The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
660 
Discrete Global Grid (DGG) is an earth simulate grid that can 
be infinitely partitioned without shape transform. This discrete 
grid can simulate the earth surface when it is partitioned to a 
certain degree [5]. It has hierarchy and it is continuous in global 
area. So it is hopefully to solve the problems such as data fault, 
distortion and topology inconsistent which will happen when 
flat gird model is adopted in global multi-scale data 
management. It has become a new research hot point in the 
international GIS academe. 
Nowadays the Discrete Global Grid is not mature. Instead many 
different solutions using inscribed polygon are adopted. Some 
are based on octahedron, such as the hierarchal structure model 
of global data, the continuous index model, the spatial data 
quality and hierarchal integrated model. Some are based on 
dodecahedron, such as the global DEM and image data 
compression model. Some are based on icosahedrons, such as 
global hierarchal data index model, the global navigation model 
and the global grid positioning system. Some are based on 
generic icosahedrons, such as global environment detection 
model [6]. 
But these solutions above have disadvantages. They organize 
and partition data regions on the basis of triangle unit. So the 
spatial operations should be carried out on the basis of triangle. 
For the geometry structure of triangle grid is complex, it 
doesn’t have the confirmed orientation and symmetrical shape. 
So the spatial index is much complex and it is difficult to meet 
the demand of contiguity analysis, spatial query, data updating 
and visualization. Otherwise the data provided by data 
producing department is strictly partitioned according to 
longitude and latitude grid. It needs large mount of work to 
partition and organize data according to inscribed polygon. 
2.2 The simplified partition and organizing method 
adopted 
This paper adopts a simple earth partition method according to 
the demand of global terrain visualization. 
Plate Carree projection is to directly map the longitude 
coordinate of the earth sphere to x axis and map the latitude 
coordinate of the earth sphere to y axis. This projection method 
results in the plane orthogonalization. The result of the 
projected sphere is a rectangle with the ratio of length and 
width 2:1. The demonstrative figure is Fig 1. 
Figure 1. Plate Carree projection and geographic subdivisions 
of the globe 
Although the Plate Carree projection and geographic 
subdivisions of the globe has distortions at the poles, it is more 
simple and familiar to us. And it can meet the demand of 
visualization for the two reasons below: 
The first reason is that when we view earth from the space, the 
view point is far away from earth. So we can treat the earth 
sphere as a regular sphere and the elevation fluctuation of the 
earth's surface can be ignored. It is not necessary to display the 
coarsest terrain data on the earth surface. When the view point 
gets close to the earth, we only need to display a pseudo 
background sphere. When the view point is close enough to the 
earth surface, the elevation fluctuation need to be displayed. At 
this time the earth surface can be treated as an infinite flat 
surface. In this small scale the distortion along the longitude 
can be ignored. 
The second reason is that when we browse the globe, most of 
the action still occurs at the middle latitudes where the majority 
of the world’s population is concentrated. In these places the 
distortion is small and cannot affect our visual effect. 
So in this paper Plate Carree projection and geographic 
subdivisions of the globe is adopted instead of polyhedral 
subdivisions. After the global surface is subdivided the pyramid 
model and linear quad tree are used to manage and organize 
DEM data. 
Figure 2. Quad tree partition and organizing method 
Some of the important parameters need be introduced in linear 
quad tree and pyramid structure. 
> Pixel: the elevation point in block 
> Block (ID): the data in every leaf node can be presented as 
block. The serial number of each leaf node begins from the 
left-bottom comer of the entire terrain data. The origin is at 
geographic coordinate (-180° ,-90° ). At the ID number 
increase from left to right, from bottom to top. 
> Layer: Refers to the layer of quad tree. The top layer is 
defined as the 0 layer. This layer stores the original DEM 
data; the data resolution is the highest. 
The Meta data of layer in the pyramid is as below 
struct CVGEDemLayerlnfo 
{ 
int XBlocksNum; 
//the Block number in x direction of current layer 
int YBlocksNum; 
// the Block number in y direction of current layer