Arctic Ocean Model Intercomparison Project: Key outcomes. Boundary condition considerations, readiness of regional models for coupling. Arctic System Model

Arctic Ocean Model Arctic Ocean Model Intercomparison Project: Intercomparison Project:

Key outcomes. Boundary condition considerations, readiness of regional

models for coupling.

Arctic System Model workshop IARC, UAF, Arctic System Model workshop IARC, UAF, Fairbanks, August 6-7, 2007Fairbanks, August 6-7, 2007

Andrey ProshutinskyAndrey Proshutinsky

Woods Hole Oceanographic InstitutionWoods Hole Oceanographic Institution

Regional AOMIP ModelsRegional AOMIP Models

Global AOMIP modelsGlobal AOMIP models

AOMIP model ID LANL UW NERSC UCL

Home Institute Los Alamos National Laboratories

University of Washington

Nansen Environmental and Remote Sensing Center

Universite Catholique de Louvain

Ocean Model Pedigree

POP POIM MICOM OPA

Coupled Sea-Ice Model

Yes Yes Yes Yes

POP – Parallel Ocean Model; MOM - GFDL Modular Ocean Model

POIM – Parallel Ocean Ice Model; MICOM – Miami Isopycnal Ocean Model

POM – Princeton Ocean Module Model;

OPA – Ocean General Circulation modeling System

Ten AOMIP modeling groups are presented here:

AWI - Alfred Wegener Institute - Bremerhaven, Germany; CNF - Frontier Research Center for Global Change, Japan, with InternationalArctic Research Center, USA GSFC - NASA Goddard Space Flight Center, USA; ICMMG - Institute ofComputational Math. and Math. Geophysics, Russia; IOS -Institute of Ocean Sciences, Canada; LANL -Los Alamos National Laboratory, USA; NPS - Naval Postgraduate School, USA; POL - Proudman Oceanographic Laboratory, Liverpool, UK; UL - Universite Laval, Quebec, Canada; UW - University of Washington, USA.

Arctic Ocean Model Intercomparison Project (AOMIP)

Common model domainCommon model domain

The AOMIP Grid is The AOMIP Grid is defined over a geographic defined over a geographic domain that includes the domain that includes the Arctic Ocean, the Bering Arctic Ocean, the Bering Strait, the Canadian Arctic Strait, the Canadian Arctic Archipelago, the Fram Archipelago, the Fram Strait and the Greenland, Strait and the Greenland, Iceland, and Norwegian Iceland, and Norwegian SeasSeas.

AWI - Alfred Wegener Institute, POL –Proudman Oceanog Lab., UK Holloway, G., et al. (2007), Water properties

CNF – FRCC/IARC , GSFC - Goddard Space Flight Center and circulation in Arctic Ocean models, J. Geophysical

ICMMG – Russia , IOS -Institute of Ocean Sciences Res., 112, C04S03, doi:10.1029/2006JC003642.

LANL -Los Alamos , NPS - Naval Postgraduate School, USA; UL - Universite Laval , UW - University of Washington, USA.





















































Model validationModel validation

The first group of studies has focused on the The first group of studies has focused on the analysis of differences among model results analysis of differences among model results and between model results and observations.and between model results and observations.

This was important for determining model This was important for determining model errors and model uncertainties, and this was errors and model uncertainties, and this was the first step in the process of model the first step in the process of model improvement.improvement.

GOAL 1

Sea Sea levellevel

Ice Ice driftdrift

Ice Ice thicknessthickness

Ice Ice concentrationconcentration

Water Water circulationcirculation

HydrographyHydrography

Model Model validation validation

parametersparameters

GOAL 1

Model improvement Model improvement activitiesactivities

Restoring andRestoring andFlux correctionFlux correction

Atmospheric loadingAtmospheric loading

Vertical and lateralVertical and lateralmixing mixing

Tidal ocean & iceTidal ocean & iceeffectseffects

Bering Strait Inflow Bering Strait Inflow and river runoffand river runoff

Neptune effectNeptune effect

New advection New advection schemesschemes

Data assimilationassimilationtechnology

Vertical and Vertical and lateral resolutionlateral resolution

Forcing biasesForcing biases

Land-fast iceLand-fast ice

GOAL 1

Wind- and Wind- and convection-convection-

drivendrivenmixing mixing

Wind- and Wind- and convection-convection-

drivendrivenmixing mixing

Heat contentHeat contentvariability variability and role of and role of

differentdifferentfactorsfactors

Heat contentHeat contentvariability variability and role of and role of

differentdifferentfactorsfactors

FreshwaterFreshwatercontentcontent

variability andvariability androle of differentrole of different

factorsfactors

FreshwaterFreshwatercontentcontent

variability andvariability androle of differentrole of different

factorsfactors

Reconstruction Reconstruction of hydrography of hydrography and circulationand circulation

