Progress Report:
June 2005  -  Workshop 9
(AOMIP)

The Ninth AOMIP Workshop (June 6-7), was held at McGill University in Montreal, Canada. Details about the workshop, including downloadable versions of meeting presentations, session reports, and minutes are available at the AOMIP website (see http://fish.cims.nyu.edu/project_aomip/workshops/workshop_9/overview.html).

Introduction

The AOMIP workshop was conducted at McGill University in order to allow university students to attend the workshop sessions and to meet and discuss major problems of modeling of the Arctic Ocean and sea ice dynamics and thermodynamics with the international Arctic Modeling community.  In total, 29 scientists and students from the USA, Canada, Germany, Sweden and Russia took part in the workshop.

Different in style from the organization of the Seventh AOMIP workshop, where Day 1 was dedicated to the science presentations and Day 2 to the discussions, the agenda of Ninth workshop included five major sessions with major AOMIP topics. These sessions/presentations were immediately followed by discussions in order to concentrate on the most pressing problems, without delay.

AOMIP modelers are engaged in exploring model characteristics in a number of directions, including resolution, accuracy of advection schemes, salinity restoring, and other parameterizations, such as lateral-ocean mixing.  A common thread in these studies is that models developed and tuned for certain regions do not often work as well in other regions.  This is a major difficulty facing global climate modelers.  It is also becoming clear that Arctic Ocean models are resolving neither the proper spatial scales nor the mean physics.  Various models show that Arctic simulation results change as the model resolution is changed.  Different models employ different parameterizations representing the same physical effect. The differences between the parameterizations often have huge repercussions for an Arctic simulation.  AOMIP participants are currently working toward understanding such model behavior.

 

Workshop themes and discussions

The major workshop themes included:

1.      Model improvements and improved results

2.      Model experiments

3.      Data for model forcing, validation and calibration

4.      Modeling with data assimilation

5.      Processes, mechanisms, and physics

6.      Internal project issues

Respectively, each workshop theme/session was followed by discussions:

1.      Model improvement plans

2.      Future experiment plans

3.      Data sources for validation and calibration

4.      Role of data assimilation in AOMIP

5.      Fresh water balance, Atlantic water circulation, tides, sea ice and mixing problems

6.      Project coordination, funding opportunities, research themes, and publications

 

Major topics and key issues

In total, 24 talks presented results and analysis of AOMIP model runs. As usual, throughout the talks, several key topics were repeatedly discussed from various viewpoints because of their importance for climate studies and modeling. The key topics were:

• Propagation and transformation of Atlantic waters in the Arctic Ocean
• Variability of the Arctic Ocean freshwater content
• Advection schemes
• Data assimilation techniques and philosophy
• Role of tides in the Arctic Ocean and in sea-ice dynamics and thermodynamics
• Arctic climate variability in global and regional models
• General model intercomparison issues
• AOMIP organization, goals and objectives.

 

Workshop conclusions and recommendations

The workshop conclusions are based on discussions of the 9^th Workshop in Montreal, as well as the 8^th Virtual Workshop held during February, 2005.  The conclusions reiterate the most important conclusions and recommendations of the 7^th AOMIP Workshop, held in Princeton, June, 2004.  This repetition/reminding of conclusions of the 7^th and 8^th Workshops is necessary in order to re-start AOMIP activities, largely because the bulk of AOMIP activities have been frozen since March 2005.

1. AOMIP’s working plan for the next 2 years (June, 2005 to February 28, 2007) includes:

 

A.  Completion of analyses of coordinated 50-year experiments:

 

·         further intercomparison of model output

·         identification of key differences

·        determination of causes of differences among models

·        testing of proposed model improvements

·        formulation of major recommendations for model improvements

 

B.  Formulation of coordinated 100-year model runs:

 

·        analysis and intercomparison of model output

·        investigation of Arctic Ocean and sea ice variability based on model results and observations

 

2. Several scientific and modeling themes identified by AOMIP need serious attention in order to understand and to model Arctic Ocean behavior successfully. These themes are:

 

