STUDY ON GPS STATION’S ZENITH DELAY TO MITIGATE THE INSAR 
ATMOSPHERE EFFECT 
LI Tao a , ZAHNG Shiyu a , Zhang Shuangcheng 3 , LIU Jingnan 3 , XIA Ye b 
3 GNSS Research Center of Wuhan University, 129 Luoyu Road,Wuhan, 430079, China - taoli@whu.edu.cn 
b GeoForschungsZentrum Potsdam, Germany - xia@gfz-potsdam.de 
Commission VII, WG VII/2 
KEY WORDS: Atmosphere, SAR, GPS, Temporal, Correlation, Climate 
ABSTRACT: 
Both InSAR and GPS can detect tropospheric delay of radar beam propagation caused by water vapour fluctuation. What InSAR 
detected is the differential results of two SAR images taken at two epochs whose spatial resolution can be tens of meters. GPS can 
continuously detect tropospheric delay above the station and average the final results into zenith direction. But as GPS stations spans 
tens of kilometres, the final interpolated tropospheric delay in space is very poor and it depends on the distribution of GPS stations. 
This paper studied the tropospheric delay on Tandem interferogram in Tianjin area. Using surface fitting method in baseline error 
correcting, both the baseline error and the whole scene tropospheric delay were suppressed. Meanwhile, 13 GPS continuous 
operating stations in Tianjin were used to calculate the zenith delay over each station. GAMMIT software is used to get the zenith 
delay according to the Tandem SAR image acquiring season and epoch. The spline method was used to interpolate the zenith delay 
to the SAR image area and corresponding results were helpful to mitigate the tropospheric delay on Tandem interferogram. This 
study shows that GPS zenith delay may be useful to mitigate the tropospheric delay in InSAR. 
1. INTRUDUCTION 
2. TROPOSPHERIC DELAY 
InSAR technique is widely used in topography mapping and 
surface displacement. However, the results were sensitive to the 
satellite orbit errors, tropospheric effect, decorrelation, and so 
on. 
In 1994, Massonnet firstly found the tropospheric effect in 
interferograms while studying Landers earthquakes. After that, 
many researchers paid attention to this field. Hanssen used 26 
Tandem ERS interferogram to study the tropospheric delay 
effect in Netherlands and the results showed that tropospheric 
delay in interferograms may be 0.3 to 2.3 circles. In 2004, Li 
Zhiwei, et.al testified that the tropospheric errors in 
interferogram can not be regarded as Gaussian distribution 
which means we can not use averaging method to mitigate this 
kind of errors. In 2005, Li zhenhong, et.al proposed a new 
method to interpolate the GPS zenith day in large area 
considering the topography and correlation among the GPS 
stations which is called GTTM( Topography-Dependent 
Turbulence Model). Using this model the MODIS data can be 
rectified and the fusing data of GPS/MODIS turned to be more 
accuracy than before. However, MODIS data is not so 
appropriate for InSAR because it is sensitive to cloudy 
condition. 
This paper used 13 GPS continuous operation stations and the 
tandem interferogram in Tianjin area to study the correlation 
between the two data. The results showed that both methods can 
measure the tropospheric delay very accurately, but as GPS 
stations are located sparsely its final resolution may not satisfy 
the InSAR correction. 
2.1 Tropospheric delay 
The reflection rate of the microwave can be regarded as a 
variable along the propagation path. The tropospheric delay can 
be accumulated by the following function: 
S = 
10 6 cos 9. 
-(ioX^+f N W ',dh)=S„ + S„ 
(1) 
where H = the height of troposphere 
= the incident angel of the radar beam, 
N hyp = the hydro reflection rate 
N wet = the wet reflection rate, 
V = dry delay 
s ~” - the wet delay 
Using the observations of the local temperature, atmospheric 
pressure and the atmospheric humidity, tropospheric delay can 
be calculated. Using this result as an initial one, GPS software 
carried out more accurate zenith delay in every one hour or 
every one quarter. Every one minute zenith delay may not be 
reliable for this may arise too much variables to the function 
and cause false results. 
2.2 Tropospheric delay and interferogram phase 
In interferogram, the tropospheric delay can be derived by 
phase variation: