Lead Supervisor: Bob Plant, Department of Meteorology, University of Reading

Email: r.s.plant@reading.ac.uk

Co-supervisors: Chris Holloway, Department of Meteorology, University of Reading; Ian Boutle, Met Office

The numerical models used for weather forecasting and climate projection perform their calculations on a grid and simulate the flow on spatial scales larger than the grid. To obtain realistic simulations is imperative that the models include parameterization schemes, which determine the effect on the simulated flow of processes taking place at scales smaller than the grid. Models have parameterizations for various important processes, an especially important example being that for moist convection.

Model parameterization schemes, which estimate effects of processes occurring at scales smaller than those resolved on the model grid, have always been applied on the same computational grid as that for the resolved flow. However, with the newly developed Met Office “LFric” model, that no longer needs to be the case, and there could be some good theoretical and practical reasons for choosing to apply them at other scales.The aim of this project is to establish  the prospects for a multi-scale representation of convective processes and to explore for the first time how such a representation should be constructed. It will focus in particular on the “convective grey zone,” where we would anticipate the approach having maximum impact.  The grey zone is a regime where model grid lengths are of order 1-4 km and thus of similar scale to the convective updrafts in thunderstorm. It is notoriously problematic in both theory and practice but ignoring the regime is not an option: current short-range weather forecasts operate in this regime and high-resolution climate modelling is beginning to approach it. Thus, new parameterization approaches are sorely needed.

Our starting point will be the state-of-the-art CoMorph convection parameterization recently developed by the Met Office with university collaborators including the supervsiors. We will first use this diagnostically on data from simulations in both the grey zone and at very high-resolutions (order 100-500 m). Simulations could be based on idealized configurations of archetypal convective scenarios and/or using cases from the WesCon observational campaign of summer 2023. The analysis will allow us to answer questions such as:

  1. How well does CoMorph predict properties of the convection in the high-resolution model?
  2. To what extent does this depend on the spatial and temporal scales at which it is used?

These questions are important to establish what information about the finer scales is reliably captured.  Then we ask:

  1. How do the CoMorph results from the grey zone simulations compare with those of the sub-km simulations?

The issue here is to establish the extent to which the information that could in principle be reliably captured by CoMorph is actually captured in practice when its input suffers from mean-state errors typical of those produced by a grey-zone simulation.  Depending on results from the first half of the PhD, concepts for multi-scale representations of convection will be developed and tested.

The novelty of the multi-scale concept means that there are a wide range of possibilities and so the balance of activities in later parts of the project can be moulded by the student according to their developing aptitudes and interests, and the initial results obtained. As examples, it might consider: (i) what structural characteristics are most important for a useful grey-zone multi-parameterization, (ii) the development of a stochastic approach for feeding back additional tendencies from a coarse to a fine grid;  (iii) formulating an appropriate use of multi-scale CoMorph for convection that is relatively shallow; (iv) exploring interactions with other parameterized processes; or, (v) consideration of a machine-learning approach to estimate and correct differences between the grey-zone and high-resolution simulations, conditional on the CoMorph outputs for the mean state.

Over the coming 5-10 years, LFric is expected become established as the main numerical modelling tool for weather forecasting and climate applications at the Met Office, and at universities in the UK and elsewhere. This research project will be (one of) the first to use it outside of the Met Office and thus offers the student the chance to develop skills and experience that will be in increasingly high demand.

Training Opportunities

This project comes with a Met Office CASE award, which provides additional support for visits to the Met Office in Exeter. The Met Office supervisor will provide training on their latest state-of-the-art weather and climate modelling framework, LFric. The student will develop hands-on experience with the model at km and sub-km scales, learning from its developers, and improving skills in model workflows and data processing.

The student will also interact (e.g. via plenary workshops) with the NERC-funded ParaChute programme on turbulent processes and convection, which includes several dozen investigators and researchers at several UK universities and the Met Office.

Student Profile

This project would be suitable for students with a degree in physics, mathematics or a closely related environmental or physical science.  Knowledge of python or a similar programming language is desirable. Full training and support will be given for the use of the LFric model, but temperamentally it is important that the student should not be scared to get their hands dirty with coding and with careful analysis work.

Funding Particulars

This project has CASE support from the Met Office.