Stochastic daily precipitation models: 2. A comparison of distributions of amounts

Publisher Website
Google Scholar
Abstract
Chain‐dependent and independent exponential, gamma, and mixed exponential distributions are compared as models for the distribution of daily precipitation. Parameters for each distribution are estimated by maximum likelihood techniques for 14‐day periods. The Akaike information criterion is used to select the most appropriate distribution for each period and for the entire year. For the five U.S. stations studied, the independent mixed exponential distribution was the best on the basis of the Akaike information criterion, and the independent gamma and chain‐dependent gamma ranked second and third, respectively. Fourier series are fit to the parameters by least squares to provide starting values for subsequent numerical maximum likelihood estimates of the Fourier coefficients. According to the Akaike information criterion, the Fourier series description of model parameters for the mixed exponential model is superior to the specification of parameters for each 14‐day period.