Lead Supervisor: Ross Bannister, National Centre for Earth Observation (NCEO) and Department of Meteorology, University of Reading

Email: r.n.bannister@reading.ac.uk

Co-supervisors: Alison Fowler, National Centre for Earth Observation (NCEO) and Department of Meteorology, University of Reading; William Ingram, Visiting Research Fellow, University of Reading

Are you a physicist, mathematician, engineer, meteorologist, computer scientist or other quantitative scientist who likes the idea of earning a PhD working with some of the most fascinating and complex physical models, like those that forecast the weather?

State-of-the-art weather forecasting systems use detailed models of the atmosphere with millions of notional degrees of freedom. Despite the chaotic physics, models already do considerably better than used to be thought possible.  A big issue is their starting-point (initial conditions) as they cannot be run starting from observations alone, as these are not complete or accurate enough.  Instead they combine the observations for ‘today’ with the forecast for ‘today’ using a technique called data assimilation (DA).

The forecasts used by DA are, however, never perfect. Some of their error is in the ‘amplitude’ of features (e.g. a storm), and some is in the ‘location’.  We have all experienced a forecast that was correct apart from the location (or timing) being wrong.  Existing DA methods are designed to directly correct for errors in amplitude and so do not explicitly consider errors in location. The latest generation of DA systems, “4DVar”, can correct errors of location by altering the winds so they move features differently, but these wind changes may themselves be incorrect, and no research yet has tried to quantify how and how well 4DVar deals with errors of position.

This project is about how to quantify and correct for errors of location in analyses and forecasts, using field warping, an established technique that so far has not been exploited fully in atmospheric science. Field warping is a technique that could achieve such location correction, and will introduce a new perspective into weather prediction.

You will look at how this varies across different fields (temperature, rainfall, etc.) and will estimate how much forecasts could be improved if the error of location could be removed from the analyses. They will also look in detail at the challenging case of small-scale rainfall forecasts over the UK, covered by rain radar observations.  This is a particularly interesting because the model typically produces plausible features but they may not correspond well to particular features in reality.  This is a stimulating task investigating a novel approach at the intersection of meteorology and data science.

As a student on this project, you will be a member of the Department of Meteorology at the University of Reading, the Data Assimilation Research Centre (DARC, research.reading.ac.uk/met-darc), and the National Centre for Earth Observation (NCEO, www.nceo.ac.uk).  You will be based at Reading University, an attractive campus with good facilities.  Reading is a prosperous large town with all the amenities you’d expect, with some beautiful countryside around yet easy access to London.  The Department of Meteorology is highly-ranked for atmospheric science, so you will be exposed to a wide range of stimulating related science. It is an Academic Partner of the UK Met Office and also has close ties with its neighbour the European Centre for Medium-range Weather Forecasting.

Training Opportunities

Training will be offered in data assimilation, meteorology, and numerical weather prediction in the form of masters-level modules, summer schools, and special courses.  The student will have opportunities to present their work locally, at national and international conferences, and during visits to the Met Office HQ in Exeter.

Student Profile

The ideal student would have an excellent degree in physics, engineering, mathematics, meteorology, computer science, or another highly quantitative and analytical discipline. The student must be willing to learn the mechanics of data assimilation, show creative/innovative thinking, and be highly driven to gain new knowledge in this challenging area.  The ability to adapt and run complex computer code is essential.