Combining real-time airborne bioaerosol spectrometry with neural network algorithms to quantify different bioaerosol emissions from agriculture.

点击翻译按钮获取中文摘要

This studentship will combine state of the art real-time single particle integrated optoelectronic bio-aerosol and dust-aerosol spectrometry combined with neural net data analysis and micrometeorological flux measurement techniques, augmented by laboratory wind tunnel studies, and high resolution dispersion models, to transform our understanding and quantification of bioaerosol emission fluxes from agricultural landscapes under different atmospheric conditions and due to the influence from agricultural activities. Experiments will be carried out at the Rothamsted Institutes agricultural sites who will provide access to multi-site turbulence and meteorological measurement systems.Questions addressed will include understanding the contributions of different farming activities to airborne bioaerosol dispersion and deposition for different agricultural ecosystems.Below canopy transmission, deposition and airborne dispersion of bioaerosol particulates will be measured using novel, field deployable optoelectronic instruments comprising multi band UV excitation-fluorescence spectrometers to identify bioaerosol classes, single particle fluorescence lifetime to discriminate between biological and non-biological as well as bio-dust mixtures from soils and plants. This work will support ongoing collaboration with bioaerosol instrument manufacturers including DMT-USA, PLAIR CH and Swisens-CH as well as instruments developed by the University of Hertfordshire with which we have long-standing collaboration.Discrimination between different bioaerosol classes will be improved using AI training data sets generated from instrument characterisation studies in the Eurochamp Chamber for Aerosol Modelling and Bio-aerosol Research (ChAMBRe), Genoa. Additional training experiments at the ChAMBRe are planned for 2024 which will contribute to this project.During natural pollen emission events a standard portable single particle, holographic- imaging spectrometer will be used to quantify airborne pollen concentrations of different species in real-time to monitor and compare natural emission events with farming activity generated bioaerosols. New studies have shown the influence of rainfall on enhancing emissions of specific classes of bioaerosols and these will also be monitored for soil, seeding, threshing and planting activities.. A range of different emission flux methodologies will be examined using the new boaerosol spectrometer databases collected under different farming activities to assess future reference standards to improve monitoring and mitigation of such emissions relevant to human, animal and ecosystem health. If time and instruments are available, these measurements will be supplemented by analysis of spray pesticide dispersion and leaf scale deposition efficiencies within agricultural canopies using high speed liquid droplet spectrometers. Finally local scale 3D plume mapping techniques using drone-based aerosol measurements will also be investigated collecting data for inverse emission flux model studies to improve understanding of pesticide deposition efficiencies and plume dispersal across field scales. The results will be used to improve quantification of bioaerosol emissions generated by different farming activities compared to natural emission mechanisms and their inter-relationships with different environmental factors by applying novel dimensional reduction algorithms combined with robust, outlier resistant AI clustering techniques. The aim will be, for the first time, to include directly measured bioaerosol fluxes into new multi-dimensional bioaerosol and micrometeorological databases that may be used for testing a range of new analytical approaches to monitoring emissions from agricultural ecosystems.