ABSTRACT
Digital Terrain Models (DTMs)
has become a very significant tool in extracting geomorphological information
from various land areas. Their use is in the fields of mapping, landscape
planning, urban design and many more. Automatic stereocorrelation has been used
to generate a DTM from ASTER stereo image pair (3N and 3B) using ENVI software.
Elevation values were extracted and used with four different interpolation
algorithms. The resulting surfaces when compared with those from the
topographic map showed that the Inverse Distance Weighting (IDW) can achieve a
Root Mean Square Error (RMSE) of ±10.773m and Mean Absolute
Error (MAE) of ±8.714m for flat terrain
as compared to RMSE of ±11.035m, MAE of ±8.999m for spline; RMSE of ±11.121m,
MAE of ±9.102m for Natural Neighbour (NN) and RMSE of ±12.108m, MAE of ±9.979m
for kriging interpolation method using a point density of 61.49 points per km2.
Hence, IDW is best for this surface type. For undulating terrain, IDW again
gave the least RMSE of ±13.549m and MAE of ±10.789m in comparison to RMSE of
±13.711m, MAE of ±10.963m for NN; RMSE of ±13.717m, MAE of ±11.028m for spline
and RMSE of ±14.835m, MAE of ±11.658m for kriging interpolation method for
point density 62.30 points per km2 and hence, IDW is again best for
this surface type. For mountainous terrain, NN interpolation method with RMSE
of ±19.044m and MAE of ±13.909m gave best results than the other interpolation
types. RMSE of ±21.167m, MAE of ±15.241m was obtained for kriging; RMSE of
±21.632, MAE of ±14.687m for IDW and RMSE of ±21.721m, MAE of ±14.544m for
spline for point density 141.64 points per km2 and so NN works best
for mountainous terrains. It is therefore recommended that IDW interpolation
algorithm should be used for both flat and undulating terrains whereas NN
should also be used for mountainous terrains.
CHAPTER ONE
INTRODUCTION
1.1 Background of
Study
Land surfaces are generally represented in a computer
environment as DTMs (Weibel and Heller, 1999). These digital representations
are modeled from terrain reliefs through collections of data samples and
algorithms which can interpolate elevations of intermediate unknown points.
DTMs have several possible applications such as in the field of military where
usage ranges from surveillance and intelligence gathering to strategic planning
in battle field as a guide in missile launching. DTMs also play an integrate
part in creating relief maps. Accurate elevation data helps geologists to
determine and extract various geomorphological information from various terrain
characteristics.
Advancement in technology has increased extensively the
capability of DTMs generation from satellite images to more accurately
represent terrains, making it useful in the field of civil engineering,
landscape planning, urban design and road traffic engineering. Integration of
DTM data with Geographic Information System (GIS) provides opportunity to model
terrain relief, analyze and visualize phenomenon related to topography.
Over the year’s digital representation of terrains have been
denoted severally as DTM, Digital Elevation Model (DEM) or Digital Surface
Model (DSM). Although these terms are mostly used synonymously, the difference
or meaning basically lies in its mode of application (Oksanen, 2006). DSM data
includes low rise and high rise buildings, roads, bridges, forest trees and
structures that can be found on the surface of the earth (Maune et al.
2007). DEM data does not necessarily include objects or manmade features on the
surface of the earth, but mostly represents the bare ground with natural
phenomenon like rivers (Oksanen, 2006; Maune et al. 2007). A DTM on the
other hand is a continuous or smooth surface which aside from the values of elevations
grids, also consists of other elements that describe the topographic surface
such as slope, aspect, curvature, gradient, skeleton (pits, saddles, ridges,
peaks) and others (Podobnikar,
2005). The DEM is often
used generically for DTM (Maune et al. 2007; Li et al, 2005).
