Physical Downscaling & Climate Modelling

One of the main objectives of the Climate Programme is to run numerical models to give a future forecast of the ocean state over the coming decades (with a probable time horizon of 100 years from the present). The models that will be used for this work are similar to those that have been used for 7 day forecast operational predictions and are based on the Regional Ocean Modeling System (ROMS). The model domain extends over a large part of the North East Atlantic, covering Ireland's Exclusive Economic Zone.  The boundaries are set by the requirement to place them in deep water away from the Rockall and Hatton Banks.

The major differences from our operational model runs are that the climate runs cannot use boundary conditions from present day data and large scale models and they cannot use the output from the weather models that give us our weather forecasts. Instead we must use the output from an Atmosphere-Ocean General Circulation Model. They are integrated over part of the 20th century and the 21st and 22nd centuries using estimated conditions of carbon flux for different scenarios. These models have a very crude scale for both the atmosphere and ocean. Because of the large grid size, the ocean bottom is poorly represented. They do not include the major driving force of the tide.  The reason for this is simply the availability of computing power.  They do however include the major processes generating and using greenhouse gases.

The purpose of this note is twofold, firstly to discuss the relevant issues in deciding how the Marine Institute should downscale from the Atmosphere-Ocean General Circulation Model, and secondly to outline the main products of the climate change modelling as we perceive them now. In mid 2008, the Marine Institute will receive delivery  of a significant computer on which to run its numerical models. Even so the 2.5 km North East Atlantic model will only run at 80 times 'real' time. Using this model we could only expect to perform one simulation of a few decades. If we use a lower resolution 5 km model, we might expect the speed to increase to 400 times real time, so that we can perform just a few centennial model runs - including the recent past for comparison with measured data.

Potential Products of the Climate Models

The climate models generate a huge amount of data. It is necessary to process the data outputs to examine the main climate changes that are forecast to occur and to evaluate the stability of the model solution.

1. Volume and heat transport by the shelf edge current:
2. Meridional heat flux
3. Volume integrated heat reservoirs above banks:
4. Transport of heat and salt from deep water, across the shelf edge onto the shelf.
5. Position of the Irish Shelf front
6. Sea surface and seabed temperature and salinity
7. Extremes of storm surge sea level
8. Volume integrated heat, kinetic energy and momentum.

Physical Downscaling Model

A model of the Arctic and Atlantic (see below) implemented with ROMS will be used to downscale hindcast (20C3M) and climate scenario (A1B) runs from UK Met Office Hadley Centre global climate model HadCM3. The model has an average grid size of 12.6 km for the whole domain and ranges between 10 and 15 km around Ireland. The hindcast and climate runs will be performed on the new Marine Institute cluster that is expected to be operational by July 2008. The hindcast will cover the 1980-2000 period; while, the scenario runs will be performed for 2046-2066 and 2080-2100.