PART IV — DYNAMICS OF THE ATMOSPHERE-OCEAN SYSTEM 



ent" curve in Figure IV-9, a vast 

 World Weather Watch (WWW) pro- 

 gram to acquire an adequate network 

 of observations has been set in mo- 

 tion by the WMO (World Meteoro- 

 logical Organization) and a compan- 

 ion research arm, GARP (Global 

 Atmospheric Research Program), un- 

 der IC5U (International Council of 

 Scientific Unions) and WMO. The 

 aims, rationale, and scope of these 

 undertakings have been well docu- 

 mented in many reports and will not 

 be reiterated here; suffice to say that 

 a satisfactory solution of long-range 

 forecasting problems is not likely to 

 come about without them. 



Statistical Aggregation — Neverthe- 

 less, the future of long-range weather 

 forecasting does not and should not 

 depend solely on the possibilities in- 

 herent in the iterative approach de- 

 scribed earlier. Virtually every group 

 of meteorologists that has attacked 

 this problem over the past century 

 has done so by working not with 

 short time-step iterations but, rather, 

 by studying statistical ensembles and 

 the evolution of average wind and 

 weather systems — e.g., from month 

 to month and season to season. The 

 long-range forecasting services of the 

 Soviet Union, England, Japan, and the 

 United States operate with statistical 

 aggregates as well as physical meth- 

 ods. In the statistical approach, it is 

 taken for granted that the average 

 prevailing wind and weather patterns 

 for one month, together with the 

 associated abnormalities of sea tem- 

 peratures and land surfaces (e.g., 

 covered or free of snow), largely de- 

 termine how the general weather pat- 

 terns are going to develop during the 

 following month under the influence 

 of the solar radiation appropriate to 

 time of year. A small effort in nu- 

 merical modeling using this philoso- 

 phy has begun. 



How good are long-range predic- 

 tions by conventional non-iterative 

 methods? This is a question of scien- 

 tific as well as practical importance, 

 because any positive skill over and 

 above climatological probability im- 



plies knowledge that ought to funnel 

 into further research and thereby lead 

 to more reliable prediction. The pres- 

 ent skill at forecasting departures 

 from normal of average temperature 

 at 100 cities over the United States 

 for 5-day, 30-day, and experimental 

 seasonal forecasts may be roughly 

 given as 75, 61, and 58 percent, re- 

 spectively, if chance is defined as 50 

 percent. Similarly, for precipitation, 

 5-day, 30-day, and seasonal forecasts 

 average roughly 5<5, 52, and 51 per- 

 cent, respectively. While these skills 

 are far from perfect they do indicate, 

 particularly for temperature, that the 

 methods contain some knowledge of 

 long-term atmospheric behavior. The 

 5- and 30-day forecasts that are re- 

 leased to the public appear to be of 

 definite economic value, judging from 

 hundreds of comments by users and 

 also from their reaction when the 

 forecasts are not received on time. 



Despite the work and performance 

 of many groups around the world 

 along these practical lines and the 

 fact that their forecasts show some 

 small but definite skill in long-range 

 prediction (contrasted with the utter 

 failure, up to now, of dynamical 

 iterative models at periods up to 

 a month), the statistical-physical- 

 synoptic (synoptic here meaning an 

 over-all view with the help of maps) 

 approach has been relatively neg- 

 lected by meteorologists in the United 

 States. 



The Role of Oceanography — 

 Oceanographers may see the long- 

 range problem more clearly than me- 

 teorologists as one in which statistical 

 aggregates play an important part — 

 both in forecasting general thermal 

 conditions in the sea and in forecast- 

 ing its long-period interaction with 

 the atmosphere. Perhaps this is be- 

 cause large-scale changes in the sea 

 take place much more slowly (about 

 ten times more slowly) than in the 

 atmosphere and the reasons can 

 therefore provide a sort of memory 

 bank for the atmosphere. 



In the past decade, research has 

 shown that the thermal state of the 



oceans, especially the temperatures in 

 the upper few hundred meters, varies 

 considerably from month to month 

 and year to year, and that these vari- 

 ations are both cause and result of 

 disturbed weather conditions over 

 areas thousands of miles square. By 

 complex teleconnected processes, the 

 effects of these disturbed conditions 

 are transmitted to areas thousands of 

 miles distant. Thus, the prevailing 

 wind systems of the globe — the 

 westerlies, the trade winds, and the 

 jet streams — may be forced into 

 highly abnormal patterns with con- 

 comitant abnormalities of weather. 

 Because these reservoirs of anoma- 

 lous heat in the ocean are deep, often 

 up to 500 meters, and may last for 

 long periods of time, the atmosphere 

 can be forced into long spells of 

 "unusual" weather, sometimes re- 

 sulting in regional droughts or heavy 

 rains over periods ranging from 

 months to seasons, and even years or 

 decades. 



Potential Lines of Action — The 

 interface between meteorology and 

 oceanography is thus a promising 

 area which should receive more at- 

 tention. Several items are needed: 



1. A network of observations for 

 both air and sea measurements 

 over the world's oceans, or at 

 least over the Pacific Ocean 

 where much of the world's 

 weather appears to be gener- 

 ated. This network can be a 

 mix of ocean weather ships, spe- 

 cially equipped merchant ships, 

 and — particularly — un- 

 manned, instrumented buoys 

 which have now been demon- 

 strated to be feasible. A net- 

 work of observations about 

 500 kilometers apart would be 

 adequate as a start; later, the 

 data gathered could indicate 

 whether a finer or coarser grid 

 is necessary. Satellite measure- 

 ments can supplement but can- 

 not replace these observations, 

 particularly the subsurface ones 

 which monitor the heat reser- 



100 



