Operated by multiple sedimentary processes and dynamics, the clastic particles are mixed by corresponding multiple populations of grain sizes within different settings, which is finally recorded in the grain-size distribution (GSD) of the sediments. Many statistical models, such as log-normal, Rosin, log-hyperbolic, Weibull, log-skew-Laplace, gamma, and tangential hyperbolic, have been used to unmix the distributions successfully. However, they all have inherent deficiency in dealing with the skewness implied in the GSD. Adding a skewness property into the normal distribution, skew normal (SN) distribution offers a new technique to process GSD which is quantified by the sum of SN sub-distributions with four parameters, volume percentage, location, scale, and shape. In the help of deconvolution algorithm, numerical solutions of these parameter sets are resolved to estimate the mean, variance, skewness, and kurtosis of individual SN sub-distribution. GSD of eight samples from the swampy lakeshore of Lake Ulungur are unmixed utilizing SN distribution in this paper. The number of components SN sub-distributions is ranged from four to eight in individual GSD, and the dominant component percentage is from 30.93 to 71.90%. Cumulative probability of component SN sub-distributions in probability scale could be used to discuss the modes of clastic sediment transport further: component SN sub-distribution with flection-point of cumulative probability curves located less than − 1.0 ϕ and skewness larger than 0.4 suggests the surface-creep population; component SN sub-distribution with flection-point located between − 1.0 and 4.0 ϕ and skewness closed to zero implies the saltation population; component SN sub-distribution with flection-point larger than 4.0 ϕ and skewness less than − 0.75 infers suspension population.