To Abstract List
The rationale for WOCE consisted of a number of central ideas, including the need to produce global-scale data sets to address a global-scale problem: climate change. In 1979, when discussion began, a major assumption was that General Circulation Model (GCM) improvement would occur even without a major international organized effort, but that global data sets would never emerge without such an effort. Thus the organizers placed a much heavier priority on the field programs than on the model development, with the understanding that by the late 1990s, attention could be turned to further model improvement and synthesis.
As we come into the late 1990s, the initial assumptions are seen to have been reasonably realistic. We do have global or near-global data sets, and near-eddy resolving GCMs exist. Based upon the history of oceanographic field programs, one can guess that over the next 10 years these data will be analyzed and interpreted. But a major premise of WOCE was that the most definitive interpretations and understanding would emerge from combining the information in the observations with that encompassed by GCMs. Here there is a close meteorological analogy: with adequate data and models, one often obtains the best understanding of the atmosphere by analyzing neither the model nor the data in isolation, but by analyzing the sensible combination of the two in the form of analyzed fields.
The meteorological experience is of limited utility for oceanographers. In this talk I will review where we stand in the problem of combining models with observations on large to global scales, a field in which there seems to be a lot of confusion. Several points are worth understanding:
Synthesis (or "estimation," or "fusion" or "assimilation") always begins by quantitative comparison of models with observations. It is reasonable to combine a model with data only if a number of consistency tests are first passed. On the other hand, synthesis can always be done - there is no such thing as a model or data set which is too poor to be used. There may be little gained by the synthesis, but a knowledgeable user can always make a sensible methodology work. But the essential ingredient is a quantitative understanding of the information content of both model and data. One must also distinguish the solution of a problem in principle (estimation theory is highly developed and a great deal is known about it) from the problem of practical implementation of algorithms for systems of enormous dimension. The latter, and not the former, is the oceanographers' problem. I will summarize some of the work that has been done in implementing true synthesis, and lay out what seem to be the most immediate and longer term, problems. The WOCE data and models raise a number of distinct problems of which the most difficult is probably that of using observations which reflect very long oceanic time scales: specifically the long hydrographic lines.
To Pacific Workshop
To WOCE Home Page