Computer Modeling

Computer modeling is a general term that describes the use of computers to simulate objects or events. As such, it is sometimes known as computer simulation. Forensic applications of computer modeling can produce purely graphical results (for example, the face of an unknown murder victim reconstructed from a skull) or mathematical idealizations of physical, chemical, biological, or geological processes (for example, calculations performed to estimate the speed of a vehicle before an accident). Most forensic computer models are extensions of graphical and mathematical techniques that have been used by forensic scientists for many years, but which have become much more complicated and visually compelling because of continuing advances in computer technology.

Craniofacial reconstruction (re-building the shape of the skull and face) is one example of a purely empirical graphical forensic technique that is adaptable to computer modeling. The traditional approach is to shape layers of clay placed on a cast of a skull in order to produce a likeness of an unknown person. The thickness of clay on different parts of the skull is constrained by information from tissue thickness databases, which were originally obtained from cadavers but now measured using techniques such as computerized tomography (CT) scans, magnetic resonance imaging (MRI), or ultra-sound imaging of living subjects. This has been an important advance, because cadaver measurements represented only a small segment of the general population. In computer-assisted craniofacial reconstruction, a virtual representation of the skull is created using a laser scanner or stereo photography to produce a three-dimensional mesh of points. Tissue thickness at selected points on the skull is specified mathematically, often using statistical relationships derived from large CT scan or MRI database, and the shape of the face is modeled as a smooth three-dimensional surface that passes through the measurement points. The main weakness of any craniofacial reconstruction technique is that the soft tissue thickness is always an estimate and it is difficult to infer facial characteristics reflecting age, weight, sex, and ethnicity from skull shape (although this information can be inferred from a complete skeleton). Superficial characteristics such as hair color and skin texture are impossible to infer from skull shape and are only artistic embellishments. Therefore, a general resemblance between a craniofacial reconstruction and a deceased person is the best that can be achieved.

Process-based forensic computer models combine equations describing physical or chemical processes with empirical information in order to reconstruct sequences of events. One widely used computer program for automobile accident reconstruction, known as SMAC (Simulation Model of Automobile Collisions), was originally developed by the National Highway Traffic Safety Administration. It uses Newton's laws of force and motion to simulate colliding automobiles as moving bodies in much the same way that one might simulate the collision of billiard balls. Factors such as road condition and tire type are incorporated using empirical coefficients, and the model input is adjusted until the output agrees with observations made at the accident site. Whereas this kind of computer model might calculate the energy at impact, it would not explicitly simulate the crumpling and deformation of the automobiles. Computer animation can be used to visualize the results of process-based models by depicting the automobiles as specific makes, models, and colors rather than nondescript masses or by incorporating realistic topography and scenery to simulate the accident scene. This kind of animation, in which variables such as vehicle position and speed are the result of scientific analysis and inference, is known as forensic animation.

A more sophisticated kind of process-based computer modeling involves the detailed simulation of physical or chemical processes in two or three dimensions (and often over time) in order to reconstruct an event or process. For example, a sophisticated accident model might simulate the bending and buckling of each structural member in an automobile rather than just the total amount of energy absorbed by one moving mass colliding with another. Another example is the use of computer models to simulate the two and three-dimensional movement of chemicals contaminating an aquifer. In order to obtain accurate results using this kind of model, geologists must collect detailed information about the materials comprising the aquifer by drilling test wells, taking samples of the aquifer materials, and conducting a variety of tests. The velocity and chemical composition of the groundwater are then calculated at many thousands, and perhaps even millions, of points within the simulated aquifer and the model is calibrated by adjusting the input until the results agree with observed conditions. Experts can use this kind of model to infer the source of the contaminants or the time that they entered the aquifer, which can be important in legal proceedings such as the well-known lawsuit concerning groundwater contamination in Woburn, Massachusetts. Fire scientists likewise use computational fluid dynamics models to simulate the spread of fires in buildings, and other computer models can be used to simulate the mechanics of solid objects, the flow of fluids, and chemical reactions. As computer models become more complicated, however, they also become more difficult to apply because the quality and quantity of input increase dramatically. As is the case for simple process-based models, the results of multidimensional can be visualized using static and animated computer graphics.