InSAR Ionospheric Error Correction Model And Method Based On Low-frequency SAR Split-spectrum Interferometry And Offset Estimation

Interferometric Synthetic Aperture Radar(InSAR)is a space remote sensing technology that can accurately measure tiny surface deformation.InSAR has been widely used in urban surface subsidence investigation,crustal deformation measurement,surface material migration,and change monitoring.In recent years,several SAR satellites with different radar wavebands have been launched gradually,providing rich data sources for the further promotion and application of InSAR technology.Interestingly,compared with highfrequency(short-wave)SAR systems(such as X/C band),low-frequency SAR systems(such as L/P band)can guarantee high coherence between SAR image pairs due to their strong penetration and have unique advantages in monitoring large gradient deformation,dense vegetation areas,and long time series of surface deformation.However,the impact of the SAR signal passing through the ionosphere is one of the primary error sources of InSAR deformation measurement.Remarkably,the ionospheric delay of the SAR signal is inversely proportional to the radar carrier frequency due to the free electron dispersion in the ionosphere.Meanwhile,since the density of the ionospheric free electrons varies significantly with latitude,the ionospheric influence on SAR azimuth deformation measurement is significantly higher than that of line of sight(LOS).At low or high geomagnetic latitudes,for low-frequency SAR systems,the ionospheric delay effect may cause deformation measurement errors of tens of centimeters or even meters in the line of InSAR.In contrast,as an important supplement of InSAR,the azimuth-direction deformation may lead to measurement errors of tens of meters in its measurement results.Although there are several conventional InSAR LOS direction ionospheric error correction models,the single correction models are affected by SAR system parameters,polarization mode,external conditions,or the model’s defects,while the joint model depends on prior information.InSAR azimuth ionospheric error correction mainly relies on the directional filtering and RSSI method.The former is difficult to accurately select empirical parameters,while the latter is limited by SAR bandwidth.It is worth noting that time-series InSAR is one of the crucial methods for high-precision dynamic deformation monitoring.The ionospheric delay correction of the time-series InSAR mainly adopts the RSSI method with poor robustness and the multi-subband RSSI method with low efficiency,which focus on LOS direction.As an essential supplement to time-series InSAR LOS direction,time-series InSAR azimuth deformation is more susceptible to ionospheric errors.However,no scholars have proposed an estimation and correction method for time-series InSAR azimuth ionospheric errors.To solve the above problems,this paper systematically analyzed the spatial distribution and spatial-temporal variation characteristics of the ionosphere and studied the ionospheric influence on the LOS direction and azimuth direction of InSAR measurement.Based on these studies,the ionospheric error correction theory of conventional InSAR LOS direction and MAI azimuth direction deformation is mainly studied.We proposed a combined ionospheric correction model for the InSAR LOS direction,integrating the RRSSI(Reformulating RSSI,RRSSI)and azimuth offset methods.Afterward,we constructed an ionospheric error correction model for the InSAR azimuth direction based on SAR azimuth Split-Spectrum Interferometry(Azi SSI).Simultaneously,considering the existing ionospheric correction methods for time-series InSAR LOS are low computational efficiency and robustness.We developed a robustness and calculation efficiency method for time-series InSAR LOS ionospheric correction.Besides,because the ionospheric correction for time-series InSAR azimuth direction has been blank for a long time,we proposed a new method for time-series azimuth ionospheric errors correction based on deformation model constraints.Aiming at the problem that the existing single models of conventional InSAR LOS ionospheric correction are affected by SAR system parameters,polarization mode,external conditions,or the model’s defects,while the joint model depends on prior information.We focused on the RSSI and azimuth shift(AS)methods.According to the properties of the RSSI and azimuth shift(AS)methods,this study constructed a posterior ionospheric joint correction model based on variance component estimation(VCE),RRSSI,and AS methods.ALOS-1PALSAR interferometric pairs with 28 MHz and 14 MHz bandwidths of track 471 covering the Wenchuan earthquake and track 103 covering the northwest coast of Chile were used to test the proposed model.Besides,we obtained 39 GPS observations within the coverage area of track 471 before and after the Wenchuan earthquake to verify the accuracy of the proposed model.The results show that the ionospheric error correction performance of the proposed model is better than that of the single models and the weighted least-squares method.The root mean square error(RMSE)between InSAR and GPS decreased from 11.80 cm before correction to 7.70 cm,and the improvement rate reached 34.75%,which was higher than31.86%,32.63%,and 32.63% for the AS method,RRSSI method,and weighted least-squares method.The existing ionospheric error correction methods of conventional InSAR azimuth direction are limited by SAR system parameters or empirical model parameter selection.To address this problem,we constructed an azimuth split-spectral interferometry(Azi SSI)model.ALOS-1 PALSAR interferometric pairs with a bandwidth of 28 MHz and 14 MHz of track476 covering the Wenchuan earthquake and track 103 covering the northwest coast of Chile were used as the test data.Also,the accuracy of the proposed model is verified by using the simulated azimuth deformation using the existing fault model and the 36 GPS observations within the coverage area of track 476 before and after the Wenchuan earthquake.Experimental results show that compared with the RRSSI and directional filtering methods,the proposed model can effectively remove the ionospheric errors in InSAR azimuth deformation and retain the real deformation information.Compared with the forward azimuth deformation and GPS data,it is found that the corrected results of the proposed model are pretty close to them.The RMSE of InSAR and GPS azimuth deformation decreases from 0.77 m before correction to 0.28 m after correction,with an improvement rate of 63.64%.The correlation between InSAR and GPS increases from 0.14 before correction to 0.59 after correction.Aiming at the poor robustness or low computational efficiency of the current time-series InSAR LOS direction ionospheric correction methods,we developed a time-series InSAR LOS direction ionospheric error correction method.To test the proposed method,ALOS-1PALSAR time-series images covering the Lazufre volcano region in Chile at low latitude and the Anaktuvuk River tundra fire region in Alaska at high latitude were used.The data from two GPS sites near the Lazufre volcano area were obtained for validation analysis.The results show that the correction effect of the proposed method is significantly better than that of the RSSI method.Compared with GPS data,the RMSEs between the InSAR and GPS data of the two GPS stations after correction are 6.22 cm and 0.39 cm,respectively.Compared with17.17 cm and 6.89 cm before correction,the improvement rate is 63.77% and 94.31%,respectively.Significantly higher than 44.50% and 87.17% after RSSI method correction.Because of the lack of current correction methods for time-series InSAR azimuth direction ionospheric errors and the difficulty of developing the existing ionospheric correction methods for conventional InSAR azimuth direction into a time-series model,we proposed the idea that is using the deformation model constraint to construct a time-series InSAR azimuth direction ionospheric error correction model.ALOS-1 PALSAR time-series images of track 103 and 104 covering the northwest coast of Chile from November 2007 to April 2011 were obtained to test the proposed model,and GPS observations within the image coverage area were obtained to verify the accuracy of the proposed model.Experiments show that the model can effectively remove ionospheric errors in azimuth deformation.Compared with the GPS time-series data,the maximum difference between InSAR and GPS time-series is about 2.8 m before correction.However,after correction,the time-series results of the two kinds of data are very close.The RMSEs between InSAR and GPS decreased from 0.77 m and 0.51 m before correction to 0.01 m and 0.01 m after correction,and the improvement rate was 98.70% and 98.04%,respectively.