Study On Atmosphere Delay Correction Methods In Differential Interferometric Synthetic Aperture Radar

In recent years, geological disasters frequently occurred in our country, such asearthquake, ground subsidence, landslides, debris flows and so on, which have caused aserious threats to people’s life and property security. In order to ensure the smoothimplementation of the sustainable development strategy and establish people-orientedharmonious society, a series of geological disaster investigation and prevention projectshave been carried out in our country. As an effective tool in a wide range of geologicaldisaster survey and monitoring, InSAR plays an important role. However, due to theexistence of water vapor, the radar signal will get propagation path delay and propagationspeed slow down when it through the atmosphere. those will lead an additional delay valuesin our observation phase, i.e. atmospheric delay phase, which affects the accuracy of theInSAR technology and limits its application in the landslide, dangerous rock massdeformation monitoring and so on. Especially with the subway and high-speed railconstruction meeting the rapid economic development needs, it is an urgent need to acquirea large range of high precision surface deformation information around subway andhigh-speed rail line. So study on how to correct the atmospheric delay error and improve thelarge-scale InSAR monitoring accuracy has great significance not only to expand InSARapplication fileds and monitoring ability, but also to geological hazard survey and economicconstruction.For atmospheric delay error, two kinds of the atmospheric delay correction methodshave been studied in this paper, namely based on SAR image themself methed and based onexternal data method, such as using MODIS, GPS, ECMWF data. The principle and methodof two kinds of the atmospheric delay correction methods have been described in n detail.and some new algorithms have been recommended. Some useful conclusions have beenobtained basing on the case study of all kinds of methods.Through the study, this paper made the following main achievements and innovation:1) For the problem that the MODIS water vapor retrieval algorithm is affected byvarious reflection errors, the MODIS water vapor error correction model has been studiedby using GPS and cloud free MODIS data cover xian area. The conversion coefficient between zenith wet delay and precipitable water vapor value has been studied. The studiesshow that MODIS water vapor value larger than GPS water vapor value about1.127times,which there is no relation with the elevation. The conversion coefficient considering theseasonal changes has a high precision. For Xi’an area the conversion coefficient is6.0insummer and6.52in winter.2) The atmospheric delay correction method by using MODIS water vapor has beenstudied. The interpolation problem for the cloud of pollution pixels are discussed, and theatmospheric delay correction ability and extend have been proved by the correctionexamples. At the same time, this paper points out that under the conditions of atmospherechanges greatly, using MODIS data to correct even will introduce observation error, sousing MODIS image for correction of D-InSAR results must be cautious.3) For the problem of space interpolation by using GPS to correct atmospheric delay,GPS wet delay interpolation algorithm considering topographic effects and new algorithmKriging interpolation lgorithm considering terrain factors are proposed. Studies show thattwo kinds of interpolation methods proposed in this paper can reflect the topography effecton atmospheric distribution and detail information better than that of the traditional Kriginginterpolation method and IDW. At the same time, the proposed IKriging method can also beused for multiple factors influencing interpolation.4) For atmospheric delay correction by using ECMWF data, low spatial resolution andtime resolution are two important limited factors. For the disadvantage of the ECMWF datawith lower spatial resolution, the strected boundary layer model (SBLM) has been studied andcarried out, and the corresponding software is compiled. When they are compared to morethan5600sampling points GPS and surface temperature, pressure data for two months insummer of Europe, the results showed that the correlation coefficients between the SBLMcalculated PWV, P and T and true values were as high as0.94,0.99and0.95, the precision ofthe model were2.38mm and2.97hpa and2.42K. Study on SBLM characteristics showedthat the precision of the model were more affected by elevation change than the latitude.When SBLM was used to correct two pairs tripe SAR interferograms atmospheric delaycorrection cover taiyuan basin, part of the terrain related delay phase and long wave phasewere corrected, and the correction of amplitude were up to28%and43%respectively. 5) When there is no available external data for the atmospheric delay correction, theatmospheric correction algorithm for single differential interferograms were presented. Thesignificant level of terrain correlation wet delay model were studied by Akaike informationcriterion, which shows that exponential model is more consistent with differential delayphase and topography regression relationship compared with linear model. In order to avoidthe effect of noise on the differential delay phase and topography regression model, theSection Median Method (SMM) is presented, which simplified and refined the regressionmodel. The method is used for the actual data of the atmospheric delay correction showsthat the correction effect is very obvious and up to70%. The method provides a train ofthought for single interferogram atmospheric delay phase correction when the absence ofany external data for atmospheric delay phase correction.6) The reliability of the atmospheric phase separated from InSAR time series analysismethod and the correction effect are researched and evaluated for the first time in this paper.The study shows that there are certain relationship between the atmospheric phase separatedfrom InSAR time series analysis method and height, which is consistent with the spacedistribution feature atmospheric delay.The study shows that the maximum correction ratecan reach up to10mm before and after the atmospheric delay correction, and the resultsshow more smooth after the atmospheric correction than that befor. The results has highconsistency between PS-InSAR and SBAS-InSAR after the atmospheric delay correction,and the average discrepant values is within5mm. The deformation trend of the resultsbefore and after the atmospheric delay correction is basically the same. The methods ofInSAR time series analysishas a very good inhibitory effect on the atmospheric delay phase.