Temporal trends of discrete extreme events: A case study

Investigating trends in discrete events is essential for the study of changing patterns of extreme events. Temporal trends in the inter-arrival times of occurrence of drought events were examined for 21 selected stations across Victoria, Australia. In the present study, the Standardize Precipitation Index (SPI) was applied for 12-month time scale to identify drought. A drought event here is defined as a period in which the SPI is continuously negative and reaching a value of -1.0 or less. Often, nonparametric tests are commonly used to test for trends including in discrete events. However, discrete events are not constant because of the presence of zero values or non-normality of data. The methodology applies to long-term records of event counts and is based on the stochastic concepts of Poisson process and standard linear regression. Overall, of the 21 stations, 15 showed statistically significant increasing frequency indicates those events are becoming more frequent. Only one station gave insignificant result. The remaining 5 stations showed the time between events was significantly increasing designates droughts are becoming less frequent.