and an estimate of the (standard error of estimate)? 
"g ve "E 
y 2-023 
2 ei 
v.c n- 1 
( g ) 
Application 
The four different sampling systems were applied to the combination 
conifers (spruce + fir) of age-class => 41 years, occurring on a 10,000 
hectare forest area in Austria. The major part of the area is occupied by 
pure spruce; on the remaining part spruce grows together with smaller 
groups or single trees of fir and beech. Pure stands of beech occupy only 
small areas. 
The area is covered by black and white infra-red aerial photographs at a 
scale of 1:30,000 on which the conifers can easily be separated from the 
beech because of their darker tone. Infra-red photographs at a scale of 
1:10,000 are also available, but only for part of the area. From the 
1:30,000 scale photographs, three sheets of orthophoto mosaics, each with 
a scale of 1:10,000, were constructed: These sheets were mounted together, 
and using existing management maps as À guide, a delineation was made of 
the boundaries between forest and non-forest, properties, and age-classes 
of < 40 and > 41 years. In addition, one hundred per cent pure beech 
stands were delineated. For the stratified sampling, the conifers were 
further subdivided into two strata, Called Ópen and dense. For the two-stage 
PPS sampling the total 10,000 ha was covered by a square grid, each cell 
of which occupied 225 ha. The conifer areas of each cell served as primary 
units. 
In all sampling methods, the sample plots (for the two-stage sampling 
method, the sample plots are identical with secundary units) were randomly 
pricked and numbered on the orthophoto mosaic. The plots were circular and 
covered an area of 0.05 ha each. 
In all sampling methods ten samples were taken, from which the calculation 
of standard deviations and total volumes were made. Taking meañs of ten 
samples, gave a more reliable figure to be used for comparing the different 
methods. In addition the ten sample results allow the calculation of a 
rough estimate of the coefficient of variation of the standard deviation 
(CVs) which accordíng to Hansen, Hurwitz and Madow (1953) should not 
exceed 13 per cent. 
In the calculations of total volume and standard deviation, the area 
occupied by the conifers is the same for all methods, i e 5917.9 hectare. 
This area is assumed to be free of error. 
For the unrestricted sampling method, estimates of total volume and 
standard deviation resulted from ten samples, in each of which 40 plots 
were taken (see Table 1) 
For the stratified sampling, the area proportion of the strata open and 
dense was respectively Po - 0.21 and Pd - 0.79. In each of the ten samples 
here, 25 plots were taken, proportionally distributed over the two strata. 
Estimates for total volume and standard deviation are given in Table 1. 
For the two-stage sampling please refer to Stellingwerf (1978). 
Stellingwerf demonstrates that, in the same forest area as described in 
this article, for a two-stage PPS sampling, optimum results were obtained 
   
with gl 
total ! 
Samplii 
(apply: 
volume 
For the 
were Te 
percen! 
other 1 
calcul: 
The rej 
The (s! 
the vai 
The me: 
and the 
The vai 
and the 
For thi 
These : 
Table : 
Estima 
the st: 
Method 
Unrestricte 
Stratified 
Two-stage 
Two-phase