[ES_JOBS_NET] PhD Opportunity in Stochastic Convection Parameterization at University of Reading, UK

Todd Jones t.r.jones at reading.ac.uk
Wed Jan 31 02:52:50 MST 2018

Application Deadline extended to February 5
New stochastic representations for the uncertainties in operational numerical weather prediction
Lead Supervisor: Robert Plant, University of Reading, Department of Meteorology, r.s.plant at reading.ac.uk
Co supervisors: Todd Jones, University of Reading, t.r.jones at reading.ac.uk<mailto:t.r.jones at reading.ac.uk>, Sarah-Jane Lock, ECMWF, sarah-jane.lock at ecmwf.int<mailto:sarah-jane.lock at ecmwf.int>
Detail<http://www.met.reading.ac.uk/nercdtp/home/available/desc/entry2018/SC201834.pdf> (http://www.met.reading.ac.uk/nercdtp/home/available/desc/entry2018/SC201834.pdf)
Video (https://www.youtube.com/watch?v=U44-kRvfrLM&list=PLZWYaq_mWwsEM5dH1abHjYIgU2EVaegT9&index=3)
This project has co-supervision and placement opportunities from the ECMWF.
Training opportunities: Training will be given and experience acquired in performing numerical weather forecasts using arguably the best probabilistic forecasting system in the world. Specialist courses and support will be available at ECMWF to complement core training in the fundamentals of atmospheric science at the University of Reading. Student profile: This project would be suitable for students with a degree in physics, mathematics or a closely related environmental or physical science. http://www.reading.ac.uk/nercdtp

Funding Eligibility: UK and EU nationals. More details: http://www.met.reading.ac.uk/nercdtp/home/apply.php

The main interview day will be 14 February 2018 although a few supervisors may interview on different days in early February.

Project Description:
Numerical weather prediction (NWP) has advanced enormously over recent decades, at the rate of around 1 day of predictability per decade. That is, a forecast made today for five days ahead is typically as accurate as a forecast would have been for two days ahead if made 30 years ago. However, advances have not only been in improving accuracy: some of the greatest societal benefits have been obtained by improving our understanding and ability to predict the uncertainties in the forecasts. NWP is now treated as fundamentally a probabilistic problem, in which the consideration of uncertainties plays a central role.
Uncertainties occur because atmospheric motions are chaotic, with subtle differences in initial conditions leading to much larger differences at later times. Uncertainties also occur because the equations of atmospheric motion cannot be perfectly solved and because the equations for some important physical processes within the atmosphere are not fully known. The effects of several small-scale physical processes must be represented indirectly via parameterization schemes. Key among these is moist convection, an essential component of the tropical climate and much high-impact weather.
Traditionally, parameterization schemes were assumed to be deterministic, producing representations of small-scale processes unique to the resolved-scale atmospheric state. That assumption becomes increasingly problematic as computational advances permit more accurate simulations at finer resolutions where small-scale processes exhibit more stochastic qualities. Small-scale convective fluctuations often interact strongly with the non-linear flow dynamics, with substantial repercussions for large-scale model evolution. Producing accurate and reliable probabilistic predictions therefore depends critically on representations of parameterization uncertainties.
The European Centre for Medium-range Weather Forecasting (ECMWF) has pioneered simple and effective stochastic representations of model uncertainty, with major benefits for forecasting practice. Though undeniably effective, the simple treatments do have some physical deficiencies (e.g. undesirable impacts on detailed heat and moisture budgets). Improved stochastic representation of uncertainty is an important scientific topic: it should be based on a judicious combination of theory, observations and supporting simulations. For example, uncertainties should depend in systematic ways on the current state of the atmosphere.
This project will identify and quantify the physical deficiencies in the current methods for treating uncertainty in operational NWP and will develop new methods designed to eliminate or ameliorate those deficiencies. We will focus especially on convection, as a key small-scale process that induces larger-scale uncertainties. We will also pay particular attention to atmospheric states that are close to transitions (e.g. from clear-sky to convectively active situations). These are important aspects of overall model uncertainty but we believe that they may not be well handled by current approaches.
Ultimately the goal of the project is to construct and to demonstrate the value of new methods of uncertainty representation that produce the right uncertainty for the right physical reasons. We expect the project to be both challenging and rewarding. If the project goes well, the close links with ECMWF mean that is perfectly feasible for solid physical insights from the PhD to be translated in future operational practice.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/es_jobs_net/attachments/20180131/73dde466/attachment-0001.html>

More information about the Es_jobs_net mailing list