A. Circulation and properties of the Arctic Ocean Atlantic Water Layer (AWL):

• There are striking differences in the circulation of AWL and its transformations among AOMIP models. Approximately half of the models show cyclonic circulation of Atlantic water and the other half show the opposite circulation. It is extremely important to identify the physical mechanism underlying these differences. It is equally important to validate the models based on observational data.
• The nature and relevance of the AWL circulation is largely not appreciated by the global climate modeling community. AOMIP is fostering an awareness of this issue by holding joint meetings with global ocean and global climate modelers.
• There is an urgent need for more observational data on the AWL. AOMIP is performing 50 and 100 year simulations, and thus we are especially interested in an historical perspective on AWL characteristics. Such data sets are currently collected by several observational projects concerning low-frequency climate variability at the University of Washington (UW), International Arctic Research Center (IARC), Alfred Wegener Institute (AWI), and Arctic and Antarctic Research Institute (AARI). It is recommended that all historical data characterizing water temperature and salinity, chemical and radionuclide tracers, circulation parameters including currents at different levels, depths of the upper and bottom boundaries of AWL, integrated characteristics such as heat and salt content, and fluxes of AWL through different sections have to be archived at a Live-Access Server (LAS) and available for AOMIP partners. Such data can be used by AOMIP for model validation, calibration, and finally, improvement.
• The AOMIP, in turn, based on numerical experiments, can provide observational projects (especially for the International PolarYear (IPY) March 2007 – March 2009 period) with recommendations about the most representative locations for AWL monitoring.
• In order to further integrate AOMIP and AWL-related currents and future IPY studies, we recognize the need for a 2006 AGU Spring or Fall special session, followed by a workshop specifically dedicated to the AWL circulation and transformation.

 

B. Arctic Ocean freshwater dynamics and trends:

There are striking differences in the Arctic Ocean freshwater content and its seasonal, interannual, and decadal variability among different AOMIP-regional and AOMIP-global models. Variability of the freshwater content, and mechanisms of fresh water accumulation and release, is extremely important for understanding major issues relating to global-ocean meridional overturning circulation and climate change. Workshop participants recognized several important deficiencies in the current understanding of the problems of fresh water simulations. These problems include uncertainties in the choices of boundary conditions for rivers, straits, and open boundaries (volume, diffusion or precipitation inflows/outflows); restoring or not restoring alternatives; and forcing data (rates of precipitation/evaporation and river runoff). For example, existing models use seasonal climatology for precipitation, which a priori contradicts the basic assumption of climate change, and leads to an imbalance in fresh water inputs and outputs. As a result, practically all regional models (without restoring) have significant salinity drifts. Additional observational data collection and interpretation, for the specific purpose of ocean and climate modeling in the Arctic, is needed.

 

C. Model improvements

There are several ways to improve AOMIP models and they include improvements in physics and improvements in numerics:

(i) Improvements in model numerics (numerical advection problems were discussed during workshop)

• The quality of numerical advection appears to be important to carrying appropriate AWL properties without excessive diffusion.  We have explored the use of a second order moment (SOM) method as compared with traditional centered difference or flux-corrected transport schemes.  Previous efforts in AOMIP have provided SOM for sea ice advection, and this method is now extended to 3D ocean tracers.  Resulting improvements to AWL properties and transports may enabling advances in physics-based mixing process representations which, hitherto, were contaminated by numerical diffusion.

 (ii) Improvements in model physics:

• AOMIP researchers have already proposed several improvements for Arctic models, including implementation of tidal dynamics into ocean and ice models. Preliminary experiments have shown that tidal and inertial sea-ice motion leads to additional generation of sea-ice mass and a redistribution of sea-ice deformations and fracture-zone formations. Inclusion of tidal dynamics into the ocean component of coupled models could improve mixing and heat transformations in the ocean and help understand better the heat release mechanism from the AWL to the bottom of sea ice and into the lower atmosphere. Although tidally-enhanced ocean heat transport contributes toward thinning sea ice, tidal energy also fractures the ice cover, promoting ridging and increase of ice volume.  As well, by fracturing ice cover, tides mobilize ice allowing freer redistribution.  The result is that the impact of tides on ice in any region is a complex interplay of gains and losses.
• Another direction of improvements is an inclusion of processes responsible for establishing and breaking of the landfast ice. In the existing models regions of landfast ice are treated as pack ice, which drifts. As a consequence, more momentum is transferred to the ocean from wind stresses than in the real ocean. Absence of landfast ice in the models means an improper representation of coastal polynyas, of sea-ice production, and possibly of deep-water production. This is because some of these processes occur along the coastline in the models, instead of along the continental, shelf break as in the real ocean.

