Evaluation of ocean circulation models for the Bering Sea and Aleutian Islands Region

Evaluation of ocean Evaluation of ocean circulation models for the circulation models for the Bering Sea and Aleutian Bering Sea and Aleutian

Islands RegionIslands RegionAlbert J. HermannAlbert J. Hermann11 and David L. Musgrave and David L. Musgrave22

National Oceanic & Atmospheric Administration, Pacific National Oceanic & Atmospheric Administration, Pacific Marine Environmental Laboratory, 7600 Sand Point Way Marine Environmental Laboratory, 7600 Sand Point Way

NE, Seattle, WA 98115 USA, (206) 526-6495, NE, Seattle, WA 98115 USA, (206) 526-6495, [email protected]@noaa.gov.

2 School of Fisheries and Ocean Sciences, Insitute of 2 School of Fisheries and Ocean Sciences, Insitute of Marine Science P.O. Box 757220, Fairbanks AK 99775-Marine Science P.O. Box 757220, Fairbanks AK 99775-

7220 USA, (907) 474-7837, [email protected] USA, (907) 474-7837, [email protected]

This workshop explored the present and This workshop explored the present and future state of ocean circulation modeling future state of ocean circulation modeling and biological modeling of the Bering Sea and biological modeling of the Bering Sea and Aleutian Island (BSAI) and the North and Aleutian Island (BSAI) and the North

PacificPacificMajor topics:Major topics: 1) the present state of knowledge concerning the 1) the present state of knowledge concerning the

BSAIBSAI 2) the various types of circulation models which 2) the various types of circulation models which

could be applied to the BSAI, with some could be applied to the BSAI, with some assessment of their strengths and weaknessesassessment of their strengths and weaknesses

3) existing physical and biological models 3) existing physical and biological models 4) adequacy of present forcing and bathymetry 4) adequacy of present forcing and bathymetry

datasetsdatasets 5) current status and future prospects for data 5) current status and future prospects for data

assimilationassimilation 6) modeling needs of managers for this region6) modeling needs of managers for this region 7) a timetable over which we might expect the 7) a timetable over which we might expect the

development of improved modelsdevelopment of improved models

1. Present state of knowledge of 1. Present state of knowledge of the BSAI regionthe BSAI region

Highly productiveHighly productive A big, broad shelf Passes: wide, narrow, shallow,

deep Canyons Cross-shelf flux Powerful tides A deep basin Ice

A few highlights….A few highlights….

Flows through passes – spatially variable, Flows through passes – spatially variable, strongly mixed, very important to biologystrongly mixed, very important to biology

Ice-edge blooms with possible “Oscillating Ice-edge blooms with possible “Oscillating Control”Control”

Distinct shelf regimes via tidal mixingDistinct shelf regimes via tidal mixing Getting warmer, less ice!Getting warmer, less ice! PDO was significant, but now other modes PDO was significant, but now other modes

more importantmore important

2. Classes of Ocean Circulation 2. Classes of Ocean Circulation ModelsModels

Pure tidal modelsPure tidal models Quasi-gestrophic models (simpler physics)Quasi-gestrophic models (simpler physics) Primitive equation models (hydrostatic but Primitive equation models (hydrostatic but

otherwise include all physics)otherwise include all physics) Terrain-following coordinatesTerrain-following coordinates Z-coordinateZ-coordinate Layered coordinateLayered coordinate Unstructured gridUnstructured grid

3. Existing physical and 3. Existing physical and biological modelsbiological models

3.1 Atmospheric models 3.1 Atmospheric models National Center for Environmental Prediction (NCEP) hindcastsNational Center for Environmental Prediction (NCEP) hindcasts

Assimilated atmospheric dataAssimilated atmospheric data Easy to get!Easy to get! Biases include shortwave radiation (not enough stratus clouds)Biases include shortwave radiation (not enough stratus clouds)

ECMWF hindcastsECMWF hindcasts Better for Europe, not necessarily better for BeringBetter for Europe, not necessarily better for Bering Commercial productCommercial product