DTMs can be represented and stored in several ways. The
commonly used data formats for DTMs are, (i) the regular grid (raster) and (ii)
the Triangulated Irregular Network (TIN) (Weibel and Heller, 1991; Peng et
al, 2004). The TIN transforms an irregularly spaced points data thus (x, y,
z) values to form contiguous, non-overlapping, triangles that represents the
surface. The TIN model allows extra data in complex areas and less data in
non-complex areas thereby reducing redundancy. This therefore enables it to
represent information about altitude, slope and aspects. However, they can be
quite demanding towards memory space and computing time and also the algorithms
involved could be sophisticated (De Wulf et al, 2012).
DTMs grid according to Weibel and Heller (1991) gives a
matrix structure that records topological relation between data points stored
as a two-dimensional array of elevations. Although the raster format has a
number of setbacks which involves a rectangular data array irrespective of the
morphology of the terrain, it remains the most popular format in the
foreseeable future (Pike et al, 2009). This is because, it represents a
terrain in a more technically controlled manner of grid cells where each cell
could have its own property (Hengl, 2006). Grid DTMs ensures simplicity of the
models and low memory space requirements whilst allowing for fast and
straightforward data computations (De Wulf et al, 2012).
In DTMs elevations are
presented as surface values on the land surface in areas of interest. Shi et
al (2005) recognized that the overall accuracy of a generated DTM depends
on both the propagation error and the model error. Leberl (1973) also
asserted that a DTM performance depend on the terrain and the method used in
interpolating the new points from the existing measurements. This therefore
suggests that, apart from a good sampling of points required to improve the
quality of any DTM, a good modelling of the surface would also depend on the
appropriate DTM interpolation method chosen and used.
Many research works have been conducted on the various
interpolation algorithms, however, an understanding of the terrain conditions
upon which the interpolation is performed have largely been ignored. Hengl et
al (2009) therefore claimed that, an inexperienced user would mostly be
confused as to which technique to select in order to produce a DTM that would
best suit a particular purpose. There are various data sources for DTMs. These
data sources are severally aerial photography, satellite imagery, cartographic
maps and measured terrestrial points.
1.2 Problem
Statement
Many techniques exist for
interpolating to approximate a surface from elevation data exists. The accuracy
of the resulting surfaces depends on the nature of the landform and the
interpolating algorithm used for interpolating the surface. There is no technique
defined for different landforms but the user has to experiment with different
techniques to select the best that will fit each landform type or just use any
randomly irrespective of whether it is the best for the circumstance or not.
The implementation and determination of which interpolation type is best for
each landform type poses a problem which is the objective of this research
1.3 Aim and
Objectives
1.3.1 Aim
The aim of this thesis is to
implement various interpolation algorithms on different terrain types and
determine which algorithm is most suitable for which type of terrain.
The objectives of
the research are;
1.
To generate DTMs from ASTER stereo
imagery.
2.
To investigate how the various
interpolation algorithms perform with different terrain characteristics.
3.
To determine the quality of DTM
generated.
1.4 Research
Questions
1.
How many interpolation points are
needed for DTM generation for each interpolation type?
2.
What interpolating algorithm is best
suited for the different terrains?
3.
What is the quality of the generated
surface?
1.5 Organization of
Thesis
The work
represented here are structured into five chapters.
Chapter 1 is an introductory chapter that includes a background
to the study and problem statement. The main aim and objectives are also laid
out here. A number of research questions are posed to answer the objectives.
In chapter 2, DTM generation and interpolation methods are
discussed. This chapter contains literature about DTM sources and the
generation of DTM from ASTER. Interpolation types and methods, as well as the
errors associated with interpolation and previous work that have been done are
also reviewed.
The
materials and the methodology applied in the current study are discussed in
chapter
This
include the processes involved in DTM generation, issue of contour derived DTM,
DTM modelling, study area and dataset preparation for interpolation.
The
results obtained are stated in chapter 4. This chapter also contains a
discussion of the results.
The
main findings are stated in chapter 5 as conclusions. Some recommendations are
also made towards further research in this chapter
For more Geomatic Engineering Projects Click here
===================================================================
Item Type: Ghanaian Topic | Size: 56 pages | Chapters: 1-5
Format: MS Word | Delivery: Within 30Mins.
===================================================================
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.