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

Report of the CSM Polar Climate Working Group Meeting

Fourth Annual Climate System Model Workshop

by Dick Moritz, Co-chair

June 23, 1999

The CSM Polar Climate Working Group (PCWG) met on Wednesday, 23 June 1999, in Breckenridge, Colorado, as part of the Fourth Annual CSM Workshop. A list of attendees is at the end of this report (33 people attended this meeting). The agenda for the PCWG meeting included presentations, discussion of elements of the polar region physics in the CSM, and discussion of the application of CSM to studies of polar climate.

1. Sea ice model development

The elements of sea ice model physics under development for incorporation into CSM fall into three categories: (i) Plastic rheology using an elliptical yield curve (versus the cavitating fluid model in CSM 1.2); (ii) Ice thickness distribution to specify the ice mass balance (versus the mean ice thickness and ice concentration in CSM 1.2); and (iii) Enhanced ice thermodynamics, including the representation of snow cover, melt ponds, surface albedo, internal ice temperature, and enthalpy.

The status of CSM PCWG activities relevant to each category is as follows:

A. Two elliptical yield curve rheologies have been developed for the CSM application. J. Zhang (University of Washington) has developed a viscous-plastic (VP) rheology that uses the ADI solution method. This code has been applied to several ice-ocean model simulations. B. Briegleb (NCAR) has implemented this VP ADI code at NCAR, and it is being used in a CSM fully-coupled integration now underway. E. Hunke (Los Alamos National Laboratory, LANL) has developed an elastic-viscous-plastic (EVP) model as part of the LANL CICE model. This code has been applied to ice-ocean simulations in collaboration with Y. Zhang (Naval Postgraduate School).

B. An ice thickness distribution model has been developed by C. Bitz (University of Washington) and has been implemented in the University of Victoria general circulation model (GCM). Two-dimensional and three-dimensional simulation results reported by M. Holland (NCAR) and C. Bitz indicate that the resolution of more than one ice category significantly affects the seasonal and spatial features of simulated Arctic climate, and that the ice thickness distribution also affects the simulated climate interactions between the Arctic and lower latitudes. W. Lipscomb (LANL) has developed an ice thickness distribution model and has reported results from one-dimensional sensitivity studies. The results suggest that, with a resolution in the range 4 to 10 categories of ice thickness, the simulated mass balance begins to converge. These studies indicate that useful improvements in both ice thickness distribution and enhanced thermodynamics are possible in the context of less than a factor of two increase in computational load from the sea ice portion of the CSM domain.

C. Before ice thickness distribution and enhanced thermodynamics are implemented in CSM, some additional experiments are needed. For example, ice/ocean models forced by high-quality atmospheric forcing data are required to judge alternate model parameterizations. Analogous to AMIP, these experiments should aim to fix all experimental conditions except the alternate parameterizations to minimize ambiguity in the interpretation of the results. (Bitz, Briegleb, Hunke, Zhang, Moritz, Weatherly)

2. The new CSM Ocean Model and new NCAR computers

In light of the decision to move to the POP ocean model for CSM, there was discussion of the need to run the CSM sea ice model on the POP ocean grid. The status of these efforts is as follows:

* EVP Model: Done, currently runs on POP grid.

* ADI VP Model: Conversion to general orthogonal coordinates (including POP) to be completed by autumn, 1999 (Briegleb).

It was also noted that the change from vector to parallel computers at NCAR will require that the sea ice model codes be ported and parallelized.

The status of these efforts is as follows:

* EVP Model: Parallelization already underway; porting to NCAR machines to begin late summer, 1999 (Hunke).

* ADI VP Model: Porting and parallelization to follow conversion to general orthogonal coordinates (Briegleb, J. Zhang).

3. CSM performance in polar regions

A. Surface incident radiation fluxes in CCM 3.6

A. Rivers and R. Moritz used input data from the SHEBA experiment to prescribe the radiatively active components of the arctic atmospheric column and performed Instantaneous Radiative Flux (IRF) experiments using the CCM 3.6 Column Radiation Model (CRM). The simulated broadband incident longwave and shortwave irradiances were compared with simultaneous measurements. The results show:

* The simulated incident shortwave under clear skies matches the observations to within the irradiance measurement error.

* CCM 3.6 undersimulates incident longwave under clear skies by 15 W/m2.

* CCM 3.6 undersimulates incident shortwave under overcast skies by 15 W/M2.

Sensitivity studies indicate that the discrepancies may stem from the water vapor absorption formulas (clear sky LW) and either the parameterization for CWP or the prescribed (constant) liquid water equivalent radius (overcast SW).

During summer, 1999, additional studies will be performed using the SHEBA data with CRM codes that are potential candidates for use in CCM4 (Moritz, Rivers, Collins).

B. Meridional sea ice transport

B. Briegleb presented results from CSM simulations showing oversimulation of the equatorward transport of sea ice on the poleward side of the Antarctic Circumpolar Current (ACC) and a general unrealistic pattern of ice transport throughout the Arctic Ocean and on the polar margins of the North Atlantic.

The problem near the ACC could be related to the ocean model, the atmosphere model, or a combination of both, not necessarily the sea ice model. During summer, 1999, additional diagnostic studies of the CSM formulations of atmosphere/ice stress and ice/ocean stress will be performed with an eye to this problem (Weatherly, Briegleb, Moritz).