based on based on modeling with modeling with

data data assimilationassimilation

Reconstruction Reconstruction of hydrography of hydrography and circulationand circulation

based on based on modeling with modeling with

data data assimilationassimilation

Atlantic water Atlantic water circulationcirculation

origin, origin, variability , variability , sense of sense of rotationrotation

Atlantic water Atlantic water circulationcirculation

origin, origin, variability , variability , sense of sense of rotationrotation

Investigation of Investigation of sea level rise: sea level rise: its rate, role of its rate, role of different factors,different factors,

model errorsmodel errors

Investigation of Investigation of sea level rise: sea level rise: its rate, role of its rate, role of different factors,different factors,

model errorsmodel errors

Process and Process and arctic change arctic change studiesstudies

GOAL 2

Model improvements recommendations:

Tidal forcing is important for Arctic Ocean modeling; Tidal and inertial dynamics has to be included in the sea

ice models as well; Inverted barometer effect is an important component for

simulations of synoptic variability; Variable river runoff and Bering Strait inflow are important

parameters influencing Arctic climate and have to be taken into account;

Land-fast ice is an important regulator of dynamics and thermodynamics because it influences upwelling and downwelling, sea ice production and brine rejection, shelf water properties.

GOAL 2

Model improvementsModel improvements

Model improvement includes several phases:

- Identification of problems;

- Search for solutions/improvements;

- Testing improvements based on one or two models;

- Recommendations to others; and

- Introduction and testing of new ideas.

GOAL 1

Differences for each model between the mean model sea ice concentration and the mean sea ice from GSFC for 1979–1999. The dark line is the 0.001 concentration contour from the GSFC data. Models from left to right and top to bottom are AWI2, AWI1, UW, NPS, IOS, ICM, LANL, GSFC, and RAS. Scale is from −0.4 (red) to +0.4 (blue) with values nearer zero having less color saturation. Saturated colors indicate larger differences from the observations with red below and blue above the observed.

GOAL 1Sea ice concentrationSea ice concentration

Sea ice concentration model Sea ice concentration model validation conclusionsvalidation conclusions

Differences among the sea ice concentrations Differences among the sea ice concentrations computed by the AOMIP models are greater than computed by the AOMIP models are greater than differences among four observational data sets. differences among four observational data sets.

Regardless of the different model physics and Regardless of the different model physics and parameters, the results show that the models have parameters, the results show that the models have more variability than observed, and that, compared more variability than observed, and that, compared to observations, almost all the models to observations, almost all the models underestimate the September sea ice concentration underestimate the September sea ice concentration in the central Arctic Ocean. in the central Arctic Ocean.

This underestimation may have important This underestimation may have important implications for sea ice forecasts. implications for sea ice forecasts.

GOAL 1

Sea ice thickness model validation Sea ice thickness model validation conclusionsconclusions

There are considerable errors in sea ice thickness in IPCC results. There are considerable errors in sea ice thickness in IPCC results. (too simple sea ice rheologies in some of these models. Better (too simple sea ice rheologies in some of these models. Better models tend to pile up ice in the center of the ocean).models tend to pile up ice in the center of the ocean).

Errors may have important consequences for the atmospheric Errors may have important consequences for the atmospheric

circulation. Too large ice cover and thickness in the European circulation. Too large ice cover and thickness in the European sector could be significant in ocean-atmosphere interactions and sector could be significant in ocean-atmosphere interactions and long term variability. long term variability.

The AOMIP results are dominated by an accumulation of sea ice The AOMIP results are dominated by an accumulation of sea ice in the mid-1960s and a return to values before that event in the mid-1960s and a return to values before that event in the last decade of the 20th century. in the last decade of the 20th century.

The IPCC results show a negative trend in Arctic sea ice volume The IPCC results show a negative trend in Arctic sea ice volume over the 20th century. The AOMIP simulation shows no trend over over the 20th century. The AOMIP simulation shows no trend over that period. This suggests that the internal multidecadal variability that period. This suggests that the internal multidecadal variability of the real climate system is underestimated in IPCC models. of the real climate system is underestimated in IPCC models.

GOAL 1

Sea ice driftSea ice drift Gridded observational ice drift fields are used from two Gridded observational ice drift fields are used from two products: NSIDC (Fowler, 2003); CERSAT (Ezraty, and Piollé, products: NSIDC (Fowler, 2003); CERSAT (Ezraty, and Piollé, 2004) 2004)

GOAL 1

Difference Difference between between model model CERSAT ice CERSAT ice drift speeds drift speeds (gray shade) (gray shade) and and direction direction (black (black outline). outline).

