WEATHER 



diction is an impossibility. Where 

 does the truth lie? 



First we might ask: Are there valid 

 grounds for attempting long-range 

 prediction? Here the answer is defi- 

 nitely "yes." If weather changes were 

 due exclusively to migratory synoptic- 

 or smaller-scale weather systems, it 

 is known from the GARP experiments 

 cited previously that prediction would 

 not be possible beyond two or three 

 weeks. But it has long been recog- 

 nized that there are larger-scale pat- 

 terns in the atmosphere which tend to 

 persist or recur over periods of weeks, 

 months, or seasons. Drought episodes 

 and prolonged spells of warm or cold 

 weather may be cited as examples of 

 such patterns. They are associated 

 with abnormal features of the circu- 

 lation — unusual displacements of the 

 jet stream, the semi-permanent high 

 and low pressure centers, and so 

 forth. 



Theories of Causation — The cause 

 of long-period weather changes is a 

 debatable subject. Many investiga- 

 tors have sought to connect them to 

 extraterrestrial events — to variations 

 of solar radiation, in particular — but 

 the evidence in favor of an extrater- 

 restrial origin is not impressive. Other 

 investigators have suggested that they 

 are caused by complex feedback 

 mechanisms within the atmosphere. 

 This hypothesis cannot be discounted. 

 In laboratory experiments with rotat- 



ing fluids, it has been found possible 

 to generate long-period (on the time- 

 scale of the model) circulation fluctua- 

 tions even when external conditions 

 are kept rigidly constant. 



A final theory, which has steadily 

 gained support, attributes long-period 

 weather variations to interactions of 

 the atmosphere with surface features. 

 Anomalies of sea-surface temperature 

 and of snow cover are examples of 

 conditions that are believed capable 

 of producing and perpetuating abnor- 

 mal weather situations. 



Forecasting Metlwds — Though 

 some physical reasoning may enter 

 into the formulation of a long-range 

 forecast, the methods currently in use 

 do not have a physical basis. The 

 numerical methods applied at shorter 

 ranges are not, as presently formu- 

 lated, appropriate to long-range pre- 

 diction. 



Thus, main reliance is put on ex- 

 trapolation, statistical, and analogue 

 methods of forecasting, and human 

 judgment plays a heavy role. The 

 results obtained from these methods 

 show at best only slight skill, and 

 there seems little or no hope of sig- 

 nificant improvement through their 

 continued use and development. How- 

 ever, in view of the great economic 

 importance of long-range prediction 

 and the growing evidence that a 

 meaningful physical understanding 



of long-period atmospheric variations 

 can be achieved, it is essential that 

 efforts to derive more suitable quanti- 

 tative methods of prediction be con- 

 tinued and strengthened. 



Needed Scientific Activity — Ac- 

 tivities of two types deserve particu- 

 lar encouragement in this respect. 

 First are programs to acquire the kind 

 of global data needed for establishing 

 the physical basis of long-range pre- 

 diction. Such programs will have to 

 endure for a long time and will not 

 only have to measure the usual me- 

 teorological variables employed in 

 numerical prediction but will have to 

 measure additional parameters such as 

 sea-surface temperature, snow cover, 

 and the like. It is apparent that ob- 

 servations from satellites will be the 

 key element in a global monitoring 

 effort. 



A second type of activity that 

 merits vigorous support is experi- 

 mental work in numerical modeling 

 of the general circulation, of the sort 

 now practiced by a number of groups 

 in the United States. From such ex- 

 periments, it may well be possible to 

 discover the underlying causes of 

 long-term weather and climatic anom- 

 alies. In fact, the modeling experi- 

 ments are essential to the observa- 

 tional effort, for without them we can 

 never be sure, until perhaps it is too 

 late, that the proper variables are 

 being measured. 



Long-Range Weather Forecasting 



Scientists who work in long-range 

 weather forecasting encounter great 

 difficulties, not only in the intricacies 

 of their chosen field but also in get- 

 ting across to other scientists and the 

 lay public the essential nature of 

 their problem and the reasons for 

 their painfully slow progress in the 

 modern-day milieu of satellites, com- 

 puters, and atomic reactors. When 

 solar eclipses can be predicted to frac- 

 tions of a second and the position of 



a satellite pinpointed millions of miles 

 out in space, it is not readily under- 

 standable why reliable weather pre- 

 dictions cannot be made for a week, 

 month, season, or even a year in 

 advance. Indeed, eminent scientists 

 from disciplines other than meteorol- 

 ogy, underestimating the complexity 

 of the long-range problem, have tried 

 to solve it only to come away with a 

 feeling of humility in the face of 

 what the late von Neumann used to 



call "the second most difficult prob- 

 lem in the world" (human behavior 

 presumably being the first). 



And yet, the potential economic 

 value of reliable long-range forecasts 

 probably exceeds that for short-range 

 (daily) forecasts. Many groups need 

 as much as a month or a season or 

 more lead-time to adjust their plans. 

 These include such diverse types as 

 manufacturers (e.g., summer suits, 



97 



