Evolution of future extreme drought frequency in two climate model large ensembles

Recent studies project a significant increase in drought frequency over most continents over the 21^st century. However, few studies have specifically looked at extreme droughts, defined here as having a return period larger than 20 years. In this work, two large climate model ensembles, the 50-member Canadian Earth System Model (CanESM2) and the 40-member Community Earth System Model (CESM1), both under the RCP8.5 scenario are used to project the evolution of the extreme drought frequency in the near (2036-2065) and far future (2070-2099) relative to the 1980-2009 historical period. The use of a large ensemble allows for a robust estimation of the frequency of very large droughts. Frequency changes for the 2, 20 and 100-year droughts were computed.

Extreme meteorological droughts were globally assessed using the short-term (1-month) and long-term (24-month) Standardized Precipitation Index (SPI). Extreme hydrological extreme droughts were assessed by the 1-month Streamflow Drought Index (SDI), using a lumped hydrological model on 5797 North American catchments to transform climate model outputs into catchment streamflows.

Results show that both climate models project increases of extreme meteorological drought frequency over many of the world’s regions, with a typical two or three-fold increase. The spatial distribution of regions with increasing meteorological drought frequency mostly matches those projected changes in future mean annual precipitation. Changes in future extreme hydrological droughts are dramatically more severe than for meteorological droughts, with up to a 27-times increase in frequency for the 100-year hydrological droughts, outlining the large impact of temperature change. The frequency change is the largest for the 100-year compared to the 2 and 20-year hydrological droughts.