Asymmetric response of global drought and pluvial detection to the length of climate epoch

Abstract

Standardized indices (SIs) are routinely used to assess hydroclimate extreme events. However, to what extent such assessment is robust considering the choice of climate epoch (that is, a specific time period with explicit data length) is still not clear especially when developing new SIs. In this study, we use multiple SIs to systematically investigate the sensitivities of global drought and pluvial detection to the length of climate epoch by separating the parameter fitting step from the entire SI calculation process. We find that the sensitivity of event classifications to data length is more similar between Standardized Precipitation Evapotranspiration Index (SPEI) and Standardized Moisture Anomaly Index (SZI) at the six-month time scale, compared to that of Standardized Precipitation Index (SPI). Moreover, despite the symmetric definition of droughts and pluvials, the choice of climate epoch has asymmetric impacts on event detection. Detection of SZI-based droughts is more sensitive to data length than pluvials and generally requires longer series (additional 10 years compared to a 30-year climatology) to accurately fit the tailed distribution, especially over a few hot spot regions, including South Africa, South Asia and Oceania.