Community Climate Systems ProgramCommunity Climate Systems Program Offers global hindcasts and climate forecastsOffers global hindcasts and climate forecasts Better shortwave radiation (b/c assimilates cloud data/climatology)Better shortwave radiation (b/c assimilates cloud data/climatology)

Regional modelsRegional models ETA – downscales NCEP nowcastsETA – downscales NCEP nowcasts NARR – downscales NCEP hindcastsNARR – downscales NCEP hindcasts MM5 – general tool for downscaling global windsMM5 – general tool for downscaling global winds

3.2. Ice models 3.2. Ice models

Hibler model and its decendents:Hibler model and its decendents: Thermodynamics:Thermodynamics:

• Melt/freeze, snow layersMelt/freeze, snow layers DynamicsDynamics

• Viscous-plastic solidViscous-plastic solid

Vary in number of ice/snow layersVary in number of ice/snow layers

3.3 Circulation models 3.3 Circulation models

Maslowski group Maslowski group N Hemisphere model based on MOM/POPN Hemisphere model based on MOM/POP

Chao groupChao group N Pacific model based on ROMSN Pacific model based on ROMS

Wang groupWang group Bering Sea model based on POMBering Sea model based on POM

Curchitser/Hermann groupCurchitser/Hermann group Northeast Pacific model based on ROMSNortheast Pacific model based on ROMS

3.4. NPZ Models 3.4. NPZ Models

3.4.1. 1D models3.4.1. 1D models NEMURO – NPZD and fish NEMURO – NPZD and fish

• Saury and herringSaury and herring Merico (2001) – NPZD Merico (2001) – NPZD

• phytoplankton successionphytoplankton succession

3.4.2. 3D models3.4.2. 3D models Run both “online” and “offline”Run both “online” and “offline” Examples:Examples:

• Wang/Diehl NPZ model of BeringWang/Diehl NPZ model of Bering• Hinckley et al NPZ models of CGOAHinckley et al NPZ models of CGOA• Powell/Hermann NPZ for Northeast Pacific Powell/Hermann NPZ for Northeast Pacific

3.5 Individual-based Models 3.5 Individual-based Models

Float-tracking plus behaviorFloat-tracking plus behavior Very useful for individual fish speciesVery useful for individual fish species Can run online or offlineCan run online or offline Examples (from CGOA)Examples (from CGOA)

Hinckley et al pollock modelHinckley et al pollock model Rand salmon modelRand salmon model

3.6 Aggregated models 3.6 Aggregated models

Consider entire food webConsider entire food web ECOPATH – look at steady stateECOPATH – look at steady state ECOSIM – add timeECOSIM – add time ECOSPACE – add space and timeECOSPACE – add space and time Can benefit from coupling with NPZDCan benefit from coupling with NPZD

3.7 Fisheries models 3.7 Fisheries models

MSVPAMSVPA Other stock assesment modelsOther stock assesment models

4. Adequacy of models, forcing 4. Adequacy of models, forcing and bathymetryand bathymetry

IDEALLY we would likeIDEALLY we would like Uniformly fine scale resolution or adaptive Uniformly fine scale resolution or adaptive

space/time resolutionspace/time resolution Numerically accurate/convergentNumerically accurate/convergent Handle tides, subtidal flows, mixing all Handle tides, subtidal flows, mixing all

togethertogether Perfectly accurate bathymetryPerfectly accurate bathymetry Perfectly accurate forcing; hincasts and Perfectly accurate forcing; hincasts and

forecastsforecasts

4.1 Numerics and resolution 4.1 Numerics and resolution Terrain-following coordinatesTerrain-following coordinates

• Tend to overemphasize bathymetryTend to overemphasize bathymetry• Need smoothed bathymetryNeed smoothed bathymetry• Great for surface and bottom boundary layersGreat for surface and bottom boundary layers

Z-coordinateZ-coordinate• Less accurate/convergent numericsLess accurate/convergent numerics• Consistent vertical spacing near the surfaceConsistent vertical spacing near the surface• Distort bottom topography into stair stepsDistort bottom topography into stair steps

