Pattern Recognition Based Software for Oil Spills Identification by Gas-Chromatography and IR Spectrophotometry

For environmental control purposes, floating oil spills in harbours, off shore areas and their sources must often be identified. Pattern recognition, applied to IR spectrophotometric data (600–2000 cm⁻¹ range), and to chromatographic data (n -alkanes) for the spill and various suspected sources such as oil and fuels from ships bunkers and harbour installations, can lead to definite conclusions; particularly after artificial weathering formula are used. The software application provides quick and accurate identification of the pollution source. The identification algorithm has a learning stage in which the user creates a minimal database. This database has a tree structure with classes (fuels, crude, etc.) and members representing samples from already known sources. A sample contains IR and chromatographic data and information of the originating source. A larger database means more knowledge, which conveys a better identification. When the origin of an unknown sample is searched for, the software looks for the best match through the database and displays the results in two lists; sorted by calculated similarity. One list displays the classes in which the unknown sample could be included and the other displays the possible sources. An extra check can be done by visual inspection of the overlapped graphics (unknown sample and each of the identified sources).