Sea Ice Modeling: A Mini-Workshop
1. Executive Summary
A mini-workshop titled "Sea Ice Role in Global Climate Variability and Change" was held at the NASA Goddard Institute for Space Studies on 11-12 March 1999. The workshop brought together polar observers, remote sensing scientists, sea ice process modelers, and global climate modelers, who collectively assessed the adequacy of existing models and data sets and attempted to map out strategies to improve upon current understanding.
A summary of the workshop, written for publication in Transactions of the American Geophysical Union, is provided in the first section below. This is followed by the workshop agenda, rapporteur summaries of the workshop sessions, and a list of participants.
The workshop was successful in initiating discussions among a diverse group of researchers with differing needs for complexity in their treatment of sea ice, and we anticipate that this will be the beginning of fruitful scientific collaborations. We hope that with continuing interactions we can highlight the importance of polar processes in global climate and help assure that appropriate observations and data are acquired and analyzed.
In the final session of the workshop we discussed priorities for developing improved understanding of the role of sea ice in climate variability and climate change. Although there was no attempt to develop a prioritized list of recommendations, the discussion highlighted the following areas:
(1) Sea Ice Data Sets: Unrealistic temporal fluctuations of sea ice cover in existing sea ice data sets, due to data gaps and inconsistencies in data collection, are a great handicap for modeling and analysis of the role of sea ice in climate variability. Apparently the only attempt at present to develop a homogenized sea ice record for the past century or even half century is an update of the Global Sea Ice and Sea Surface Temperature (GISST) record, but this has been delayed and at present climate modelers have no homogenized sea ice data set. Furthermore, for the purpose of understanding long-term climate change, it is particularly important that the process of homogenization not remove real long-term sea ice change as well as discontinuities.
It is recommended that there be a concerted effort to develop a best estimate and uncertainty limits for sea ice change from the pre-industrial era (say the late 1800s) to the present. Greater spatial and temporal detail, with lesser uncertainties, should be possible for successive subintervals of the data, say for the period just after World War II to the present, and for the period of satellite data. With these data it will be possible to carry out sensitivity studies and analyses that help clarify the role of sea ice in climate variability and climate change.
(2) Global Observations: Good global data sets are required to test adequately climate models and their ability to simulate global changes of sea ice. For state-of-the-art global climate models better data are needed particularly for global distributions of
- sea ice thickness, including seasonal change,
- surface fluxes over sea-ice, or their proxies (surface temperature, humidity, wind),
- sea ice concentration and extent (including seasonal change),
- sea ice motion.
In the case of sea ice concentration, it may be possible to improve current data sets based on passive microwave satellite observations by further analysis of the data already available, although these data sets begin only in the 1970s. In the case of the distribution of sea ice thickness, it may be possible to combine microwave radar data with model simulations to infer ice thickness. The sea ice motion will become more important in the future as the resolution of climate models improves and they become better able to simulate sea ice dynamics.
(3) Modeling: There is clearly a disconnect between the needs of large scale [general circulation model (GCM)] modelers and the priorities of modelers who work on smaller scale regional ice processes. Currently, there are few (if any) systematic efforts to extrapolate very detailed physics and parameterizations in high resolution models to coarser resolution. Since most ice physics is heavily scale-dependent, this is of some concern. It is important because GCM modelers need to know, in order to construct the GCMs appropriately, 1) what elements of the physics are most important for climate and long term variability, and 2) whether the net effects of multiple small scale processes and components can be effectively parameterized at larger scales. Are there simple rules of thumb that a GCM can use that do a reasonable job of simulating, for instance, the ratio of lateral to basal melting at the large scale, without having to consider large numbers of thickness categories and temperature profiles within a single grid box? At what point do such simple parameterizations break down?
We recommend that an effort be made to bridge these gaps by specifically funding researchers to agglomerate information from high resolution ice models to the larger scales of typical GCM grid box sizes (from 4 degree to about 1 degree resolutions). The results from this will be important in assessing the minimum specifications of resolution, physics and complexity needed for the next generation of GCMs to address climate change properly. Success in this activity depends upon further regional processes observational studies and associated modeling.
(4) Regional Process Studies: Several complementary efforts are necessary to devise and assess potential parameterizations for sea ice processes that are sub-grid scale in current and foreseeable global climate models, specifically: 1) regional observational studies and associated high-resolution modeling to improve understanding of local climate-relevant sea ice processes; 2) parameterization of the aggregate consequences of these fine-scale processes when averaged over the larger spatial scales typical of climate models; and 3) evaluation of novel sea ice parameterization ideas via inclusion in global climate models. This latter requires consensus measures of evaluation for the performance of the sea ice component of climate models (Kreyscher, M., M. Harder, P. Lemke and G. Flato. Results of the Sea Ice Model Intercomparison Project: Evaluation of sea-ice rheology schemes for use in climate simulations. J. Geophys. Res. — Oceans, submitted).
Specific recommendations in these areas are:
Observational experiments, complementary to those of SHEBA, are necessary to characterize sea ice processes in regions of unique sea ice behavior, e.g., the Southern Ocean and the marginal ice zone. Numerical simulations that begin to assess the feedbacks between highly resolved regional atmosphere/ocean/ice processes and the larger scale climate system should also be pursued. Possible approaches to the latter that require careful investigation include climate modeling on unstructured and nested spatial grids.
We recommend that an effort be made to bridge the gap between high-resolution, regional models and more coarsely resolved climate models by supporting researchers to agglomerate information from high-resolution ice models to the GCM gridbox scale, as mentioned above.
Consensus metrics for the evaluation of the sea ice component of global climate models should be developed as the basis for assessing proposed sea ice parameterizations. Systematic sensitivity studies and intercomparison projects should be conducted to document the improvements (or lack thereof) that accompany the utilization of new sea ice parameterizations and the magnitude of the impact that the revised parameterizations have on the climate simulations.