CEEH logo ceeh rapporter
<<- Tilbage til phd- og postdoc projekter
<<- Back to PhD- and postdoc projects

Implementation of Locally Mass Conserving Semi-Lagrangianschemes for precise and efficient chemical transport modeling.

PhD Project Description for Brian Sørensen

Project Aim

The aim of this PhD project will be to perform research within atmospheric chemistry modeling, given a set of scientific hypotheses defined within CEEH (Center for Energy, Environment and Health). The main hypothesis is that the Locally Mass Conserving Semi-Lagrangian scheme (LMCSL) can, while ensuring mass conservation, improve the so-called online chemical transport model Enviro- HIRLAM in an operational setup.

The main goals of this PhD work are:

1. to use the newest numerical techniques in chemical transport modeling as well as understanding the underlying meteorological, physical and chemical processes. Improved aerosol dynamics and secondary aerosol production/interaction will be considered.

2. to implement the LMCSL scheme in the operational Enviro-HIRLAM and to test and validate the new mass conserving dynamical core.

3. to complete at least one simulation of the year 2000 in the CEEH framework, comparing it to an ofline simulation.

4. if possible to implement a newly developed mass conserving Particle In Cell scheme (PIC). There will be special emphasis on reducing the so called mass-wind inconsistency problem.

Background

The work of this PhD is to be carried out within the Centre for Energy, Environment and Health (CEEH) (www.CEEH.dk) at the University of Copenhagen and at the Danish Meteorological Institute (DMI). The CEEH is funded by the Programme Comission on Energy and Environment, Danish Agency for Science, Technology and Innovation. CEEH is a strategic research centre with the mission to develop models which support planning of future energy systems in Denmark, where both direct and indirect costs related to the impacts on the environment, climate change and health are considered. This requires precise modeling of chemical species and efficient numerical models.
The online integration of ACTMs, has the benefit of being able to use the meteorological fields at each time step, and can have two-way feedbacks between the meteorological and chemical processes. There is also numerical advantages in using online coupled models, which is that everything is on the same grid with the same time step. No interpolations are needed and it is possible to have the exact same advection for both the meteorological fields and the chemical fields. Enviro-HIRLAM is an online-coupled meteorological, chemical transport and dispersion model developed at the DMI. It is based on the HIRLAM model, which is used for operational weather forecasting.

The Enviro-HIRLAM model uses a traditional semi-implicit semi-Lagrangian (SISL) advection scheme for all meteorological fields except the chemical constituents, which are advected with the BOTT scheme, which is a 3D Eulerian fluxform scheme. This combination is used to ensure mass conservation and the ability to use large time steps. It is however not consistent, since the advection is treated differently for different fields, they are not exposed to the same forcing. It has been shown to have a significant effect in long online-coupled simulations. The BOTT scheme can become very computationally expensive as the number of chemical constituents increases with more complex chemistry schemes. Since it is Eulerian it is also subject to the traditional Eulerian time step limitations, and the individual time step is split into three, to overcome this.
Enviro-HIRLAM employs the NWP-Chem scheme for gas-phase chemistry. The scheme has been developed for online models, with as few species and reactions as possible to facilitate execution in an operational environment. It is based on the quasi-steady state approximation and contains 17 species and 20 reactions. The LMCSL scheme is a semi-Lagrangian with a quasi-Lagrangian vertical coordinate. By combining modified semi-Lagrangian interpolation weights and a local and mass conserving monotonic filter, it ensures mass conservation, increased accuracy and consistency within the model. It can also be used with a Lagrangian vertical coordinate which reduces the vertical numerical diffusion.
Like the LMCSL scheme the CISL scheme is locally mass conserving and consistent. It uses cell integrals instead of modified interpolation weights. The Lagrangian vertical coordinate can also be implemented in the CISL scheme. The mass conserving particle in cell (PIC) scheme is still in development but has been tested successfully in a number of passive advection tests and in a shallow water water model. The scheme makes use of a Lagrangian set of particles. In this way it seems possible to fulfill essentially all desired numerical properties of a transport scheme. In particular the accuracy increases dramatically while the CPU costs decreases. The only expense is an increase in RAM....


Dato: 12-Nov-2009