METHODS 
A total of 65 trees were selected from three sites located in northern 
Kenya and one in semi-arid central Kenya. These included three of the most 
common growth forms of trees within the region; (a) flat topped, multiple 
stemmed (Acacia reficiens) (b) flat topped, single stemmed; (Acacia tortilis) 
A. mellifera) and (c) rounded crown, single stem (Commiphora sp. A. drepano- 
lobium). The number of trees per species ranged from 10 (A. reficiens) to 
30 (A. drepanolobium). Sample trees for each species were chosen from the 
full size range available up to a maximum height of 4 m. Sample site, size 
and date reflected logistical restraints rather than statistical requirements. 
Data were pooled from four independent sites. The crown diameter of each 
tree was measured and the tree was then cut down and weighed. Air dry weights 
were obtained after for one month drying in the dry season. Sub-samples 
were then oven dried at 90 degrees centigrade for 24 hours and weighed again. 
Air dry weights were then corrected to oven dry weights using the resultant 
oven dry to air dry weight ratio. Oven dry weight or woody biomass was then 
related to crown diameter using regression analysis. 
  
Along a transect parallel to and centered on a one kilometer stretch 
of road at the southern foot of Mt. Kulal near Lake Turkana in north- 
central Kenya, vertical black and white aerial photographs with 60$ overlap 
were taken using first a 70 mm motor driven camera with a 76 mm lens and 
second a 35 mm motor driven camera with a 35 mm lens. The two scales were 
1:2,000 and 1:1,317 respectively. A radar altimeter allowed the retention 
of a uniform flying height of 300 meters. Variation in scale was checked 
on the ground and aerial photo measurements of road width at intervals along 
the transect. These acted as control sites for any variation in photo scale. 
  
Fig. 1 
Crown diameter measurements of 40 trees were made on the aerial photo- s 
graphs (Fig. 1). The same trees were then located on the ground, identified 
as to species and crown diameter measurements made to the nearest decimeter. 
Regression analysis was then used to predict crown diameter from the photo- 
graphic measurements. The pr 
expres 
ing to 
Since it was not possible to use true double sampling in which some of 
the trees which were photographed were also destructively harvested, the V(O) -V 
following equations were developed in order to predict woody biomass from (Q = 
crown diameter as measured on aerial photographs. 
Let 
Let B = a + bC l. Then V 
where B is wood biomass in kg., C is crown diameter in m., and a and b are the and V( 
intercept and regression slope respectively. 
so tha 
Also let C 2d * eP 2. 
where P is the tree diameter as measured on the photographs in mm and d and 
e are regression coefficients. 
Then B = a + bd + beP 3. The ca 
To be useful, the variance of B, a measure of the precision with which woody es 
biomass is estimated, is also needed. penden 
Let Q = bd Then 
and R = beP and 
Then V(B) = V(a) +V(Q) +V(R) +2COV(aQ) +2COV(aR) +2COV(QR) 4, 
and 
416 
ne A A i mm sn P 
MEER SN ON RES