Satellite position 
a tree 
H=V/tan(EL) (1) 
where H is the horizontal displacement 
V is the vertical displacement 
EL is the sensor elevation angle 
Here suppose the EL at imaging time is 75 degrees, then a 
mountain peak with 1,000 meters will has a horizontally 
displacement of 267.949 meters, this displacement is important 
and inevitable. 
As we know, the changes of satellites’ parameters are capsule, 
the horizontal displacements generated by this also very small 
even with a large height. For instance, suppose the sensor 
elevation angle changes form 75 degrees by arc-second (one 
arc-second equal to 1/3600 of a degree) to 75.00028 degrees, 
then H changes from 267.949 meters to 267.944 meters, an 
insignificant change. It illustrates that vertical displacement has 
no improvement for satellite pointing calibration, so a flat area 
is more likely to be selected as a test range for it is as good as 
mountain. A flat area also more convenient to arrange GCP 
(ground control points). Good GCPs are related to the result of 
calibration also. To select culture features as GCPs (Gene Dial, 
Jacek Grodecki, 2003) is an effective way for build good GCPs 
which can easy be found on the image, be permanent, and be 
low-cost. 
2.3 Area condition 
This condition depends on satellites’ swath width. The overall 
size of a test range ought to suitable for satellite sensor so that 
geometric calibration can be done on a full scene at once. 
Choosing test ranges based on the swath of a satellite is one 
thing to be thought. 
2.4 Cost condition 
Cost condition is a subjective factor; to gain most profit with 
least input is one principle for doing everything. So if there are 
many areas satisfying the three conditions list above, the one 
with the lowest cost must be chosen. Several facts to be 
considered in this, such as traffic, GCPs, convenience. 
3. PROCEDURES 
Those four conditions are kept as rules when selecting a test 
range in China and the four analyze stages usually are carried 
out from weather analysis to terrain analysis, then consider the 
area condition, last take cost into account. 
3.1 Weather analysis 
Cloud is the emphasis in this stage. Analyzing this side needs 
meteorological datum. We obtained relative data from China 
Meteorological Administration. It is statistical data from year 
1951 to 2000. Cloud data is a picture about year average cloud 
amount distribution of the whole country. It tells us a rough 
result that where has the high probability of cloud- free skies. 
But no more detail geographic information in the picture. For 
obtain detailed geographic information, a superposable analysis 
with administrative divide map is necessary. 
First, using the ERDAS software to let the cloud amount 
distribution map has geographic information. Later in Arclnfo 
environment, it easy to do registration, and then select cloud- 
free zones with accurate geographic information. Plenty of 
zones satisfied this condition, choose some of them and then do 
terrain analysis on this result. 
3.2 Terrain analysis 
The purpose of this procedure is to judge landscape of zones 
which be selected in the previous stage and screen out flat areas 
for next analysis. In this article, according to Surveying and 
Mapping Standard Compilation, flat area is given a definition 
for zones whose topography has inconspicuously hypsography, 
and vertical displacement is less than 300 meters. (2003. 
Surveying and Mapping Standard Compilation) 
A programming is written for the judgment. When selecting a 
rectangular bound with geodetic coordinate of top left comer