Polarimetric SAR remote sensing of the 2011 Tohoku earthquake using ALOS/PALSAR

The 2011 magnitude 9.0 Tohoku earthquake caused catastrophic damage to coastal areas in northeastern Japan. It is quite difficult to obtain an immediate assessment of large-scale natural disasters by ground survey. Microwave remote sensing by airborne or spaceborne sensors is a suitable tool for monitoring near-real-time damage over large areas. However, it is not always possible to acquire datasets with sufficient spatial and temporal resolutions for the precise detection of large areas of damage; therefore, this study focuses on the application of fully polarimetric scattering observations for deduction of information about damaged areas. Earthquake-induced changes in the radar backscattering mechanism are investigated by polarimetric scattering mechanism indicators. In addition, a change detection technique using the Expectation–Maximization (EM) based thresholding approach is applied to various polarimetric features. Furthermore, a new polarimetric information fusion method based on the contextual Markov Random Field (MRF) is proposed. The mapping accuracy for tsunami-swept urban areas can be improved by about 50% with significant reduction of false- and missed-alarm rates compared with the single-polarization SAR approach.

► We examine usability of L-band ALOS/PALSAR to detect earthquake/tsunami damages.
► Earthquake-induced changes of scattering mechanism are evaluated by SAR polarimetry.
► Automatic detection of damages from different polarimetric parameters is presented.
► Significantly damaged urban areas can be detected with an accuracy of about 90%.
► A new contextual information fusion method is proposed to improve mapping accuracy.