[an error occurred while processing this directive] [an error occurred while processing this directive]

Report of the CSM Polar Climate Working Group

John W. Weatherly and Richard Moritz, Co-Chairs

24 June 1998

The Village at Breckenridge


The CSM Polar Climate Working Group (PCWG) meeting was held on 24 June 1998 at the Third Annual CSM Workshop in Breckenridge, Colorado. An initial summary of the meeting was presented in the plenary session of the CSM Workshop. The purpose of this report is to summarize the meeting for interested parties who may or may not have attended the meeting and also to make recommendations to the CSM Scientific Steering Committee (SSC) on the continuing CSM PCWG activities.

I. Summary of Presentations

John Weatherly (NCAR) provided an introduction to the PCWG meeting. During the plenary session of the CSM Workshop, he presented work on acknowledged deficiencies in the coupled-model simulation and recent ice-ocean simulations with viscous-plastic ice dynamics. He summarized the present state of the CSM sea ice model and suggested that the PCWG address the following: the implementation of the viscous-plastic (VP) and/or elastic-viscous plastic (EVP) ice rheologies; the use of a rotated, independent grid for the sea ice model; and the continued development of the thermodynamic parameterizations in the CSM sea ice model. He also discussed the continuing need for the distribution of and access to CSM polar data.

Richard Moritz (University of Washington) presented an overview of the Surface Heat Budget of the Arctic (SHEBA) program and its relation to the CSM modeling activity. Several SHEBA datasets are available now on the World Wide Web (WWW). A comprehensive SHEBA dataset will be available by October, 2000. These data will be used for general circulation model (GCM) development and validation. For these and other applications, a column version of the CSM model is needed, including sea ice, flux coupler, and the upper ocean.

Bruce Briegleb (NCAR) presented work on the implementation of the viscous-plastic ice rheology into the CSM sea ice model, using a rotated, spherical grid. This model is being used for coupled ice-ocean experiments (with the CSM NCAR Ocean Model), as well as for ice-alone sensitivity studies. He showed the significant improvement in the plastic solution of the model with 5 to 15 psuedo-timesteps. He also showed that the Arctic ice thickness pattern from the VP model driven by the Community Climate Model version 3 (CCM3) winds is similar to that of the CSM coupled run that used a cavitating-fluid (CF) model. He concluded that the arbitrary redistribution required by the CF model did not strongly affect the coupled model solution.

Jim Maslanik (University of Colorado) presented results of work (jointly with Amanda Lynch, University of Colorado) using the Arctic Regional Climate Simulation (ARCsym), a regional, mesoscale atmosphere model of the Arctic with fully coupled dynamic sea ice. The sea ice model uses the EVP ice dynamics. The simulations of the minimum 1990 ice conditions improved with the EVP dynamics compared to simulations with the CF dynamics and a thermodynamic-only simulation. There were significant feedbacks to the atmospheric pressure patterns with the EVP dynamics as well. He concluded that the CF dynamics were not a viable option for a coupled climate model.

Todd Arbetter (University of Colorado) presented comparisons of simulations of 1992 Arctic ice conditions using VP, EVP, and CF ice dynamics, using both single-thickness and 28-category thickness models. The two elliptical-curve models (VP and EVP) yielded similar results, while the CF dynamics generally produced larger velocities and thinner ice. The sensitivity of the VP and EVP models to increases/decreases in wind stress magnitude showed greater/lesser ridging in winter and greater/lesser outflow in summer. His conclusion was that the CF model is inadequate for climate modeling and that a single ice thickness representation is also insufficient.

Walter Meier (University of Colorado) compared Arctic sea ice motions simulated with the EVP dynamics to motions derived from Special Sensor Microwave Imager (SSM/I) satellite observations using correlative techniques. The correlation between model and satellite velocities varies significantly in time and is lowest when Arctic sea ice appears rigidly locked in place in spite of strong surface air pressure gradients, while the dynamic model simulates non-rigid ice motions. His study suggests that even the present elliptical yield curve models do not simulate the rigid ice correctly.

Bill Lipscomb (University of Washington) presented work on the effects of modeling energy-conserving thermodynamics and multiple ice thicknesses. His work shows the importance of conserving the enthalpy of sea ice (based on a prescribed salinity profile) during the melting process. He also showed the significant effects of representing thicker (ridged) ice in the Arctic, as well as thinner ice categories. In most cases, three to five thickness categories were sufficient for the convergence of solutions, while in some cases more that five categories were needed. The thickness-dependent albedo parameterization also was shown to significantly affect the solutions. His conclusion was a minimum of three ice thicknesses is necessary for climate modeling.

