Recognition of gasoline in fire debris using machine learning: Part II, application of a neural network

The recognition of ignitable liquid (IL) residues in fire debris is a resource intensive but key part of an arson investigation. Due to the highly diverse and heavily loaded chemical matrix of fire debris samples, combined with the broad chemical composition of IL, the interpretation of the laboratory analysis results is a very challenging task for the forensic examiner. Fire debris samples are commonly analyzed using gas chromatography coupled to mass spectrometry (GC-MS). This method delivers both the total ion chromatogram (TIC) with the individually separated compounds and the underlying mass spectrum of each of the separated compounds. In this study, a completely new approach for the recognition of gasoline in fire debris samples is presented. First, the GC-MS data, including retention time, signal intensity, and mass spectrum is converted into a bitmap image. Five different data-to-image conversion approaches are tested, and their advantages and limitations are discussed. Subsequently, a convolutional neural network (CNN) is utilized to allocate the generated images to the classes "with gasoline" or "without gasoline". The applied approaches to generate a digital image and the pattern recognition of the CNN perform very well in the classification of unknown test samples. Depending on the data-to-image generation approach used, the rate of correct sample classification in the test dataset is between 95% and 98%. The machine learning approach in this study, as well as the complementary method presented in an accompanying article, are not only useful for the recognition of gasoline in fire debris but are equally applicable to any additional areas in which the interpretation of complex chromatographic and mass spectrometric is required.

Recognition of gasoline in fire debris using machine learning: Part I, application of random forest, gradient boosting, support vector machine, and naïve bayes

The detection and identification of ignitable liquid (IL) residues in fire debris are two very challenging tasks in a fire investigation. To this day, the recognition of IL in fire debris includes the chemical analysis of the fire debris composition, followed by the examination and interpretation of the analysis result by a trained forensic examiner. Throughout the last decade, chemometrics and artificial intelligence have become increasingly important. In the present study, machine learning algorithms capable of recognizing gasoline residues in fire debris based on GC-MS data have been developed. Four methods, including random forest, gradient boosting, support vector machine, and naïve bayes are applied and used to classify fire debris samples into the two categories "with gasoline" or "without gasoline". A fifth method (logistic regression) did not converge due to well separated classes. A database comprising 360 measurements, including fire debris samples of real cases as well as fire debris samples spiked with known amounts of weathered gasoline (up to 99.6%), was available to train the machine learning algorithms (using 85% of the data) and to subsequently test the performance of the methods when classifying unknown samples (using 15% of the data). In general, the methods perform very well, as three of it succeeded to classify all test samples correctly without any false positive or false negative allocations. One (naïve bayes) was not trained enough to classify other (non-gasoline) IL correctly as "no gasoline". Furthermore, the random forest method reveals which chemical compounds are most relevant for the algorithm to classify the samples. In general, the presented approach is highly promising and could easily be extended or adapted to other types of IL. Similar to the neural network presented in the accompanying paper, such methods have the potential to serve as a fast screening technique for fire debris samples, thus supporting the forensic examiner by providing an additional independent opinion. Nonetheless, the definite identification of IL residues in fire debris always has to be accomplished by a forensic examiner.

Quantitative screening level assessment of human risk from PCBs released in glacial meltwater: Silvretta Glacier, Swiss Alps

Persistent organic pollutants (POPs) are entrained within glaciers globally, reemerging in many alpine ecosystems. Despite available data on POP flux from glaciers, a study of human health risk caused by POPs released in glacial meltwater has never been attempted. Glaciers in the European Alps house the largest known quantity of POPs in the Northern Hemisphere, presenting an opportunity for identification of potential risk in an endmember scenario case study. With methodology developed by the US Environmental Protection Agency (EPA), we provide a regional screening level human risk analysis of one class of POPs, polychlorinated-biphenyls (PCB) that have been measured in melt waters from the Silvretta Glacier in the Swiss Alps. Our model suggests the potential for both cancer and non-cancer impacts in residents with lifetime exposure to current levels of PCB in glacial meltwater and average consumption of local fish. For residents with an abbreviated 30-year exposure timeframe, the risk for cancer and non-cancer impacts is low. Populations that consume higher quantities of local fish are predicted to be at a greater risk, with risk to lifetime consumers higher by an order of magnitude. Based on the results of our screening study, we suggest that local government move to the next step within the risk assessment framework: local monitoring and management. Within the Alps, other glacial watersheds of a similar size and latitude may see comparable risk and our model framework can be adapted for further implementation therein.

Legacy organochlorine pollutants in glacial watersheds: a review

Northern Hemisphere alpine glaciers have been identified as a point of concentration and reemergence of legacy organochlorine pollutants (OCPs). In this review, we compile a selection of published literature combining long-range, global atmospheric transport and distribution-based compartmental environmental flux models, as well as data from glacial meltwater, ice core, crevasse and proglacial lake sediment studies. Regional studies of ice and meltwater in alpine glaciers of the northern latitudes show similarities in sample deposition profiles and concentration due to chemical atmospheric residence time, precipitation type and glacier flow rates. In glaciated locations near areas of extensive OCPs use, such as the Swiss and Italian Alps, glacier sample concentrations are higher, while in areas more distant from use, including Arctic nations, OCPs concentrations in glaciers are significantly lower. Our review identifies alpine glaciers co-located with regions characterized by OCPs use as a significant organochlorine pollutant distribution source, secondary in timing and location to direct deposition, with subsequent bioaccumulation and potential human risk impacts.

Release of PCBs from Silvretta glacier (Switzerland) investigated in lake sediments and meltwater

This study is part of our investigations about the release of persistent organic pollutants from melting Alpine glaciers and the relevance of the glaciers as secondary sources of legacy pollutants. Here, we studied the melt-related release of polychlorinated biphenyls (PCBs) in proglacial lakes and glacier streams of the catchment of the Silvretta glacier, located in the Swiss Alps. To explore a spatial and temporal distribution of chemicals in glacier melt, we combined two approaches: (1) analysing a sediment record as an archive of past remobilization and (2) passive water sampling to capture the current release of PCBs during melt period. In addition, we determined PCBs in a non-glacier-fed stream as a reference for the background pollutant level in the area. The PCBs in the sediment core from the Silvretta lake generally complied with trends of PCB emissions into the environment. Elevated concentrations during the most recent ten years, comparable in level with times of the highest atmospheric input, were attributed to accelerated melting of the glacier. This interpretation is supported by the detected PCB fractionation pattern towards heavier, less volatile congeners, and by increased activity concentrations of the radioactive tracer (137)Cs in this part of the sediment core. In contrast, PCB concentrations were not elevated in the stream water, since no significant difference between pollutant concentrations in the glacier-fed and the non-glacier-fed streams was detected. In stream water, no current decrease of the PCBs with distance from the glacier was observed. Thus, according to our data, an influence of PCBs release due to accelerated glacier melt was only detected in the proglacial lake, but not in the other compartments of the Silvretta catchment.