[an error occurred while processing this directive]
[an error occurred while processing this directive]
Report on Ice Modeling for the NCAR CSM Edited by Dick Moritz and Elizabeth Hunke
Report on Ice Modeling for the NCAR CSM
Edited by Dick Moritz and Elizabeth Hunke
To accelerate progress toward a suitable sea ice model in the NCAR Climate System Model (CSM), a meeting of sea ice modelers actively engaged in CSM development was held at NCAR on 27-28 September 1999. The meeting was organized by Dick Moritz and Elizabeth Hunke (co-chairs of the CSM Polar Climate Working Group) and Maurice Blackmon. Lydia Harper and JoAnne Martin of the NCAR CSM Program Office provided logistical support for the meeting.
The purpose of the meeting was to identify and prioritize the problem areas and tasks that need to be addressed to produce a good sea ice model for implementation in CSM 2.0 and to get commitments from meeting participants to work on these problems and tasks in a specified time frame.
The meeting agenda was designed to proceed from a common understanding of what we want to achieve and where we stand, through in-depth discussion of problem areas, to commitments by individuals to address specific problems and tasks.
2. Goals and Objectives
Each meeting participant gave a brief presentation of their goals and objectives related to CSM sea ice modeling. The general goals of the group, and of the CSM polar climate community, are to develop and implement a global, coupled CSM that simulates polar climate and its interactions well, and to make the model available for use by the climate research community. The community-use aspect is especially important as the unique feature that distinguishes CSM from other climate modeling activities. This aspect includes availability of model components and model output for downloading, documentation of model physics, code and experiments, and maintenance of code and data sets.
Thus, the specific goal of the group is to develop and implement sea ice models and an interface with the flux coupler for CSM 2.0 release. The target date for this goal is June, 2000. Another more specific goal of the group is to help prescribe the ice state at the surface for model validation, initializations, and for forcing partially decoupled experiments as part of the model development/evaluation process.
Several participants noted that their interest is to study polar climate, including its variability, future climate change, and interactions with the global thermohaline ocean circulation and the hydrologic cycle, and that to pursue this interest requires that CSM have an appropriate sea ice model, better than the current (CSM 1.0) sea ice model.
The specific objectives of individuals and subgroups varied, as outlined here:
- LANL CICE (Hunke, Lipscomb, Briegleb): Enhance ice physics and numerical codes. Perform validated ice/ocean integrations. Use CICE in CSM to evaluate, diagnose, and improve CICE.
- VP ADI (Zhang, Briegleb): Enhance ice physics and numerical codes. Implement in CSM. Perform data assimilation studies for validation.
- ITD (Bitz, Briegleb, Lipscomb): Develop a physically sound, well-tested model that represents the state variables and processes relevant to climate. Test, diagnose, evaluate, and improve code. Implement in CSM 2.0.
- Thermodynamics: (Lipscomb, Briegleb, Bitz): Implement energy conserving physics governing ice temperature and salinity with explicit resolution of melt ponds and their albedo effects.
- Transport (Bitz, Briegleb, Hunke, Zhang): Implement a second order advection scheme for CSM sea ice.
3. Status of CSM Ice Modeling
A. Latest CSM Coupled Model
CSM has been integrated with the NCOM "x2" ocean model and the VP ADI sea ice model (on a north polar grid) with river runoff and the Bering Strait closed. The integration was spun up for 20 years, then run for 40 years in coupled mode. Qualitatively, the sea ice and polar climates look reasonable in terms of ice area, ice volume, spatial variation of ice thickness, surface heat flux, ocean heat flux, growth rates, and ice drift. Quantitatively, there are large discrepancies between this simulation and observed polar climate. Mean simulated ice thickness in the East Siberian seas is 6 to 7 meters compared to less than 2 meters observed. The model oversimulates ice drift in the Beaufort Gyre compared to the Transpolar Drift Stream, and it transports too much ice too rapidly from the Beaufort Sea toward the Chukchi and East Siberian Sea. The katabatic/coastal winds and the ice circulations they drive around the Antarctic coast appear to extend too far offshore. Simulated southward transport of sea ice through Fram Strait is close to observed, but there are compensating errors in ice velocity (high) and ice thickness (low).
