Undergraduate research experience placement in Social Tree UpRooting DirectorY (STURDY) in Lancaster Environment Centre

Paid summer placement opportunity

The aim of this project is to compile a database of tree failures across the UK from various forms of social media.  This database will then be used to validate predictions of individual tree failure by the TREEFALL model which has been developed in LEC with NERC funding.

We have collected over 9 million tweets focusing on key phrases relating to tree failure and limb loss throughout the UK (phrases such as “tree failure” or “blown over”) from the social media platform Twitter covering the period December 2007 to August 2016.  Over 70,000 of these tweets are precisely geocoded via GPS coordinates (latitude and longitude), whilst many others contain more general indicators of location such as the 90,000 individual place names mentioned in the tweet text or the home address of 3.2 million users.  In addition, 2.9 million tweets have accompanying photos or videos which can help provide further context to each tweet. This dataset likely holds an insight into the location and impact of tree failure incidents throughout the UK as seen by the users of Twitter. The curation and exploration of this dataset involves the identification of relevant tweets, working with location data of varying spatial specificity and the processing of media files.

Specific objectives are as follows:

• Create a spatial database of historical tree failure events. This will involve solving the following issues:
  • Coping with the varying accuracy of the datasets (e.g. precise or approximate location of tree, quantitative or qualitative estimate of height, interpretation of species)
  • Working with image and text data
  • Coping with multiple reports of the same event (e.g. reported by different individuals to different degrees of accuracy)
  • Triangulating all available information to give the best estimate of location and tree parameters.

• Use the database to validate the TREEFALL model. We will run the TREEFALL for specific historical storm events (e.g. December 2011, October 2013) and compare our modelled predictions of tree failure with reported cases of tree failure held within our socially-generated database.  The database will give us the opportunity to test, refine and validate our model such that we can have confidence in its predictions for past, present or future storm events.

This project links directly to the TREEFALL project which has been funded by NERC under their Environmental Risks to Infrastructure Innovation Programme. This programme is concerned with identifying, understanding and quantifying environmental risks to the infrastructure systems in the UK.  This project is being undertaken in partnership RSK ADAS and Scottish Power.  Tree failure is a major problem for Scottish Power since failure close to overhead lines can result in significant disruption across the power supply network.  Proactive management of trees close to overhead power lines is therefore necessary in order to minimise the risks of disruption to the network.

Duncan Whyatt & Alan Blackburn

Lancaster University

Please note the following eligibility criteria. Applicants should:

• be studying for an undergraduate degree in a quantitative discipline outside of NERC’s scientific remit(e.g. mathematics, statistics, computing, engineering, physics)
• be applying for a placement in a different department to their undergraduate degree,
• be undertaking their first undergraduate degree studies (or integrated Masters),
• be expected to obtain a first or upper second class UK honours degree,
• be eligible for subsequent NERC PhD funding (i.e. UK, EU or right to remain in the UK.