The dependent variable, Y, and independent variables, X's, change as the array is 

 filled. A more extensive array using the same principle is shown below. 



The coefficients b , b, , b., and b^ are derived from the mean bulk depth and 

 o 1 2 3 ^ 



serial correlations of lags 1 and 2 of the bulk depths along linear transects 

 through the slash array (appendix IV) and not according to the cell filling for- 

 mat, shown above. The following data were obtained from field transects and are 

 used to develop the array for the slash example. 



Average depth - A. 44 in (11.3 cm) 



2 2 

 Variance = 23.01 m (148.5 cm ) 



Serial correlations 



Orientation Lag 1 Lag 2 



1 0.45 -0.044 



2 0.69 



3 0.62 +0.38 



Serial correlations are simply correlations of data pairs (Snedecor and 

 Cochran 1967) obtained from sequential transect depths of lag 1 and 2. A lag 

 of 1 designates a correlation of adjacent depths, whereas a lag of 2 designates 

 a correlation of data pairs obtained by skipping one depth in the transect se- 

 quence. The numerical value of the correlation is the correlation coefficient 

 that allows the user to quantify the similarity or dissimilarity of these data 

 p ai rs . 



The error, e, associated with each prediction of y is obtained through ran- 

 dom access of the cumulative distribution of the bulk depth (see appendix V) . 

 Because of the error term, generated fuel arrays are not identical. This is in 

 agreement with the goal to produce a pattern that has the essence of the array 

 but is not an exact reproduction. An example is wallpaper design. Your eye 

 recognizes a pattern, but may not find exact comparison. 



13 



