First meeting of the BOM users group

Wednesday December 12, 2000

Attending:

Jarle Berntsen, Tor Eldevik, Yngve Heggelund, Bjørn Ådlandsvik, Inge Eliassen, Helge Avlesen, Agathe Sørflaten, Ingvild Lygren and Jonas Palsson.

The meeting had the following agenda for the discussions:

Briefly on the background of BOM

Numerical ocean models seem to be a constantly evolving type of technical software, and the Bergen Ocean Model (BOM) is no exception! Professor Jarle Berntsen, who has been the main developer of BOM, started the meeting by giving some background on the development of BOM.

Numerical ocean modeling is a relatively new field in science, at least in Norway. After the algae incident the summer of 1988, where thousands of tons of salmon were lost, the marine science community in Norway got funding to develop a model for the prediction of such algae blooms. The Norwegian industry also was interested in ocean modeling, and funded a large model intercomparison project, MOMOP, to find the model best suited for the challenging Norwegian waters.

The model that came best out of the intercomparison was the Princeton Ocean Model, and this model became the most widely used model in Norway, and probably still is.

Jarle used this model for some years when he was working at the Institute of Marine Research in Bergen (IMR), but after he became appointed a professor in applied mathematics at UiB, he started developing the Bergen Ocean Model together with people from IMR and the university. Having a track record in producing technical software for numerical integration and acoustics he saw there was clearly room for taking new and more advanced numerical methods into oceanography.

Experiences

In the recent years BOM has been used as the main tool by around ten Master level students and four PhD students, as well as a handful of researchers in Norway. There are also a few people abroad using the model. The main focus has been on Norwegian waters, such as the circulation in the Skagerrak and North Sea, various fjord studies, and recently there has been a lot of simulation activity focusing on the Ormen Lange gas field, outside the west coast of Norway. The PhD programmes has mainly been on model development, evalutation of different methods for the computation of internal pressure, development and testing of two-way nesting approaches, convergence studies, investigation of general numerical issues linked to the C-grid and sigma-coordinate system, such as coriolis averaging and the choice of time-stepping mechanism. BOM has also been used for applications in theoretical hydrodynamics, such as simulation of short frontal waves and spiral eddies.

The list of publications reflects most of these activities.

Development

The main design goal for BOM was modularity, and this is achieved through the use of e.g. Fortran 90 and modules. Names of many variables are for instance same as in POM, and many subroutines are also strongly inspired by POM.

The discussion went on topics such as internal pressure, the choice of vertical coordinate system and time-stepping techniques. The conclusion was that the problems we have encountered using BOM are of fundamental character, and are not necessarily avoided by choosing a different model concept. Choosing a layered/isopycnal model solves the internal pressure headache, but gives resolution problems near surface and bottom. Using finite element models gives very flexible handling of irregular geometries, but current freely available models have problems with conservation. On the question of implicit versus explicit time-stepping of surface gravity waves it was disccussed whether or not one can expect a truly consistant solution using mode-splitting. Practice has shown that mode-splitting seem to be the winner for problems where tidal effects are important.

On the development front, BOM 2.0 will be readied and released early 2001. New features are the improved Coriolis averaging technique due to Berntsen & Espelid, an option to use Stelling-Van Kester type internal pressure scheme, an option to use a Superbee-limiter scheme for advection of momentum (and other scalars), some optimization and the introduction of OpenMP directives for more efficient execution on SMP machines. A parallel version using MPI (improved scalability) will also probably be available in 2001.

Future applications and Funding

We also had time for a brief discussion on possible future applications of BOM. The Ormen Lange studies has revealed several interesting effects that are interesting for both researchers and the oil-industry, so it is likely that more funding will be available.

Jarle was also telling about a long time project on diving the outlet of a river in a fjord to increase the primary production. Physical experiments are about to be started, just avaiting funding for some tubes...

The meeting was ended with a very nice dinner at Singer's!

Written by Helge Avlesen <avle @ ii.uib.no>