Product Periodic Autoregressive Processes for Modeling Intermittent Monthly Streamflows

Several attempts have been made in the past to model hydrological processes such as monthly streamflows in dry regions. One of the crucial problems in modeling this type of process is the handling of zero flows. A stochastic model is presented herein which enables the reproduction of the percentage of zero flows in each month, the monthly mean and variance, and the month‐to‐month correlation of the intermittent flows. The model considers the intermittent monthly flow process as a product of a periodic binary discrete process and a periodic continuous process. Both the discrete and the continuous processes are periodic first‐order autoregressive. Parameter estimation has been developed based on the method of moments, method of transition probability, and method of maximum likelihood.