ICHEC and Met Éireann are involved in a collaboration to develop and run both Climate and Numerical Weather forecast models. Currently this involves the HIRLAM and Harmonie weather models, as well as the EC-Earth climate model.
EC-EARTH is an Earth system model that is being developed by a number of European National Weather Services. This takes advantage of advances in computing power to use weather modelling codes for climate simulation, to give higher resolution and accurary. A prime advantage of using a weather forecasting model for climate studies is the operational infrastructure; an enormous amount of observations can be assimilated into the model and it's behaviour can be verified against observations from the daily and seasonal to decadal timescales.
ICHEC's Dr. Honoré Tapamo has ported EC-Earth to multiple platforms including the walton cluster and the BlueGene/P computer. This code is now running on the stokes system, and running CMIP5 during 2009-2010 experiments for the IPCC AR5 report
As part of the EC-EARTH project, it is anticipated that approx. 200 TB of data will be generated on ICHEC systems, to be used by EC-EARTH consortium members and the wider climate community. To store this data, ICHEC is working in collaboration with e-INIS to store this data and provide rapid light-path access to Irish and International academic partners, at speeds of up to 10 Gbps. A web portal is being added to the data in conjunction that will enable climate researchers to discover this data from around the world, and presenting the data in a number of formats for public access by tools such as Google Earth.
HIRLAM is the High Resolution Limited Area Model, developed in cooperation by a number of Weather Services including Met Éireann. It is a numerical short-range weather forecasting system, initiated in 1985, used in routine weather forecasting in a number of countries' services, including DMI, FMI, IMS, KNMI, met.no, INM, and SMHI.
The collaboration with Met Éireann includes running the operational HIRLAM forecast on ICHEC systems, but also scientific and technical development of the forecast models.
Harmonie is a weather code based on HIRLAM and ALADIN. ALADIN is a spectral limited-area model developed at Meteo-France based on the ARPEGE/IFS model, and HARMONIE uses a combination of ALADINs non-hydrostatic dynamics and HIRLAM physics to create mesoscale model with a horizontal grid-scale of 2.5km. Harmonie is currently under evaluation as a next-generation replacement for HIRLAM, and is run daily at ICHEC for a 24-hour forecast.
We are now implementing the latest version of HARMONIE, Cycle 35h1 based on AROME physics. This is expected to take advantage of the greater computing power provided by stokes for more detailed forecasts.
The HIRLAM and ALADIN consortia aim to integrate the knowledge, experience, and results from these activities, and incorporate them in an operationally feasible distributed ensemble forecasting system. The major challenge for this system is to provide reliable probabilistic forecast information, on the short range (up to 60h), at spatial resolution of 10-20 km and particularly suited for the probabilistic forecasting of severe, high impact, weather.
To do this, GLAMEPS is a Grand Limited-Area Ensemble system. Individual countries from HIRLAM and ALADIN will each produce a subset of ensemble members in a variety of manners. Results from each member will be exchanged in real-time between GLAMEPS participants and added together in a common statistic for probabilistic forecasting. It is anticipated that this will produce ensemble forecasts based on a range of perturbation methods from ~7 HIRLAM countries and ~10 ALADIN countries.
In late 2008 ICHEC was one of the first HIRLAM member countries to run GLAMEPS elements on behalf of Met Éireann, taking advantage not only of the existing operational infrastructure, but also the greater network bandwidth and availability of storage that ICHEC can provide. The large compuataional power available at ICHEC may also be harnessed when severe storms appear likely, running a larger ensemble of forecasts, and hence giving better information on the range of possible outcomes. Work is underway to run GlamEPS semi-operationally in 2009-2010.