PART IV — DYNAMICS OF THE ATMOSPHERE-OCEAN SYSTEM 



else had changed, was made. The 

 extrapolators moved in, and there 

 were prophecies of ice ages. 



Statistical and physical explana- 

 tion of the problem of climatic change 

 can be conceived, but each approach 

 has fundamental difficulties. In the 

 lirst case, existing series of climatic 

 statistics based on instrumental read- 

 ings are short — about 250 years is 

 the longest reliable record. The stat- 

 istics of these long series are not 

 stationary; there is variance at the 

 longest periods they cover. Historical 

 and geological evidence indicates 

 greater fluctuations of climate than 

 the instrumental record. Statistics in- 

 dicative of climate are not stationary. 

 There can be no test of significance 

 to separate climatic change that might 

 be associated with man's activities 

 from the "natural" changes associated 



perhaps with internal instabilities of 

 the ocean-atmosphere system or per- 

 haps with extraterrestrial change. 

 The physical approach leads to sim- 

 ilar conclusions, as Lorenz, in par- 

 ticular, has pointed out. The equa- 

 tions governing the ocean-atmosphere 

 system certainly have innumerable 

 solutions, and it may be that sets of 

 these solutions exist with very dif- 

 ferent statistics — i.e., that the earth 

 may have many possible climates 

 with its present atmospheric composi- 

 tion and external forcing function. 

 At most, changing the composition 

 of the atmosphere (e.g., by adding 

 CO2) might change the nature of 

 the inevitable change of climate. 



Indicated Future Research 



Present activity is in two direc- 

 tions — confirming and extending our 



knowledge of the changes of atmos- 

 pheric composition due to industrial 

 activity by monitoring programs, 

 and developing physical models of 

 the climate. This latter is one of the 

 major scientific problems of the age, 

 and we do not yet know whether it 

 can be resolved to any useful extent. 

 The requirement is for a model, sim- 

 ilar to existing models of the global 

 atmosphere and ocean but completely 

 independent of any feature of the 

 present climate. The most complex 

 existing models incorporate a forcing 

 function specified in terms of the cur- 

 rent climatological cloud distribution 

 and ocean surface temperature. The 

 output of these models can, therefore, 

 at best only be consistent with the 

 existing climate. The requirement is 

 for a model which will generate its 

 own cloud distribution and ocean- 

 surface temperatures. 



120 



