I. The aims of the study
The thesis approaches problems of the cartographic representation of
landforms using digital data sources, modelled and tested for an area W of Karlstad in Sweden.
During the modelling process several cartogrphical problems appeared which seemed to be the most current questions awaiting solutions both on a theoretical and a practical plane. Such problems were:
n amalgamation of data collected with different methods:
n analysis of aerial photos, field survey, maps in analog and digital form, databases and satellite images
-standardization of raw data
-databases related to different datum's
The development of the digital data technique has been driven by an actual
demand for data and data handling as a basis for planning decisions.
When the technical development advances, conditions are given for further exploitation of data. An example is the emergence of tools for visualization of the landscape both as an entire complex and in detail. this supports a better understanding of the landscape both as an entire complex and in detail. This supports a better understanding of the landscape and makes this information available to take into account in a physical planning process. Portraying of the landsurface forms have been the subject for the cartographer's attention periodically at various technical levels, already from the time of the first "map" construction. Despite of successive refinements in the terrain representation in analogue maps even the official maps of today have obvious limitations in this respect.
The digital data management have been faced with other types of questions,
Vizualization, functional and aesthetic design advanced to the centre of interest only during the last decade. Vizualization cannot be reduced to the question of "appearance", it embraces the techniques leading to the visual representation as well. The present work is focused on these techniques and their influence on the geographical and cartographical data quality, from the data collection to the final representation processes.
The aims and approach of the thesis can be formulated in the following objectives:
-To discuss possible error sources at various stage of digital geographic data handling and consequences of accumulation of such errors when the data are introduced in a GIS
-To focus on cartographic aspects in the processing and representation/visulaization of the data
-To compare different digital elevation datasets, sampled in the study, in terms of their potential to visualize medium-scale landforms in terrain types of the study area by application of interpolation algorithms and methods of data visualization
-To evaluate methods of data sampling to be used for improving upon existing less detailed datasets (time constraints, equipment needed, etc.)
These points form the backbone of the thesis with most of the objectives directed towards the main scope, the visualization. Why is that such a central point and why are the methods described and tested in the project important? The shortest way to answer the question is to state that these methods make it easier to understand the terrain. They give both a structural dissection and synthesis of the landscape elements, which for purposes of interpretation extends the traditional vertical map view and can even simulate the "being over there" sensation. This allows a planner to more easy lighlight terrain characteristics for an area or shorten the time spent on various types of field work.
II. Methods used in study
The methods used in the thesis are founded on sources available in the literature. In the choice of the computational programs and GIS utilities it was kept in mind that these must be commonly available with relatively easy-to-use interfaces. These principles were rather necessary as the target user of the methods are not necessarily GIS experts.
There was no intention to make a theoretical overview as a scope for itself.
It is almost impossible to be expert in all the issues raised here. The thesis involved data collection and surface modelling for five different medium-sized landform types, chosen in consultation with the physical and environmental planning units of the local municipality.
This reflects also the fact that a practical modelling process was to be presented in the thesis.
The cartographic aspects of digital geographic information takes the quality
and accuracy definitions as starting points. It takes account of the error sources
arising during the handling of digital map data and reviews the possible errors
during the recent project. Merging datasets using different references addresses
some questions waiting for answers, both theoretically and in practical use. The
sampling of continuous surfaces and their representation is a recurrent problem.
The generalisation procedures are a conscious reduction of the information content,
while the loss of information caused by conversion of databases is an unwanted
The overview of the cartographic aspects ends with some thoughts about visualization.
The methodical stages of the study were as follows:
- A theoretical backbone of the thesis, an overview of concepts and practices in digital geographic data handling applied in the study.
- Data representation forms including models, structures and formats.
- Review of the primary and secondary data sources concerning data collection and storage.
- Data transformation processes comprising geometric transformations, geographic-cartographic
conversions and conversions between different data structures.
- Geodetic datum's, co-ordinate system and map projections are discussed in general terms and with reference to their different standards and use in some countries.
- The description of the Hungarian reference systems is elaborated, for a comparison to the Swedish reference systems, which are treated in detail.
- Image processing methods, both the supervised and the unsupervised are described in context of their application to satellite images for land use mapping.
- Mathematical formulations and calculations described by geostatistics were applied to find attribute values between observation points measured by field survey and extracted from contour lines from aerial photos.
- The basic concept is that data are gathered at discrete points. To extend this information over a continuous surface, interpolation methods are applied.
- A point-to-point transformation takes place for reconstructing the terrain and to represent it by a digital surface model or other visualization method. This involves interpolation from the irregularly spaced data to a regular gridnet or a tessellated system. When choosing method for interpolation, no single algorithm is the best, as the distribution and the accuracy influence the results; the suitability depends on the nature of data elements, the context in which the results are to be applied and the desired output.
- The exact and the approximative methods can be used. A deterministic model is applied, if the context of the data values is well understood. The stochastic methods where the interpolated surface is one of several possible surfaces resting on the known points are flexible, with probability taken into consideration. The trend surfaces and Kriging belong to this type. The inverse distance method, the minimum curvature, the polynomial regression, the Shepard's method and triangulation are the deterministic interpolators used in recent project.
