Result of DInSAR Analysis of Zentoku Landslide, Japan, and Development of Landslide Recognition Tool Ryoichi Furuta (RESTEC), Atsushi Yashima (Gifu Univ.), Kazuhide Sawada (Gifu Univ.), Hiroshi Fukuoka (Kyoto Univ.)
Outline DInSAR Result of Zentoku Landslide Current status of Zentoku landslide estimated from DInSAR Result of 2009 Reactivated Landslide in Kochi Pref., Japan. Development of Landslide Recognition Tool for AVNIR-2 image Modification of Landslide Recognition Tool Conclusions
DInSAR Result of Zentoku Landslide (Overview) Zentoku Landslide Tokushima Pref., Japan Steep Slope slope inclination is approx. 30 degrees JERS-1, ALOS DInSAR 1992~1998 by JERS-1 SAR Negative result 2006~2008 by ALOS PALSAR Negative result Osaka Tokyo Zentoku Landslide Map around Zentoku Landslide with landslide map
Case study of JERS-1 SAR Differential Interferometry for creep-type landslides in Japan JERS-1 DInSAR 1995/04/28 1995/10/21 JERS-1 DInSAR 1995/04/28 1997/11/07 Takisaka Landslide Akasaki Landslide -5.9cm JERS-1 DInSAR 1992/09/01 1994/09/19 5.9cm Toyomaki Landslide JERS-1 DInSAR 1993/04/19 1993/07/16 Zentoku Landslide JERS-1 DInSAR 1995/09/05 1995/10/19 Itaya Landslide
Previous result of DInSAR of Zentoku Landslide (2006 2008) 2006 2007 2008 0828 1013 0716 0831 1016 0718 (a) (b) (c) (d) (e) (f) (a) (b) (c) (d) (e) (f) Az Rg -5.9cm +5.9cm METI, JAXA analyzed by RESTEC
Current status of Zentoku Landslide estimated by DInSAR result DInSAR analysis was applied using ALOS PALSAR data acquired in 2009 List of utilized data 2009/01/18 case1 2009/03/05 case2 2009/07/21 case3 2009/09/05 Software: SIGMA-SAR Referenced DEM: GSI-DEM 50m
Current status of Zentoku Landslide estimated by DInSAR result 2009/01/18-2009/03/05 2009/03/05-2009/07/21 2009/07/21-2009/09/05 Az. Rg. -5.9cm +5.9cm METI, JAXA processed by SIGMA-SAR Definitive displacement can t confirm. The result of 2009 is similar to the result of previous year.
Comparison of result of ascending path and descending path Slant range image of result of DInSAR Descending Ascending Rg. Az. Az. Rg. -5.9cm +5.9cm METI, JAXA processed by SIGMA-SAR Definitive displacement can t confirm as analysis of data from ascending orbit
Current status of Zentoku Landslide estimated by DInSAR result Definitive displacement can t confirm so far. DInSAR images of 2009 were looks noisy like previous result. Result of DInSAR of ascending path and descending path was similar. Zentoku Landslide might be stable state.
Reactivated Landslide in Kochi Pref., Japan Reactivated Landslide was detected from DInSAR result Location: Kochi Pref., Japan Very near from Zentoku Landslide Size: approx. L1,000m x W800m Detected from a pair of PALSAR data was acquired in Jan,. 2007 and Jan., 2008. Data was acquired by descending mode.
Reactivated Landslide in Kochi Pref. Zentoku Landslide Reactivated Landslide. -5.9cm +5.9cm PALSAR DInSAR result of Shikoku area Master: 2007/01/14, FBS, 34.3 Slave: 2008/01/17, FBS, 34.3 Satellite direction: Descending METI, JAXA Processed by SIGMA-SAR
Reactivated Landslide in Kochi Pref. 3D-view on GoogleEarth Subside or deformation to west direction was detected. Two reactivated landslides, upper block and lower block, were confirmed. Shape of deformation area is corresponding to landslide map. subside uplift
Development of Landslide Recognition Tool Landslide Recognition Tool Collapse pattern of collapse-type landslide have seen in similar pattern. Applying template matching (face recognition) technique to recognize collapse-type landslide area. Modification of Landslide Recognition Tool Water and cloud region mask function by blue and NIR channel Blue>threshold as cloud NIR<threshold as water Urban area mask function for process time acceleration using NDVI image Calculate averaged NDVI Classify urban area and forest area Utilization of optimized template for process time acceleration To detect landslide collapse point in multi direction To reduce number of template (from 8 to 1)
Image of Processing Band1 Band2 Make mask image (water & cloud area) Band3 Band4 Resize to smal size image Input AVNIR-2 images Original algorithm Modified algorithm Make mask image (urban & forest area) RGB RGB Template Matching (Landslide Recognition) Template Matching (Landslide Recognition)
Result with unmasked processing JAXA JAXA
Effect of water area mask NIR < 50 as water, Blue > 170 as cloud JAXA JAXA
Effect of urban area mask JAXA JAXA
Conclusions Result of DInSAR of Zentoku Landslide Definitive displacement can t confirm in result of 2009. Comparison result of ascending and descending was similar. Reactivated landslide in Kochi Pref., Japan was detected. Modification of Landslide Recognition Tool Applied mask functions (cloud, water, and urban) can be reduce a time for processing was confirmed. Modified algorithm and optimized template recognized landslide collapse points appropriately. Some errors still remained. Making more optimized template.
Dx Relationship of differential phase and slope angle differential phase Offnadir angle of acsending dφ=(4π/λ)*(dh/cos(θ i )) dφ: differential phase λ: wave length dh: surface deformation θ i : local incidence angle Offnadir angle of decsending Phase (cycle) 2 1.5 1 0.5 0-0.5-1 -1.5-2 0 10 20 30 40 50 A34.3 D34.3 slope angle (deg.) Relationship of differential phase and slope angle Dy Dy Dx Analysis condition: λ=23.8cm, 5cm subsidence Data acquired from ascending orbit is effective to detect surface deformation of east slope than descending orbit.