DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu




ŠUMARSKI LIST 7-8/2009 str. 17     <-- 17 -->        PDF

M. Vedriš, A. Jazbec, M. Frntić, M. Božić, E. Goršić: PRECIZNOST PROCJENE STRUKTURNIH ... Šumarski list br. 7–8, CXXXIII (2009), 369-379


peated measures ANOVA, due to lack of independence between methods (plot
sizes) on the same standpoints.


Estimates of number of trees by methods (Figure 2) ranged between 275.4
and 303.5 per hectare, although differences were not statistically significant at


0.05 confidence level (Repeated measures ANOVA: F = 0.6027, df = 7,
p = 0.7526). Precision expressed by relative sample error varied from 13.58 %
(K13-19-26) to 28.34 % (K5-12). Better results (lesser sample error) were obtained
on bigger plots, though concentric circles (K5-12, K7-13 and K7-13-20)
have considerably greater sample error due to fewer trees per plot.


Basal area estimates by methods ranged from 34.80 to 37.76 m2per hectare
(Figure 3), making no statistically significant differences (Repeated measures
ANOVA: F = 0.2948, df = 7, p = 0.9547). Relative precision ranged
from 10.13 % (K13-19-26) to 26.96 % on smallest plots (K7,98). Sample
error of basal area estimate on concentric circles was just slightly bigger in
spite of fewer trees per plot. Reason for that is stability of basal area on plots
regardless to fewer trees: concentric circles include fewer trees but have great
share of bigger ones that contribute to basal area more than smaller ones.


Estimate of stand volume by methods ranged from 457.93 to 496.47 m3per
hectare (Figure 4). There was no statistical difference in volume estimates between
analysed methods (Repeated measures ANOVA: F = 0.2650, df = 7,
p = 0.9661). Relative precision ranged between 10.14 % (K7-13-20) and


30.36%(K7,98). Better precision was obtained with bigger plots, due to more
trees per plot. Concentric circles produce just slight increase in sample error
while lowering the cost of measurement by reducing the number of trees per plot.


Number of measured trees per plot was computed as an indicator of plot efficiency.
Differences in number of trees per plot between plot sizes were statistically
significant at 0.05 level (Repeated measures ANOVA: F = 187.621,
df = 7, p = 0.0000), except for: K7,98 and K5-12; K7-13 and K9,77; K7-13-20
and K12,62 (Fisher LSD Post-hoc test).


Evident increasing trend of number of trees per plot by increasing of plot
size is the main cause of better precision. Although concentric circles reduce
number of trees per plot, loss of precision for basal area and volume are minimal
(Figure 5). Therefore plots K5-12 are acceptable for use in this kind of
stands, with remark that they require well trained staff and modern instruments.
Plots K7-13 do not improve precision while increasing number of trees
per plot (9), therefore are not recommended. Triple concentric circles K7-13-20
reduce sample error almost by 10 %, although by significant increase of measured
trees per plot.


Plots K11,28 reduce number of trees per plot with minimal increase in
sample error compared to K12,62 plots. That fact makes them acceptable
choice for gain in efficiency. However, K11,28 sample should be adjusted with
more plots to satisfy required sampling intensity (5 % of stand), which would
increase costs. In order to simplify the sampling plan, legislation does not require
precision rather sampling intensity (5 % of stand area), which restricts
opportunity to optimize sample size.


The choice of plot size is based on inventory goals and should depend on
cost of measurements and expected precision. This kind of research can provide
useful base for determining plot size by costs and precision of data. Exact
ratio of cost and precision could be computed by including time measurement
per plots of different sizes.


Key words: forest inventory, circular sample plots, number of trees,
basal area, volume, estimation, precision, CirConcomputer model