The Four-Parameter Kappa Distribution in Prediction of Standardized Precipitation Index

The standardized precipitation index (SPI) is used vastly for drought monitoring. Among the many of drought indices, the using of SPI for drought events in various time scales is more important. The SPI calculation is based on fitting a probability distribution that describes an observed series of precipitation. There are many models for obtaining probability distribution and parameter estimation, but choosing an appropriate approach is important. This paper employs the four-parameter kappa distribution and techniques of parameters estimation for calculation of SPI. The parameters of kappa distribution are estimated by the genetic algorithm and method of maximum likelihood, and finally the results indicate that the genetic algorithm is more satisfactory for parameters estimation and SPI prediction.