5:15 PM - 6:45 PM
[STT35-P06] Ground Subsidence Monitoring System for Kobe Port Island using Time-Series Interferometric Synthetic Aperture Radar
Keywords:Time-series Interferometric Synthetic Aperture Radar , Kobe Port Island, Sigma-SAR, GNSS
The phenomenon of ground subsidence refers to the sinking of the ground due to the extraction of excess groundwater trapped in the interstices between gravel, sand layers, etc., resulting in the squeezing out of interstitial water from clay layers, causing the clay layer to shrink. Once the ground has subsided, it does not return to its original state, and the amount of subsidence accumulates year by year. Ground subsidence poses risks such as structural damage to buildings, damage to lifelines, and increased vulnerability to tsunamis and high tides. However, ground subsidence continues to be observed in many regions through leveling surveys. This study conducted time-series interferometric synthetic aperture radar (SAR) analysis focusing on Port Island in Kobe City, Hyogo Prefecture, and the Osaka Plain, where ground subsidence has been confirmed by leveling surveys. Additionally, comparisons and accuracy evaluations were conducted with survey data obtained from GNSS points and from Kobe City Hall for Port Island.
The research procedure involved data acquisition, time-series interferometric processing of zero-baseline data, and accuracy evaluation in sequence. The study focused on Port Island in Kobe City, Hyogo Prefecture, and the Osaka Plain, where ground subsidence occurs, and analyzed data from 2014 to 2023 using Descending 35 scenes. Phase addition processing was performed on the Descending 34 pair, but it was not possible to perform the addition process properly due to the influence of the ionosphere. Therefore, ionospheric correction was applied to remove the ionosphere, resulting in improved accuracy of the interferometric images. Trends of subsidence were confirmed in both Port Island and the Osaka Plain. The analysis was performed by using Sigma-SAR.
Finally, accuracy evaluation was conducted by comparing 13 GNSS points and 4 survey points on Port Island using average residuals. The average residual of the 13 GNSS points was 2.8 cm, indicating high accuracy, but slightly larger errors were observed at Sakai, Konan, and Yao points, with average residuals of 4.9 cm, 4.7 cm, and 3.8 cm, respectively. The average residual of the 4 survey points on Port Island was 6.2 cm, indicating slightly lower accuracy. However, among the 4 points, the northern 3 points showed high accuracy with average residuals of 1.3 cm, 1.5 cm, and 2.1 cm, respectively.
From these results, errors were observed at the Sakai, Konan, and Yao points among the GNSS points, and significant deviation was observed at the southernmost point of the survey points on Port Island. However, excluding the 3 points with large average residuals among the 13 GNSS points, the average residual of the remaining 10 points was 2.1 cm, and excluding the southernmost point that showed significant deviation among the 4 survey points on Port Island, the average residual of the remaining 3 points was 1.7 cm, indicating a partial but high accuracy of measurement with SAR. Therefore, it is considered effective to measure ground movements nationwide efficiently by combining SAR with leveling surveys.
The research procedure involved data acquisition, time-series interferometric processing of zero-baseline data, and accuracy evaluation in sequence. The study focused on Port Island in Kobe City, Hyogo Prefecture, and the Osaka Plain, where ground subsidence occurs, and analyzed data from 2014 to 2023 using Descending 35 scenes. Phase addition processing was performed on the Descending 34 pair, but it was not possible to perform the addition process properly due to the influence of the ionosphere. Therefore, ionospheric correction was applied to remove the ionosphere, resulting in improved accuracy of the interferometric images. Trends of subsidence were confirmed in both Port Island and the Osaka Plain. The analysis was performed by using Sigma-SAR.
Finally, accuracy evaluation was conducted by comparing 13 GNSS points and 4 survey points on Port Island using average residuals. The average residual of the 13 GNSS points was 2.8 cm, indicating high accuracy, but slightly larger errors were observed at Sakai, Konan, and Yao points, with average residuals of 4.9 cm, 4.7 cm, and 3.8 cm, respectively. The average residual of the 4 survey points on Port Island was 6.2 cm, indicating slightly lower accuracy. However, among the 4 points, the northern 3 points showed high accuracy with average residuals of 1.3 cm, 1.5 cm, and 2.1 cm, respectively.
From these results, errors were observed at the Sakai, Konan, and Yao points among the GNSS points, and significant deviation was observed at the southernmost point of the survey points on Port Island. However, excluding the 3 points with large average residuals among the 13 GNSS points, the average residual of the remaining 10 points was 2.1 cm, and excluding the southernmost point that showed significant deviation among the 4 survey points on Port Island, the average residual of the remaining 3 points was 1.7 cm, indicating a partial but high accuracy of measurement with SAR. Therefore, it is considered effective to measure ground movements nationwide efficiently by combining SAR with leveling surveys.