325 
RESEARCH ON PRECISE GEOMETRY MODEL OF SYNTHETIC APERTURE RADAR 
INTERFEROMETRY 
Song Shujing, Liu Yihua, Jiao Jian, Zeng Qiming 
GIS and Remote Sensing Institute of Peking University, Beijing, China 
qmzeng@pku.edu.cn, enorlae@gmail.com 
KEYWORDS: InSAR, Geometry Model, DEM Correction, GCP 
ABSTRACT: 
InSAR, a new branch of Remote Sensing technology, has developed remarkably in recent years. Many researches have showed that 
high precision is always one of the most important goals in its development. In this paper, we systematically analyzed the 
disadvantages of Approximate Geometry Model (AGM) broadly adopt by InSAR. The AGM can not provide ideal level of precision 
because it does not take the curvature of the earth into account, assuming that the surface of the earth consists of planes. To achieve 
higher precision, we proposed Precise Geometry Model (PGM) which involves the curvature of the earth. The PGM has its obvious 
strengths especially when SAR images a wider area than it usually does. In fact, the precise model, to some extent, can prevent the 
errors which happen in each step from accumulating. It may decrease the possibility of total errors which affect the final output of 
InSAR. Based on the PGM, we proposed several significant steps of InSAR algorithms indicating how to flatten phase, estimate 
baseline and compute elevation for SAR interferometry. In addition, we implemented these steps in Visual Studio.Net and compared 
the experiment results with those obtained from commercial software such as EV-InSAR. Then we performed further analysis about 
results and reached a conclusion. Some suggestions were also be made for future research about PGM in more depth. 
1. INTRODUCTION 
Synthetic Aperture Radar Interferometry (InSAR), the synthesis 
of conventional SAR techniques and interferometry techniques, 
has been developed over several decades. Recently, InSAR has 
addressed several limitations in conventional SAR systems and 
been applied in numerous entirely new fields of earth science 
studies including detection of the earth’s surface change, 
topographic mapping, classification of land cover and land 
cover [1]. 
The precision of DEM (Digital Elevation Model) derived from 
InSAR highly rely on several crucial steps of the processing 
which consist of the baseline estimation, flattening phase and 
elevation computation. Since each step has its own 
characteristics, we summarized the recent researches 
respectively as follows: First, the current approach for baseline 
estimation mostly roots in parameters from satellites which are 
used with geometric parameters from InSAR to obtain the 
baselines. Now that the current approach had been proven its 
conciseness and efficiency, we estimated baselines by a very 
familiar way with a slight difference. Second, flattening phase 
can be performed in both time domain and frequency domain. 
In the time domain, the output of phase times the minus ground 
phase is equal to that of frequency spectrum transfer after 
Fourier Transform. In the frequency domain, on the other hand, 
frequencies of brightest fringes in both azimuth and range 
directions are evaluated and then corresponding complements 
are made to counteract the ground phase effects. Finally, 
methods of elevation computation can be classified into two 
categories according to whether control points are needed. 
Most algorithms base on geometry model of InSAR which 
helps establish the parameter equations regardless of whether 
control points are involved. However, the geometry models of 
InSAR are almost Approximate Geometry Model (AGM), a 
simplified model at the expense of precision. Since we 
expected high precision of InSAR output, we proposed Precise 
Geometry Model (PGM) to improve its results by 
implementing several algorithms for each significant step 
mentioned above. 
2. INSAR ALGORITHMS BASED ON PGM 
As we mentioned above, to obtain DEM is an essential 
application of InSAR techniques. In this process, two SLCs 
(Single Look Complex) in which store the information of 
phases are needed with the standard format of CEOS 
(Committee of Earth Observing System). Besides the SAR 
images, this format can also provide information about obit, 
acquisition time of each line, incidence angle of central pixels 
in each line and so on. 
2.1 Principle of InSAR 
Total phase <j> tota ,oï every pixel in SAR images consists of 
phase (j) Q which represents the characters of land objects and 
phase (j) R which is determined by a two-way route. Assuming 
that time break between two SAR images is very short, phase 
tj) Q remains the same in each image. Interferogram can be 
attained by registration of two SAR images and multiplication 
of plural conjugate images, in which (j) is eliminated and the 
left information is only related to topography and other 
transmission conditions. Then we divided the topography phase 
(f> into two parts which are respectively flat ground phase 
^ and altitude phase ^ shown in Equ. 1. 
fi total = 00 + 0R 
0 = 0flat + 0alt 
Fig.2 shows how the SAR sensor takes images twice to use 
interferometry techniques. In this figure, s ] and s 2