(American Geophysical Union, 1987-05) Singh, V. P.; Krstanovic, P. F.
The principle of maximum entropy was applied to derive a stochastic model for sediment yield from
upland watersheds. By maximizing the conditional entropy subject to certain constraints, a probability
distribution of sediment yield conditioned on the probability distribution of direct runoff volume was
obtained. This distribution resulted in minimally prejudiced assignment of probabilities on the basis of
given information. The parameters of this distribution were determined from such prior information
about the direct runoff volume and sediment yield as their means and covariance. The stochastic model
was verified by using three sets of field data and was compared with a bivariate normal distribution. The
model yielded sediment yield reasonably accurately.