NO. 7 ARGENTINE WEEKLY WEATHER FORECAST HOXMARK 3 



As Dr. C. G. Abbot had found that thei'e is a relation between the 

 visible phenomena of the sun and the solar radiation intensities, an 

 arrangement was made with the Astronomical Observatory of the 

 University of La Plata to make observations of the sun from August, 

 1920. Later on, the same class of observations were made by the 

 Magnetical Observatory of Pilar, Argentina, and the Astronomical 

 Observatory at Rio de Janeiro, Brazil. Figure i shows the mean 

 distribution of faculae in longitude on the sun following marked 

 outbreaks of faculae, and figure 2 shows the relation of solar radia- 

 tion to area and intensity of faculae (after Clayton). 



Some months after the departure of Mr. H. H. Clayton, until then 

 chief of the forecast department of the Argentine Meteorological 

 Office, when the author took charge of the weekly prediction, the 

 scope of the investigations was widened by including the daily ob- 

 servations of the magnetic components, and from February, 1924, 

 there were included the atmospheric electricity data, from the Pilar 

 Observatory. 



To calculate the weekly forecast of the weather, all the various 

 observations previously mentioned, solar radiation, solar faculae, 

 earth magnetism, atmospheric electricity and the meteorological ob- 

 servations are analysed and subjected to an exhaustive and com- 

 parative study. 



VERIFICATION— TEMPERATURE 



The difficulties offered by a weather prognostication for intervals 

 longer than 24 hours, increase considerably with the length of the 

 selected period. 



For example, the prediction for the first day of the weekly 

 forecast can be formulated with great accuracy, but as the number 

 of days augment, the technical processes and the calculations must 

 be multiplied to attain the exactitude of the first day, so that it can 

 be stated that the difficulties increase in geometrical progression. 



The verification of a forecast formulated in exact numbers, and 

 for a limited territory, does not sufifer from the same weakness as 

 that of a prediction published in words, whose true sense afterwards 

 can be made a subject of dispute. The fixed temperatures given by 

 the weekly forecast are very convenient for a mathematical com- 

 parison, absolutely free from personal bias. 



The method of finding correlation coefficients worked out by Galton 

 and Pearson is much employed at present, and therefore the following 

 equation was used to determine the relations between predicted and 

 observed values. 



