Table 2.— Distribution of sample trees by species and crown class used in estimating 

 foliage weight 



Range of measurements 



Number 

 of trees 



3.5 



=3.5 



Crown 



Number 

 of trees 





inches d.b.h. 



H h h 



Q.u.n. 



neigni 



length 



per acre 



Age 











Cant 









Years 



BP 



12 



4 





o - do 



iL - d.1 



461 



- 6324 



6-182 



WP 



17 



2 



n - 7 4 



o - bo 



o - 4d 



300 



- 4500 



4 - 55 



wc 



14 



1 1 



n - OK P, 



U - tLO.O 



2 - 1 27 



d. - 9o 



1 1 U 



- 7800 



3 - 237 



WC-INT 



11 



11 



- 10,6 



4 - 64 



2 - 42 



169 



- 3611 



16 - 162 



WL 



15 



3 



- 6.6 



3 - 50 



3 - 29 



600 



- 5100 



3 - 55 



VA/U 



Wn 



16 



4 



- 7.0 



4 - 38 



3 - 38 



848 



- 7200 



4 - 32 



AF 



15 



10 



0-12.7 



3 - 93 



2 - 70 



87 - 



15121 



4 - 201 



PP 



15 



20 



- 34.0 



2 - 144 



2 - 89 



5 



- 7200 



6 - 217 



PP-INT 



12 



9 



- 12.1 



4 - 70 



1 - 44 



142 



- 3344 



11-140 



LP 



12 







- 1.8 



2 - 13 



1 - 12 



300 



- 6000 



5 - 11 



GF 



16 



8 



- 15.6 



3 - 79 



3 - 70 



300 



- 7079 



4 - 124 



GF-INT 



13 



9 



0-11.8 



4 - 72 



2 - 60 



359 



- 5804 



15-117 



ES 



14 



3 



- 9.0 



2- 57 



2 - 44 



600 - 



12730 



7-153 



DF 



16 



10 



- 33.9 



4-143 



4 - 86 



21 



- 7200 



6 - 262 



DF-INT 



12 



10 



- 1 1 .2 



4 - 83 



1 - 47 



261 



- 4500 



12 - 145 



Total 



210 



114 















'See text page 2 for species list. 



where: 



D 

 H 



CL 

 CR 

 A 



TPA 

 BA 



OREL 



SP, 



diameter breast height, inches (cm) 

 height, ft (m) 

 crown length, ft (m) 

 crown ratio 

 tree age. years 



number of trees per acre (number per hectare) 

 stand basal area, ft^ ac (m^ ha) 

 relative diameter (tree d.b.h/quadratic mean 

 stand diameter) 



SI, = 



{ 

 { 



1 if tree is of species i, 



otherwise 



1 if tree is in site class j, 



otherwise 



predicted crown width {CW). ft (m) or 

 foliage weight {WT), lbs (kg) 



RESULTS 



Variable selection began with the model chosen in the REX 

 screening process as having the lowest mean square error/ 

 var(Y). A general model was chosen when coefficients that 

 were not significantly different from zero when tested by the 

 T-statistic at the 0.05 level, were deleted from the parameter 

 set. and all remaining coefficients were significant when the 

 model was refit. 



Refinements on the general model were investigated by 

 allowing each parameter to vary by species. Species x para- 

 meter interactions that significantly improved the regression 

 sum of squares were then included in the final model form. 



In developing the equations, the data were divided into two 

 sets consisting of those trees 3.5 inches (8.9 cm) d.b.h. and 

 larger, and trees less than 3.5 inches (8.9 cm). Each set was 

 then modeled independently. The break in diameter was 

 chosen to correspond to prognosis model methodology that 

 predicts diameter growth for large trees and height growth for 

 smaller trees. For the set of smaller trees, d.b.h. was not in- 

 cluded in the list of possible predictor variables. The funda- 

 mental reason for modeling the two size classes of trees dif- 

 ferently is that for small trees, either diameter at breast height is 

 not available, or is not as meaningful a dimension as it is for 

 larger trees. 



3 



