Co-seismic landslides automatic detection on regional scale with sub-pixel analysis of multi temporal high resolution optical images: Application to southwest of Port au Prince,Haiti

Customary methods for landslide detection are labor demanding,time consuming and costly. moreover,it becomes a real challenge to map event triggered landslides using customary techniques. various automatic and semi-automatic techniques have been proposed for landslide detection. these techniques require extensive technical knowledge and familiarity of the selected study area for visual interpretation,suitable sampling selection,manual parameterization and the trial and error approach for selection of thresholds which makes the process irreproducible. in this context,the purpose of this study is to introduce and investigate the applicability of an automatic technique of cosi-corr (co-registration of optically sensed images and correlation) for landslide identification on regional scale using high resolution orthorectified worldwies 2 images. unlike,conventional pixel based semi-automatic methods which depends on spectral properties of pixels,this method is based on pixels shift. the correlation process was performed by using the cosi-corr software. the ground displacement field measured from the correlation indicates high displacement values for landslides triggered by 2010 haiti earthquake. the method automatically identified more than 73% of co-seismic landslides at the threshold value of 3.75. the adopted procedure proposes a quick and economic way for automatic detection of co-seismic landslides on regional scale if suitable pre- and post-event satellite data is available. one limitation of this method is that it does not show any unique behavioral characteristics for different types of landslides so that they could be easily distinguished or classified. © 2018,national centre of excellence in geology. all rights reserved.