The use of soil in forensic investigations is increasing in police forces around the world, including the Federal Bureau of Investigation, the Royal Canadian Mounted Police and the Australian Federal Police. In fact, in Australia, successful soil forensic investigations have contributed evidence in cases ranging from attempted murder, cold case murder and sexual assault—and all evidence has been accepted by the nation’s supreme courts.
In a new study, researchers from the National Centre for Forensic Studies in collaboration with Geoscience Australia and the Australia Federal Police, have taken another step toward developing a method to match soil remnants found on personal items to regional soil samples to either implicate or eliminate a person’s presence at a crime scene.
In the study, published in the Journal of Forensic Sciences, the team surveyed a 260 km2 area, collecting 286 soil samples, including three “blind” samples within the area but away from the established grid.
The team then used Fourier transform infrared (FTIR) spectroscopy, X-ray fluorescence (XRF) and inductively coupled plasma-mass spectrometry (ICP-MS) to analyze the samples and identify which grids they originated from. The empirical soil provenancing method employed relies on pre-existing geochemical surveys and soil databases, as well as statistical analysis that incorporates principal components (PCs), which are calculated after imputation of censored data.
“Most developed countries have existing soil databases used for such things as mineral exploration or land use decision support,” explained Patrice de Caritat, study co-author and principal research scientist at Geoscience Australia. “We're plugging our methods into these databases to see if we can locate samples from the database information, rather than needing to collect samples specifically for each investigation.”
According to the study results, each blind sample behaved differently in terms of provenancing performance. The best results for blind sample 1 were obtained with the XRF with PCs method, with a precision of 94% and accuracy of 67%. The next highest precision came from XRF without PCs at 87%.
For blind sample 2, FTIR plus magnetic susceptibility without PCs yielded a 100% precision rate, while the most accurate results were obtained using ICP-MS with PCs (87%). The next highest precision rate came from ICP-MS with PCs (94%).
Lastly, the best results for blind sample 3 were obtained using FTIR plus magnetic susceptibility with PCs—92% precision and 60% accuracy.
Across all three blind samples, FTIR plus magnetic susceptibility with PCs yielded the highest average precision at 81%, followed closely by XRF without PCs at 80%. Combining all the methods into a single analysis yielded the highest average accuracy 61%, followed by ICP-MS with PCs (59%).
Overall, analysis of the blind samples allowed researchers to eliminate 60% of the territory under investigation.
"Much of forensics is about elimination, so being able to rule out 60% of an area is a substantial contribution toward successfully locating a sample,” said de Caritat. “You can reduce the time, risk and investment of the ongoing investigation. The more parameters we look at, the more accurate the system is. We have reached 90% detection in some cases, although we think that would involve too many factors for real-world crime detection.”
Photo: Map of Canberra, showing location of reference samples. Location of test (blinded) samples in blue. Credit: Patrice De Caritat