2:45 PM - 3:00 PM
[STT39-11] Monitoring of volcanic activity on Kuchinoerabu-jima Island using the PALSAR-2 time series InSAR analysis
Keywords:Remote sensing, Volcano
Kuchinoerabu-jima Island is a volcanic island located in the south of Kagoshima Prefecture. Recent years there is a high risk of re-eruption due to repeated eruptive activities. Long-term (5 years) volcanic activity and related ground changes were observed using time-series interference processing using JAXA's PALSAR2 data. We also investigated a relationship between Normalized difference Vegetation Index(NDVI) from the senstinel-2 and ALOS-2/PALSAR-2 amplitude images with regard to the volcanic ash coverage.
We analyzed 24 scenes of PALSAR-2 southbound data and 12 scenes of northbound data observed from March 2016 to March 2021. In the subsequent interference SAR processing, all the data was reconstructed as if they were obtained in the reference orbit so that simple InSAR phase summation could reveal the total phase variation among the whole observation period (Zero baseline processing). The total variation was obtained in two ways: first is to obtain all the fluctuations of the two adjacent period by the InSAR processing and second is to sum all the fluctuations included from March 2016 to March 2021 (Method 1). Kuchinoerabu Island has an altitude of 650m and is affected by the weather because it shapes conical-horn type mountain. Method 1 could be affected by the weather at the beginning and end of the observation. In order to correct this, we proposed Method-2 that atmospherically corrects the interference phase obtained in each period and then adds all. Atmospheric correction performed by Method 2 calculates an approximate line from the correlation between phase and altitude in each period by the least square method and creates a regression line of the influence of the atmosphere in each period.
When the interference SAR time series analysis (2.5 analysis) was performed by Method 1, the total sedimentation for 5 years was 3.81 cm. That for Method 2 was 3.92 cm. For accuracy verification, the fluctuation amount of the GNSS point was converted into the fluctuation amount in the vertical direction calculated from SAR. As a result, it was found that the sedimentation was 3.68 cm in the vertical direction in 5 years. When GNSS was used as the correct answer data, the error rate was 3.41% for Method 1 and 6.12% for Method 2, both of which were calculated with high accuracy.
As the forests are converted to non-forests by ash fall, and this change can be extracted using the NDVI. In addition, the reflection characteristics of radio waves in the SAR image change due to ash fall. Therefore, there is a possibility that the ash fall area can be estimated from the two information by accurately aligning and comparing the images obtained from the two different satellites and performing statistical processing. Therefore, NDVI was calculated from Sentinel-2 data and the correspondence with PALSAR2 images was measured. The radar backscatter of the SAR image greatly varies depending on the slope of the terrain. To reduce this variation, radiometric terrain correction was applied. The effects of ash fall appeared on both the NDVI image and the amplitude image. It was confirmed that the values of NDVI and backscattering coefficient decreased after ash fall in both NDVI and amplitude compared to the image before ash fall.
We analyzed 24 scenes of PALSAR-2 southbound data and 12 scenes of northbound data observed from March 2016 to March 2021. In the subsequent interference SAR processing, all the data was reconstructed as if they were obtained in the reference orbit so that simple InSAR phase summation could reveal the total phase variation among the whole observation period (Zero baseline processing). The total variation was obtained in two ways: first is to obtain all the fluctuations of the two adjacent period by the InSAR processing and second is to sum all the fluctuations included from March 2016 to March 2021 (Method 1). Kuchinoerabu Island has an altitude of 650m and is affected by the weather because it shapes conical-horn type mountain. Method 1 could be affected by the weather at the beginning and end of the observation. In order to correct this, we proposed Method-2 that atmospherically corrects the interference phase obtained in each period and then adds all. Atmospheric correction performed by Method 2 calculates an approximate line from the correlation between phase and altitude in each period by the least square method and creates a regression line of the influence of the atmosphere in each period.
When the interference SAR time series analysis (2.5 analysis) was performed by Method 1, the total sedimentation for 5 years was 3.81 cm. That for Method 2 was 3.92 cm. For accuracy verification, the fluctuation amount of the GNSS point was converted into the fluctuation amount in the vertical direction calculated from SAR. As a result, it was found that the sedimentation was 3.68 cm in the vertical direction in 5 years. When GNSS was used as the correct answer data, the error rate was 3.41% for Method 1 and 6.12% for Method 2, both of which were calculated with high accuracy.
As the forests are converted to non-forests by ash fall, and this change can be extracted using the NDVI. In addition, the reflection characteristics of radio waves in the SAR image change due to ash fall. Therefore, there is a possibility that the ash fall area can be estimated from the two information by accurately aligning and comparing the images obtained from the two different satellites and performing statistical processing. Therefore, NDVI was calculated from Sentinel-2 data and the correspondence with PALSAR2 images was measured. The radar backscatter of the SAR image greatly varies depending on the slope of the terrain. To reduce this variation, radiometric terrain correction was applied. The effects of ash fall appeared on both the NDVI image and the amplitude image. It was confirmed that the values of NDVI and backscattering coefficient decreased after ash fall in both NDVI and amplitude compared to the image before ash fall.