This implies that 

 (11.10) 0.54 Wro = 0,0100 



and that Wj.o- 0.0185. 



When Wj.Q is multiplied by 0.23 the result is 0.0043, This quantity 

 naust be subtracted from each value of U(r, 1) and U(rj*l) as r varies 

 from zero to 20 (before bordering). 



A total of O.OISS fft)^ times 20 is subtracted from the total E vajLue 

 of the stereo data when this correction is made. The total reduction of E 

 value is 0.372 (ft)^. 



The reanalyzed spectrum for this correction is not shown ia any 

 figure. It was used however, in subsequent analyses, and the correction 

 will be incorporated in subsequent plots. 



This correction as made to the spectrum also implies that a correction 

 must be applied to the covariance surface. The correction is to subtract 

 0,186 from Q(0, q) as q varies from +20 to *20, This correction re* 

 moves the effect of column noise from the covariance surface* 



Comparison with the wave pole spectrum 



The problem of transforming a spectrum of the form [A(iC, p)] into 



the form [(A(h,0)] in order to integrate out so that the dlreetloaal spec* 



trum can be compared with the wave pole spectrum is difficult. The «'s 



and p's are proportional to the square of the wave frequency. One sys» 



tem is in Cartesian coordinates and the other is in polar coordinates. The 



spectral estimates in both systems have considerable sam.pling variation. 



191 



