thus, while no simple parameterisation may be capable of modelling all possible states of the sea surface, we suggest that the TY2001 formula is capable of predicting the observed surface roughness under most of the conditions which, in practice, are normally observed. We also suggest that it gives a better estimate of the roughness at very high wind speeds compared to other formulas.