Summary. Wind tunnel and field experiments are reported in which continuous, synchronous measurements of grain transport rates and near-bed velocity profiles were made at one second intervals to assess mass-flux response to velocity variations. Resulting grain flux and velocity series demonstrate the variability concealed by conventional time-averaged data. In steady tunnel winds, timedependent mass transport rates are found to correlate better with fluctuations in mean velocity near the top of the saltation layer than with estimates of instantaneous shear stress. Quasi-periodic oscillation (20-30 seconds) of near-bed mass-flux and flow velocity in the lower regions of the inner boundary layer is evident in such airflows as the saltation system moves towards equilibrium with a developing bed form and confined boundary layer. This phenomenon may not occur in nature at these time-scales, however.In systematically unsteady airflows, the time constant between flux rate and velocity near the top of the saltation layer is shown to be of order one second, tentatively confirming Anderson and Haff's [2] calculations of saltation response time. Mass-flux also correlates well with large time-dependent variations in velocity in this region. Without grain replenishment, progressive surface re-sorting induces non-stationarity in grain flux under all observed flow regimes. Mass-flux and velocity histories measured on dunes show no correspondence. This difference is attributed to the stochastic nature of three-dimensional turbulence, the larger integral scales of atmospheric flows, measurement noise, and the effects of flow non-uniformity on satisfactory definition of shear velocity. Unsteady velocity profiles over a transverse dune are shown to be non-logarithmic above 20 cm but log-linear velocity segments of variable extent are found within the upper saltation layer.

No abstract

No abstract

This paper reviews some aspects of wind tunnel experiments on sand‐transporting winds. It follows previous papers that have discussed the influence of the outer region of the boundary layer on wind velocity measurements. This influence was quantified with the use Coles’Wake function. In this paper this correction is applied to six previously described wind velocity profiles. An attempt is made to calculate the profile parameter II from these measurements. The values found were not consistent with the expected II, which was determined by Coles for clean air flow. This value (II=0.55) was assumed to be valid in previous analyses for sand‐transporting winds. Evidence for a mutual dependency of friction velocity and profile parameter is presented and the difficulty in determining u. is pointed out. It is suggested that the constant stress region of the boundary layer should be kept large enough for measurements when Coles’Wake function is not to be used in the data analysis.

Experimental data are presented demonstrating the influence of boundary layer flow conditions on aerodynamic entrainment of grains in the absence of intersaltation collisions. New methods are proposed for (I) the unambiguous determination of aerodynamic threshold for any grain population and (2) approximation of the probability density function (PDF) distributions of threshold shear velocity for aerodynamic entrainment.In wind tunnel experiments, the orderly spatial development of flow conditions within a developing boundary layer over the roughened surface of a flat plate constrains the aerodynamic threshold condition in terms of both mean and fluctuating values. Initial grain dislodgements and subsequent erosion from narrow strips of loose, finely fractionated ballotini were recorded photographically as wind speed was increased. Boundary layer parameters, including average threshold shear velocity ( U.,), were calculated using the momentum integral method.Direct observations show that sporadic oscillation of grains preceded dislodgement. At slightly higher velocities most grains rolled over their neighbours before entering saltation. Initial entrainment in spatially semi-organized flurries of 50 or more grains was followed by quiescent periods at airflow velocities close to threshold. These observations provide strong circumstantial evidence linking both the nature and spatial pattern of initial grain motions to sweep events during the fluid bursting process.For each grain fraction, values of U., were found to span an unexpectedly wide range and to decrease downwind from the leading edge of the plate as turbulence intensity increased. A probabilistic entrainment model is applied to the aerodynamic threshold condition so as to incorporate the effects of changing turbulent flow regimes over the plate. Analysis of strip erosion curves gives both an objective definition of the threshold condition and usable approximations of the PDF for U,, required by the model and for future stochastic treatment of the threshold condition.