awi cmip peru_nomv

AWI Climate Model in CMIP6 simulations

Outline

• Motivation and background• FESOM/ECHAM6 (AWI-CM) description

and validation• CMIP6 and HighResMIP• FESOM setup for CMIP6• Conclusions and outlook

Motivation Models: describe climate state in space and time! Classical approach employs regular meshes:

cheapdynamics is poor

rich dynamicsexpensive

coarse fine

cheapboundary exchange

downscaling

Local refinement

MPIOM setups focused on different regions (Sein et al., 2015)

9 km resolution in the Arctic, 25 to 100 km elsewhere

Allows for multi-decadal global integrations with well-resolved Arctic

With the same number of nodes regular grids allow 60-70 km resolution

Advantageslocal refinement(s) in a global model realistic representation of small-scale features, e.g.

• narrow straits, polynias and overflows• topography

Unstructured mesh approach

Finite Element Sea Ice Ocean Model (FESOM)

Solves:hydrostatic primitive equationssea ice equations

Uses Finite Element method: continuous linear basis functionstriangles in horizontaltetrahedra (or prisms) in vertical

Influence of local refinement in the Arctic

9 km

25 km

kinetic energy temperature at 300m

90,000 surface nodesT63

ECHAM6 FESOM

1990 constant radiative conditions

Coupling with ECHAM6 (AWI CM)

Coupling with ECHAM6

Model validation: 2m Temperature

Model validation: Total precipitation

90,000 surface nodes 130,000 surface nodes

ocean resolution! WORKS !

ECHAM6–FESOM: role of resolution

1990 constant radiative conditionsbias in potential density at 1000m depth

! the deep bias is reduced !

before after

ECHAM6–FESOM: role of resolution

Summary 1

• Same biases as in CMIP5 models• What is the source of the bias?• Local refinement helps• Key regions to focus on?

CMIP6: Overarching questions

– How does the Earth system respond to forcing?– What are the origins and consequences of systematic model biases? – How can we assess future climate changes given internal climate variability, predictability, and uncertainties in scenarios?

Eyring et al., 2016 (GMD)

CMIP6

CMIP6: Organization

A handful of common experiments are the entry card to CMIP6:

DECK (Diagnostic, Evaluation and Characterization of Klima)CMIP historical simulations (1850 to 2014)

Common forcing and data standards, coordination, documentation, infrastructure

Eyring et al., 2016 (GMD)

CMIP6: Organization• Ensemble of CMIP-Endorsed Model

Intercomparison Projects (MIPs)• Depending on scientific interest modelling groups may or

may not take part in some or all of them

• AWI:• ScenarioMIP• OMIP (Ocean)• PMIP (Paleo climate)• HighResMIP (High Resolution)• CORDEX (Coordinated Regional Climate Downscaling

Experiment) – only diagnostic• SIMIP (Sea Ice) – only diagnostic

CMIP6: Entry card experiments• DECK

• AMIP: 1979-2014 with observed SST and SIC (Sea Ice Concentration) and observed forcings (anthropogenic including greenhouse gases and aerosols, volcanic, solar) MPI

• Control: 500 years 1850 forcing (after spin-up)• Abrupt 4*CO2 (150 years)• 1% CO2 increase / year (150 years)

• CMIP6 historical• Observed forcings 1850-2014

Entry card for CMIP6

CMIP6: ScenarioMIP• RCP‘s 2.6, 4.5, 6.0, and 8.5 • 2015 to 2100

CMIP6: HighResMIP• Short spin-up from EN4 climatology

(averaged over 1950 to 1954): ≈ 50 years with constant 1950 forcing

• 100 years with constant 1950 forcing as a reference

• 100 years with observed forcing up to 2014 and RCP8.5 forcing up to 2050

CMIP6 standard resolution and high resIn PRIMAVERA extra high res

HighResMIP

• 17 international modelling groups• Atmospheric resolution: T127 (~100km), T255 (~50

km), T359(~35 km), T511(~25km), T799 (~16km)• Ocean resolution from 1º to 0.25º. Except for

FESOM• Two coupled simulations for 1950-2050 (control and

transient)

º

CMIP6: HighResMIP• Short spin-up from EN4 climatology

(averaged over 1950 to 1954): ≈ 50 years with constant 1950 forcing

• 100 years with constant 1950 forcing as a reference

• 100 years with observed forcing up to 2014 and RCP8.5 forcing up to 2050

CMIP6 standard resolution and high resIn PRIMAVERA extra high res

• The flexible layout of AWI-CM (Sidorenko et al. 2015, Rackow et al. 2016) allows to use eddy-resolving resolutions in key ocean areas. We exploit this capability in the North Atlantic (NA) in order to reduce long-standing biases, specifically the deep (~1000m) biases

Hierarchy of ocean model grids

reso

lutio

n

[km

]

REF87KCORE

AGUV

GLOB

top: Hierarchy of different ocean model grids. REF87K and CORE use ~1° resolution and moderate refinementto about 25km in the tropics and in the Arctic. AGUV and GLOB focus on the Agulhasand North Atlantic current region, with different weighting between those regions.

Improvements in NA deep-ocean hydrography

REF87K

CORE

AGUV

GLOB

REF87K

CORE

AGUV

GLOB

right: Difference of potential temperature and salinity at 1000m to the WOA2005 climatologybelow: Global profiles of potential temperature and salinity, difference to the WOA2005 climatology

T127-GLOB

T127-AGUV

T127-CORE2

T63-REF87K

temperature

salinity

pot. temperature salinity

What are the ocean key regions?

• Oceanic fronts • Regions of eddies activity• Deep water production• Polar regions (sea ice)• Straits• ???

Fronts. Observed (AVISO) SSH

Eddies activity: SSH variance (AVISO)

FESOM mesh resolution

FESOM HR local examples (Europe)

FESOM HR local examples (GoM)

SSH variance. Agulhas system

North Atlantic Ocean deep bias

Temperature (K)

Outlook. Frontier meshRossby radius (Hallberg, 2013)

Conclusions• AWI-CM with T63 (~200km) atmospheric resolution

shows results similar to most of the CMIP5 climate models

• The model development for the CMIP6 simulations requires farther validation with T127 (~100km) and T255 (~50 km) atmosphere (ECHAM6)

• The ocean model resolution can play a crucial role in reduction of model biases

• The flexibility of FESOM (in the sense of horizontal resolution) could answer the question where and how to choose the ocean resolution in climate models.

Thank you!