Geoscience Reference
In-Depth Information
3.1.2 D16 Algorithm Program Coding
In order to automatically generate the result from D16 algorithm formula, the
Ruby programming language is used to write a compact and simple D16 algorithm
structure as shown in Fig.
8
into an execution file. Ruby is a dynamic program-
ming language; reflective, general-purpose specialized in object-oriented concept
that combines syntax inspired by Perl with Smalltalk-like programming language
(Wikipedia
2010
).
3.2 Comparison of D8 and D16 Algorithm Result
There are many methods and approaches used to compare the results between D16
algorithm and the existing D8 algorithm. Small resample size pixels, large scale
area comparison, 3D modelling technique (3D view) and etc. were used to verify
the accuracy increment of the D16 algorithm from the D8 algorithm. In general,
these comparison methods can be classified based on the data type itself:
• Topographic map
: 3D Visualization
• SRTM
: Comparison using visual interpretation
: Generalization technique
: Edge reliability comparison
• ASTER
: Swamp area detection
Comparison of small pixel size with larger pixel size is a necessary method
because small pixel resolution provides more accurate results in situation where
the data is stored for each smaller pixel size in details as in a large-scale map. As
compared to a bigger resample pixel size, the information or data had been gen-
eralised and the output information is less accurate which contains fewer details as
stored in large small-scale map. A smaller pixel resolution is used and acts as a
schema to a bigger pixel resolution in comparing the accuracy of the final result for
both algorithms.
Wilson (
2002
) highlights D8 algorithm as good at valley area but cannot model
the flow divergence at ridge areas. Thus, a test on SRTM data will be used to
generate a general visual interpretation for both in valley and hilly areas. STRM
data also will be used in generalisation comparison for both algorithms, which use
a resampled pixel of 30 m resolution toward 120 m resolution.
A 3D model will be created from ArcScene module (ArcGIS software) for
visualisation purposes for both results. This testing technique is organized to
determine which algorithm can provide a better reliability result in the real
modelling scheme according to the nature of water flow direction (able to flow