 

(iii) Improvements in model forcing and model validation and calibration technology

• The AOMIP workshop participants recognize the importance of model validation and calibration for robust decisions and estimates of results of model improvement. In order to facilitate the carrying out of such work, it is necessary to prepare standard model validation data sets, which include observational data organized for specific AOMIP tasks. AOMIP researchers agreed to prepare and distribute among AOMIP investigators the necessary data. The data will be posted on the AOMIP Live-Access Server (LAS), and thus be available to the entire modeling community. It is important to complete this work by October 1, 2005, in order to continue model validation. The data sets, with instructions on how to use these data for validation procedures, are:

• Water temperature, salinity, and circulation (with a focus on AWL circulation and transformation, it is extremely important to have these data sets). AOMIP requests this information from NABOS (Nansen and Amundsen Basin Observing System) and hopes to have wider discussions with IARC scientists. (Karcher, Steele)
• Radioactive tracer data (iodine, cesium) exist from 1980's. (Karcher)
• Sea-ice drift and sea-ice deformation. (Hibler)
• Sea-ice extent and concentration based on passive microwave data. (Zhang)
• Monthly sea level at the most representative tide gauge stations. (Proshutinsky)
• Sea-ice thickness from submarines (Hunke) and ULS (Up-Looking Sonars) (Johnson).
• Long-term T and S time series from hydrographic sections (Kola Section, Faroe-Shetland Section, Fram Strait, SCICEX, etc. (Steele)
• Current meter data from all possible moorings. (All, based on institutional sources)
• Long-term data sets from ASOF. (Karcher)

• The AOMIP workshop participants recognize the importance of model forcing and recommend organizing a virtual workshop to formulate conditions for a new coordinated model run with a major focus on discussion of new and improved model forcing parameters.

 

3. Collaboration with other MIPs, especially ARCMIP (Arctic Regional Climate MIP) and OMIP (Ocean MIP), is an important activity of the AOMIP project. During previous AOMIP and OMIP workshops, both projects exchanged information about project goals, objectives, problems, and possible solutions. It was reported that OMIP forcing data set does not well represent arctic conditions, which could lead to unrealistic sea-ice thickness and concentration. Inappropriate forcing data could influence the simulation by global models of variability of water circulation, heat content, freshwater content, and rate of overturning circulation. Participants agreed that R. Gerdes will serve to coordinate information, between the AOMIP and OMIP communities, about such problems as well as accomplishments.

 

4. Internal AOMIP issues were discussed at the end of workshop and focused on publication needs. AOMIP recommends:

• Each AOMIP team/scientist will prepare a paper for joint AOMIP publication by December 1, 2005. These publications will be submitted to Journal Geophysical Research (JGR). Papers will be dedicated to AOMIP intercomparison of model-model and model-data results for different environmental parameters; model validation and calibration; and interpretation of the Arctic Ocean changes based on model results and observations.

 

5. The AOMIP activity in 2007/2008 will be our final analysis of Arctic-Ocean variability over a 100-year time scale and to make recommendations to the global-ocean modeling community on Arctic-Ocean modeling. In parallel, several AOMIP groups (AWI, UW and IARC) will be involved in the developing of regional Arctic Ocean models able to assimilate observational data. These groups will inform the rest of AOMIP community about their results and will formulate their recommendations on how to improve Arctic models using results from data assimilation.

 

6. This and previous workshop experiences showed that a 2-day workshop period is not enough to analyze project results (presentations and discussions) and to discuss all project issues and plans. Two options were suggested to solve this problem:

·        all talks have be limited to 10 minutes (just motivation and conclusions) and all in the morning, leaving the entire afternoons for discussions. 

·        Have a 3-day workshop with the major focus on project discussions, recommendations, and plans.

 

7. The 10^th AOMIP workshop is planned for June 2006 at the Village of Breckenridge, Colorado, in order to interact with participants of the 11^th Annual Community Climate System Model Workshop.  This will allow AOMIP researchers to meet directly with key representatives of the global-ocean modeling community (similar to 7^th workshop where AOMIP met with key OMIP scientists at GFDL, Princeton in 2004). Participants of both workshops will meet jointly to discuss questions of mutual interest, such as recommendations for improving of global models in the Arctic. Proshutinsky and Steele will be responsible for coordination and organization of joint sessions and probably for a special session at CCSM workshop specifically focused on modeling of the Arctic Ocean climate variability.