Factors that may affect discrepancies between the CSM control simulation of sea ice and observations include:
B. LANL CICE
This model now runs coupled to the POP ocean model on generalized orthogonal displaced pole grids. The model uses the Winton 3-layer thermodynamics, represents frazil ice formation, and computes surface fluxes based on surface roughness. Transport is accomplished by the second order MPDATA scheme, and the LANL remapping advection scheme is being tested. The EVP stress states now converge more closely to the elliptical yield curve. The model communicates with the CSM flux coupler, and CICE documentation is being kept up to date and will be made available on the LANL CICE web site.
C. VP ADI
This model has now been formulated for general orthogonal coordinates, and a code for this has been written but not tested. Data assimilation studies with the VP ADI demonstrate the sensitivity of results to including/excluding observations of ice motion. There are different perceptions about the ultimate numerical efficiency that would be achieved by VP ADI in a parallelized, CSM implementation on the POP grid, compared to LANL CICE. Some participants think it would be more efficient; some think it would be less efficient.
D. Bitz ITD
The Bitz ITD model is being documented as an APL Technical Memo and will be made available for download from a University of Washington web site. The user base for this model has grown to include the Naval Postgraduate School, Canadian Climate Centre, University of Victoria, and researchers at the University of Illinois and the U.S. National Ice Center. The current plan is to implement this in the Parallel Climate Model (PCM) within the CICE framework, which will transfer directly to CSM.
E. CICE Thermodynamics and Mass Balance
A model of ice state is under development that will resolve a number of thickness categories and in each category will distinguish ridged and unridged ice and ponded and unponded ice. The model is now running in 1-D mode.
F. Ocean Heat Flux
The current CSM ocean model computes a freezing/melting potential as the product of seawater density, specific heat capacity, and upper ocean water temperature departure from the freezing point. It is recommended that the CSM ice model be changed to use the McPhee and Maykut parameterization for oceanic heat flux, which depends on the friction velocity in the upper ocean. Also needed is a parameterization that realistically apportions this heat among basal and lateral ablation during melt.
The Parallel Climate Model (PCM) group at NCAR and collaborating institutions emphasizes the use of modern climate models to perform ensemble integrations and is not aiming to provide community models. Currently the PCM uses the POP ocean on 1/3 or 2/3 degree grids, the Y. Zhang sea ice model, and the CCM3, T42 atmosphere. In the future, PCM will incorporate the CICE EVP sea ice model and the Bitz ITD model (approximate time frame: 2 months). These components will be made available to CSM if desired. In the more distant future, PCM hopes to run the CCM at T85 resolution.
4. Key Problems of CSM Ice Modeling
Under development at LANL and NCAR is a CSM standard coupler, which will replace the CSM flux coupler used in CSM 1.0. Lists of variables have been developed to be passed to/from the coupler and the ocean and the atmosphere models. Elizabeth Hunke presented these lists (Tables I and II).
The order of computation in a time step was discussed. The order currently envisioned in the CSM standard coupler is indicated in Table III. Some modifications to this order were discussed. A potential problem arises when the ice fraction changes between zero and non-zero values in a grid cell, associated with passing quantities per unit ice area to/from the flux coupler. It was suggested that the group consider passing quantities per grid box area instead. The coupler operates in the context of different time steps: 20 minutes for CCM dynamics, 1 hour for ice-atmosphere coupling, 2 hours for radiation, and 24 hours for the ocean model. E. Hunke and C. Bitz will propose changes in the coupler to address the ice area problem.
B. Ice Physics
The group discussed aspects of representing more complicated physical states, and associated processes, in the CSM sea ice model. These include: salinity profile in the ice, melt ponds, albedo depending on ice thickness, ocean heat flux formulation, lateral/basal melt formulation, snow depth distribution, conductivity of the snow cover, horizontal ice transport, ice strength formulation, and ice ridging formulation.
5. Priorities for CSM Ice Modeling Work
The discussion of priorities was motivated by the desire to focus on the variables and processes most important for climate and to assign priority to tasks that are feasible within a time horizon of about 6 months.
