188 MAEIOX EXPEDITIOX TO DAVIS STEAIT AXD BAFFIX BAY 



land, is the most important in causing yearly variation in the number 

 of icebergs. The agency of next importance, about one-third of the 

 Greenland-Newfoundland factor, is the pressure difference. Stykkis- 

 liolm-Bergen. The other yariates exert so little effect that they 

 warrant no further attention. 



Both factors (<?') and (c/) were found to l)e real as shown by the 

 following partial correlations: 



= -0.32 r„A.. = -0.39 



acf.d 



This leads to the development of a regression equation for fore- 

 casting purposes, on the basis : 



(«) = -0.08 (c')-0.12 (^/)+4.8 



When the calculated value of (-7) was plotted for the -t" years. 

 18S0-1026, against the actual value of recorded bergs it was found 

 that only in 8 instances did the calculated value differ by more than 

 ±2. on a scale 0—10. The years exhibiting the greatest departures 

 were 188.5, 1901, 1905, 1908, "1909, 1912, 1911, and 1924. Such occa- 



|/^^•»56789§/^J4i67898/^J•<5«7895/2^<'567a9^/^JI^J6789'n 



Icebergs South of Newfoundland 1880 1930 



FiGUKE 118. — The full line reijreseut.s the iiuiiilier uf icebergs smith of NewfouiulUinJ 

 each year on a scale of 0-10, mean value 4.8. The dashed line is the forecasted 

 number of berg.s obtained by substitution of the necessary meteorological data in 

 the regression equation (p. 188). 



sional abnormalities may be due to any one or all of several causes, 

 for instance, ice jams in the Arctic Archipelago; variations in pre- 

 cipitation; winter storms: sunimer calms; the breaking away of the 

 fast ice of ^Melville Bay and. })ossibly, of that fronting the great 

 Humboldt Glacier. 



The foregoing correlations, and the regression equation con- 

 structed therefrom, have been employed by the international ice 

 patrol to forecast the number of icebergs that may be expected to 

 (h-ift into the North Atlantic during the spring. The recent estab- 

 lishment of meteorological stations in (ireenland and northern Can- 

 ada, and the prompt receipt of their reports by radio permits the 

 nse of monthly mean values and the immediate construction of iso- 

 baric maps. In using snch in the ecpiation (r?) = — 0.08(r') — 0.12 

 («:/)+4.8, Tsafjord (Iceland) has been substitnted for Stykkishohn, 

 and Julianaluiab ((ireenland) for Ivigtut ; c', therefore represents: 



^, ^ 6[(2XDec.J-'"-t-Dec.i)-|-(2XJan.J-B'-t-Jan.')-KFeb.J-Bi+Feb.') +(Mar.J-»'+M:>r. ' ) ) 



5 



^ (Oct.'-B+Nov.'-B-|-2XDec.i-B+2XJan.i-B) 



Where'=Isan<)nrily prBsstire value for Julianahaab; BIsIsanomaly pres.sure value for Belle Isle; 

 I = Isanoinaly pressure value for Isafjord; and B = Isaru)inaly pressure value for Hergeii. 



