Icon, Combine - COSMOS

The COSMOS framework
ICON
COMBINE
COSMOS annual meeting, Helsinki, 2008
Marco Giorgetta

.
Content
� ...
� Motivation
� Status
� Developments
� Model names

.
Motivation
Which are driving forces in model development?
� New scientific questions (Projects)
� Different scales in time and/or space
� New process models
� New numerical techniques
� New experimental designs
� New environment where to use models
� Portability
� Optimization
� New software techniques
� Data structures
� Parallelization
� Domain decomposition

.
Status
COSMOS “Millennium”
� SCE+SRE:
� Models:
� “cosmos-a” � ECHAM5 =
atmosphere
� “cosmos-o” � MPIOM = ocean
� “cosmos-ob” � MPIOM-HAMOCC= ocean + biogeochemistry
� “cosmos-ao” � ECHAM5/MPIOM = atmosphere ocean
� “cosmos-asob” � ECHAM5-JSBACH/MPIOM-HAMOCC
= carbon cycle
� Resolutions:
� T31L19/GR3L40
� T63L31/GR1.5L40 (tests, spin-up)
� Computers/Sites
� DKRZ, FMI, …
� Codes:
� ECHAM5.4.01, ECHAM5-HAM, ECHAM5J
� MPIOM
� HAMOCC
� OASIS3

.
Developments
Ongoing
� Higher resolution version (T63 L31 / GR1.5 L40) of the Millennium
model
� Inclusion of HAM in standard ECHAM5 code
� Implementation of CDNC cloud scheme of U. Lohmann for aerosol
clouds interaction
� Chemistry aerosol cloud coupling (“HAMMOZ”)
� SALSA aerosol scheme
� Dynamical vegetation and land use
� Tri-polar higher resolution ocean grid TP0.4, TP0.1

.
Developments
Further developments
� Radiation: Test new radiation package of ECMWF including
RRTM-LW, RRTM-SW and McICA cloud generator?
� Fast chemisty (family concept)
� Methane processes (wetlands, permafrost)
� Nitrogen cycle
� Isotopes
� Compiling system (Autoconf)
� Running environment
� Coupler: OASIS3 � OASIS4

.
Priorities
� Technical Documentation
� High resolution carbon cycle model � Millennium
� Aerosols � Volcanoe
� Fast chemistry for century time scale experiments
� Detailed chemistry for shorter time scales / campaigns
� Optimization for use on new DKRZ computers

.
Other presentations related to specific models:
� Johann Jungclaus Millennium
� Harri Kokkola Salsa
� Martin Schulz ECHAM5-HAMMOZ
� Malte Müller MPIOM
� Christian Reick JSBACH
� Joachim Segschneider HAMOCC

.
Model names (1/2)
� As more configurations become possible, names become longer
and more complicated
� Names are used in the configuration process to construct model
and scripts
� Current names: “COSMOS-XYZ”
“XYZ” shows components
� A: Atmosphere dynamics+physics � ECHAM5
� S: Surface, soil and vegetation � JSBACH
� O: Ocean dynamics and physics � MPIOM
� B: Ocean Biogeochem. + sediments � HAMOCC
� Example: COSMOS-ASOB is the carbon cycle model configuration
used for the Millennium project

.
Model names (2/2)
� The model complexity is growing further
� Aerosols HAM, SALSA, SAM2
� Chemistry MOZART (, MECCA, (CHEM2))
� Add more letters for model configuration?
� M for aerosol microphysics
� C for chemistry
� COSMOS-[A, C, M; S; O, B]
� Problem: single letters not sufficient to describe specific model
setups, e.g. Because of different aerosol and chemistry schemes
� Alternative: Use project names to tag and name model
configurations for specific projects
Examples: “IPCC model”, “Millennium model”

ICON
Icosahedral Non-hydrostatic General Circulation Models
A joint MPI-M DWD project
ICON will provide future dynamical cores and
atmosphere and ocean GCMs for COSMOS

MPI-M & DWD collaboration
� Benefits of MPI-M from DWD
� operational mode (daily testing & verification),
� comparing against observations: atmospheric data assimilation
� Benefits of DWD from MPI-M
� seasonal prediction (including ocean model)
� experience with stratosphere
� Joint pool of physics packages

