A global perspective on the sub-seasonal clustering of precipitation extremes

Abstract

The occurrence of several precipitation extremes over sub-seasonal time windows can have major impacts on human societies, leading for instance to floods. Here, we apply a simple statistical framework based on Ripley’s K function, at a global scale and for each season separately, to identify regions where precipitation extremes tend to cluster in time over timescales of a few days to a few weeks. We analyze several observational and reanalysis datasets, as well as output from CMIP6 Global Climate Models (GCMs). Good agreement is found on the spatio-temporal clustering patterns across datasets. Sub-seasonal temporal clustering is largely concentrated over the tropical oceans, where it can be detected year-round. It is also significant over certain tropical lands, like Eastern Africa, and seasonally outside the tropics in several regions, most notably around the eastern subtropical oceans (Iberian Peninsula and Western North America during the DJF and MAM seasons) Southwest Asia (especially during JJA and SON) and Australia (in SON). We also find that CMIP6 models generally correctly reproduce clustering patterns, paving the way for an assessment of trends in sub-seasonal clustering under climate change. Clustering of present-day extremes increases in many areas under climate change. Changes diagnosed by comparing present day and future extreme percentiles are positive and negative and strongest in the tropical areas.

Keywords

• Precipitation extremes
• Temporal clustering
• Time series statistics