Launch of the DSI Seminar Series 2017

Data assimilation with finite element air pollution models

Presented by Dr David Fairbairn

Monday 27 March 2017


DSI Boardroom, South Kensington Campus


Dr David Fairbairn research background has involved data assimilation applied to geophysical models. He completed an engineering doctorate at the University of Surrey focused on the development of ensemble data assimilation methods for Numerical Weather Prediction (2009-2013). This was followed by a 3-year post-doc at Meteo-France in Toulouse, working on land surface data assimilation. David hope’s to build on this experience for the MAGIC project at Imperial College, where he will be assimilating urban sensor data into air pollution models.
It is difficult to predict with precision the future of cities, but there will be significant adaptations and changes by 2050, due to advances in technology, changing populations, social expectations, and climate change. Managing Air for Green Inner Cities (MAGIC) project is using advanced fluid dynamics and energy analysis to reduce energy use and improve air quality. A roadmap is needed to ensure that decisions taken as the city evolves lead towards a sustainable future. My role in the MAGIC project is to lead research into advanced techniques for data assimilation and sensor optimization for urban flow and air pollution problems. It is well known that pollutants such as nitrates and ozone are linked to respiratory diseases. During certain weather conditions, large concentrations of these gasses can become trapped in urban street canyons, which increases the risk of dangerous exposure. It is important for local authorities to be able to effectively monitor the dispersion of these pollutants. Pollution sensors are generally accurate but are limited by their sparse coverage, while a pollution dispersion model offers a good spatial coverage but may have problems with accuracy. Data assimilation can spread the observed information through space and time using the fluid dynamics of the model. This enables a much better picture of the pollution dispersion than observations or a model individually can provide. In this presentation I will demonstrate some simple examples assimilating pseudo-observations in to a finite element pollution dispersion model of a street canyon. I will also discuss our plans to assimilate sensor data into a model of a region in Central London with half a mile radius.



London Behaviour Analytics Meeting

Behaviour Analytics Lab - Data Science Institute

Venue: GDO Meeting Room 

Scientists interested in behaviour are increasingly adopting automated imaging and quantitative analysis into their work. Regardless of the species, this entails solving related technical problems from hardware, to computer vision, and data visualisation as well as conceptual problems such as how to identify representations and features that capture the essence of behaviour without over simplifying. There are many groups in London working on behaviour across the full range of model organisms and humans but there is not currently a good way of brining them together. An informal workshop-style meeting at the DSI would be an excellent opportunity to share expertise, make new connections and, start collaborations.



EuSpRiG 2017 (European Spreadsheet Risk Interest Group Conference- in association with the Data Science Institute)

Thurs July 6 2017Prof Mike Ward

Blackett Laboratory 
Prince Consort Road, Building 6
Imperial College London

EuSpRIG runs this well established annual conference to provide somewhere for researchers, practioners, vendors, consultants, auditors, etc, to network and share experiences in the field of spreadsheet risks research with international leaders. EuSpRIG  offers authoritative & comprehensive web based information describing the current state of the art in Spreadsheet Risk Management. Read more about the confence and EuSpRIG here. 


Advances in Data Science 2017 

 Monday May, 15 2017 to Tuesday May, 16 2017

Advances in Data Science will take place in Manchester, UK.

A two-day meeting to present recent developments in data science, with a focus on advanced analytics (machine learning, Bayesian statistics, scalable algorithms), privacy, visualisation, software and diverse applications. Speakers include leading data scientists from industry and academia (see programme).

Registration will open soon and there will be an opportunity to submit abstracts for consideration as short talk or poster presentations at the meeting.

Sponsored by University of Manchester Data Science Institute.


Data Science for Cyber-Security:

 25-27th September 2017

Imperial College will be holding this conference on the use of data science for cyber-security applications. These applications include deployment of statistical methodology, machine learning, and Big Data analytics for network modelling, anomaly detection, forensics, risk management, and more. Organised by Niall Adams and Nick Heard from Imperial and sponsored by the DSI, this conference aims to showcase cutting-edge research in statistical cyber-security in academia, business and government. Read more and register here.

To keep up to date with our news and announcements for all of our events, please sign up to our mailing list and follow us on Twitter and Facebook.