Model characteristics
� Joint grid – joint numerics (at least operators, solvers)
� Structure-preserving-discretization
� conserve relevant quantities on discrete level
(energy, Ertel’s potential vorticity, total mass, tracer mass)
� Zooming ability
� Modern software architecture
� object-orientated programming
� Keep the model “open”
� Encourage/facilitate model adaption & extension by users
� High scalability for next generation computers

.
Grid generator
Concept of patches
� for refinement
� for domain decomposition
Unstructured grid
� minimize distance between neighbors
in memory
� only relationships between neighbors
are stored: no traditional array data
structure

Basic operators on the C-grid: div-grad-curl
Rotation for hexagons:
Stokes theorem
The velocity is given counterclockwise.
Application of integral theorems
Divergence for triangles:
Gauss theorem
The velocity points outward.
Gradient at an edge:
given by the relation

ICON Models, work in progress
Grid generator
– Global or regional triangular grids derived from icosahedron
– Optimization to improve accuracy of numerical operators
Discretization and operators
- C-grid discretization
- Div, Grad, …
Infrastructure
- Data structure allowing regional refinements, domain composed of patches
- Parallelization (MPI & OpenMP)
- Asynchronous and parallel I/O
- Encoding (NetCDF, GRIB)
- Postprocessing (CDO)
Dynamical cores
- Shallow water model (ICOSWM)
- Vertical slice model
- Hydrostatic atmosphere
- Hydrostatic ocean
- Non-hydrostatic atmosphere
Earth system model / COSMOS
- OASIS
- ICOxAM
- ICOHOM
General circulation models
- Hydrostatic atmosphere (ICOHAM)
- Hydrostatic ocean (ICOHOM)
- Non-hydrostatic atmosphere (ICONAM)

COMBINE:
Comprehensive Modelling of the
Earth System for Better Climate
Prediction and Projection
Call: FP7, 2nd call (29 Nov.-25 Feb.)
Topic: New components in Earth System modelling for better climate
projections
Funding scheme: Large-scale integrating project, 4-8 M€

.
COMBINE Abstract
� The European integrating project COMBINE brings together research groups to
advance Earth system models (ESMs) for more accurate climate projections and for
reduced uncertainty in the prediction of climate and climate change in the next
decades. COMBINE will contribute to better assessments of changes in the physical
climate system and of their impacts in the societal and economic system. The
proposed work will strengthen the scientific base for environmental policies of the
EU for the post-2012 climate negotiations.
� COMBINE proposes to improve ESMs by including key physical and
biogeochemical processes to model more accurately the forcing mechanisms and
the feedbacks determining the magnitude of climate change in the 21st century. For
this purpose the project will incorporate carbon and nitrogen cycle, aerosols
coupled to cloud microphysics and chemistry, proper stratospheric dynamics and
increased resolution, ice sheets and permafrost in current Earth system models.
COMBINE also proposes to improve initialisation techniques to make the best
possible use of observation based analyses of ocean and ice to benefit from the
predictability of the climate system in predictions of the climate of the next few
decades. Combining more realistic models and skilful initialisation is expected to
reduce the uncertainty in climate projections.
� Resulting effects will be investigated in the physical climate system and in impacts
on water availability and agriculture, globally and in 3 regions under the influence of
different climate feedback mechanisms. Results from the comprehensive ESMs will
be used in an integrated assessment model to test the underlying assumptions in
the scenarios, and hence to contribute to improved scenarios. COMBINE will make
use of the experimental design and of the scenarios proposed for IPCC AR5.
Therefore the project will be able to contribute to the AR5, by its relevant research
and by the contribution of experiments to the IPCC data archives.

.
COMBINE – Abstract (1)
� COMBINE proposes to improve ESMs by including
key physical and biogeochemical processes to
model more accurately the forcing mechanisms and
the feedbacks determining the magnitude of climate
change in the 21st century. For this purpose the project
will incorporate carbon and nitrogen cycle, aerosols
coupled to cloud microphysics and chemistry,
proper stratospheric dynamics and increased
resolution, ice sheets and permafrost in current
Earth system models.

