Quantifying the relative importance of potential evapotranspiration and timescale selection in assessing extreme drought frequency in conterminous China

Abstract

Extreme drought frequency (EDF), which is typically assessed by drought indices, is an essential property for reflecting extreme drought events. Generally, EDF varies greatly depending on the consideration of potential evapotranspiration (PET) and the selected timescale for drought indices. However, there is little understanding of how PET and timescale individually and interactively alter EDF at multiple temporal and spatial scales. Based on 40 years of weather station data (1970–2010), we designed a factorial analysis of drought indices with/without PET (i.e. Standardized Precipitation Index; SPI, Standardized Precipitation Evapotranspiration Index; SPEI) at 3-month and 12-month timescales. Specifically, we quantified the relative contribution of potential evapotranspiration (PET), timescale and their interaction to EDF at annual and monthly scales in conterminous China. We found that: (1) PET individually had an effect that was more than 25% greater than the effect of timescale on EDF at annual scale. Interaction between PET and timescale had a dominant effect (above 50%) at the monthly scale, except in the humid region; (2) PET had a strong effect on the temporal variability of EDF and timescale selection accounted for spatial variability of EDF. (3) EDF was more sensitive to the temperature than precipitation, based on the results of sensitivity analyses; (4) The PET and timescale effects on EDF showed a different pattern in the highest (first level) terrain region that is centered on the Tibetan Plateau and delineated by the altitude greater than 3500 m. This study highlights the necessity of considering the interaction of PET and timescale, and the importance of regional condition (i.e. precipitation, temperature and elevation) in selecting the optimal drought indices and timescale combinations to detect extreme drought. Accurately quantifying the linkages among drought indices for estimating EDF on different timescales and terrain condition can help us determine the accuracy of EDF assessments and guide future research.