AWIAWI GSFCGSFC IOSIOS

NPSNPS UWUW AWI-2AWI-2

Sea ice drift model validation resultsSea ice drift model validation results

One class of models has a mode at drift speeds around 3 One class of models has a mode at drift speeds around 3 cm/s and a short tail toward higher speeds. Another class cm/s and a short tail toward higher speeds. Another class shows a more even frequency distribution with large probability shows a more even frequency distribution with large probability of drift speeds of 10 to 20 cm/s. Observations clearly agree of drift speeds of 10 to 20 cm/s. Observations clearly agree better with the first class of model results. better with the first class of model results.

Reasons for these differences lie in discrepancies in sea ice Reasons for these differences lie in discrepancies in sea ice model characteristics and sea ice-ocean coupling. model characteristics and sea ice-ocean coupling.

In general, the models are capable of producing realistic In general, the models are capable of producing realistic drift pattern variability. drift pattern variability.

The winter of 1994/1995 stands out because of its maximum The winter of 1994/1995 stands out because of its maximum in Fram Strait ice export. While export estimates of some in Fram Strait ice export. While export estimates of some models agree with observations, the corresponding inner Arctic models agree with observations, the corresponding inner Arctic drift pattern is not reproduced. drift pattern is not reproduced.

GOAL 1

Sea surface height/sea levelSea surface height/sea level

Several improvements are needed to Several improvements are needed to reduce model errors: reduce model errors: • Minimum model depth has to be less Minimum model depth has to be less than 10 m. than 10 m. • Take into account: atmospheric Take into account: atmospheric loading, fast ice, and volume water loading, fast ice, and volume water fluxes representing Bering Strait fluxes representing Bering Strait inflow and river runoff. inflow and river runoff.

Sea level rise for 1954–2006 is Sea level rise for 1954–2006 is estimated as 0.250 cm/yr. estimated as 0.250 cm/yr. The sea level dropped significantly The sea level dropped significantly after 1990 and increased after the after 1990 and increased after the circulation regime changed from circulation regime changed from cyclonic to anticyclonic in 1997. In cyclonic to anticyclonic in 1997. In contrast, from 2000 to 2006 the sea contrast, from 2000 to 2006 the sea level rose despite the stabilization of level rose despite the stabilization of the AO index at its lowest values after the AO index at its lowest values after 2000. 2000.

GOAL 1

Tidal dynamics of Tidal dynamics of water and sea icewater and sea ice

A three-dimensional coupled ocean/ice A three-dimensional coupled ocean/ice model, intended for long-term Arctic climate model, intended for long-term Arctic climate studies, is extended to include tidal effects. studies, is extended to include tidal effects. From saved output of an Arctic tides model;From saved output of an Arctic tides model;

We introduce parameterizations for:We introduce parameterizations for:

(1) enhanced ocean mixing associated with (1) enhanced ocean mixing associated with tides; and tides; and

(2) the role of tides fracturing and (2) the role of tides fracturing and mobilizing sea ice. mobilizing sea ice.

GOAL 1

Circulation patterns and tidal effectsCirculation patterns and tidal effects

Holloway, G., and A. Proshutinsky (2007), Role of tides in Arctic ocean/ice climate, J. Geophys. Res., 112,

C04S06, doi:10.1029/2006JC003643.

GOAL 1

Upper left: Potential temperature (°C) is shown at 320 m during December 1999 from a case without tides. Upper left: Potential temperature (°C) is shown at 320 m during December 1999 from a case without tides.

Upper right: Temperature (°C), without tides, is shown on the vertical section marked by a green bar in the upper left Upper right: Temperature (°C), without tides, is shown on the vertical section marked by a green bar in the upper left panel.panel.

Lower right: Temperature (°C) is shown on the same vertical section, with the same color scale, as upper right but here Lower right: Temperature (°C) is shown on the same vertical section, with the same color scale, as upper right but here including effects of tidesincluding effects of tides

Lower left: The difference of temperature (°C) with tides and without tidesLower left: The difference of temperature (°C) with tides and without tides

Temperature (°C), without tides

Temperature (°C), with tidesTemperature difference with tides and without tides

Holloway, G., and A. Proshutinsky (2007), Role of tides in Arctic ocean/ice climate, J. Geophys. Res., 112, C04S06,

doi:10.1029/2006JC003643.

GOAL 1

Tidal effect resultsTidal effect results

Results show tides enhancing loss of heat from Results show tides enhancing loss of heat from Atlantic waters. Atlantic waters.

The impact of tides on sea ice is more subtle as The impact of tides on sea ice is more subtle as thinning due to enhanced ocean heat flux competes thinning due to enhanced ocean heat flux competes with net ice growth during rapid openings and with net ice growth during rapid openings and closings of tidal leads. closings of tidal leads.

