Science Service System
You are here : Home : Proposals_Summary

Summary of Proposal LAN2752

TitleGeneration of urban area and density maps of global mega-cities using X-band polarimetric data
Investigator Susaki, Junichi - Kyoto University, Graduate School of Engineering
Team Member
Associate Professor Susaki, Junichi - Kyoto University, Graduate School of Engineering
Summary[Objectives] We will develop a method to produce urban areas and densities by using TerraSAR-X images. [Method] (1) Urban area extraction We have already developed a method to estimate urban areas and urban density by using ALOS/PALSAR images. In the first stage, the method uses total power and volume scattering derived by four component decomposition method. As a result, urban areas and parts of mountains are extracted. In the second stage, it uses the randomness of polarimetric orientation angle (POA) calculated from fully polarimetric data. Urban areas show relatively homogeneous POAs whereas mountains show heterogeneous POAs. By calculating the variance of POAs, urban areas are clearly discriminated from mountains. It was demonstrated that the method was effective to extract urban areas by using L-band ALOS/PALSAR images. We applied the above method to images obtained by using X-band fully polarimetric Polarimetric and Interferometric Airborne Synthetic Aperture Radar System (Pi-SAR2), developed and operated by National Institute of Information and Communications Technology (NICT), Japan. Interestingly, it was confirmed that the first stage of the above-mentioned method was effective whereas the second stage generated invalid results. Then, we proposed the different approach that uses the polarimetric coherency between the scattering of HH and that of VV. The results of extracting urban areas in Kobe and Kyoto, Japan were acceptable with an overall accuracy of approximately 80 to 90% at 100-m spatial scale. Now, we want to apply our current method to X-band TerraSAR-X images. In addition, our method is assumed to use fully polarimetric images. However, fully polarimetric images are very limited. Therefore, we’ll develop another method based on dual polarimetric images (HH + VV). (2) Urban density estimation We have proposed a method to estimate urban density by using a fully polarimetric PALSAR image. The method corrects the POA effect against total power and four components, and generates the normalized statistical indices (Susaki et al., RSE, 2014). The results was validated with GIS data in five Japanese cities and two other cities (Munich and New York), and it was demonstrated that the performance was acceptable. We’ll apply this method to X-band TerraSAR-X images. In very high resolution (VHR) SAR images, foreshortening problem cannot be ignored. We’ll develop a method to estimate urban density after eliminating the foreshortening effects, and compare the statuses of urban development in global mega-cities. [Data requirements] Fully (if possible) and dual polarimetric data [Funding] Minister of Education, Culture, Sports, Science and Technology of Japan

Back to list of proposals

© DLR 2004-2016