Abstract Reliable predictions of wind erosion depend on the accuracy of determining whether erosion occurs or not. Among the several factors that govern the initiation of soil movement by wind, surface moisture is one of the most significant. Some widely used models that predict the threshold shear velocity for particle detachment of wet soils by wind were critically reviewed and evaluated. Wind-tunnel experiments were conducted on pre-wetted dune sand with moisture contents ranging from 0·00 to 0·04 kg kg−1. Sand samples were exposed to different wind speeds for 2 min. Moisture content was determined gravimetrically before and after each experiment, and the saltation of sand particles was recorded electronically with a saltiphone. Shear velocities were deduced from the wind speed profiles. For each moisture content, the experiments were repeated at different shear velocities, with the threshold shear velocity being determined by least-squares analysis of the relationships between particle number rates and shear velocity.
Within the 2-min test runs, temporal changes in particle number rates and moisture contents were detected. A steep increase in the threshold shear velocity with moisture content was observed. When comparing the models, large differences between the predicted results became apparent. At a moisture content of 0·007 kg kg−1, which is half the moisture content retained to the soil matrix at a water tension (or matric potential) of −1·5 MPa, the increase in ‘wet’ threshold shear velocity predicted with the different models relative to the dry threshold shear velocity ranged from 117% to 171%. The highest care should therefore be taken when using current models to predict the threshold shear velocity of wet sediment. Nevertheless, the models of Chepil (1956; Proc. Soil Sci. Soc. Am., 20, 288–292) and Saleh & Fryrear (1995; Soil Sci., 160, 304–309) are the best alternatives available.