Loading measurements . --Using the planar intersect technique (Brown 1974; Brown and 

 Roussopoulos 1974) , loading was measured for at least 15 randomly preselected sample 

 points along each primary transect. Sometimes the random selection resulted in more 

 than 15 points out of 50 for measurement of loading. For 1/4- to 1-inch particles, two 

 2-foot planes, crossed perpendicularly, were vertically oriented and intersections 

 counted. For 1- to 3-inch and greater than 3-inch pieces, the 2-foot planes were ex- 

 tended to 4 feet. Particles to 1/4 inch were not tallied because of the counting work 

 involved and the fact that the information appeared unnecessary to meet objectives.. 



Depth measurements . --High intercept and bulk depth measurements were recorded at 

 each sampling point. High intercept depth was measured as the vertical distance from 

 the bottom of the litter layer to the highest 0- to 3-inch diameter slash particle 

 intersecting each 2-foot plane. Pieces greater than 3 inches in diameter were omitted 

 in determining depth because they occur infrequently as the highest particle and have 

 considerably less influence on rate of spread than smaller pieces. 



Bulk depth was measured in each of four pie-shaped quadrants of a 2-foot diameter 

 cylinder whose central axis was vertically oriented at each sample point. The two 

 perpendicular sampling planes for tallying intersections of 1- to 3-inch particles, 

 delineated the cylinder into quadrants. The top of fuel was the average height of an 

 imaginary pliable sheet draped over the fuel particles. The bottom was at the base of 

 the litter layer. Vertical gaps free of fuel for more than 1 foot were subtracted from 

 each quadrant's depth. Gaps of less than 1 foot were assumed to maintain vertical con- 

 tinuity of flames, thus were included in the depth measurements. Depths of the four 

 quadrants were averaged to obtain a bulk depth estimate for each sample point. 



Lopping . --After loading and depth were measured initially, the slash along the 

 transect was lopped so that all branches were within 2 feet of the ground and boles 

 within 1 foot. High intercept and bulk depths were remeasured at all sample points 

 where depth had changed. 



Analysis 



Rationale 



To understand the rationale behind the method of analysis applied here, bear in 

 mind the objectives of the effort and the nature of the variables. The first objective 

 was to establish an equation that can be used to predict the mean bulk depth of a slash 

 fuel bed from quantities that describe the amount of slash on the area. The sampling 

 procedure provided an estimate of the bulk depth and the variables for quantifying the 

 slash loading at each sample point: the number of intercepts of 1/4- to 1-inch and 1- 

 to 3-inch-diameter fuel particles. The expected number of such intercepts can be 

 predicted from slash loading, tree species, and d.b.h. simply by parsing the loading 

 according to the fractional weight distribution of the individual tree crowns (Brown 

 1978). But the number of intercepts (regardless of size class) is subject to great 

 sampling variability, since it should be approximately Poisson distributed. 



A Poisson-distributed variable has a variance equal to its mean, so if N is its 

 expected value, samples from N - M through N + M should occur with approximately equal 

 frequency. This intrinsic variability makes it impossible to distinguish with certainty 

 between a change in the mean value of the number of intercepts and simple data scatter 

 when moving from one sample point to the next. To alleviate this problem, the data can 

 be aggregated, combining measurements that lie within a range ±/N of each other. We can 

 use the average values to discern the underlying trend of the average bulk depth with 

 the average number of intercepts; otherwise, the trend would be largely obscured due to 

 the great scatter in the individual samples. Sampled bulk depths also exhibited 

 significant scatter, indicating a need to aggregate data. 



1 



