PART IV — DYNAMICS OF THE ATMOSPHERE-OCEAN SYSTEM 



Figure IV-8 — DATA REQUIRED FOR FORECASTS 



LATITUDE 



20 



The diagram gives an indication of the data necessary for forecasts in the middle 

 latitudes for varying lengths of the forecast period. It is important to note that both 

 atmospheric and oceanographic data are needed for all forecast periods. 



ena selected on the basis of their 

 practical importance: severe local 

 storms (thunderstorms, hailstorms, 

 and tornadoes); hurricanes; and syn- 

 optic disturbances (cyclones, anticy- 

 clones, and fronts and their asso- 

 ciated upper-level troughs, ridges, 

 and jet streams). 



Severe Local Storms — These 

 storms develop with extreme rapidity 

 and seldom have lifetimes of more 

 than a few hours. On the basis of 

 the large-scale temperature, moisture, 

 and wind fields, and their expected 

 changes, it is possible to delineate 

 areas in which severe storms are 

 likely to occur 6 to 12 hours in ad- 

 vance, or sometimes even longer. But 

 there is at present no way of predict- 

 ing when and where an individual 

 storm will develop. Once a storm 

 has been detected, extrapolation and 

 steering methods can be used to 

 predict its motion with fair accuracy, 

 but in view of the short lifetime of 

 the typical storm, the forecast rarely 

 holds for more than a few hours. 



Statistical Forecasting — Though 

 statistical methods have wide appli- 

 cation in forecasting, the term, as 

 applied here, refers to any of a num- 

 ber of techniques in which past data 

 samples are employed to derive sta- 

 tistical relationships between the 

 variable being forecast and the same 

 or other meteorological variables at 

 an earlier time. The statistical method 

 is particularly valuable in forecasting 

 local phenomena that are too complex 

 or too poorly understood to be 

 treated by numerical or physical 

 methods but that experience has 

 shown to be related to identifiable, 

 antecedent causes. 



The Analogue Method — The aim 

 of this method is to find a previous 

 weather situation which resembles 

 the current situation and to use the 

 outcome of the earlier case to deter- 

 mine the present forecast. The 

 method has the advantage of sim- 

 plicity, but its usefulness is extremely 

 limited since sufficiently close ana- 



logues are difficult to find, even when 

 long weather records are available. 



Mixed Methods — Combinations 

 of the foregoing methods are quite 

 common. Thus, surface temperature 

 is customarily forecast by a combina- 

 tion of numerical and statistical tech- 

 niques in order to obtain better 

 predictions than would be obtained 

 from use of the numerical method 

 alone. 



Short-Range Prediction 



The problems encountered, meth- 

 ods employed, and the time period 

 for which accurate predictions can 

 be made differ according to the phe- 

 nomenon or scale of motion being 

 forecast. It is therefore convenient 

 to discuss the subject on the basis 

 of different types of weather systems 

 involved. To keep the subject within 

 reasonable limits, the discussion will 

 be limited to the following phenom- 



Weather radar is the most valuable 

 tool in severe-storm detection, and 

 it is only since the introduction of 

 radar that adequate monitoring of 

 severe storms has been possible. 

 Geostationary satellites also have 

 great potential usefulness in identify- 

 ing and tracking these systems. Until 

 there is full radar coverage of the 

 United States and permanent surveil- 

 lance by geostationary satellite with 

 both visual and infrared sensing ca- 

 pability, short-range prediction of 

 severe storms will not have reached 

 the limits of accuracy allowed by the 

 present state of the art. 



Ultimately, one may hope that the 

 methods of numerical weather pre- 

 diction used so successfully with 

 larger-scale storms will be applied to 

 thunderstorms and other small-scale 

 phenomena. But there seems no clear 

 way of achieving this hope in the 

 foreseeable future. To forecast these 

 phenomena by numerical methods re- 

 quires observations of the basic me- 

 teorological variables — wind, tem- 



94 



