PhD Studentship joint between The University of Bristol and The Proudman Oceanographic Laboratory
Topic: Applying Rule-Based Models to Sea Level Forecasting Supervisors: Jonathan Lawry (University of Bristol), Kevin Horsburgh (Proudman Oceanographic Laboratory) Project Details (<http://www.enm.bris.ac.uk/ai/enjl/positions.html>) Storm surges are the response of the sea surface to the meteorological forces of wind and atmospheric pressure. They represent an important component of total sea level and have been the subject of much scientific investigation. Marine scientists and engineers need to understand the statistics of surge occurrence and tide-surge interaction over long timescales in order to provide estimates of extreme sea level for design purposes. Numerical models of tides and surges (based on hydrodynamic equations) have a long history in coastal flood warning. Whilst these models have been very successful, and form the backbone of current operational forecast procedures, they are inherently limited by inaccuracies in bathymetry, meteorological forcing and parameterisations of sub-grid scale processes. There are now real opportunities for alternative, data-driven methods of surge prediction using artificial intelligence (AI) and in particular rule-based models (RBMs). RBMs can provide a high-level linguistic representation of the mapping between input and output variables in a prediction problem, allowing for more understandable models which give an insight into important underlying relationships. Such models can also be extended to incorporate both the fuzzy and probabilistic uncertainty typically present in hydrology and oceanography applications. This PhD will allow the student to demonstrate the use of rule-based models to an important environmental problem. The student will develop rule-based models combining probabilistic and fuzzy uncertainty and compare these with both deterministic forecasting techniques and alternative time series methods. The student will apply and extend techniques such as the LID3 algorithm for learning probability estimation trees incorporating fuzzy description labels. The new models will then be applied to the improvement of tidal forecasts in regions of extreme tidal range, and to the prediction of storm surges at key sites around the UK. The models will also be used to examine the principal physical causes of extreme sea level events, and the logical rules governing tide-surge interaction. Finally, the project will demonstrate the possibility of using artificial intelligence for the interpretation of ensemble forecasts. The successful applicant will have a good numerical degree in mathematics, statistics, computer science, engineering or physical science. Some knowledge of artificial intelligence, probability theory and fuzzy logic would be an advantage, although full training will be given. The student will be provided with training in coastal oceanography and numerical modelling. All necessary data from tide gauges and from the deterministic numerical models will be supplied by POL and the Met Office. Funding This project is funded by the EPSRC Flood Risk Management Consortium. This position is open to anyone, but non-EU candidates must find alternative means to fund the difference between UK and overseas tuition fees. Contact Please contact either Jonathan Lawry or Kevin Horsburgh directly for any informal inquires. The Closing date for applications is 30 June. Application forms are available online. All applications and references should be sent directly to: Emma Weeks Dept. Engineering Maths. University of Bristol, Bristol, BS8 1TR, UK _______________________________________________ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai _______________________________________________ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai
