Determining the region of origin of blood spatter patterns considering fluid dynamics, statistical uncertainties
De Brabanter, Kris
Trajectory reconstruction in bloodstain pattern analysis is currently performed by assuming that blood drop trajectories are straight along directions provided by stain inspection. Recently, several attempts have been made at reconstructing ballistic trajectories backwards, considering the effects of gravity and drag forces. Here, we propose a method to reconstruct the region of origin of impact blood spatter patterns that considers fluid dynamics and statistical uncertainties. The fluid dynamics relies on defining for each stain a range of physically possible trajectories, based on known physics of how drops deform, both in flight and upon slanted impact. Statistical uncertainties are estimated and propagated along the calculations, and a probabilistic approach is used to determine the region of origin as a volume most compatible with the backward trajectories. A publicly available data set of impact spatter patterns on a vertical wall with various impactor velocities and distances to target is used to test the model and evaluate its robustness, precision, and accuracy. Results show that the proposed method allows reconstruction of bloodletting events with distances between the wall and blood source larger than ˜1 m. The uncertainty of the method is determined, and its dependency on the distance between the blood source and the wall is characterized. Causes of error and uncertainty are discussed. The proposed method allows the consideration of stains indicating impact velocities that point downwards, which have typically been excluded from trajectory reconstruction. Based on the proposed method, two practical recommendations on crime scene documentation are drawn.
This article is published as Attinger, Daniel, Patrick M. Comiskey, Alexander L. Yarin, and Kris De Brabanter. "Determining the region of origin of blood spatter patterns considering fluid dynamics, statistical uncertainties." Forensic Science International (2019). DOI: 10.1016/j.forsciint.2019.02.003. Posted with permission.