Abstract
A mathematical model which can be used to generate monthly time series of precipitation and temperature that resemble the historic weather records in terms of their mean, variance, skewness and month to month correlations was tested. The basic model is the lag one Markov model with a Gamma distributed random variable. The model was tested in three locations within the continental United States (Omaha, Nebraska; Ithaca, New York; and San Francisco, California) and in one location in Puerto Rico (San Juan). Test results indicate that the model does very well in reproducing the first two moments, mean and variance, of the historic records when both precipitation and temperature are generated. The model was capable of reproducing the coefficient of skewness for 50% of the records at the four locations tested. The lag one monthly coefficient of correlation was well reproduced for 65% of the records. Although the work on the model was motivated by the need for synthetic precipitation and temperature data for the preliminary design of land treatment systems for the disposal of wastewater, its use is not limited to this type of project.