Computational Framework for Multi-Scale Environmental Modelling

Lead Research Organisation: Imperial College London
Department Name: Earth Science and Engineering

Abstract

Convection is one of the most challenging problems in atmospheric science today. It covers highly energetic processes, like volcanic eruptions or biomass burning plumes, as well as fields of cumulus clouds or single deep thunderclouds. Standard atmospheric models, such as those used in weather forecasting or climate prediction, are generally not able to resolve the scales involved in convective activity. While cloud or plume sizes may well reach the 1km scale in the horizontal and 10km scale in the vertical, key processes, such as turbulent mixing that lead to extremely important entrainment (mixing environmental properties into the convective column) are of the scale of a few to tens of metres. Such convective processes are often responsible for fast vertical transport of pollutants from the boundary layer to higher atmospheric layers (in volcanic eruptions, or very deep convective clouds, up to the stratosphere), and therefore their correct simulation is highly crucial. Similarly, less vigorously convective but highly turbulent flows in complex topographies (urban or mountainous environments) are important for the dispersion of hazardous chemical species or for the development of wild fires. Such situations are challenging for the current generation of environmental numerical models. The overall purpose of this project is to couple and optimise two existing computational models (Imperial-FLUIDITY and Cambridge-ATHAM). ATHAM is a high-resolution atmospheric model with physical parameterisations for a wide range of plume and cloud relevant applications. ATHAM has successfully simulated atmospheric processes with high spatial resolution within a limited area for problems where topography and the interaction with the flow outside the computational domain are of secondary importance. FLUIDITY contains state-of-the-art parallel adaptive mesh methods that are able to optimally resolve flows, whilst being able to represent key force balances (geostrophic and hydrostatic) exactly which is important for accuracy and stability, and has been developed in its oceanographic guise of ICOM. FLUIDITY lacks the physical parameterisations for atmospheric problems that ATHAM will supply; however, it provides a general framework for CFD problems for a wide range of computational domains and resolutions. ATHAM-FLUIDITY will combine the best elements from both models. That is, the flexible adaptive mesh and balance maintaining finite element methods of FLUIDITY and the advanced physical models of ATHAM, allowing a new range of problems associated with global atmospheric models to be investigated. An important example is convection, which often develops within frontal systems that are part of the large-scale flow with topography and differential heating due to surface inhomogeneities often providing the perturbation that can trigger convection. The combined model will be able to capture large-scale flows as well as fine-scale features in areas of interest allowing a more efficient interaction of scales in one single model. Over the last decade, the programming paradigm has changed from structured to modular and object-oriented programming, in which any set of modern languages may be widely used. Therefore, combining a number of open-source codes and libraries with the physics and advanced numerical technologies contained within FLUIDITY and ATHAM offers an excellent opportunity to develop the combined ATHAM-FLUIDITY model as a next-generation environmental flow model. The main advantages of the resulting open-source model will be: (a) flexibility of the problem formulation; (b) multi-physics modelling to optimally represent the physics using parallel mesh adaptivity and; (c) modular design of the advanced component technologies (e.g. CAD-geometry and mesh generation, linear and non-linear solvers etc). ATHAM-FLUIDITY will be linked produce computational results that are more realistic and accurate than the existing software codes.

Publications


10 25 50
 
Description During the course of the project a number of numerical techniques for additional features were developed and integrated into the Fluidity codebase (https://launchpad.net/fluidity):

1. a pressure-correction based algorithm to solve the fully compressible Navier-Stokes equations for gaseous fluid flow under gravity and rotation for arbitrary equations of state was developed (previously only an incompressible form of this algorithm was present in the code);

2. an extension of the above was developed to allow for a multi-material bulk formulation approach to systems with multiple constituents, in both compressible (gaseous) and incompressible (droplet/particulate) phases. This extension was necessary to implement the 'Active Tracer' component of the ATHAM (http://www.geog.cam.ac.uk/research/projects/atham/model.html) scheme, in which the heat capacities of spatially varying constituents are considered explicitly;

3. an extensible framework for coupling micro-physics with the Fluidity dynamic core was developed. This allows for essentially adiabatic reactions between constituent species (such as chemical reactions between air pollutants) as well as for diabatic phase changes by the species themselves (such as the ice-liquid, water-water, vapour exchange in a moist atmosphere);

4. an automated method for exact hydrostatic pressure splitting was developed, to ensure the numerical stability of unbalanced finite element pairs by preventing the spurious excitation of fast gravity waves;

5. validation and benchmarking test cases were considered including dry and moist bubble experiments, inertia gravity waves, and lee wave generation behind topography. In all cases comparisons between fixed and adaptive mesh methods were considered. These test cases, as well as the above numerical developments, form the basis for two publications close to being submitted.

The new capabilities, coupled to Fluidity's pre-existing automated parallel mesh adaptive routines produce a code well suited to high-resolution modelling of geophysical multi-material flows, especially those with pronounced transient multi-scale features such as fronts or jets. Specific examples of possible applications include convective scale cloud formation, volcanic plumes, wild fires and point source air pollutant release events.

In addition, the developments of this project have been valuable in contributing to a PhD project which has been developing Fluidity's full multi-phase modelling capabilities (doi:10.1093/gji/ggs059). In particular this project has allowed for the rapid development of compressible phases and the re-use of the hydrostatic pressure splitting algorithm discussed above. These projects will result in simulations of volcanic plumes in the very near future.
Exploitation Route through the Open Source nature of the developed code, and as is happening by some of the researchers under a new EU FP7 project.
Sectors Environment
 
Description EC FP7
Amount € 4,998,851 (EUR)
Funding ID http://www.pearl-fp7.eu/ 
Organisation European Commission (EC) 
Sector Public
Country European Union (EU)
Start 01/2014 
End 12/2018
 
Description EPSRC Grand Challenge project MAGIC
Amount
Funding ID EP/N010221/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom of Great Britain & Northern Ireland (UK)
Start 12/2015 
End 12/2020