Climate change scenarios and impacts assessments for North Atlantic from ICES

and other European groups

Ken DrinkwaterInstitute of Marine Research and Bjerknes Center for Climate

Research, Bergen, Norway

NW Atlantic Ocean Climate Change Workshop

Bedford Institute, Dartmouth, Canada

February 16, 2010

ICES•Produced some reviews on CC, most recently for OSPAR

•Presently writing a more complete review of climate change and its biological impacts

•ICES wrestling with what its role should be. CC is in the ICES scientific plan. However, most of the work on climate change and its impacts financed by EU and national funding agencies. They are determining more what the priorities are.

•Few physical oceanographers in ICES, and fewer working on global or even regional models for future forecasting.

•Few ICES scientists involved in IPCC process.

EU (Frameworks)•Concerned about CC

•Have been funding work on climate variability and change over last 10+ years. Funding is increasing.

•Concerned about climate scenarios and effects on fisheries and underlying ecosystem. Acidification also becoming of interest. (Humboldt Institute advertised 10 post-doc positions)

•Call in 2008 on Tipping points in the Arctic

•EU Framework 7, more concern expressed about economic opportunities, adaptation, governance issues especially for the Arctic (e.g. call for proposals on economic effects for the Arctic under climate change; interested in transport, resources and fisheries; 11 million euros over 3-4 years; will be 1 successful proposal)

Nordic Council•Concerned about CC

•Funding 5 new Centers of Excellence on climate change; each for 5 years; covers both marine and terristrial environments (guess maybe 1, perhaps 2 in marine environment)

•Funds for post-docs, Ph.Ds, networking

Research Council of Norway•Very concerned about CC, especially on fisheries

•Funding Centers of Excellence on climate change; Bjerknes Centre for Climate Change in Bergen funded for 10 years (runs out in 2012) but recently has been given funds from Norwegian government for unlimited time frame of equivalent funds to CoE.

•Bjerknes Center dealing mostly with physical and chemical aspects of climate change, climate models, regional modelling, some impacts in fisheries, also on energy, coastal regions.

•Many relevent calls for proposals on climate change and its impacts. Funds for post-docs, Ph.Ds, scientist’s time. Need collaboration with many institutions.

Predicting Future Ecological Changes

1 to 2°C Temperature Increase

Year 1

Year 30Year 20

Year 10 Likely Polar cod retreats from subarctic into the Arctic

Chung et al., 2008, UBC Report

Problems with Bioclimate Envelope Models

•Assumption that everything else stays the same

•Only deal with one species at a time – no consideration of prey, predators, competitors

•Does not deal with connectivity of life history stages.

Future zooplankton production -Barents SeaFuture zooplankton production -Barents Sea2045-20541995-2004

Production increases in Atlantic Waters

Ellingsen et al. (2008)

Production decreases in Arctic Waters

Capelin Spawning in Response to Climate Change Capelin Spawning in Response to Climate Change

Huse and Ellingsen, 2008

Present Spawning

Future Spawning

Direction of distributional shift of adult feeding migration

North Atlantic and Arctic Regional Model

• Based on ROMS• Downscaled from GISS model (used in IPCC (2007) and recommended by Overland and Wang (2007) because of reasonable representation of sea ice)• Took atmospheric forcing from last 20 years of 20th Century control run from GISS AOM with 3 x 4 deg lat-long grid. Model Domain

Melsom et al., 2009

Observed Modelled

T

S

North Sea Regional Model• Based on ROMS• Downscaled from Bergen Climate Model (used in IPCC (2007)• Took atmospheric forcing from last 20 years of 20th Century control run from GISS AOM with 3 x 4 deg lat-long grid.

Mean winds from BCM 1972-1997 (blue) and 2072-2097 (red).

Lessons learned from downscaling

• Need to choose GCM to downscale from carefully – one that does a reasonable job in hindcasting recent conditions.

• Need to downscale from more than one GCM, preferably several and create ensemble mean from the results.

• Need to couple atmospheric and oceanic models.

• Regional results will reflect GCM results (Jacob et al., 2007).

•Therefore, need to improve GCMs.

• For projections over the next 30+ years need to start with present conditions.

• Dr. Kevin Trenberth (head of Climate Analysis Section, NCAR) states:

(...) None of the models used by IPCC are initialized to the observed state and none of the climate states in the models correspond even remotely to the current observed climate. In particular, the starting state of the oceans, sea ice, and soil moisture has no relationship to the observed state at any recent time in any of the IPCC models.(...)

http://blogs.nature.com/climatefeedback/2007/06/predictions_of_climate.html

IPCC GCMs: Limitations

No initialisation to the present state (particular problematic for the ocean)

(…) I postulate that regional climate change is impossible to deal with properly unless the (global climate) models are initialized (to the current state).

Some Present IPCC GCMs Limitations

Small to large (1-3) scales: mixing and turbulence, friction, waves, clouds, marine optics

Arctic sea ice conditions not well represented

Tides, tidal variability has the potential to impact significantly on climate: e.g. (W.Munk et al., 2001)

No variability considered (temporal, spatial) in tidal forcing (IPCC, 2007)

Generally poor representation of ENSO, NAO, etc.