Layered coordinateLayered coordinate• Good for the deep oceanGood for the deep ocean• bad for the shallow ocean (can’t do tidal mixing)bad for the shallow ocean (can’t do tidal mixing)

Unstructured gridUnstructured grid• Potentially powerfulPotentially powerful• Hard to implementHard to implement• Relatively untested in the Bering SeaRelatively untested in the Bering Sea• Danger of predetermining answer with choice of gridDanger of predetermining answer with choice of grid

Crucial elements to get right…Crucial elements to get right…

Inflow/outflow through the Aleutian passesInflow/outflow through the Aleutian passes sets conditions in the Southeast BSsets conditions in the Southeast BS

Outflow through the Bering StraitOutflow through the Bering Strait Tidal mixing on the shelfTidal mixing on the shelf Ice!Ice!

4.2 Ice 4.2 Ice

Hibler-based models are probably Hibler-based models are probably sufficient for the Bering Sea (no multiyear sufficient for the Bering Sea (no multiyear ice)ice)

Major uncertainties arise from shortwave Major uncertainties arise from shortwave radiation forcing; need to improve radiation forcing; need to improve

4.3 Atmospheric forcing 4.3 Atmospheric forcing

NCEP probably OK for winds (except for NCEP probably OK for winds (except for Aleutians)Aleutians)

NCEP shortwave is badly biasedNCEP shortwave is badly biased CCSM is promisingCCSM is promising Need better bulk flux algorithms (e.g. to relate Need better bulk flux algorithms (e.g. to relate

wind speed to wind stress)wind speed to wind stress) Extended range mesoscale forecasts are Extended range mesoscale forecasts are

impossibleimpossible Long range forecasts/scenarios are usefulLong range forecasts/scenarios are useful Downscaling is needed!Downscaling is needed!

4.4 Freshwater discharge 4.4 Freshwater discharge

Important in a few areasImportant in a few areas Data is essentially nonexistent!Data is essentially nonexistent!

4.5 Tides 4.5 Tides

Existing models can handle tidal and Existing models can handle tidal and subtidal dynamics simultaneouslysubtidal dynamics simultaneously

This is crucial for the Bering Sea, as the This is crucial for the Bering Sea, as the two interacttwo interact

Tidal phasing may be biologically Tidal phasing may be biologically important, so want to get it right.important, so want to get it right.

4.6 Bathymetry 4.6 Bathymetry

OK on the shelf OK on the shelf Need more data in the canyonsNeed more data in the canyons Need much more data in the passesNeed much more data in the passes USGS eventually digitizing the Bering Sea USGS eventually digitizing the Bering Sea

chartscharts

4.7 NPZ models4.7 NPZ models

Need to get:Need to get: Pelagic/benthic gradients, north-south and Pelagic/benthic gradients, north-south and

cross-shelfcross-shelf Green Belt at the shelf breakGreen Belt at the shelf break Important prey species for fishImportant prey species for fish Jellyfish?Jellyfish? IRON and other nutrientsIRON and other nutrients

4.8 IBM models 4.8 IBM models

Need better data on fish movement and Need better data on fish movement and behaviourbehaviour

Need better data on space/time Need better data on space/time distributiondistribution

Groundfish surveys have been very useful Groundfish surveys have been very useful for modelers.for modelers.

4.9 Aggregated models 4.9 Aggregated models

Could benefit from NPZ resultsCould benefit from NPZ results Aggregate NPZ by space, water mass, or Aggregate NPZ by space, water mass, or

biological regime?biological regime?