Present model results are compared with an Present model results are compared with an ensemble of nine AOMIP models. ensemble of nine AOMIP models.

Among results from AOMIP is a tendency for models Among results from AOMIP is a tendency for models to accumulate excessive Arctic Ocean heat to accumulate excessive Arctic Ocean heat throughout the intercomparison period 1950 to 2000 throughout the intercomparison period 1950 to 2000 which is contrary to observations. Tidally induced which is contrary to observations. Tidally induced ventilation of ocean heat reduces this discrepancy. ventilation of ocean heat reduces this discrepancy.

GOAL 1

Atlantic Water circulationAtlantic Water circulation

There are several scientific questions associated with the origin, direction, and variability of the Atlantic water layer circulation in the Arctic Ocean. Observational studies suggest that this circulation is cyclonic and its intensity may change depending on Arctic Oscillation or North Atlantic Oscillation regime. How surface forced ocean regulates circulation in deep layers is not clear. Figures above suggest that deep circulation does not change significantly when surface circulation changes from anticyclonic to cyclonic.

GOAL 2

Models with cyclonic Models with cyclonic circulation of Atlantic watercirculation of Atlantic water

MOM high MOM high resolutionresolution

POMPOMMOM low MOM low resolutionresolution

MOM MOM Global, Global, OPAOPA

AOMIP studies showed that some models generate cyclonic circulation which intensity changes in time insignificantly. Other model results show that circulation changes and even may reverse its direction. What is the origin of these reversals?

GOAL 2

Models with anticyclonic Models with anticyclonic circulation of Atlantic layercirculation of Atlantic layerMOM high resolutionMOM high resolution Finite elementsFinite elements

MOMMOM

Several models showed that the Atlantic water circulation is very stable and is anticyclonic!!! Note that model forcing, initial conditions, bathymetry, etc. were identical in the models reproduced cyclonic and anticyclonic motion of the Atlantic water.

GOAL 2

Monthly mean potential temperature (°C) is shown as a function of depth and time for models AWI, CNF, GSFC, ICMMG, IOS, LANL, NPS, UL and UW averaged over subdomain “E”.





Monthly mean potential temperature (°C) is shown as a function of depth and time for models AWI, CNF, GSFC, ICMMG, IOS, LANL, NPS, UL and UW averaged over subdomain “A”.





Symbols identify the models.





Total heat referenced to 0°C, integrated over the volume of subdomain “E”, is plotted in units of 1022 J. Horizontal lines during the 1970s and 1980s are

decadal mean heat for subdomain “E” from EWG [1997, 1998] summer and winter atlases.

Total heat referenced to 0°C, integrated over the volume of subdomain “A”, is plotted in units of 1022 J. Horizontal lines during the 1970s and 1980s are decadal mean heat for subdomain “A” from EWG [1997, 1998] summer and winter atlases





Monthly mean salinity is shown as a function of depth and time for models AWI, CNF, GSFC, ICMMG, IOS, LANL, UL, NPS and UW averaged over sub-domain “E” (left) and “A” - right

E sub-domain A sub-domain









Total freshwater referenced to S = 34.8 (including negative contributions when S > 34.8) is plotted in units of 10^13 m^3. Horizontal lines during the 1970s and 1980s are decadal mean freshwater from EWG [1997, 1998] summer and winter atlases integrated over the volume of sub-domain “E” (left) and A (right)

Neptune: Circulation and Topostrophy

Flow (left) and topostrophy (right) is shown over bathymetry during December 1987 from IOS model, using neptune (top) and without neptune (bottom).

Red and yellow – positive topostrophy and cyclonic motion.

Blue – anticyclonic motion

Monthly mean normalized topostrophy is shown as a function of depth and time for sub-domains “E” (left) and “A” (right).





Normalized topostrophy (dimensionless) averaged over the volume of sub-domains “E” (left) and “A” (right)





Speed of monthly mean flow averaged over sub-domains “E” and “A”





Total kinetic energy of monthly mean flows is plotted in units of 1014 J integrated over the volume of subdomain “E”.





Averages of models' summer (July, August, September) and winter (March, April, May) temperatures, averaged from 1980 through 1989, are compared with the average of EWG summer and winter atlases (solid trace), 1980s decadal mean.





Lateral area (1012 m2) is plotted vs. depth (m) for the Amerasian and Eurasian basins.





Root mean square slope is plotted vs. depth (m) for the Amerasian and Eurasian basins.





Documents

Arctic Ocean Model Intercomparison Project: Key outcomes. Boundary condition considerations, readiness of regional models for coupling. Arctic System Model