NO. 6 SOLAR RADIATION AND WEATHER CLAYTON 59 



Brazil, while simultaneously there is a rise of pressure and a fall of 

 temperature in southern Argentina. Progressive changes coming 

 from the south frequently overlapped those forming in the north. 



(4) There are other variables which are probable, but the effect 

 of which has not yet been determined. For example, any distribution 

 of pressure resulting from solar changes, if long continued over 

 ocean areas, must set in motion ocean currents which in turn cause 

 surface changes in temperature and pressure. 



It is gratifying to know that notwithstanding these difficulties 

 forecasts have been successfully carried on for a year. A rigid 

 mathematical method of verification proves them to be better than 

 chance forecasts for an important station in the United States. There 

 is every reason to suppose that these forecasts will go on increasing 

 in accuracy as the data on which they are based increases in complete- 

 ness and accuracy, and the knowledge of how these solar changes 

 affect our atmosphere increases. 



PREDICTION OF SOLAR RADIATION CHANGES 



Figure 34 brings out clearly that, in order to predict weather 

 changes which follow solar radiation changes so closely as they do 

 at Williston, it is necessary to anticipate the solar changes. In order 

 to test the accuracy with which this could be done from visual obser- 

 vations of sun spots and faculae, beginning in May, 1924, a forecast 

 was made of the solar radiation to be expected five days ahead of 

 the observed phenomena. At first these were filed, but after July i, 

 1924, were mailed as soon as made to Dr. Abbot. Figure 43 gives 

 a plot of the predicted and observed values of solar radiation. A 

 dotted curve joins the forecasted values of solar radiation from May 

 to September, inclusive, on all the days on which observed values 

 were obtained and a continuous curve joins the observed values. All 

 days on which there were no observations or on which there were 

 doubtful observations were omitted. 



Besides the absolute values of solar radiation, a forecast was 

 also made each day as to how much the solar radiation would depart 

 from the general trend as determined by 27-day means of consecutive 

 observations. 



These forecasts were verified by Mr. Eliot C. French,^ and checked 

 by Dr. Abbot in the same way as were the temperature forecasts 



* I am also indebted to Mr. French for assistance in the preparation of the 

 data presented in this paper and for making forecasts during periods when 

 I was absent, and to Misses H. V. Miller and M. I. Robinson for assistance 

 in the computations. 