4.10 Fisheries models 4.10 Fisheries models

As with aggregated models, could make As with aggregated models, could make more spatially explicitmore spatially explicit

4.11 Model coupling 4.11 Model coupling

IDEAL integrate model might include:IDEAL integrate model might include: IBMs of Multiple species and life stagesIBMs of Multiple species and life stages NPZ with multiple size classesNPZ with multiple size classes Feedback between IMB and NPZ!Feedback between IMB and NPZ! Long time scale simulationsLong time scale simulations Web-accessible output and graphicsWeb-accessible output and graphics

5. Status, Needs and Prospects 5. Status, Needs and Prospects for Data Assimilationfor Data Assimilation

T,S, nutrients in passes would be powerful constraintT,S, nutrients in passes would be powerful constraint Skill assesment is difficult to do wellSkill assesment is difficult to do well Existing physical assimilation capabilitiesExisting physical assimilation capabilities

3D variational assimilation (ROMS, Chao et al.)3D variational assimilation (ROMS, Chao et al.)• ““Weak constraint” blend of data and modelWeak constraint” blend of data and model• Useful for nowcasts, not as good for dynamical analysisUseful for nowcasts, not as good for dynamical analysis• Inexpensive to runInexpensive to run

4D variational assimilation (ROMS, Moore et al.)4D variational assimilation (ROMS, Moore et al.)• ““Strong constraint” adjustment of IC and BC for hindcastsStrong constraint” adjustment of IC and BC for hindcasts• Can be used for sensitivity analysis, indices!Can be used for sensitivity analysis, indices!• Can be expensive to runCan be expensive to run

Possibilities for biological model optimizationPossibilities for biological model optimization 4D variational assimilation (in ROMS)4D variational assimilation (in ROMS) Genetic algorithmsGenetic algorithms

Data sourcesData sources

SMMR sattelite for ice coverSMMR sattelite for ice cover TOPEX/POSEIDON/AVISO altimetryTOPEX/POSEIDON/AVISO altimetry

No information on the shelfNo information on the shelf Long term mooringsLong term moorings

Nice long time series should be continuedNice long time series should be continued Long spatial correlation scales make these representative of Long spatial correlation scales make these representative of

broad areas on the shelfbroad areas on the shelf XBT data very sparse prior to the 70sXBT data very sparse prior to the 70s Hydro/mooring data sparse for the Western Bering SeaHydro/mooring data sparse for the Western Bering Sea BASIS program in the Eastern Bering BASIS program in the Eastern Bering Global circulation models for ICs and BCsGlobal circulation models for ICs and BCs

6. Needs of managers6. Needs of managers Mandates fromMandates from

National Environment Protection ActNational Environment Protection Act Marine Mammal Protection ActMarine Mammal Protection Act Endangered Species ActEndangered Species Act

• Stellar sea lionStellar sea lion• Sea ottersSea otters• Fur sealsFur seals• Right whalesRight whales• Fin whalesFin whales

Predictions of 5-10 years are of special interestPredictions of 5-10 years are of special interest Issues includeIssues include

BycatchBycatch Indirect effects of fishingIndirect effects of fishing Phys-bio interactionsPhys-bio interactions

Hindcasts of circulation and biology can help establish likely Hindcasts of circulation and biology can help establish likely response to future changeresponse to future change

Need better indices!Need better indices!

7. Estimated timetable of new 7. Estimated timetable of new model products and projectsmodel products and projects

See the report!See the report!

Summary ISummary I The ideal circulation model would adequately and The ideal circulation model would adequately and

simultaneously resolve all the relevant scales of motion simultaneously resolve all the relevant scales of motion and phenomena in the BSAI, e.g. and phenomena in the BSAI, e.g.

flows through the Aleutian Passesflows through the Aleutian Passes seasonal iceseasonal ice tidal mixing on the shelves. tidal mixing on the shelves.

None of the present modeling approaches can rapidly None of the present modeling approaches can rapidly and simultaneously capture all of these features for and simultaneously capture all of these features for extended time periods on today’s computersextended time periods on today’s computers

continuing advances in computer technology are expected to continuing advances in computer technology are expected to expand the limits of feasible simulations, at least doubling the expand the limits of feasible simulations, at least doubling the possible spatial resolution for such runs before 2010. possible spatial resolution for such runs before 2010.

Both nested approaches with structured grids, and Both nested approaches with structured grids, and variable resolution approaches with unstructured grids, variable resolution approaches with unstructured grids, appear promising ways forward. appear promising ways forward.

