Statistical technique for assessing trends in drought occurrence - a case study

Drought is an extreme weather event affecting human wellbeing and the natural environments. In addition, climate change is exacerbating the situations by making drought conditions more common. Identifying and quantifying trends in drought events thus pose a real challenge in managing water resources systems. This study evaluated the trends in the frequency of occurrence of drought events in the Yarra River catchment in south-east Victoria, Australia. The Standardized Precipitation Index (SPI) and the Standardized Hydrological Drought Index (SHDI) were applied for 6-months timescale to identify meteorological as well as hydrological drought events, respectively. Drought indices are the most common and effective tools for the detection of drought periods. This study considered a drought event to commence as soon as the SPI/SHDI value becomes less than zero. Generally, trends in any data series are investigated by non-parametric statistical techniques. However, the time series of rare and extreme weather events are not suited to these traditional approaches due to the presence of zero values and non-normality of data. In this study, the changing pattern in the inter-arrival time of successive drought events were assessed with a statistical parametric method resulting from the concepts of Poisson Process and standard linear regression technique. The SPI and the SHDI were employed to detect drought incidences for two rainfall and two streamflow stations, respectively, within the study area. Overall, all the stations showed statistically insignificant decreasing trends in the rate of inter-arrival times of drought events which indicates that the drought events are becoming more frequent. This study will assist water managers to assess and develop appropriate mitigation strategies to overcome the future drought impacts.