The U.S.-Mexico border crisis is a mass disaster that has claimed the lives of thousands and led to an overwhelming increase of unidentified remains along the border. In Texas, these issues coupled with a decentralized Justice of the Peace system has led to the mismanagement of the dead and the burial of hundreds or more individuals in poorly marked or unmarked graves in cemeteries throughout the South Texas region. Finding these graves is the first step toward identification and repatriation and doing so noninvasively is crucial to minimizing community impact. Many graves have been located through memory recall from community members, local officials, and groundskeepers, however, the success of the geophysical techniques currently used to identify subsurface anomalies associated with these burials has been inconsistent. This study evaluated the effectiveness of predictive modeling, ground penetrating radar (GPR), and drone-based infrared (IR) imaging for the detection of unmarked graves in South Texas cemeteries.
Six cemeteries located along the southern Texas border were selected for this research. Each cemetery has a history of burying the unidentified remains of presumed migrants and at minimum, initial exhumation efforts have occurred at each location allowing for ground truthing of the geophysical data. Using ArcGIS Pro software and a DJI Mavic 2 Enterprise Advanced (M2EA) drone system equipped with a dual visual and thermal infrared camera, high resolution maps were generated including the locations of all known grave markers and coordinate data from the exhumation efforts conducted at each cemetery. From these maps, maximum entropy modeling (Maxent), a presence-only modeling approach, was applied to the cemetery with the highest number of unidentified burials to assess whether this model could predict other areas in the cemetery likely to contain unmarked graves. GPR data were collected in areas of interest identified through previous pedestrian surveys and interviews to assess the effectiveness of this method for detecting subsurface anomalies consistent with graves in moderate to poor soil conditions. Finally, IR imaging was collected and evaluated in the same areas of interest to determine if graves could be detected through the differential emission of thermal radiation in the disturbed soil associated with the burials compared to the surrounding undisturbed, compact soil.
Maxent demonstrated moderate model performance, accurately predicting many of the geographic locations of the unidentified burials, however these results are tempered by significant gaps in its prediction area, biased sampling, and correlated spatial variables, reducing its generalizability beyond the one cemetery selected. The results of the GPR outputs revealed an inconsistent pattern with some graves producing strong signals, others producing voids, and many more remaining undetected entirely. The results of the IR imaging suggest that the drone system and camera sensor selected do well to identify topsoil variation and vegetation differences, however, fail to identify deep thermal differences associated with the burials.
Overall, this research demonstrates that grave detection in the highly active cemeteries in South Texas where soil conditions are poor, is very complex. Maxent, in its current form is inadequate for the prediction of possible burial locations of the unidentified and is not generalizable to other cemeteries. While GPR appeared to identify the most graves of all the methods selected for this study, results were inconsistent and unpredictable. Finally, IR imaging, as acquired with the M2EA system is ineffective for identifying graves despite its ability to locate faint topographical variation in the imagery. The results of this research, while mostly inconsistent and predominantly negative, are important to the field of forensic anthropology and to the grave detection literature for the area. Alternative methods should be considered, and multiple geophysical methods should be applied wherever possible to account for a lack of ideal survey conditions, to provide coverage where others fail, and to corroborate findings as valid.
(Publisher abstract provided.)
Downloads
Similar Publications
- Longitudinal Examination of the Bullying-Sexual Violence Pathway Across Early to Late Adolescence: Implicating Homophobic Name-Calling
- Improving Aptamer Affinity and Determining Sequence–Activity Relationships via Motif-SELEX
- Witnessing community violence and its consequences: Changes across middle school