DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu



ŠUMARSKI LIST 5-6/2010 str. 56     <-- 56 -->        PDF

D. Klobučar: PRIMJENA GEOSTATISTIKE U UREĐIVANJU ŠUMAŠumarski list br. 5–6, CXXXIV (2010), 249-259
Uuttera,J., M.Maltamo, S.Kurki, S.Mykrä,vestigations of Forest Ecosystems Using Re1998:
Differences in forest structure and land-mote Sensing Imagery. Silva Fennica 39(4):
scape patterns between ownership groups in599–617.
Central Finland. Boreal Env. Res. 3: 191–200.


Osnova gospodarenja za g. j. “Banov Brod”.
ISSN 1239-6095.


Pravilnik o uređivanju šuma (NN 111/06; NN 141/08).
Zawadzki,J., C. J.Cieszewski, M.Zasada, R.


C.Lowe,2005:Applying Geostatistics for In-
SUMMARY: The possibilities of forest measurements have been significantly
improved nowadays, by using georeferenced maps, implementing remote
sensing, developing artificial intelligence, using the global positioning
system and geographical information system. Moreover, the exact position (x,
y) of the measurement (of variables) of the specific location (Z) in the forest
allows the monitoring of the information and the analysis of the so called continuous
model of spatial variation, as opposed to the discrete model of spatial
variation which is assumed to be homogeneous.


Ever since geostatistics was introduced to geoscinces (Krige 1951, Matheron
1965), it has been implemented in many areas whose interest lies in analyzing
spatial data. Geostatistics is based on the concept of regionalized
variable (which means that the value of the variable depends on the sampling
area).


The goal was research and presentation of using geostatistics in the forest
management, with the aim of improving the present approach to using and
mapping the forest inventory data for Croatia. The geostatistical analysis was
performed on a part of an management unit “Banov Brod”, Pitomača forestry
administration, for three structural elements (variables): number of trees (N),
basal area (G) and volume (V). The research included the compartments /subcompartments
9a, d, e, 10 a, b (Figure 1), with the total area of 69, 57 ha.


In order to determine the anisotropy, semivariogram surface maps of each
of the elements were made. The semivariograms were used as a measure of
spatial dependence, and experimental and theoretical semivariograms were
calculated. The experimental semivariogram for each structural element was
calculated after multiple fitting of number and width of lags. The parameters
used for Ordinary Kringing interpolation of each of the structural elements
were obtained from the theoretical semivariogram model.


The interpolation of structural elements was also conducted by using the
inverse distance method. The testing of the interpolation model was done by
using a numeric cross-validation approach. Furthermore, the usefulness of
making a variogram cloud in the spatial structural elements’ analysis was
shown. Three programs were used during this project: VARIOWIN 2.21; SURFER
8.0™, and STATISTICA 7.1 ™.


Semivariogram surface maps for the three analyzed structural elements
did not indicate the presence of anisotropy (Figure 2). As anisotropy was not
determined and omnidirectional experimental semivariogram were calculated
(Figure 3). All experimental semivariograms can be considered reliable because
they contain a great number of pairs of data. What they have in common
is the existence of hardly explainable high nugget, that is the difference in the
values of close samples or measurement errors, as well as the range, which is
bigger than the sampling interval. The omnidirectional experimental semivariogram
of the tree volume and basal area (Figures 3a, b) start oscillating
very soon, which shows that there is no large range of these two structural elements
in any direction. The omnidirectional experimental semivariogram of