Changes in characterising extremes

The results of studies of extreme events are strongly influenced by the choice of methods used, as well as by the extremeness of the event. Impressive progress has been made in the ability to characterise past changes in weather, climate and ocean extremes, as well as to describe their future evolution. In addition to methodological advances, there is now improved clarity in the definition of these extremes, as well as in the associated metrics. Another significant advance has been the creation and regular updating of many comprehensive, quality-controlled datasets covering global land and ocean areas. Advanced calibration and interpolation techniques, the assimilation of data and blending of datasets enhance their use for detecting and interpreting changes in extremes. Studies that use an ensemble of in situ, multi-model and reanalysis datasets, and undertake both comparative evaluations and sensitivity analyses, can often deliver more optimal results. The availability and reliability of information on extremes in the pre-instrumental era has improved tremendously, but the spatial coverage of in situ observations continues to be highly uneven, including for ocean areas. The outputs of numerical models are the only source of robust information on the future characteristics of weather, climate and ocean extremes. The level of skill is highly dependent on the model, its implementation and on the specific extreme of interest. International efforts to understand and reduce the sources and levels of uncertainty have led to major achievements in simulating future changes in atmosphere and ocean conditions, including extremes.