A multi-temporal approach to using multispectral remote sensing for the prospection of clandestine mass graves in temperate environments.

Abstract

Recent years have seen mass graves resulting from human rights abuses and violations become a stimulant for criminal proceedings and investigations. As a consequence the prospection for clandestine mass graves is at the forefront of international forensics and is considered of ever increasing importance. This study investigates the utility of archive orbital multispectral imagery for the detection of clandestine mass graves specifically in temperate environments, using spatial variations in vegetation stress as indicators of disturbance, indicative of the presence of a mass grave. Three studies were undertaken. The first monitored vegetation stress at the surface of two small-scale controlled mass graves, one containing mammalian remains (pigs) and the other empty, using a hand held field spectrometer for the first 121 days post burial. It was found that vegetation stress was not influenced by the presence of mammalian remains over this short time frame, however, the disturbance caused by creating the grave was detectable. Meteorological factors including temperature and dew point were found to affect the NDVI recorded more than other meteorological variables such as precipitation. Geophysical surveys were also undertaken, using temporal electrical imaging and electromagnetic profiling, to provide a spatial measure of the state of decomposition within the grave and to indicate the detectability of both a full and an empty grave throughout the study. By undertaking regular geophysical surveys it was found that both graves were able to be detected using both techniques up to 121 days post burial. The percentage difference in geophysical response from both techniques, between the pig and empty grave compared to the undisturbed ground peaked at 70 days post burial. This peak corresponded to the period where climatic events were recorded, including increased precipitation and decreasing ambient temperatures. Therefore, it is clear that the results obtained from the geophysical surveys were influenced heavily by climatic factors and in the case of this study, enabled the graves to be more readily detected. The second study was a long term decadal investigation of variations in NDVI of a large- scale, mass grave site resulting from the 2001 Foot and Mouth epidemic. This provided a pre-operational proof-of-concept for clandestine human mass grave detection using multi-temporal (18 day repeat period or multiple thereof), multi-spectral (visible to infra-red –0.45-12.5µm), medium and fine spatial resolution (30m and 5m ground sample distance) orbital remote sensing of a known burial location. A dense time-series of archive satellite images was used to study variations in NDVI of the vegetation directly above the grave, compared to the undisturbed vegetation surrounding it. By calculating NDVI using fine spatial resolution imagery from RapidEye (5m), the time frame within which significance was detected between the grave surfaces and the undisturbed ground, was able to be extended to 9 years, 11 months post burial. The third study concerned the active prospection of clandestine human mass graves in Bosnia where differences in vegetation stress, on a small spatial (~15m) but large temporal scales (exceeding a decade post burial), were examined for a number of known mass burial locations. In turn, graves were prospected for on the basis of appropriate spatial and temporal variations in vegetative stress and health. It was found that medium spatial resolution imagery (30m) was not useful in locating human mass graves on Cancari Road, however fine spatial resolution imagery (5m) was successful in detecting the disturbance caused by the exhumation of Cancari Road Grave 1 (CR01) in 2009. To conclude, variations in vegetation stress, derived through vegetation indices, can be used to locate areas of disturbance that may be indicative of a mass grave and other man made changes over time. However, it is important to consider the spatial resolution of the orbital platform and its appropriateness in relation to the extent of the target.