Atmospheric Correction Technique in SAR Interferometry for Monitoring Volcanic Activities

Abstract

Ground deformation in volcano is a consequence of changes in magma chambers volume. Magma storage, migration and volume change is closely associated phenomena with the ground deformation. Therefore, measuring ground deformation provides important information to understand the volcanic activities. For some specific volcanoes, such as Shinmoedake volcano, ground deformation of even a few centimeters can occur before eruption. Thus, measuring ground deformation needs to be fairly accurate. SAR interferometry is a potential technique to measure the ground deformation accurately. One of the limitations in SAR interferometry, however, is atmospheric phase delay effects, which are induced when microwave propagates into the atmosphere. In this aspect, various methods for mitigating atmospheric phase delay effects have been developed. This study aims to mitigate the atmospheric phase delay especially in volcano because the stratified and turbulent atmospheric phase delay effects could severely contaminate the deformation patterns. First method used in this study is the atmospheric correction technique using MODIS data. Multispectral observation can measure the integrated water vapor in the atmosphere by analyzing ratios of water vapor absorbing channel and atmospheric window channel. It can be directly used for calculating the tropospheric phase delay effect caused by water vapor. Recent researches using multispectral datasets are restricted to approach using ENVISAT. Therefore, new approach is necessary in application using ALOS PALSAR. This study evaluates the applicability and possibility. In adequate temporal difference and cloud coverage, available datasets of MODIS successfully converted to the atmospheric phase delay corresponding to SAR acquisition time. However, there are some limitations in application into all dataset because of the cloud cover and temporal difference between the SAR acquisition time and MODIS acquisition time. In spite of limitations, the use of MODIS data in atmospheric correction yield better results and minimize misinterpreted errors. The WRF model complements the limitations of MODIS data. In this respect, an application of the WRF model in atmospheric correction of differential interferogram was carried out in the second methods. The estimated APS from the WRF model can explain the stratified APS involved in differential interferograms. However, the accuracy of model prediction should be evaluated. The direct use of the WRF model predictions for atmospheric correction yield errors for mitigating the turbulent APS and the small-scaled APS. Final approach is a time-series analysis. In model experiments, several properties of atmospheric phase screen (APS) are found out. The first is that APS could remain in a time-series analysis and mainly comes from the stratified APS. The second is that it is possible to estimate and minimize the stratified APS by using sufficient WRF models. In the case of the turbulent APS, time-weighting low pass filtering is capable to reduce it. Therefore, the main idea of the atmosphere corrected time-series analysis adopt the stratified APS and turbulent APS correction method using WRF model and time-weighting methods. In comparison with observational dataset such as GPS and MODIS dataset, the estimated ground deformation and APS from the atmosphere corrected method have low rms errors, and high correlation. Therefore, this method can be believed as an accurate approach for measuring the ground deformation in volcanic region.