This document describes the algorithms that are used in the new random displacement method (RDM) option in the ADPIC model to parameterize atmospheric boundary layer turbulence through an eddy diffusivity, K. Both the new RDM version and previous gradient version of ADPIC use eddy diffusivities, and, as before, several parameterization options are available. The options used in the RDM are similar to the options for the existing Gradient method in ADPIC, but with some changes. Preferred parameterizations are based on boundary layer turbulence scaling parameters and measured turbulent velocity statistics. Simpler parameterizations, based solely on Pasquill stability class, are also available. When eddy diffusivities are based on boundary layer turbulence scaling parameters (i.e., u, h, z and L ), {open_quotes}turbulence parameterization{close_quotes} is an appropriate term. In other cases, this term is used loosely to describe {open_quotes}sigma curves{close_quotes}. These are semi-empirical relationships between the standard deviations, {sigma}z(x) and {sigma}y(x), of concentration from a point source and downwind distance. Separate sigma curves are used for each of six Pasquill stability classes, which are used to categorize the diffusive properties of the atmospheric surface layer. Consequently, sigma curves are more than parameterizations of turbulence since they also prescribe the final concentration distribution (formore » a point source) given a Pasquill stability class. In the ADPIC model, sigma curves can be used to calculate the eddy diffusivities, K{sub Z} and K{sub H}. Thus, they can be used to {open_quotes}back out{close_quotes} parameterizations for K which are consistent with the dispersion associated with the particular sigma curve. This results in eddy diffusivities which are spatially homogeneous, but travel time dependent.« less