11:30 AM - 11:45 AM
[STT36-10] Comparison of atmospheric delay correction methods in InSAR: Case study in Japan
Keywords:InSAR, Atmospheric delay, Tropospheric delay
Interferometric synthetic aperture radar (InSAR) is widely used to measure surface displacements using the phase differences of two acquisition dates. Various Earth phenomena have been studied using InSAR such as volcanic activity, landslides, earthquakes and variations in groundwater levels. The delay of the microwave in the atmosphere is a major source of error in InSAR measurements. In the tropospheric layer, temperature, pressure, and relative humidity vary spatially and temporally, creating tropospheric delay masking out the surface displacement signals. Previous studies suggested methods to reduce the atmospheric noise by calculating the delay using atmospheric data and models. In this study, we compared the results of atmospheric delay calculation obtained by the RINC software (Ozawa et al., 2016) and GACOS (Yu, et al., 2017). The former directly integrates the delay along the propagation path using the atmospheric refractivity model of Thayer (1974) and the Meso-Scale Model (MSM) of Japan Meteorological Agency. The latter uses the atmospheric refractivity model of Berrada-Baby et al. (1988) combined with a decomposition model of the stratified and turbulent components, applied on the European Centre for Medium-Range Weather Forecasts (ECMWF) model and optionally the delay obtained from GNSS.
We calculated the delay on the timing of nine ALOS-2 acquisitions in the Tohoku region and fifteen ALOS-2 acquisitions in the Osaka region in Japan. We found that the amount of tropospheric delay tended to be larger for GACOS compared to RINC (Fig.1). To investigate the reason, we calculated the delay by applying the two refractivity models of Berrada-Baby et al. (1988) and Thayer (1974) on identical MSM data sets, and confirmed that the former systematically gives larger delay compared to the latter. For the topography-correlated stratified delay, some differences between the RINC and GACOS results were observed in the Tohoku region, whereas the topography-correlated delay was consistent in the Osaka region. As for the turbulent (topography-uncorrelated) component, the difference between the RINC and GACOS results was larger than 20 cm for half of the cases (12 out of 24), suggesting the difficulty in accurately estimating the turbulent contributions.
We applied the delay correction calculated by the both methods on the ALOS-2 interferograms processed for the two regions. The mean standard deviation of the LOS displacements did not reduce after applying the correction of RINC and GACOS, suggesting that the ionospheric delay contribution is dominated and thus indicating the importance of the ionospheric correction.
We calculated the delay on the timing of nine ALOS-2 acquisitions in the Tohoku region and fifteen ALOS-2 acquisitions in the Osaka region in Japan. We found that the amount of tropospheric delay tended to be larger for GACOS compared to RINC (Fig.1). To investigate the reason, we calculated the delay by applying the two refractivity models of Berrada-Baby et al. (1988) and Thayer (1974) on identical MSM data sets, and confirmed that the former systematically gives larger delay compared to the latter. For the topography-correlated stratified delay, some differences between the RINC and GACOS results were observed in the Tohoku region, whereas the topography-correlated delay was consistent in the Osaka region. As for the turbulent (topography-uncorrelated) component, the difference between the RINC and GACOS results was larger than 20 cm for half of the cases (12 out of 24), suggesting the difficulty in accurately estimating the turbulent contributions.
We applied the delay correction calculated by the both methods on the ALOS-2 interferograms processed for the two regions. The mean standard deviation of the LOS displacements did not reduce after applying the correction of RINC and GACOS, suggesting that the ionospheric delay contribution is dominated and thus indicating the importance of the ionospheric correction.