Elizabeth Forde

We present results from a study evaluating the utility of supervised machine learning to classify single particle ultraviolet laser-induced fluorescence (UV-LIF) signatures to investigate airborne primary biological aerosol particle (PBAP) concentrations in a busy, multifunctional building using a Multiparameter Bioaerosol Spectrometer. First we introduce and demonstrate a gradient boosting ensemble decision tree algorithm’s ability to accurately classify laboratory generated PBAP samples into broad taxonomic classes with a high level of accuracy. We then develop a framework to appraise the classification accuracy and performance using the Hellinger distance metric to compare product parameter probability density function similarity; this framework showed that key training classes were sufficiently different in terms of particle fluorescence and morphology to facilitate classification. We also demonstrate the utility of including advanced morphological parameters to minimise inter-class conflation and improve classification confidence, where relying on the fluorescent spectra alone would likely result in misattribution. Finally, we apply these methods to ambient data collected within a large multi-functional building where ambient bacterial- and fungal-like classes were identified to display trends corresponding to human activity; fungal-like classes displayed a consistent diurnal trend with a maximum at midday and hourly peaks correlating to movements within the building; bacteria-like aerosol displayed complex, episodic events during opening hours. All PBAP classes fell to low baseline concentrations when the building was unoccupied overnight and at weekends.

Measurements of primary biological aerosol particles (PBAPs) have been conducted worldwide using ultraviolet light-induced fluorescence (UV-LIF) spectrometers. However, how these instruments detect and respond to known biological and non-biological particles, and how they compare, remains uncertain due to limited laboratory intercomparisons. Using the Defence Science and Technology Laboratory, Aerosol Challenge Simulator (ACS), controlled concentrations of biological and non-biological aerosol particles, singly or as mixtures, were produced for testing and intercomparison of multiple versions of the Wideband Integrated Bioaerosol Spectrometer (WIBS) and Multiparameter Bioaerosol Spectrometer (MBS). Although the results suggest some challenges in discriminating biological particle types across different versions of the same UV-LIF instrument, a difference in fluorescence intensity between the non-biological and biological samples could be identified for most instruments. While lower concentrations of fluorescent particles were detected by the MBS, the MBS demonstrates the potential to discriminate between pollen and other biological particles. This study presents the first published technical summary and use of the ACS for instrument intercomparisons. Within this work a clear overview of the data pre-processing is also presented, and documentation of instrument version/model numbers is suggested to assess potential instrument variations between different versions of the same instrument. Further laboratory studies sampling different particle types are suggested before use in quantifying impact on ambient classification.

Primary biological aerosol particles (PBAPs) are an abundant subset of atmospheric aerosol particles which comprise viruses, bacteria, fungal spores, pollen, and fragments such as plant and animal debris. The abundance and diversity of these particles remain poorly constrained, causing significant uncertainties for modelling scenarios and for understanding the potential implications of these particles in different environments. PBAP concentrations were studied at four different sites in the United Kingdom (Weybourne, Davidstow, Capel Dewi, and Chilbolton) using an ultraviolet light-induced fluorescence (UV-LIF) instrument, the Wideband Integrated Bioaerosol Spectrometer (WIBS), versions 3 and 4. Using hierarchical agglomerative cluster (HAC) analysis, particles were statistically discriminated. Fluorescent particles and clusters were then analysed by comparing to laboratory data of known particle types, assessing their diurnal variation and examining their relationship to the meteorological variables temperature, relative humidity, wind speed, and wind direction. Using local land cover types, sources of the suspected fluorescent particles and clusters were then identified. Most sites exhibited a wet discharged fungal spore dominance, with the exception of one site, Davidstow, which had higher concentrations of bacteria, suggested to result from the presence of a local dairy factory and farm. Differences were identified as to the sources of wet discharged fungal spores, with particles originating from arable and horticultural land at Chilbolton, and improved grassland areas at Weybourne. Total fluorescent particles at Capel Dewi were inferred to comprise two sources, with bacteria originating from the broadleaf and coniferous woodland and wet discharged fungal spores from nearby improved grassland areas, similar to Weybourne. The use of the HAC method and a higher fluorescence threshold (9 standard deviations instead of 3) produced clusters which were considered to be biological following the complete analysis. More published data and information on the reaction of different speciated biological particle types to fluctuations in meteorological conditions, such as relative humidity and temperature, would aid particle type characterisation in studies such as this.

Primary biological aerosol particles (PBAP) are an abundant subset of atmospheric aerosol particles which comprise viruses, bacteria, fungal spores, pollen, and fragments such as plant and animal debris. The abundance and diversity of these particles remain poorly constrained, causing significant uncertainties for modelling scenarios and for understanding the potential implications of these particles in different environments. PBAP concentrations were studied at four different sites in the United Kingdom (Weybourne, Davidstow, Capel Dewi, and Chilbolton) using an ultra-violet light induced fluorescence (UV-LIF) instrument, the Wideband Integrated Bioaerosol Spectrometer (WIBS), versions 3 and 4. Using hierarchical agglomerative cluster (HAC) analysis, particles were statistically discriminated between. Fluorescent particles and clusters were then analysed by assessing their diurnal variation and their relationship to the meteorological variables, temperature and relative humidity, and wind speed and direction. Using local land cover types, sources of the suspected fluorescent particles and clusters were then identified. Most sites exhibited a wet discharged fungal spore dominance, with the exception of one site, Davidstow, which had higher concentrations of bacteria, suggested to result from the presence of a local dairy factory. Differences were identified as to the sources of wet discharged fungal spores, with particles originating from arable and horticultural land at Chilbolton, and improved grassland areas at Weybourne. Total fluorescent particles at Capel Dewi were inferred to comprise two sources, with bacteria originating from the broadleaf and coniferous woodland and wet discharged fungal spores from nearby improved grassland areas, similar to Weybourne. The use of HAC and a higher fluorescence threshold (9SD) produced clusters which were considered to be biological following the complete analysis. More knowledge of the reaction of speciated biological particles to differences in meteorology, such as relative humidity and temperature would aid characterisation studies such as this.