Report of the CCSM Polar Climate Working Group Meeting
8 February 2002
NCAR Chapman Room
Co-Chairs: Dick Moritz and Elizabeth Hunke
Julie Schramm provided an update on the Community Sea Ice Model Version 4 (CSIM4). CSIM4 code is organized using CVS (Concurrent Version System), which allows users to track changes in the code by revision, tag, and date. CVS is being used on all components of CCSM. Julie is in charge of checking in versions on the main CCSM trunk in CVS. As of 8 January 2002 version csim4_6_8 had been checked in.
The code for CSIM4 will be released with the other components of CCSM in May, 2002. The planned release of CSIM4 will include the following elements: code, Model Description, User's Guide and Reference Guide. The Model Description will document the fundamental equations and physics incorporated that comprise the model. The User's Guide will provide the information required to run the model, including lists of input files. This guide will also document the overall CCSM scripts that a user has to know. The Reference Guide will show how the subroutines branch. A nearly complete Model Description has been drafted, and is posted on the web at http://www.cesm.ucar.edu/models/ice-csim4/. An outline of the User's Guide has been drafted. The User's Guide and the Reference Guide remain to be written before the release date. Julie Schramm is working on the User's Guide, and Bruce Briegleb is working on the Reference Guide. The target date for completing these documents is 15 March.
The fundamental physical equations in the CSIM4 code have not changed for a long time. Changes since July 2001 include: addition of an ocean mixed layer model; addition of metric terms; addition of terms for calculating saturation vapor pressure over ice; increasing the albedo for dry snow surfaces; and changing the parameter that governs patchiness of snow cover. Also a number of non-physics changes have been made to enhance accuracy, and efficiency, to eliminate bugs, and to bring CSIM4 into compliance with the emerging CCSM2 standards and conventions.
Bruce Briegleb and Marika Holland presented aspects of polar climate as simulated by the CCSM2 in the ongoing 300+ year coupled run (designated as experiment b20.001). The simulation of polar atmospheric climate continues to differ in significant ways from observations, including too much wintertime cloud and downward longwave irradiance in the Arctic; summer clouds too optically thick for shortwave irradiance in the Arctic; arctic anticyclone displaced in winter driving sea ice away from the Canadian Archipelago and piling up ice on the coast of Eastern Siberia. The mean arctic sea ice extent is about right but the mean thickness is lower than observed, and the amplitude of the annual cycles of ice amount is larger than observed. In the Antarctic, especially the Weddell Sea, the simulated ice is thicker than observed. The Arctic Ocean has a halocline, maintained by the runoff model. Also the transport of ice and water volume through Fram Strait is within about 30% of the observational estimates.
Uma Bhatt presented results of ensemble experiments in which CCM3 is forced by maximum and minimum anomalies of monthly sea ice extent at the lower boundary. The simulated atmospheric response is statistically significant and exhibits dipole spatial patterns in the Atlantic and Pacific sectors of the Northern Hemisphere high latitudes. The results are qualitatively similar to those of Deser and Saravanan, who studied anomaly patterns associated with long term trends in sea ice. The ensemble of simulations was carried out at the Arctic Regional Supercomputing Center (www.arsc.edu).
Cecilia Bitz described the implementation of the sea ice physics from portions of CSIM4, into a standalone version of the Community Atmosphere Model (CAM). This implementation is thermodynamic only, there is no motion or transport of sea ice and no ridging. This provides the opportunity to prescribe ice concentration as well as ice thickness in experiments with the CAM. A concern of the AMWG has been that the prescribed sea ice model runs inefficiently (because of the way the coupler is coded currently), hence they wanted their own, more efficient version that includes the new sea ice thermodynamics. Marika Holland indicates that with a modest amount of effort, the efficiency of the prescribed sea ice model could be improved a lot.
Aaron Rivers summarized results from studies of cloud formation and radiative transfer using a single column model (SCM) version of CAM, forced by analyses and observations from the SHEBA project. The SCM simulation of SHEBA shares several features with the 3D simulation of arctic climate: too much cloud in winter; too much downward longwave at the surface; and not enough downward shortwave at the surface. Focusing on the oversimulation of wintertime low cloud, Aaron showed that the inclusion of a prognostic cloud water (PCW) model reduces the amount of winter cloud, except in the lowest model layer.
Bill Lipscomb presented a new transport scheme for climate models, called Incremental Remapping. This scheme enforces conservation by identifying as Lagrangian points, the locations at the previous time step of the current corner points of a grid cell, then integrating the mass or concentration of conserved quantities over this quadrilateral. Conservative transport then requires that this amount equal what's inside the grid cell at the current time step. This scheme becomes attractive compared to MPDATA because it is more accurate, and it is efficient if the number of conserved quantities becomes large. In the near future Bill will be setting up and running test cases with this new scheme, using the active ice only sea ice model.
Frank Bryan asked if anyone has run the CSIM on the earth simulation (vector) computer in Japan. The NCAR ocean group wants to port the code to the Fujitsu vector machine in Japan.
Jeff Kiehl announced that Phil Merilees has accepted the new CCSM liaison post and will begin working full time in September. Jeff also announced that Cecilia Bitz and Elizabeth Hunke will receive the CCSM Distinguished Achievement Award at the 2002 Workshop.
Bill Large asked whether we want to consider even higher (than X1) resolution sea ice modeling in the future? Discussion was inconclusive, but it was noted that there are few sea ice phenomena of interest in climate modeling that occur on scales smaller than the ocean and atmosphere forcing. Also there is concern that at scales smaller than 10 km, the basic model of sea ice as a continuum consisting of many discrete floes becomes unrealistic. Nevertheless, some notable problems in the existing CCSM could be addressed with sea ice at higher resolution, e.g., absence of motion in Denmark Strait when there is only one grid point there.
In general discussion, it was suggested that the PCWG form some small, focused study groups to address key problems that persist, such as Arctic winds; Arctic ice too thin; Antarctic ice too thick; and Arctic clouds.