Autocorrelation functions contain information about turbulence, including eddy sizes and strengths. In order to make that information available to Gaussian diffusion models, the autocorrelation function must itself be modeled by reasonably simple mathematical terms. The paper postulates a way to describe turbulence which leads to a method for fitting the autocorrelation function by a series of exponential terms. It shows that this procedure leads to simple, mathematically and physical consistent definitions for the transformed autocorrelation and the spectra. An example is given to show the utility of the method.