- Stochastic modelling predicts values of a variable distributed in space or time. These methods are used in 3 steps; exploration of data, structural analysis of data and Kriging estimation. The Kriging method with a weighted moving average around the point minimises the error variance of the predicted values.
- The critical selection of the acceptable interpolated surface controlled by the error measurement was based on the program packages offering different interpolation methods.
- The trend surfaces and Kriging belong to that. Inverse distance method, the minimum curvature, polynomial regression, the Shepard's method and triangulation are the deterministic interpolators used in recent project.
- Visualization techniques are essential parts in the application of GIS. A visual representation is the pictorial presentation of some mental imagery of data in media varying from paper maps to images on a computer display. Visualisation is a process of forming a mental picture, serving different purposes, including visual analysis. The visualization focuses attention on the use of computer graphics for acquiring a deeper understanding of data. The visualization process implies transformations when converting raw data into a displayable image.
- The function of visualization is exploring data and communicating results through their presentation. In modern cartography the whole palette of graphic tools are embraced by the visualization. This thesis is focused on the visualization of relief elements, by exploring methods of showing the terrain with a 3D impression.
III. Results of the study
Following the description of the theoretical framework, the main results of the thesis can be summarised as follows:
A general description is given of the situation of the study area, presenting its geology and bedrock geomorphology as well as the quaternary geology and geomorphology.
A review of the data from geographical and cartographical databases, data collected and handled by analysis of aerial photos or processing of satellite images, and data from a field survey.
An orthophoto mosaic was established allowing a "birds eye view" in an orthogonal manner, serving as localisation base for visualization of the studied landforms.
A land use map was produced by supervised classification of satellite scenes, after generalisation of the classified image.
Five different landform types were object for detailed study:
a terrain characterized by long parallel low ridges,
a part of Lake Vänern's shore with roche moutonnés
a glaciofluvial hill with shore terraces of shingle.
Statistical analysis of the databases collected with the different measurement
methods was applied for the chosen landforms. One of the aims was to quality the
" measuring methods" for visualization of medium sized landforms. Another
was to compare the collected dataset with the elevation data from the National
Elevation Databases, using Ordinary Kriging with linear option. The results of
the interpolation processes were visualized with 3D surface models.
Another type of visualization which was also tested is the Shaded relief method. The illumination was chosen with 45° vertical and 135° horizontal angle. The gradient method was used in controlling the smoothness of the shaded relief. The Lambertian reflection gives an ideal surface. The scale for the elevation values was exaggerated with a factor from 9 to 40.
As a measure for the reliability of the interpolated models, calculations of the residuals were made. Residuals have positive values when the measured data values are above the interpolated gridded surface and are negative when the measured elevation values are lower then the gridded surface. The differences between the elevation values from aerial photos, the field survey measurements and the values selected from the national Elevation Databases were calculated by grid mathematics. Volumetric computations gave numerical information about the "difference surface". To explain the deviations, the contour line maps of differences were evaluated.
Negative elements get a negative sign. Absolute values for the difference were calculated. Contour lines representing the absolute differences quality the measurement methods.
Profiles were produced in each study area along a line emphasizing the character of the landform in question, comparing the digitized contour lines from the Ekonomiska Karta, the elevation databases from the aerial photos and /or geodetic field survey and from the National Elevation Databases along a profile.
Triangulation of the already interpolated databases was needed to import them into the GIS project.
In order to represent the results of the study, six GIS projects were constructed. The most comprehensive project gives an account of the entire study area, showing the databases for geological conditions and the land use conditions. The elevation databases serves as a 3-dimensional surface to drape the above themes on.
The single projects for the studied landforms are structured with the content as follows:
digitized contours fitted to the National Elevation Databases models
digitized contours fitted to the model extracted and calculated after the aerial photos/field survey
path/traffic network mounted on the National Elevation Database models
path/traffic network mounted on the model extracted and calculated after the aerial photos/field survey
TIN model of the digitized contour lines
TIN model of gridded National Elevation Database selection
TIN model of gridded database extracted and calculated after the aerial photos/field survey
profiles mounted on the TIN model of the National Elevation Database and on the model extracted and calculated after the aerial photos/field survey.
IV. List of author's publications within the field of the thesis
Gyimóthy, B. - Lipcseyné, T. E.-Csapó, L.-Dévényi, M.-Maulik, P.: A területi információrendszer továbbfejlesztése, Beszámoló jelentés, Budapest, Tsz.: 420/79
Lipcseyné, T. E.: Harmadrendu spline-ok (Third grade splines), MSc thesis,
Lipcsey, E.: The application of Computer Cartography in Research and Teaching
at University Level, Progress in Geographical Research, Ahmadu Bello University,
Zaria, Nigeria, 1986
Lipcsey, E.: Cartographic Aspects of the Use of Digital Land Information in
Geographic Information Systems for Planning Purposes, in Geographic Information
Systems and Computer Cartography as Tools in Research and Education at the
University of Karlstad, 1996
Lipcseyné, T. E.: A térinformatika geodéziai alapjai Svédországban (Geodéziai dátum, kordináta rendszerek és vetületek), Gedézia és Kartográfia, accepted for publication.