Geoscience Reference
In-Depth Information
excessive occurrence of pixel values corresponding to the values of input contour
lines - i.e. pixels with the remainder zero. The Hammock plot (i.e. the frequency of
particular modules) can also be represented as a histogram.
In Fig. 14.3c, d the red colour represents the areas of module 0, i.e. areas where
the values of elevation after rounding off to whole numbers correspond to the values
of the input contour lines. The higher the percentage of red colour, the higher influ-
ence of values of contour lines during the interpolation and creation of the DEM
(Fig. 14.3d). The result of a high-quality interpolation should be a balanced fre-
quency of pixels in all modules (Fig. 14.3c) which represents a better quality (more
even) calculation of values of DEM pixels even in areas more remote from the input
contour lines.
The Hammock plot is the basis for the calculation of the
hammock index
(H),
another characteristic of DEM quality which points out the balance of modules. Its
value falls into the interval <-1, i - 1>. Ideal values representing a balanced module
(i.e. balanced frequency in all intervals) tend to stay around zero. The formula for
the calculation of the hammock index is (Wood, 1996):
n
i
=
1
−
1
(
nf
0
)
−
f
i
H
=
(14.2)
n
−
1
f
i
i
=
0
where
n
is contour line interval,
f
0
is frequency of mod0 (frequency of pixels with
the remainder 0), and
f
i
is frequency of other mod.
The resulting value of the hammock index based on the hammock plot in
Fig. 14.3c equals the value 0.48, which means balance of all modules (Fig. 14.3e).
However, the value of the hammock index based on the hammock plot in Fig. 14.3d
equals the value 2.29, which in the given scope <-1, 4> means a considerable imbal-
ance of modules (Fig. 14.3f ) and points out a considerable influence of the values
of the original contour lines.
14.2.2 Analysis of Parameters of Interpolation Methods
Used for the Creation of DEM in the Czech Republic
On the basis of the above-mentioned assessment it is possible to determine the
most suitable interpolation method and its parameters for each type of relief. The
following recommendations are based on several works by the author (Kadlcíková,
2007b; Kadlcíková & Tucek, 2008) in which contour lines with the interval 5 m
from the digital land model DMÚ 25 (i.e. of the corresponding scale 1:25.000) were
used as the input data for testing. The output of the testing was represented by raster
grids of individual types of relief, the pixel size being 5 m. The testing regions
(Table 14.1) were selected so that they represent all basic relief types on the level
of districts as stipulated in the Regional geomorphologic segmentation of relief in
Search WWH ::
Custom Search