Experience with ROMS for Downscaling IPCC Climate Models

2008 ROMS/TOMS European Workshop, Grenoble, 6-8 October

Bjørn Ådlandsvik, Paul Budgell, Vidar Lien

Institute of Marine Research,Bjerknes Centre for Climate Research

Bergen, Norway

Contents

Background Climate models and downscaling Critique North Sea downscaling Intermediate scale downscaling and the

Barents Sea Conclusions

Background

Norway's two largest export industries, petroleum activity and fisheries, are based on the continental shelves and are heavily influenced by climate.

There is therefore considerable interest in the development of future climate scenarios for ocean climate in the North Sea and Barents Sea.

Global climate models

AOGCM Atmosphere/Ocean General Circulation Model Atmosphere/Ocean Global Climate Model

IPCC AR4, approx. 20 AOGCMs

Regional use of AOGCMs

AOGCMs lack necessary resolution for shelf and coastal seas Topographic features Coastal upwelling Mesoscale eddies Exchange shelf sea – deep ocean

AOGCMs lack important physics for coastal seas Tidal mixing

Dynamical downscaling

Force a regional model with atmospheric forcing and lateral boundary description from an AOGCM.

Presently: One-way coupling, no feedback to AOGCM

Intent: Provide consistent high resolution regional

scenarios of present and future climate, for use in marine ecological effect studies

Forcing scenarios

20C3M 20th Century Climate in Coupled Models Historical greenhouse gas concentrations Used for validation and control

A1B “Moderate” greenhouse forcing for 21th century The most widely used IPCC scenario

Critique

The results from the AOGCMs are not good enough for regional studies

Downscaling can't improve errors in atmospheric circulation affecting the regional scale

Downscaled results may therefore be badly biased, and may give misleading input to further assessment

North Sea Downscaling Validation: 20C3M, 1972-1998

Ådlandsvik and Bentsen, Ocean Dynamics 2007 Future scenario: A1B, 2072-2098

Ådlandsvik, TellusA, 2008 AOGCM:

Bergen Climate Model, BCCR-BCM Ocean component: MICOM Atmosphere: ARPEGE

Regional Ocean Model ROMS

Model domain

Model setup Atmospheric forcing

Daily averaged BCM surface fluxes

Ocean lateral boundary forcing Monthly averaged BCM fields 8 tidal constituents Boundary scheme: FRS + Flather/Chapman

Fresh water Climatological run-off modulated by BCM precipitation Baltic = large river, salinity = 18 Relaxation of Sea Surface Salinity towards BCM

Sea surface temperature average for March 1978

BCM ROMS Climatology

ROMS SST Change

North Sea Conclusions I

BCM does a good job with integrated values for the North Sea

Some problems due to low resolution but also isopycnal coordinates on shallow shelf sea.

Downscaling works technically, with a factor ten in resolution.

North Sea Conclusions II

Downscaling provides added value by improving the BCM results where most needed Improved regional details, incl. Coastal Current Improved Atlantic Inflow Improved winter temperature Improved vertical structure, incl. surface salinity:

Conclusions III

Future scenario: Warming of the North Sea, maximum in winter

Yearly mean: BCM +1.0°C, ROMS +1.4°C Slightly increased Atlantic Inflow,

max increase in August Yearly mean: +0.2 Sv = +15 %

Problem in the Barents Sea

Most AOGCMs have too much ice in the 20th century climate

When the ice dissapeares in future climate, it gives an unrealistic warming

The atmosphere behaves very differently if it has sea ice or open water at surface boundary

With no feedback to the atmosphere, the fluxes lead to strong cooling and too much ice in the regional model

Intermediate downscaling

Atlantic-Arctic domain Get the open

boundaries far away from ice-infected area

Stretched coordinates Resolution 10 km in

Nordic Seas

Model set-up

Daily atmospheric forcing from GISS AOM, one of the three best models for ice in Arctic and Barents Sea. (Overland and Wang, 2007)

Control: 20C3M, 1981-2000 (incl. 5 y spinup) Initial and boundary states from SODA, climatology

for 1981-2000. Future: A1B, 2046-2065 (incl. 5 y spinup)

Initial and boundary = SODA + GISS future - GISS present

IMR

GISS AOMGISS AOM

Temperature at 100 m - MarchTemperature at 100 m - March

ROMSROMS SODASODA

GISS AOMGISS AOM

Sea Ice Concentration - MarchSea Ice Concentration - March

ObservedObserved

GISS AOM - ObsGISS AOM - Obs

ROMS ROMS - Obs

Barents Sea Winter Temperature

20C3M A1B A1B - 20C3M

Barents Sea Summer Temperature

20C3M A1B A1B - 20C3M

Barents Sea Ice Concentration

Barents Sea Conclusions

Downscaling provide added value by More realistic temperature and ice cover But: Still problems with ice in east

Future climate Slight warming in western Barents Sea Unrealistic strong warming in eastern Barents Decreased inflow from 2.2 to 2.1 Sv, Small change in heat transport

Overall experience

Downscaling provides added value regionally to results from AOGCMs.

Regional results must be validated for 20th century.

Results must be used with care, in particular where the validation fails.

For robust conclusions we need to downscale from several AOGCMs.