A STRATEGY TO BUILD A SEAMLESS MULTI-SCALE TIN-DEM DATABASE 
Xiong Hanjiang 3 , Tang Limin 3 , Sun Long 3 
a State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, WuHan University, 
HuBei Province, P.R.CHINA-xionghanjiang@163.com 
KEY WORDS: Geographical Information Science, Triangulation, Terrestrial photogrammetry, Spatial modeling, Topographic 
mapping 
ABSTRACT: 
With the development of 3D GIS, the visualization of the earth’s surface is more and more important. In order to improve the 
efficiency and the precise of the terrain visualization, it is essential to build a seamless multi-scale database. At present, building a 
global seamless multi-scale database is a challenge to many scholars. This paper discusses the strategy to build a seamless multi-scale 
terrain model database. Though triangulated irregular network has a lot of advantage in the terrain representation, there are several 
bottlenecks to solve. In this paper, it describes the technology of constructing the Delaunay Triangular and the data structure. Storage, 
integration and update strategy is also given. 
1 INTRODUCTION 
How to represent the earth’s surface is a basic problem of the 
geographic and the spatial information science. With the 
development of the digital earth, how to represent the spatial 
information of the earth has been the subject of extensive 
research in recent years. Digital Elevation Model (DEM for 
short) of the global terrain is an important component of the 
digital earth. Unfortunately, the present models are only suit for 
a local area. As for global area, there are no proper models and 
algorithms. A seamless DEM database can be a solution to the 
problem. It is a challenge to many scholars and it will be 
discussed in this paper. 
DEM was brought out for auto-engineering of highway 
originally in the late 1950’s. It is the digital expression of the 
earth’s surface, including both spatial and property information. 
After a half century, there have been kinds of methods to model 
the terrain, including physical models and digital models. 
Digital models can be modeled by mathematics and geometries. 
Kinds of functions are the mainly mathematics description, 
while the geometries methods are described by points, lines and 
areas. Grid, Contours and Triangulated Irregular Network (TIN 
for short) are the three primary geometric models for terrain 
representation. 
Comparing to the other models, TIN model has a lot of 
advantages in the expression of the terrain information as the 
unit of the model is the triangle. First, it is not only suit for 
regular distributed data points, but also suit for irregular 
distributed data points, while grid model is only suit for regular 
distributed data points. Generally speaking, the data points 
acquired from the field work don’t distribute regularly. TIN 
model offers a more flexible model. Second, because the earth’s 
surface is seldom absolutely flat, the data density is changing 
with the different terrain complexity. Grid model use the regular 
square meshes to represent the earth’s surface, while TIN use 
the irregular triangle meshes. TIN model can reduce the 
redundance data of Grid model and especially excel at the 
regions where the terrain is complicated and changes sharply. 
Third, the precision'of the TIN model is higher than other 
models on the terrain representation. Therefore, it does better in 
the precision and the dfficiency of calculating the elevation than 
contours model. Having so many advantages, TIN-DEM has 
been applied in hydrography and highway engineering 
successfully in recent years. 
Every coin has two sides. The disadvantages of TIN model are 
obvious, which limit the applying of the TIN model. First, 
complexity of the data structure makes it difficult to manage 
TIN data expediently and record the topologic relationships. 
And it is also hard to find a data structure that can manage all 
kinds of TIN data properly. Second, it takes a good while to 
generate the TIN meshes when there are millions of points. 
Though there are many algorithms to produce DEM based on 
the TIN model, few algorithms are high efficient. Third, 
nowadays especially in China, due to the main formats of the 
final DEM products are still based on Grid model and contours 
model, their middle product (TIN) is abandoned. It is a waste of 
time and money. What’s more, the update procedure of these 
products is inefficient and costly. As for TIN-DEM, it needs 
geographical feature lines and some other constrained edges to 
update the triangular meshes. The algorithm of the update 
process has to be optimized, too. 
In order to build a global seamless multi-scale TIN-DEM 
database, finding solutions to these disadvantages is highly 
important. This paper will discuss the possible solutions, 
including the TIN generation algorithms and the corresponding 
data structure, storage, integration and update solutions. 
2 ALGORITHMS AND DATA STRUCTURE 
The well-known Delaunay triangulation and its duality, Voronoi 
diagram, are becoming increasingly important and have found 
extensive applications in various fields (Aurenhammer F., 1991). 
It optimizes the minimal interior angle of constructed triangles, 
which makes it convenient for different engineering 
applications. In this paper, we use Delaunay triangulation. 
2.1 Delaunay Triangulation 
Let S be a set of non-collinear points in the plane. Triangulation 
T(s) is the maximal division of a plane into a set of triangles 
with the restriction that each triangle edge, except those 
defining the convex hull of S, is shared by two adjacent