.
COMBINE – Abstract (2)
� COMBINE also proposes to improve initialisation
techniques to make the best possible use of
observation based analyses of ocean and ice to benefit
from the predictability of the climate system in
predictions of the climate of the next few decades.
� Combining more realistic models and skilful initialisation
is expected to reduce the uncertainty in climate
projections.

.
COMBINE – Abstract (3)
� Resulting effects will be investigated in the physical
climate system and in impacts on water availability
and agriculture, globally and in 3 regions under the
influence of different climate feedback mechanisms.
Results from the comprehensive ESMs will be used
in an integrated assessment model to test the
underlying assumptions in the scenarios, and hence
to contribute to improved scenarios. COMBINE will
make use of the experimental design and of the
scenarios proposed for IPCC AR5. Therefore the
project will be able to contribute to the AR5, by its
relevant research and by the contribution of
experiments to the IPCC data archives

.
New components Experimental design
New components
(C1) Carbon an nitrogen cycle
(C2) Aerosols, clouds and chemistry
(C3) Stratosphere
(C4) Cryosphere
(C5) Initialisation
ESM combinations Differences
(E1) ESM D Α (C(i)) = (E2) – (E1)
(E2) ESM + C(i) D Ω (C(i)) = (E4) – (E3)
(E3) newESM – C(i) D Σ (Σ j C(j)) = (E4) – (E1)
(E4) newESM
CMIP simulations, centennial and decadal
Centennial Simulation Decadal Simulation
(CS1) Pre-industrial control (DS1) Climate prediction (2005-2035)
(CS2) 20 th century (DS2) Climate hindcasts
(CS3) 21 st century scenario
(Representative Concentration Pathway scen.)
(CS4) +1% CO 2 / year to 4xCO 2

.
ESMs and main developers
ESMs Main developer
(M1) COSMOS MPG
(M2) HadCM,
HadGEM
METO
(M3) IPSL-ESM CNRS
(M4) CMCC CMCC
(M5) CNRM-CM MF-CNRM
(M6) EC-EARTH ECMWF, EC-Earth consortium
(M7) NORCLIM UiB
Work packages
New components CMIP/AR5 + Evaluation
(WP1) C and N cycle (WP6) Decadal climate
prediction
(WP2) Aerosols, clouds and chemistry (WP7) Climate projections
and feedbacks
(WP3) Stratosphere (WP8) Impacts and Scenarios
(WP4) Cryosphere
(WP5) Initialisation

.
COMINE Pert diagram
.

.
Partners 1-11 of 23
Number* Short
name
1
(Coord.)
Organisation name Cou
ntry
MPG Max-Planck-Gesellschaft
(MPI-M + MPI-BGC)
2 METO Met Office UK
3 CNRS Centre National de la Recherche
Scientifique
4 CMCC Centro Euro-Mediterraneo per i
Cambiamenti Climatici
5 MF-CNRM Météo-France - Centre National de
Recherches Météorologiques
6 KNMI Het Koninklijk Nederlands
Meteorologisch Instituut
7 UiB University of Bergen NO
8 DMI Danish Meteorological Institute DK
9 ECMWF European Centre for Medium-Range
Weather Forecast
10 ETHZ Eidgenössische Technische
Hochschule Zürich
11 FMI Finnish Meteorological Institute FI
DE
FR
IT
FR
NL
UK
CH

.
Partners 12-23 of 23
Number* Short
name
Organisation name Cou
ntry
12 MNP Netherlands Environmental Assessment
Agency
13 SMHI Swedish Meteorological and
Hydrological Institute
14 WU Wageningen University & Research
Centre
15 UHEL University of Helsinki FI
16 CERFACS European Centre for Research and
Advanced Training in Scientific
Computation
17 UCL Université Catholique de Louvain BE
18 UNIVBRIS University of Bristol UK
19 Uni Kassel University of Kassel - Center for
Environmental Systems Research
(CESR)
20 TUC Technical University of Crete GR
21 CYI Cyprus Research and Educational
Foundation
22 INPE Instituto Nacional de Pesquisas
Espaciais
NL
SE
NL
FR
DE
CY
BR

. � Status:
� Evaluation done
� So far no offical message, but some rumours
� Invitation for negotiation in September or October
� Perspective:
� Chance for interesting research
� Progress in the development of the COSMOS ESM

.
END