Results - Response of Flood Events to Land Use and Climate Change - page 127

Fig. 6.1 Sample area; comparison between the modeled DEM applied different interpolation
techniques. a IDW, b Topo to Raster (ANUDEM), c Spline, d Kriging, Circular, e Kriging,
Exponential, f Kriging, Gaussian, g Kriging, Stable, h Kriging, Spherical
were selected randomly. Then, a DEM was created using different interpolation
techniques based on this 70 % data point set.
The DEMs hill-shade images (Fig. 6.1 ) show a comparison of the final results
between different applied interpolation techniques and it focus on a small part of
the area.
Statistical analysis: Based on the first assumption, it was assumed that the
source DEM has some errors. Due to this reason, there was no observed data to
compare the interpolated results statistically.
Hydrological analysis: To compare the accuracy of the original DEM and
Modeled DEMs, the stream network for each interpolated surface and original
DEM was derived using ArcHYDRO tools in ArcGIS. The modeled stream net-
work was compared with the observed stream network (''true'' data). By visual-
ization analysis of all achieved results, it seems that IDW, Spline, Circular,
Exponential, and Stable produced the poorest results mainly in the upstream,
where a small part of the main channel is missing.
The results obtained from Topo to Raster, Gaussian, and spherical derived
drainage networks coincided better with the reference stream network. Still, there
is some discrepancy in the modeled network in some parts and the river follows a
wrong path. On the other hand, the drainage networks obtained from the above-
mentioned DEMs allow a much more accurate delineation of the stream network.
As shown in the Fig. 6.2 , the agreement between drainage networks based on
IDW and the observed drainage network is not perfect. In particular, a small part