Mathematical Modelling of Atmospheric Convection for better Regional Climate Change Prediction

Lead Research Organisation: University of Reading
Department Name: Meteorology


The simulation of clouds and precipitation is one of the weakest aspects of weather and climate models and is known to contribute to errors in predicting the regional impacts of climate change. Clouds and precipitation are controlled by atmospheric convection which usually occurs on spatial scales smaller than the grid scale and consequently must be represented by a parameterisation. Questionable assumptions are made in convection parameterisations such as neglecting mass transport by convection.

A new theoretical approach to convection parameterisation uses conditional averaging to calculate two values of momentum, temperature and density per atmospheric grid box: one for the convectively stable part of the flow and one for the convectively active part. This approach will enable transport of mass by convection. However, numerical solution of the resulting equations could have severe time-step restrictions.

This project will involve designing a numerical technique for solving the conditionally averaged equations for representing convection that will not require a very small time-step. Support will be provided for the student to implement their numerical method in OpenFOAM, a C++ toolbox for solving differential equations.

This project is part of the ``Revolutionizing Convective Parameterization'' project at Reading which is part of the NERC/Met Office programme ``Understanding and Representing Atmospheric Convection across Scales''. The student will therefore be part of a national network. This is an opportunity to contribute to model development that will directly feed in to weather and climate forecasting.


10 25 50
publication icon
McIntyre W (2020) Numerical methods for entrainment and detrainment in the multi-fluid Euler equations for convection in Quarterly Journal of the Royal Meteorological Society

publication icon
Weller H (2020) Multifluids for Representing Subgrid-Scale Convection in Journal of Advances in Modeling Earth Systems

Studentship Projects

Project Reference Relationship Related To Start End Student Name
NE/N013743/1 01/08/2016 31/05/2020
1787146 Studentship NE/N013743/1 01/10/2016 30/09/2019 William McIntyre
Description Atmospheric convection is a process responsible for many of the circulations in the atmosphere as well as phenomena like rain, hail, lightning and storms.
Current methods to model shallow and deep convective clouds are one of the largest sources of uncertainty in weather prediction and climate models. This is partly because these processes are smaller than the data cells used in models (the cells are about 10km across but vary in different models).
We set out to model the atmosphere with the multi-fluid approach, where convective regions are a separate fluid from the rest of the atmosphere. This means that more information is available for the convective regions which allows them to be modelled more accurately.
Creating a model for multi-fluid convection is much more difficult than a conventional single-fluid method as one has to decide how the fluids interact (in a physically realistic way). If this is not done properly then the equations & simulations become unstable, meaning they break down and can not represent the convective processes.
We found that the fluids should be coupled together via a mean pressure plus a pressure anomaly. The pressure anomaly is proportional to the convergence of velocity in each fluid.
An additional coupling is the entrainment or detrainment - the mass transfer between fluids. It was found that fraction of one fluid should be transferred to the other fluid if it converges (i.e. bunches up in some region of space). This method accurately reproduced the behaviour seen in the convection-based "rising bubble" test case.
More experiments were conducted on how the fluid temperatures change when mass is transferred but no method had results which were obviously better than others.
Using the above results, a 2 fluid model was created which could model convection better than a purely single-fluid model at all coarse resolutions.
Further work must be done to fine-tune the method and to include the representation of moisture.
Exploitation Route Further research in improving the multi-fluid method for convection.
Using multi-fluid convection to improve existing models (dependent on point above).
Sectors Aerospace, Defence and Marine,Environment,Other

Title Development of AtmosFOAM 
Description AtmosFOAM is a library of utilities and applications built with OpenFOAM. AtmosFOAM is specifically designed to model processes the atmosphere and was created by Hilary Weller. Recent additions to AtmosFOAM include multi-fluid modelling framework, multi-fluid shallow water equations and the advection of moisture. 
Type Of Technology Physical Model/Kit 
Year Produced 2019 
Impact Software used to conduct multi-fluid convection simulations needed for the Research project. 
Description Invitation to talk at the Royal Meteorological Society 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Members of the Royal Meteorological Society attended talks by the 2 winners (including myself) of "Quo Vadis" at the University of Reading.
Quo Vadis is an anual event in the University of Reading Meteorology department where 2nd-year PhD students present their research projects and results thus far.
Year(s) Of Engagement Activity 2018