The land subsidence of Tianjin area

mainly induced by withdrawal of ground water

has been measured by leveling measurements for many years.Although the monitoring results of leveling technique are reliable

the sparse leveling data grid in the area and the long time span on a single bench mark prevents us from understanding the process of deformation.Spaceborne differential radar interferometry（DInSAR） has been proven a remarkable potential for mapping ground deformation phenomena over tens-of-kilometers-wide areas with centimeter-scale accuracy on a more dense space grid and time series than leveling and GPS technique.As well known

geometrical and temporal decorrelation is an important factor that prevent DInSAR from being an operational tool for displacement monitoring.Moreover

atmospheric inhomogeneities produce an atmospheric phase screen（APS） on every SAR image

which can contaminate the results of the deformation monitoring.Interferogram stacking is a technique to improve the relative accuracy of SAR interferometric surface displacement mapping based on a combination of multiple interferograms.Under the assumption of statistical independence of the atmospheric distortions

the displacement terms add up linearly whereas the error term increases only with the square root of the number of pairs considered.Using ASAR images and the approach of interferogram stacking

the subsidence phenomena of Tianjin area has been mapped.9 ENVISAT ASAR images covering the period from October 2003 to August 2004 have been selected to retrieve the process of subsidence in Tianjin area.To reduce the geometric decorrelation and topographic errors

13 pairs with perpendicular baselines minor than 300 meters are chosen from the possible combinations.In the process of removing topographic component from the inerferometric images

SRTM DEM data are used.Since the orbit data of ENVISAT have been referenced to the WGS84 ellipsoid

the SRTM DEM height values the EGM96 geoid as the reference choose.The geoid height in the area has been compensated.Among the 13 differential interferograms

only one of them is believed to be severely affected by the atmospheric artifacts

the others show almost the same deformation phase model over the work area.Since it is difficult to discriminate displacement phase contributions from the atmospheric signature only by using individual interferogram

the approach of interferogram stacking is used.5 unwrapped phase images are summed and the time span is 350 days.Before phase unwrapped

the coherence values of pixels in all the interferograms are taken into consideration while coherent targets are selected.For the problem at hand

where in a large area high coherence urban patches are surrounded by vegetation-covered field with low coherence

the method of phase unwrapping firstly triangulates the unmasked pixels and then unwraps the phases based on the minimum cost flow algorithm with the averaged coherence map as a weight file.The geocoded subsidence map of test area during 90 days shows the distribution and the relative deformation value of the displacement field.Compared with previous measurements from DInSAR and in-situ GPS

the subsidence cones characterized by the method are reliable.But the temporal decorrelation and the atmospheric distortion are not completely overcome

and the deformation estimations derived from this method still need validation and correction by in-situ measurements mainly observed in Tianjin urban area.In the future work

with more ENVISAT ASAR data acquired

we plan to apply the PS InSAR and derived methods to exploit the data with long spatial and temporal baseline as much as possible.