PhD Thesis Defence by Brian Sørensen
New Mass Conserving Multi-Tracer Efficient Transport Schemes
The main objective of the PhD project is to improve the performance, both concerning efficiency and accuracy, of the current numerical weather prediction and climate models using integrated chemical schemes.
Numerical weather prediction models have been constructed and developed for over half a century. Their aim was initially to describe the short term evolution of the local weather. Since then the community has grown in size, which has brought new ideas, questions, and challenges to the research field. This means that the once "simple" goal of describing the evolution of the weather a couple of days has grown to include, e.g. long term climate simulations, chemical transport models, chemical weather models, and air quality models. With each of these research areas, new requirements to the models follow.
In the present thesis three new transport schemes for integrated numerical weather prediction are developed. The schemes are all mass conserving, shape preserving, and multi-tracer efficient. Two of the schemes have been implemented in the online integrated Enviro-HIRLAM model at the Danish Meteorological Institute. Both schemes are based on the Locally Mass Conserving Semi-Lagrangian method developed by (Kaas2008).
The first scheme, the LMCSL-LL (Lagrangian Levels), uses a two-dimensional LMCSL method in a quasi-Lagrangian, i.e. cascade like, setup with the horizontal and vertical advection handled independently, this enables the scheme to conserve the forecasted Lagrangian levels and thus decreases the numerical vertical diffusion significantly.
The second scheme, the LMCSL-3D, is utilizing a three-dimensional implementation of the LMCSL method. This scheme extends the conventional semi-Lagrangian advection scheme in Enviro-HIRLAM to efficiently conserve mass while being shape preserving.
The last advection scheme, the Hybrid Eulerian Lagrangian (HEL), is yet to be implemented into a fully dynamic model, but has, through a comprehensive transport scheme test case suite, been compared to some of the most advanced methods in numerical transport modelling. The HEL scheme combines an Eulerian representation and a Lagrangian representation of the prognostic variables to either remove or minimize the disadvantages of the individual methods, thus effectively fulfilling most if not all of the so-called desirable properties for numerical transport schemes.
Eigil Kaas, Professor in Meteorology, University of Copenhagen
Alexander Baklanov, Adjoint Professor, Danish Meteorological Institute
Ulrik Smith Korsholm, PhD, Danish Meteorological Institute