A stochastic model demonstrates low-cost meteorological support to environment dependent simulation studies. Only the univariate problem is demonstrated, but extensions to multivariate applications are feasible. This application models atmospheric visibility. The model simulates time series at a given geographical point. A historically derived cumulative frequency distribution is converted to its equivalent normal deviate and fitted by a rational approximation. A Markovian time dependence is assumed. Seasonal and diurnal cycles are modeled. A run is initiated via a random draw from a unit normal distribution. Alternately, an initial value may be specified. Subsequent values are generated using a time-correlation weighted linear combination of the prior value and a unit normal random input. This produces a Brownian Movement phenomena within the appropriate probability density functions. A given run tends to regress toward the climatological mean while exhibiting random fluctuations. A run of sufficient length recreates the probability density functions. Results presented verify model performance against historical data and include a few applications. The FORTRAN computer subroutine is published in an appendix.

ADA 058336 Approved for public release; Distribution is unlimited. The opinions and conclusions expressed herein are those of the student-authors and do not necessarily represent the views of the U.S. Army Command and General Staff College or any other governmental agency. (References to these studies should include the foregoing statement.)