Some general statementsSome general statements

•Earlier migrations northward, later retreats southward

•Species at southern limits of their geographic distribution will decrease while those at northern limit may do well

•Growth rates expected to increase in general but depends upon what happens to prey

•Difficult to forecast because of new interactions between species, lack of knowledge of processes

Conclusions 1

• Uncertainty due to global, regional and biolgical

models is (much?) larger than the signal to study

• Need to develop uncertainty estimates of future

scenarios

• Improved understanding of processes and better

paramaterization of the models

• Develop models that include fish and fisheries

Conclusions 2

• IPCC scenario model predictions (and consequently the

RCMs based on these) are only of limited use for regional

climate change assessment

• Presently might be able to learn as much performing

controled sensitivity tests with validated regional models

• Decadal scale predictions from GCMs might provide

improved forcing data, but loose performance after approx.

1 decade (??)

• Inspite of difficulties need to“get on with it“.

Other points

1. We should not worry about failure – it points to our lack of understanding or incorrect paramaterization. We only learn when we are making mistakes.

2. Expect surprises!

...and for BergenThank you!

Bergen expects more of this under climate change!

Cod Recruitment and TemperatureCod Recruitment and Temperature

Mean Annual Bottom Temperature11

10

9

8

7

6

4

3

2

Temp

Warm Temperatures

decreases Recruitment

Warm Temperatures

increases Recruitment

Recruits

Planque and Fredou (1999)

R2 = 0.75

-1.5

-1

-0.5

0

0.5

1

1.5

2

0 2 4 6 8 10 12

Bottom Temperature

d(R

ecru

itm

ent)

/dT

If BT < 5° and T warms stock recruitment generally increase

If BT between 5° and 8.5°C little change in recruitment

If BT >8.5°C recruitment generally decreases

If BT 12°C we do not see any cod stocks

GB

Effect on abundance of 1°C increase

Increase

No change

Decrease

Collapse

?

2-3°C Temperature Increase

3-4°C Temperature Increase

Possible effect of global warming and shut down of the Atlantic thermohaline circulation

Wood et al. 2003

Low probability – high impact

What do the models suggest?• Large

uncertainty

• Most climate models produce 20-30% reduction in the strength of the AMOC

• The associated reduction in the poleward transport of heat is less than the atmospheric warming

Food web in Atlantic waterFood web in Arctic water

Changes in ecosystem function (Barents)Changes in ecosystem function (Barents)

The food web changes may be far more dramatic for the higher compared to the lower trophic levels

Falk-Petersen et al. 2007

There is the possibility that boreal species from the Arctic and Pacific will move into the Arctic and may start to mix.

What do we need to Improve Predictions of

Future Ecological Changes?

What do we need?

•More emphasis on quantitative estimates (mechanistic modelling)

•Better understanding of the processes

•Improved parameterization of the models

•Regional models

•Measure of uncertainty

What is the role of observationalists?

- Observationalists and Modellers need to work closer together

- Modeller’s to help determine what, where and how often observationalists should measure.

- Observationalists should provide more feedback on model results (requires available model results, positive criticisms)

- All motherhood statements but not generally done (improving but do we need formal procedure?)

Need Comparative Model Studies

1. Ecosystems are complex – helps determine what is a fundamental process and what is unique.

2. Provides insights that one cannot obtain by looking at a single ecosystem

3. Single model applied to several ecosystems, different models applied to single ecosystem

4. Sharing modelling approaches

Need to include Fishing Effects

Models need to be used to determine interaction between climate and fishing and explore effects of management strategies on stocks under climate change scenarios.

Fishing and ClimateMultivariate autoregressive models of Baltic cod under increasing salinities.

Martin Lindegren, Ph.D. Student, DTU Aqua, Copenhagen

ST error [o C] Mod-Obs Chapter 8, IPCC, 2007

Surface Temperature error IPCC model ensemble

Solomon et al., 2007

Validation of global climate models

IPCC, 2007

Drift problems?

Sea-ice problems?

Solomon et al., 2007

Regional Climate Models

One main conclusion: RegCM are critically affected by the driving large-scale fields from GCM and are very similar to GCM results

Regional downscaling

• Only very few coupled ocean-amtosphere models on the regional scale

• Fewer models do ensemble runs using different global models

• Regional models that are available mostly use previous IPCC global model assessments

• Few full dynamics ocean model none reported in the last IPCC

• Few coupled ecosystem models none reported in the last IPCC report

Recently: Improved global decadal predictions by improving initialisation • Doug Smith et al., 2007, Science

• Noel Keenlyside et al., 2008, nature

Related publications indicated the importance of AMO for the NA region - Knight et al. (2005) and Knight et al. (submitted)

Keenlyside et al., 2008

Next IPCC Runs

•Going to Earth System Models only

•Many of the models will be total new with increased number of paramaterizations

•Will this lead to increased spread in model ensemble?

Higher trophic level

dynamics

Futurescenari

oGCMlow

resolution

Regional Modelhigh

resolution

hydrodynamics

Lower trophic

level dynamics

Need Regional Models

IPCC provides Climate Change scenarios from GCMs: present day from 1960-to present and future up to 2100. Multi-Model Dataset (ensemble runs) of climate scenarios (ipcc-data.org)

Use IPCC senario’s for ‘downscaling’ GCMs output by regional models for hydrodynamics (and biota) in regional seas

Developing Regional Impacts of Global Change

Huse & Ellingsen, 2008

Fronts will moveFronts will move2000 2047

Modelled currents and Polar Front (thick black line).