Topics that were discussed in some detail include: ITD, surface albedo parameterization, transport algorithm, resolving the ice temperature and salinity profiles, explicit melt ponds, snow ice and flooding of ice top surface depressed below sea level, lateral/basal melt formulation, and oceanic heat flux. In Section 6 (below), the results of these discussions are presented as action items assigned to individuals, with a priority level of 1 or 2.
Special attention was given to the importance of a test-bed framework at NCAR, within which candidate formulations for ice model components can be tested. It was agreed that such a framework should support ice-only simulations, forced by prescribed atmosphere and ocean fields. The forcing fields should be selectable among observations (e.g., NCEP, ECMWF, IABP data sets) or model output (e.g., CCM, NCOM, POP). The framework should include the coupler, so that the forcing fields are passed to the sea ice model in the same manner as the atmosphere and ocean fields pass in the full CSM.
6. Action Items and Commitments
In this section, each task is given a priority level and is listed with the individual(s) who will do the work. Priority level "1" work has started, or is to start immediately, and aims for significant results by January, 2000, and implementation in CSM 2.0, target date June, 2000. Priority level "2" is research and development that may be important for sea ice modeling and merits effort as part of CSM; but because of remaining uncertainties about approach or formulation or relevance to climate, this work will proceed so as not to interfere with the priority 1 tasks. It is likely that some priority "2" model developments will be ready for CSM 2.0 and some will not. A special, highest priority of "0" was assigned to the ice-only model diagnosis framework at NCAR. The initials of the lead person for each item are listed first.
|*Set Up Ice-Only Framework at NCAR||BB, EH, RM|
|Implement Bitz ITD in CSM||CB|
|Test/Evaluate Remapping Advection||EH, BL, CB|
|Review/Evaluate CSM Surface Albedo Over Sea Ice||RM, BB|
|Implement Lipscomb ITD/Thermo in CICE and PCM||BL, EH|
|Propose Changes in Ice Area/Grid Passing of Variables in Coupler||CB, EH|
|Implement M&M Ocean Heat Flux||MH, BB|
|Analyze CSM Ice-Ocean Drag Formula||PG, RM|
|Implement Fixed Ice Salinity Model||BL, EH, CB, BB, MH|
|Analyze Hibler/Bryan Ice-Ocean Stress Problem||CB, RM, EH|
|Test/Evaluate ADI VP General Ortho Coordinate Model||JZ, BB, EH|
|Provide First Updated Sea Ice Model for NCAR/LANL Ice-Ocean Spinup||PG, EH|
|Implement Explicit Melt Ponds in CICE||BL|
|Implement Lateral/Basal Melt in CICE||BL, EH, CB|
|Adjust Snow Conductivity||BL|
|Adjust Ice Strength to Improve Simulated u, h||JZ, BB, EH, BL, CB, MH|
|Simulate Sensitivity to Ocean Heat Flux, Wind Stress||CB, MH|
|*This item includes data sets.|
|BB - Bruce Briegleb|
|BL - Bill Lipscomb|
|CB - Cecilia Bitz|
|EH - Elizabeth Hunke|
|JZ - Jinlun Zhang|
|MH - Marika Holland|
|PG - Peter Gent|
|RM - Dick Moritz|
7. Follow Up to This Meeting
This report will be edited by Dick Moritz and Elizabeth Hunke, then submitted to the CSM Scientific Steering Committee and posted on the CSM Polar Climate web page. An e-mail announcement of the report will be sent to all attendees of the PCWG meeting. Dick Moritz will contact groups with sea ice modeling programs to alert them to the CSM activities, including CRREL, University of Colorado, University of Alaska-Fairbanks, Ohio State University (BPRC), University of Wisconsin (CRC), and University of Washington (PSC, At Sci and JISAO).
The co-chairs will help organize another CSM PCWG meeting, probably in January, 2000, and as part of that activity, will plan a session to report results of the action items and to update/augment the table of action items.
Appendix I: CSM Ice Modeling Meeting Agenda
Director's Conference Room
National Center for Atmospheric Research
MONDAY, 27 September 1999
0800 - Coffee and Pastries (Available in Conference Room)
0830 - Approval of Agenda, Introduction (Dick Moritz)
0850 - CSM-Ice Modeling: Goals, Objectives, Milestones (max 10 minutes each)