m puce WY 
  
Table 2. Linear regression equations predicting tree wood biomass, Y, (kg dry 
  
  
  
  
  
and wt) from crown diameter, Y(m). 
and 
h 
SPECIES SAMPLE SIZE REGRESSION CORR. COEFF. 
EQUATION 
e 
ned 1. Acacia drepanolobium 28 Y=-9.73+11.44X 0. 80* 
he 2. A. mellifera 14 Y=-4,17+3,55X 0.82* 
S of 3. KA. reficiens 9 Y=-2.68+2. 79X 0.62 
Town 4, A. Xorüls 12 Y=-13.54+7.48X 0.81* 
5. Commiphora sp.* 20 Y=-9,40+9.57X 0,81* 
the pecies X 35 Y--8.7545.64X 0. 78* 
Species 1,5 48 Y=-7,73+9.40X 0, 85* 
All species 83 Y=-7.76+7,49X 0. 70* 
eter 
S * Data from Walther & Herlocker ( in prep.) 
on of * Significant ( P<.05) 
at 
however. Thus, two different groups of trees were found for which data from 
the component tree pecies can be combined into a single predictive equation- 
for one for A. mellifera, A. reficiens and A. tortilis and another for A. drepano- 
ry to lobium and Commiphora Sp. These regression are given in Table 2 . along wi 
1 a Regression combining all five species. 
ata 
-Cies. Development of a single regression equation capable of predicting tree 
For wood biomass from crown diameter for a large number of tree species would 
the be useful in that it would reduce or eliminate the necessity to (a) develop, 
through destructive sampling, individual predictive equations for each new 
species or species groups encountered and (b) identify individual species 
f on aerial photographs before applying the regression. In the latter case 
imilar large scale normal colour and/or colour infrared must then be use for greater 
>rmly accuracy (Heller et al. 966; Aldred,1976; Miller et al., 1976; Sayn-Wittgenstein 
. et al.,1978; Aldrich ,1979). 
For the test stand the estimated mean biomass per tree was 33.38 kg. 
based on the use of statistics in Table 3 in equation 1. The variance of 
this estimate was 15.96 as derived by equation 4. Thus the estimated biomass 
m and its standard error was 33.38 kg. * 3.96. 
A variety of terms contributed to the total V(B) (Tables 4 and 5). Of 
T the six terms given in Table 4, only V(R) is directly dependent on V(P) as 
EF. seen in equation 6. This term comprises 209% of V(B). The other five terms 
Table 3. Regression statistics required in calculating the average woody 
biomass and the standard error of this estimate for the test stand 
  
  
  
STATISTIC VALUE STATISTIC VALUE 
a "7,76 V(d) 0.108 
eT b 7.49 V(e) 0.005 
d 1.04 p 4.078 
d e 1.09 V(P) 0.032 
V(a) 7.07 COV(a,b) - 11.542 
S V(b) 1,33 COV(d,e) -10.021 
ps, 
419