ice-age changes, but regional effects on water availability may have had large effects on humans and ecosystems.

Documentation of past abrupt climate changes is quite good in certain regions such as from Greenland ice cores and some of the pollen and other records summarized above. But no reliable global maps of climate anomalies are available for any of the abrupt changes of the past. Well-dated records with high time resolution are especially scarce in the Pacific, the tropics, and southern high latitudes, but no ideal record yet exists and additional useful information can be obtained almost everywhere. Past changes in the hydrological cycle are especially poorly characterized relative to its importance to humans and ecosystems.

The unavoidable incompleteness of the sedimentary record (spatially, temporally, and in variables recorded) means that changes from before the deployment of widespread instrumentation will never be understood as well as more recent changes. Instruments have not yet observed any of the large, global abrupt climate changes (a statement that some day may require modification in hindsight with respect to human-induced changes). However, regional events that satisfy the definition of abrupt climate change have been observed instrumentally. These are important in several ways: among others, they highlight observations required to characterize events and provide early warning of event onset and termination; they point to gaps in the observational system; and, they show the important role of coupled modes in such changes (and, by inference, perhaps also in the larger changes further in the past).

Instrumental study of abrupt changes has drawn on a broad-based range of observations collected for many purposes. However, few of these data sets are targeted on those parts of the climate system that are believed to have participated in past abrupt changes or that are likely to exhibit abrupt and persistent changes when thresholds in the climate system are crossed. Documentation remains sketchy of ocean circulation, especially in regions of deepwater formation, and interactions with sea-ice processes and with freshwater fluxes from the atmosphere and from land hydrology including glaciers and permafrost. Gaps also remain in documentation of land-surface processes linked to hydrology, and of modal behavior of the coupled climate system.

Much work is required to place all of the available observations into a coherent framework, capable of assimilating new observational data and contributing to an understanding of abrupt climate change. A start on such a framework is described in the next section.