The challenge of aeolian sediment transport prediction – insights from simple statistics

Equilibrium models of aeolian sediment transport are based on the presumption that a direct relationship exists between attributes of the wind field (e.g., speed, shear stress) and sediment flux. This monotonic, time-independent parameterization is conceptually elegant but has proven problematic in practice because of the lack of correspondence between measurements and predictions. The basic relationship between a time-varying wind field and resultant sediment pulses is surprisingly poorly understood, and as a consequence the averaging periods used to provide estimates of mean speed, shear velocity, and sediment flux are more a function of our capacity to measure, record, and analyze transport data rather than any inherent dynamics of the turbulent boundary layer or saltation system. In this study, a 40-minute time series of wind speed (measured at 30 cm above the sand surface) and sediment transport (measured approximately 2-3 cm above the surface using laser counters) was analyzed using conditional averaging approaches. Within the overall time series, there are discrete blocks when transport is continuous, when no transport occurred at all, and when transport was intermittent. Separating the time series into two populations (with transport; without transport) shows that the difference in the median (wind speed) values is greater than would be expected by chance (Mann-Whitney Rank Sum Test) despite approximately 50% overlap in the two frequency distributions. The degree of overlap is even greater during the intermittent periods, which implies that transport does not occur with some very large wind speeds and conversely transport does occur with some very small wind speeds. Blocks of active transport were then analyzed using different conditional averaging criteria to isolate separate populations according to transport intensity (count number). Neighbouring bins were compared sequentially and in many cases the difference in median values was not statistically significant. The implication is that the statistics of the wind are no different despite significantly different transport rate, which undermines the utility of the equilibrium models.