On the use of standardized drought indices (SPI and SPEI) for assessing future climate change impacts on drought: introducing a dynamic approach

Drought is frequently monitored using standardized indices, such as the Standardized Precipitation Index (SPI) and the Standardized Precipitation-Evapotranspiration Index (SPEI). The latter was specifically designed to incorporate climate variability in terms of temperature. Consequently, by definition, it is more suitable for assessing variations in drought frequency and magnitude induced by climate change across various potential future scenarios. 

However, standardization presents a challenge when employing indices to evaluate the potential impacts of future climate change on drought. This is because, by definition, these indices are drawn from a standard normal random variable (null average and unit variance). The assessment of these impacts involves comparing occurrences in a future period and scenario with those in a historical control period. If the indices are separately calibrated for each period (one calibration for the future period and one for the control period), any differences observed may result solely from the sampling variability of a series drawn from a standard normal random variable. Numerous studies have assessed climate change impacts on droughts using this imperfect approach. Conversely, an alternative approach involves computing future indices using parameters from the control period. This represents a "worst-case scenario" as it overlooks potential climate change adaptation measures that could mitigate the impacts. To address this issue, our study introduces a dynamic approach wherein future changes are evaluated by computing climate normals using moving time windows. This approach enables an understanding of how impacts change with the timing of the implementation of adaptation measures. We apply this approach to Sicily and Calabria in Southern Italy, considering various climate change scenarios (Representative Concentration Scenarios). The results suggest that the region is likely to experience an increase in drought events due to climate change. These findings underscore the necessity for revised drought identification strategies that consider the non-stationarity in climate.