Summary IISummary II

Present ice model algorithms appear adequate for the Present ice model algorithms appear adequate for the Bering Sea. Bering Sea.

The accuracy of circulation hindcasts for the BSAI are The accuracy of circulation hindcasts for the BSAI are limited by the paucity of data, especially as regards the limited by the paucity of data, especially as regards the passes.passes.

Long-term moorings and systematic hydrographic surveys, in Long-term moorings and systematic hydrographic surveys, in conjunction with altimeter data, will help rectify this deficiencyconjunction with altimeter data, will help rectify this deficiency

Effective mathematical approaches are now available in Effective mathematical approaches are now available in community model codes for effective assimilation of such data community model codes for effective assimilation of such data into hindcasts and nowcasts. Computer resources are still a into hindcasts and nowcasts. Computer resources are still a limiting factor in the application of some of these codes. limiting factor in the application of some of these codes.

The atmospheric forcing datasets also have outstanding issues The atmospheric forcing datasets also have outstanding issues (e.g. biased shortwave radiation estimates), which limit the (e.g. biased shortwave radiation estimates), which limit the hindcast skill of BSAI simulations, and of ice in particular.hindcast skill of BSAI simulations, and of ice in particular.

Summary IIISummary III The ideal scientific/management biological model might The ideal scientific/management biological model might

include include multiple species and multiple life stage componentsmultiple species and multiple life stage components specific species treated using spatially explicit IBMsspecific species treated using spatially explicit IBMs coupled to multi-compartment NPZ and circulation modelscoupled to multi-compartment NPZ and circulation models

Proper feedback among different components especially Proper feedback among different components especially challengingchallenging

Intermediate step: focus on coupling spatially explicit NPZ with Intermediate step: focus on coupling spatially explicit NPZ with spatially aggregated food web models. spatially aggregated food web models.

For all models, longer time scales are needed to aid in For all models, longer time scales are needed to aid in ecosystem-based management. ecosystem-based management.

Data gaps are even larger for the biology than for the Data gaps are even larger for the biology than for the physics of the BSAIphysics of the BSAI

sustained surveys (e.g. the NMFS groundfish surveys) have sustained surveys (e.g. the NMFS groundfish surveys) have yielded much useful data for the quantification of food webs. yielded much useful data for the quantification of food webs.

Summary IVSummary IV

More collaborative development of both physical and More collaborative development of both physical and biological models is recommended, as they will require biological models is recommended, as they will require substantial human resources. substantial human resources.

Human time to examine and interpret the output can be Human time to examine and interpret the output can be just as limiting as computer hardwarejust as limiting as computer hardware

One way to ease the development and interpretation of One way to ease the development and interpretation of such multi-investigator models is to provide easy access such multi-investigator models is to provide easy access to model output through web-based software. to model output through web-based software.

FIN!FIN!

http://halibut.ims.uaf.edu/SALMON/BSIAModelWorkshophttp://halibut.ims.uaf.edu/SALMON/BSIAModelWorkshop

•A big, broad shelf•Passes: wide, narrow, shallow, deep

•Canyons•Cross-shelf flux•Powerful tides•A deep basin•Ice•High production•Lots of fish!•Climate change

Features of the Bering Sea

Foci of the workshopFoci of the workshop

reviewreview existing modeling efforts in the BSAI existing modeling efforts in the BSAI assessassess strengths and weakness of the various types of strengths and weakness of the various types of

ocean circulation models in accurately representing ocean circulation models in accurately representing circulation, mixing and exchange due tocirculation, mixing and exchange due to

forcing mechanisms (winds, tides, ice formation, river runoff) forcing mechanisms (winds, tides, ice formation, river runoff) topographic features (coastline, shelf break, Aleutian Island topographic features (coastline, shelf break, Aleutian Island

passes)passes) evaluate evaluate various monitoring and process studies that various monitoring and process studies that

would improve the accuracy of the modelswould improve the accuracy of the models describedescribe pathway for using these models to develop pathway for using these models to develop

products that would be useful for resource managers products that would be useful for resource managers and users.and users.