Seite - 811 - in Book of Full Papers - Symposium Hydro Engineering

In addition to SAR data, 2 m Digital Elevation Models (DEM) of study areas by Laser Profiler (LP) were used as initial dam body shapes to improve external deformation accuracy. 2.2. EXTERNAL DEFORMATION MONITORING USING SATELLITE SAR DATA Fig. 3 shows external deformations of nineteen rockfill dams by satellite SAR in the line-of-sight direction from the first SAR observation to the last SAR observation day shown in Table 1. The red arrows in Fig. 3 show the observation directions of the ALOS-2. From the results of preliminary examinations, the multi- look [10], [11] for noise reduction of SAR data was set at 2 x 2 pixels (a spatial resolution of about 5m) in this paper, and the distributions of external deformations in Fig. 3 are expressed with a spatial resolution of about 5 m. Negative values in Fig. 3 mean that the dam surface deforms away from the satellite and it is almost the same direction of settlement of the dam. In the case of Dam N in Fig. 3, which showed a relatively large amount of external deformation, it was clarified with electro-optical survey after the Kumamoto earthquake in April 2016 that there had been settlement of several centimeters due to the earthquake, and earthquake-induced settlement was well measured by satellite SAR. In order to evaluate the accuracy of satellite SAR based external deformation monitoring, Fig. 4 shows root mean square error (RMSE) distributions at existing GPS or electro-optical survey measurement points. RMSE is calculated using Eq. [1]. 𝑅 𝑀 𝑆 𝐸 =√ 1 𝑛 ∑ (𝑢 𝑆 𝐴 𝑅 −𝑢 0)2 𝑛 𝑖 =1 [1] Where n is the number of satellite SAR data, uSAR is the external deformation measured by satellite SAR, and u0 is the external deformation measured by GPS or electro-optical survey, respectively. If GPS data has been obtained, RMSE in Eq. [1] is calculated from the differences of external deformations between SAR and GPS measurement values at the GPS measurement points. If no GPS data has been obtained, external deformations by electro-optical survey at the SAR observation days were estimated by linear interpolation of two electro-optical survey data and RMSE was calculated from the differences of external deformations between SAR and electro-optical survey at the electro-optical measurement points. Most RMSE values show less than 5 mm in Fig. 4 and it means that external deformations by satellite SAR have good accuracy by comparing to GPS or electro-optical survey. Fig. 5 shows the comparisons of temporal changes of external deformations by SAR and GPS/electro-optical survey at the existing geodetic measurement points on the crests in the maximum cross sections of nineteen studied dams. Although there were small variations among the results between 811