Testing Climate Models
It is essential that a collaborative effort be developed to link a new class of high accuracy benchmark observations to the high-end, long-term climate forecasts with the objective of analyzing and improving climate model forecasts. The task of systematically applying observed data to climate model improvement is a field that demands innovation in both the mathematical structures selected to test the high-end forecasts as well as the observations central to the testing strategy. The development of sophisticated statistical analyses, of 4-D assimilation and of inverse modeling have now reached an advanced stage for weather models, but this has taken a lot of effort. Analogous climate activities will prove challenging, but fundamental advances will emerge from this union. It is necessary that we push forward with both the experimental and modeling methodologies in parallel, both because they demand technical integration and because they possess similar time scales to converge.
It is necessary to investigate the path from monitoring to prediction with the following objectives: Concentration on the use of resolved radiances observed from satellites. The power of spectrally resolved, absolute radiances as a method for systematically defining the state of the climate, as a language for adjudicating the cause of differences between high-end forecasts, and as a means for dissecting feedbacks on critical timescales requires this focus. In the development of model testing strategies, absolute spectrally resolved radiance data sets must be combined with other benchmarks such as GPS, solar radiance, etc.