The unrealistic pattern of ice drift in the Arctic seems to be due at least in part to differences between the CCM simulated surface winds and the observed surface winds. During summer/fall 1999, Target Data Sets for the fields of sea level pressure, surface geostrophic wind velocity, and ice velocity will be prepared, documented, and made available through the CSM web site to support diagnostic studies of CCM simulations of arctic winds (Moritz). Ice/ocean simulation experiments will be performed using these Target Data Sets as forcing functions to develop performance metrics for the CCM surface winds that should assure they will drive a realistic circulation of the sea ice (and the wind driven Arctic Ocean circulation (Bitz, Briegleb, Moritz). Also the PCWG will evaluate new CCM simulations of arctic sea level pressure and winds as they become available and in collaboration with CCM investigators. The goal is to define the factors necessary for CCM to provide a simulation of arctic surface winds (SLP) good enough to drive an accurate ice/ocean circulation. Candidate factors include: CCM spatial resolution and the resolution of orography and CCM surface boundary conditions.

It will be important to resolve any outstanding questions/problems concerning the accuracy of CSM and CCM representations of air/ice and ice/ocean stress to adequately address this set of problems.

4. Impact of planned CSM changes

It was widely agreed that to perform meaningful studies of the Arctic Ocean circulation, stratification, and freshwater balance, it is essential to have an open Bering Strait (and perhaps even an open Canadian Archipelago). Since this will be possible with the new POP ocean model, we can look forward to intensified use of CSM for studies of polar climate dynamics.

The freshwater budget of the Arctic Ocean, and its influence on global thermohaline circulation, depends on the large freshwater input from the rivers of Eurasia and North America. Thus, the inclusion of a runoff model in the next version of CSM is also very important for polar climate dynamics.

5. Interannual-Interdecadal variability involving the Arctic climate system

This is a topic of growing interest, spurred by observations of significant interdecadal changes in the circulation and water mass distribution in the Arctic Ocean, together with pronounced differences in the surface atmospheric circulation between the 1980's and the 1990's. These and other phenomena appear to be related in a mode of variability known as the Arctic Oscillation with coherent variability extending from the stratosphere into at least the mid-depths of the Arctic Ocean.

A. Proshutinsky reported studies with an ocean-ice model, forced by observed winds, that indicate regime behavior with warm/cold (cyclonic anticyclonic) regimes persisting on the order of 10 to 15 years over the Arctic. These regimes, as characterized by EOF analyses of SLP, are also seen in some manifestation in the CSM, although the amplitude is smaller than observed, and the higher EOF's are not reproduced as well as the leading EOF.

The observed variability, which is known variously as the Arctic Change, Arctic Oscillation, and North Atlantic Oscillation, presents a new challenge to understand how the climate system functions. Furthermore, simulating the variability accurately with the CSM would enhance our confidence in CSM simulations of future climate change due to anthropogenic and other forcing.

6. New hydrographic data set

M. Steele and R. Morley of PSC presented a new version of the global Levitus temperature and salinity data set with much improved data coverage and accuracy in the Arctic. This new data set incorporates Soviet hydrographic survey measurements made in the 1950's to 1990's and recent measurements made by western countries. Differences between the new, revised data set and the Levitus data set are significant in the Arctic Ocean. The new fields are much more realistic there. The Levitus data are reproduced outside the Arctic. The data set is available through the PSC web site at: http://psc.apl.washington.edu/Climatology.html.

7. Timeline for PCWG to do items (some items tentative, depending on developments at NCAR)

July, 1999 - January, 2000

1. Target data sets SLP, G, Uice.

2. Extend SHEBA CRM IRF analysis to candidate CCM4 Radiative Transfer Models.

3. Perform diagnostic analyses of CCM output and ice/ocean comparison studies to determine performance criteria for CCM simulations of arctic winds that would be adequate for driving the sea ice model.

4. Convert ADI VP model to generalized orthogonal coordinates and put on the POP grid.

5. Porting and parallelization work for ice models.

6. Code and test full ice/ocean models that incorporate the ice thickness distribution and the enhanced thermodynamics.

January, 2000: POP ready to conduct ice/ocean experiments

January - June, 2000

1. Conduct ice/ocean model intercomparison and diagnostic studies taking advantage of the modularity.

2. Produce code that implements one or more enhanced ice models within the CSM 2.0 framework.

3. Work with CCM4 developers to investigate the factors that control those aspects of the simulated arctic winds that are crucial for driving the ocean/ice model.

June, 2000

1. Recommendation to the SSC concerning the ice model for CSM 2.0.

List of Participants

Uma Bhatt, International Arctic Research Center (IARC)
Cecilia Bitz, University of Washington
Maurice Blackmon, NCAR
Bruce Briegleb, NCAR
Frank Bryan, NCAR
John Davis, Los Alamos National Laboratory
Peter Eltgroth, Lawrence Livermore National Laboratory
Peter Gent, NCAR
Cecile Hannay, IARC, Frontier University of Alaska
William Hibler, University of Alaska
Marika Holland, NCAR
William Holland, NCAR
Elizabeth Hunke, Los Alamos National Laboratory
Steven Jayne, NCAR
Philip Jones, Los Alamos National Laboratory
Brian Kauffman, NCAR
William Lipscomb, Los Alamos National Laboratory
Robert Malone, Los Alamos National Laboratory
Mathew Maltrud, Los Alamos National Laboratory
Richard Moritz, University of Washington
Norikazu Nakashiki, NCAR and CRIEPI
Linda Peters, Lawrence Livermore National Laboratory
Andrey Proshutinsky, University of Alaska-Fairbanks
Aaron Rivers, University of Washington
Daniel Robitaille, Lawrence Berkeley National Laboratory
Albert Semtner, Naval Postgraduate School
Jacob Sewall, University of California, Santa Cruz
Richard Smith, Los Alamos National Laboratory
Michael Steele, University of Washington
Stephen Vavrus, University of Wisconsin-Madison
John Weatherly, U.S. Army, CRREL
Jinlun Zhang, University of Washington
Yuxia Zhang, Naval Postgraduate School