the frequency content of the measured field. We now consider an alternative 

 approach of determining mean anomalies which utilizes the bicubic spline 

 method of de Boor (1962). In this approach, a two-dimensional cubic poly- 

 nomial surface is obtained by use of the bicubic spline, and the mean value is 

 generated directly by integration. The algorithm presented here is based on 

 de Boor's development, modified to utilize the preceeding cubic spline formula- 

 tion and to compute mean anomalies by integration of the resulting bicubic 

 surface. The development of the algorithm proceeds as follows. 



Assume that we have available a set of data points with coordinates 



(x.,y ), i = . . . I, i = . . . J, and with data value U.. at the intersections of 

 x i ' I ' 1 1 



a rectangular grid covering an area R (Figure 1). 



UOJ 



Vi 



t 



U03 







xi.yj 





U 2 





R ij 







U I 











Uoo 



U| 



u 2 o 



U30 





'10 



Xi 



FIGURE 1 . THE R.. GRID OF DATA VALUES FOR INPUT INTO THE 

 M 

 BICUBIC SPLINE ALGORITHM. 



Utilizing this data set with an appropriate set of boundary conditions specified 

 around the perimeter of R, a two-dimensional cubic polynomial of the form 



11 



