MI-POM

Numerical ocean models: MI-POM

History

MI-POM is a three-dimensional hydrodynamic ocean model which computes the time evolution of sea surface elevation, currents, salinity and temperature. It is a baroclinic, primitive equation model containing conservation equations for mass (continuity) and momentum. In addition, the thermodynamics are described by similar conservation equations for salinity and temperature. MI-POM has been developed at the institute as an enhanced version of the Princeton ocean Model (POM). POM was developed by Prof. G.Mellor and Dr. A.F.Blumberg at the Princeton University at the end of the 1970's.

Sea surface temperature in the Norwegian Sea and adjacent oceans (snapshot).

The model has a free surface and a time splitting procedure is used in which the two-dimensional part of the dynamics (sea surface elevation and depth mean currents) is calculated separately with a short time step, while the three-dimensional part containing the vertical distribution of currents, salinity and temperature may be computed with a much longer time step (usually 30-80 times longer).

Some other major features in POM are:

advanced computation of turbulent mixing by applying a 2.5-order turbulent closure submodel
terrain following vertical coordinate (sigma-coordinate)
orthogonal curvilinear staggered horizontal grid (Arakawa C)
implicit scheme for time integration which allows for high vertical resolution

From POM to MI-POM

The version of POM which was implemented at DNMI in 1990 was called Estuarine, Coastal and Ocean Model (ECOM). Since 1993, the model has provided operational forecasts of storm surges and currents in Norwegian waters. Besides the production of daily forecasts on a routine basis, MI-POM has been applied in several studies and assignments, e.g., for the public administration and the oil companies engaged in offshore activity in Norwegian waters. Lately, the model has also been applied in climate change studies.

The model is continually being upgraded and improved. The modifications of the original ECOM version that has been performed at DNMI may be divided into three areas: Physical processes, numerics, and technical advances. As a consequence of the extent of these modifications, the model is now called the Meteorological Institute's POM (MI-POM). The most significant changes are listed below.

Physical processes:

Development of an integrated model system where the ocean model is coupled with a two-dimensional sea-ice model.
Implementation of a biogeochemical module for the purpose of forecasts of eutrofication and algae bloom in the North Sea and Skagerrak/Kattegat (in cooperation with the Institute of Marine Research).
Implementation of more efficient computation of the vertical diffusion of the turbulent kinetic energy in the turbulence closure sub-model.
Modification of the turbulent kinetic energy dissipation in the turbulence closure sub-model, by introducing a cut-off for the dissipation for a "critical" Richardson number in the case of stable stratification.
Implementation of an isoneutral diffusion scheme (Redi diffusion), and an eddy-induced tracer transport scheme (Gent-McWilliams scheme) to better conserve water masses (i.e. salinity and temperature) in climate change simulations.
A new version of the calculation of the bottom stress for 2-D simulations.
Implementation of tidal forcing at the open boundaries.
Implementation of fresh and brackish run-off from rivers and/or estuaries.

Numerics:

Code modifications for execution of the model on massive parallel platforms.
Introduction of the Flow Relaxation Scheme (FRS) as the lateral open boundary condition for all dynamic variables.
Implementation of the Multidimensional Positive Definite Advection Transport Algorithm (MPDATA) for salinity and temperature (replaces the leapfrog advection scheme)

Technical advances:

Adaptation to operational forecasting: all time parameters are initiated according to the time for which the atmospheric forcing (mean sea level pressure and surface winds) is valid.
Relaxation of the thermodynamic properties (salinity and temperature), in deep layers and/or at the surface (optional).
Implementation of one-way nested simulations (using the FRS scheme) in which results from a coarse mesh run are used as boundary values for a fine mesh simulation.
Implementation of a hot start (restart) procedure.
Implementation of general routines for interpolation between different types of curvilinear horizontal grids.
Simple routines for assimilation of sea surface elevation and surface currents.
Modification of the date-time counting variables in order to fit climatological simulations (optional).

The model system is fully portable, i.e., no geographical information, such as grid orientation, size and location of computational domain, are hard-coded in the model. All such information is given as input data along with the chosen model domain and topography.

Other ocean and ice models at the institute: | MICOM | MI-IM |

Related discipline: | Oceanography |

Related method: | Numerical prediction models |