A successful global climate monitoring system must fulfill clear societal objectives. For some aspects of climate monitoring, the societal goals are understood and are clearly stated, but long-term, decadal/centennial climate predictions have, in the past, been judged more in terms of curiosity-led criteria. A curiosity-led climate program is not, however, the effective way to achieve the required societal objective, which is to produce the best possible long-term climate projections. In terms of the universal use of numerical models for climate projections, this leads to the need for monitoring programs that provide data to test model output against reliable observations. This requires an operational climate model (which the United States does not now have), and observations that emphasize accurate and reproducible data designed to provide critical tests of model output. The priorities for specific monitoring programs can be formulated in terms of these requirements, which can also provide metrics of progress. A schematic program that combines monitoring, modeling, and research is described. This program has, as its end point, the provision of demonstrably improved long-term climate projections. Europe appears to be advancing more rapidly along this path than is the United States.
