2016 年 62 巻 2 号 p. 79-88
I investigate how well a numerical weather model can correct atmospheric-related noise included in an InSAR image through application to the 2015 Sakurajima magma intrusion event. Two clear phase changes are observed in the Minami-dake summit area and the Nabeyama area in the In-SAR image acquired just after the event. However, GNSS data, other InSAR results, and a constructed source model do not support the idea that the phase change in the Minami-dake area is true crustal deformation. Applying noise reduction using a numerical weather model, the phase change in the summit area has been reduced to approximately 1 cm. With the data applying the noise reduction, a nearly vertical dike opening is inferred as the optimal source model, which is consistent with the source geometry deduced from GNSS data or seismicity. On the other hand, with uncorrected data, a quite different source model that is a crack opening with a low dip angle is determined. It strongly suggests that an appropriate atmospheric noise reduction is critically vital for InSAR-based volcano observations. I further demonstrate that a numerical weather model can reconstruct well the noise correlating with elevation, while it can hardly function to reduce turbulence-induced noise.