The infra-red-photography considerably facilitated the separation of 
conifers and broadleaves. 
As the experiment deals with conifers up to and older than 41 years, only 
the areas of these conifers in pure stands and in mixtures within the total 
area, and within the grid cells were considered. Their areas within the 
grid cells formed the PU's. Within the PU's circular 0.05 ha plots formed 
the SU's, the locations of which are randomly selected. These plots are 
usually enumerated in the field, but in our experiment an aerial volume 
table was applied instead of field enumerations. The areas of the PU's 
per grid cell were determined on the mosaic by planimeter. In the test 
two variables are used in the PPS sampling of the PU's: 
l. conifer area, including pure conifers and group mixtures 2 41 years, 
2. total forest area, including category 1 plus pure broadleaves 2 41 years. 
The grid cells were 225 hectares and, since all PU's were taken, the 
formulae 4 and 5 were used for the calculation of total volume and variance. 
This was done in order to have all PU's with their respective weights in 
one sample. 
The following results were obtained, randomly locating 20 SU's (0.05 ha 
circular plots) in each PU. 
Table 1 
Estimates of mean volumes (y) and standard deviations (8 ). Grid 225 ha. 
PPS sampling PU's, random sampling SU's, 20 SU/PU. 
  
  
  
PU variable | area mean total standard deviation PU 
volume volume of total volume (sy) 
3 
ha m? /ha m? m? % M 
1) conifer 5917.9 325.54 | 1,923,763 265,972 13.8 62 
forest area 
2) total forest 6122.9 313.64 | 1,920,395 307,095 15.99 62 
area 
  
  
  
  
  
  
  
  
x) Dealing with the second PU variable: total forest area, SU's (= plots) 
falling in pure broadleaves have a zero volume. For category 1, 
n3/ha conifers area and for 2: m3/ha total forest area. 
As conifers forest area is more closely correlated to conifer volume, this 
variable is expected to result in a lower standard deviation. In the given 
example pure broadleaves cover 205 ha or 3.35 per cent of the total forest 
area. This results in a slight reduction in the standard deviation. If pure 
broadleaves had have covered a larger area, the reduction would have been 
more striking. Stratification into the two groups of species is therefore 
advisable when information for each of the two groups of species is wanted. 
In Table 2, results are given of volumes and standard deviations for the 
variable conifer area only, using three sizes of grid cells: squares 225 ha, 
and rectangles 450 aud 675 ha. Different numbers of randomly selected SU's 
per PU conifer area are taken, as indicated in coiumn 2. Total volumes and 
standard deviations are calculated again using Formulae 4 and 5 respectively. 
In practice, when comparisons between methods are not made, Formulae 2 and 3 
are used, requiring only a limited number of PU's to be taken for the 
calculation of the standard deviation. AÀ summation list of the PU conifer 
  
  
  
   
  
  
  
  
  
  
   
   
  
  
  
  
  
   
   
   
   
      
       
  
    
  
  
  
  
  
  
  
  
  
  
   
  
  
   
   
    
aret 
devi 
(inc 
to L 
valu 
the 
Tab] 
Esti 
PPS 
numl 
34 
29 
  
  
The 
in t: 
are: 
voli 
zer 
SU': 
Thi: 
For: 
Of | 
resı