High Throughput Forensic Palynology

This proposal develops a new system to track the past location of persons and materiel, based on rapid and accurate collection and classification of pollen grains. Plant pollen stands out among naturally occurring geolocation markers for the following reasons: Pollen is abundant and collects on exposed surfaces in all humanly habitable land areas; morphological and spectroscopic properties of pollen differentiate species and express a wealth of genetic information which can be translated into a geographical trail; human subjects are mostly unaware of the information potentially divulged by the dust they carry. The system consists of an advanced sample collection subsystem, an automated microscope and multispectral imager, and software able to identify, classify and geolocate pollen grains found in the sample. Clustering algorithms applied to parameters extracted from multispectral images are used to identify grains of common origin; this raises the reliability of identification significantly relative to classification from a single image. Geolocation of subjects is derived from known distribution of classified pollen types collected from their clothing or other surfaces. The collection and analysis systems are designed for high throughput and portability with no user interaction required. No knowledge of pollen science is required to operate the system.