EC-EARTH: developing a European Earth System model based on ECMWF modelling systems

Principal Investigator(s)

Dr W. Hazeleger
Royal Netherlands Meteorological Institute (KNMI)
PO Box 2013730
AE De Bilt
The Netherlands

Wilco.Hazeleger@knmi.nl

Other researcher(s): ECMWF: Simona Stefanescu. KNMI: Camiel Severijns, Frank Selten, Richard Bintanja, Bart van den Hurk, Andreas Sterl. EC-EARTH consortium representatives: (SMHI: Klaus Wyser; Met Eirean: Tido Semmler; IM: Pedro Viterbo; ETH: Trude Storelvmo)

Project description

Climate and forecasting applications share a common ancestry and also build on the same physical principles. Nevertheless, climate research and forecasting are commonly seen as different applications. The concept of “seamless prediction” (cf. WCRP) is emerging to forge forecasting and climate change into a joint topic. The concept also extends the focus of the physical climate system toward a comprehensive view of the Earth System in which feedbacks with the biosphere are included.

The need for an Earth System model is recognized by various ECMWF Member States (MS). A consortium with representatives from 11 MS is formed in which the seasonal forecast system of ECMWF (system 3) is developed into an earth system model. The project, which is named EC-EARTH, has as general objective to develop a global Earth System model consisting of a state-of-the-art atmospheric general circulation model, a state-of-the-art ocean general circulation model, a sea-ice model, a land model, and an atmospheric chemistry model (see http://ecearth.knmi.nl).

Currently, the consortium is extending System 3 (based on IFS cycle31r1, T159L62) with the NEMO ocean model (1 degree horizontal resolution), OASIS3 coupler and the LIM sea ice model. Also, the H-TESSEL land module is being merged back from cycle32 in cycle31 and alternative components are being added (new sea-ice modules, TM chemistry transport module etc). Furthermore, improvements of the atmospheric component are being developed, such as indirect aerosol effects. The EC-EARTH model will be used for various national and European projects (e.g. EU-FP7 THOR). A prime outcome of the system will be experimental decadal predictions, as deliverable for European Framework Program projects and input for national adaptation strategies, and data for Coupled Model Intercomparison Projects that form the basis of climate change assessments.

Two activities are currently carried out in a distributed fashion

• The consortium is tuning the current model system to reduce biases in the simulated climate
• The consortium is developing new components to be added to the EC-EARTH System in a coordinated fashion

A standard version of the EC-EARTH system is maintained at ECMWF in a central repository using ECMWFs version management system Perforce . Modifications from different MS are collected and combined to new versions. This requires extensive experimentation to merge the modifications and additional tuning to reduce the biases and to optimize the code. Prime biases which will be addressed are: temperature biases over midlatitude continents, sea level pressure biases over the North Pacific and North Atlantic and associated bias in the zonal flow, bias in precipitation over the warm pool regions (see http://ecearth.knmi.nl/CY31_obssstT159.html). A typical run that is needed for testing impact of modifications is 10-years long (to rule out internal variability) when the sea surface temperature is prescribed and much longer when the system is coupled to an ocean model and sea ice model.

Within this Special Project we will perform the following activities:

• 2008: Test modifications, merging and testing new components into EC-EARTH in an iterative strategy
• 2009: Test modifications, merging and testing new components and start with implementation of System 4 as next base version of EC-EARTH
• 2010: Modifying System 4, test modifications, merging and testing new components in a iterative strategy

Currently, a T159L62 IFS cycle31r1 run without the ocean component for 10 years costs 15000 SBU. We estimate that at least 1000 years (ie 100 runs per calendar year) is necessary for tuning runs: 1500000 SBU. We request 1oooooo per year out of Special Project resources. The Global Climate Division of KNMI, which leads the project, has 600000 SBU of Member State resources at its disposal and will use part of these resources as well for this project. We expect to run at higher resolution in 2010 with the prototype System 4 as new base version of EC-EARTH, hence the increase in requested SBUs.

It is expected that the efforts will be beneficial for the performance of later cycles of the ECMWF modelling system. Also, new physical aspects will be brought into the ECMWF system by the MS by developing EC-EARTH.

Additional information

Project started in 2008

Would accept support for 1 year only, if necessary.