Improved algorithmic methods for identification of biological airborne particulate matter
**This project is expected to start before March 31st 2015**
The School of Earth, Atmospheric and Environmental Science within the University of Manchester offers this fully-funded 4-year studentship.
Aerosol particles, or particulate matter suspended in the atmosphere, are ubiquitous components of the earth system. Primary biological aerosols (PBA) consist of fungal and plant spores, fragments of plant or animal matter, and pollens. The emission and impact of biological aerosols on the Earth system is highly complex and uncertain. They impact many aspects of the environment via: ecosystem and crop damage; the potential for significant negative human and animal health impacts: spreading infectious disease; toxic and allergenic properties and have also been shown to influence precipitation patterns. State of the art instruments designed to detect the presence of PBA work on the principle that PBA contain biofluorophores such as NAD(P)H, riboflavin, and tryptophan which auto-fluoresce when excited with UV radiation. Once exposed to this radiation, the instrument then records this fluorescent response, over several wavelengths, as well as the particle shape and size.
What is the problem?
How do we turn this spectral response into classification of PBA types? Unfortunately, there is a technological bottleneck that limits our ability to accurately identify specific PBA types generally and in real-time. Typical approaches revolve around the use of 'traditional' unsupervised cluster and neural network analysis methods applied to entire data sets. However, a range of questions remain unanswered. For example, how might we develop appropriate technologies to be able to identify a specific bacterial species in real-time and thus develop appropriate mitigation or response strategies. This is the crux of this Phd studentship funded by the Defence Science and Technology Laboratory (DSTL). Whilst traditional unsupervised methods give us some insights into the potential range of disparate sources in the atmosphere, a range of supervised methods as trained to laboratory data of known PBA types remain unvalidated. Addressing this challenge will not only benefit the immediate issue of atmospheric PBA but have useful application for a wide range of other atmospheric constituents.
What will you do during this PhD:
In this project you will evaluate new supervised learning algorithms for real-time discrimination of bio-aerosol types in order to assess future PBA sensor networks, improve emission parameterisations and provide forecast validation. This project will use new laboratory and field databases based on technological advances in real-time bioaerosol detection hardware including on-the-fly bioaerosol identification and collection within urban and agricultural aerosol populations. You will apply these tools to a wide range of datasets, including developing dispersion simulations and scenarios. You will work within a highly multidisciplinary team, including project partners in DSTL and the Laboratoire des sciences du climat et l'environnement in Paris. The multidisciplinary skills you will learn during this project are highly attractive to employers, including the ability to perform data analysis within the realm of big data.
Deadline for Applications: 25th February 2016
Interview dates: week commencing 7th March 2016
Contact: Dr David Topping [Email Address Removed]
For more details on how to apply and eligibility criteria please visit