II. PCWG Administration

Both John Weatherly and Richard Moritz are willing to continue as co-chairs of the PCWG. The CSM SSC may desire to appoint another co-chair on the NCAR staff due to John Weatherly's departure from NCAR. This issue was acted upon by the joint SSC/Scientific Advisory Council meeting following the workshop, and William G. Large was appointed as a third PCWG co-chair.

The co-chairs will contact other PCWG members to determine their interest and availability for an interim PCWG meeting in 1998/1999.

A separate list of the PCWG meeting attendees is not available. For reference, see the list of CSM Workshop participants (http://www.cesm.ucar.edu/news/ws.1998/partic.html).

III. Recommendations

A. Infrastructure

The CSM ice model needs to be a true community model, developed and supported by and for the greater scientific community. Unlike the other CSM components, there are neither a NCAR sea ice model nor the in-house expertise to develop one.

Although sea ice clearly has large effects on many aspects of the CSM simulations and requires a distinct physical model for treatment, there is no dedicated "CSM sea ice scientist" at NCAR. CSM ice model development and key CSM sea ice experiments have been pushed ahead by John Weatherly, Bruce Briegleb, and others at NCAR and in the community, but this activity has been squeezed in along with other main projects. This situation is becoming critical with the departure of Weatherly for the Cold Regions Research and Engineering Lab (CRREL) and the increasing interaction with the university community. To develop, test, and implement the new CSM sea ice models will require a coordinated effort that brings together knowledge of numerical modeling and sea ice physics from within and outside NCAR, enhanced datasets, and a commitment by all concerned to see the efforts all the way through to incorporation in CSM and performance of new coupled climate experiments. The PCWG sees the following three elements as crucial for the success of the effort:

1. Enhanced engagement of the community outside NCAR in the modeling and diagnosis of polar climate and sea ice in the CSM.

2. Increased NCAR staff time for the coordination, setup, and implementation of key model experiments involving polar climate and the sea ice model.

3. A scientist at NCAR, specializing in the physics of sea ice, whose main project is coordinating the community sea ice modeling effort in CSM.

These elements could take a variety of forms. For example, a postdoc or visiting scientist, who in principle might be funded under a number of different programs, could address item 3 effectively. Item 1 could be advanced by a combined effort involving the PCWG co-chairs and members, and CSM support for web site development, to significantly increase the engagement of the extra-NCAR community. Also proposals to work with CSM as the "flagship coupled climate model of the polar regions" might find support in the growing polar research programs of NSF and other agencies.

B. Model Development

1. The CSM ice model should abandon the CF rheology in favor of one of the elliptical-yield curve rheologies. Both the VP and EVP rheologies are good candidates for use in CSM, as they give similar results. The choice between them may depend on computational issues (performance, intended platforms) and the grid chosen for the ocean model.

2. The capability of multiple ice thickness categories must be added to the CSM ice model, with appropriate albedo, snow, and energy-conserving thermodynamics included as well. These physics may have significant impact on the model's sensitivity to changes in climate. A minimum of 3 ice thickness categories has been shown to be required for accurate representation of ice growth.

3. A column model of CSM that includes sea ice, flux coupler, and upper ocean is needed and should be part of the community development effort. A community-based workshop would be required to develop the specifications of a column model.

C. Needed/Planned Experiments

1. Baseline runs using VP and EVP dynamics using observational data for forcing and validation.

2. Sensitivity of CSM to the melting-ice albedo change.

3. CCM3 runs with specified spatially and seasonally-varying ice thickness distribution, as well as ice concentration now used.

4. Sensitivity runs of multiple-thickness, new albedo, and new thermodynamics implementations, using a hierarchy of models; a column CSM air-ice-ocean model; and full CCM3 with prescribed surface conditions.

D. Data Sets

1. Better observational datasets are needed (as is available from community) for forcing baseline ice and ice-ocean experiments and for validation.

2. Initial column-model (SCM) datasets for forcing and validation are needed and will be made available by SHEBA, the Atmospheric Radiation Measurement (ARM) program, and the First International Satellite Cloud Climatology Program (ISCCP) Radiation Experiment (FIRE).

3. More comprehensive CSM polar output is needed for community diagnostic studies.