Remote Detection of Clandestine Graves


The positive location of clandestine mass graves is the first step to the organization of a recovery team and the subsequent exhumation and identification of the remains for the purpose of evidence collection for prosecution and for return of individuals to families. The development of a spatial/spectral model for discriminating graves from airborne and satellite hyperspectral imagery is a novel technique development and application that would greatly increase chances for forward detection of clandestine mass graves, monitoring production of mass graves during conflict, and even retrospective assessment. The fundamental question we propose to explore is probabilistic nature of detecting mass graves that are clandestine in nature by means of remote sensing (i.e. spectromentry and image analysis.We propose to investigate the spectral reflectance of known volatile compounds (e.g. toluene, dimethyl disulfide, cadaverine, putrescine, etc.) in soil matrices collected by means of a handheld spectrometer to define unique absorbance features. We will use pattern recognition techniques (using MatLab) to locate the most significant areas of the reflectance spectrum. We will apply those findings to the spectral reflectance of a known test mass grave site that have been collected over a longitudinal study to discriminate which unique features of the known volatile compounds (identified above) are visible in the spectral reflectance of mass graves collected in the field. Finally, we will analyze hyper and multi spectral imagery using a combination of pattern recognition techniques, subpixel analysis and classification techniques to scale up our findings to the image level.