Investigation of the Effectiveness of Earthquake Damage Evaluation Methods Using Space Images (Case Study: Sarpol-Zahab Earthquake)

The earthquake is the most dangerous natural hazard, which can target thousands of people and cause massive destruction of infrastructure or buildings.After a earthquake,recognizing the location and extent of damage buildings is necessary to take emergency measures and temporary accommodation operations.Remote sensing is an excellent tool due to reasonable costs and rapid data capture.The recent earthquake in Sarpol-Zahab is one of the most recent destructive earthquakes that destroyed infrastructures.The purpose of this study was to use multi-time detection techniques to investigate the changes caused by this earthquake.In this research,different image classification methods are used to compare their results with the ability to identify damaged buildings on high-resolution satellite data (Pleiades1) and with mid-resolution (Landsat-8 and Sentinel-2). The first goal of this study is to obtain the best map of the proper destruction for the structure of Iran by comparing the output change maps with the reference data and localization of the change detection algorithms. In the final step, we examined the various applications of the map of changes. The results show that the SVM has been more accurate in comparison of image classification and also showing changes in the neural network method. Overall accuracy in using the SVM classification for the change maps obtained in the Sentinel is 0.85 and for the Landsat is 0.64. However, the overall accuracy using the neural network for change maps The obtained results are 0.80 for Sentinel and 0.61 for Landsat, which indicates the effectiveness of Sentinel images along with SVM techniques in Iran.