Sea Ice Modeling: A Mini-Workshop
Session 1. Introductory Session
(Rapporteur: Michael Steele)
Jim Hansen opened the workshop. He noted the interest in polar climate change and the need to represent it well in global climate models. He emphasized the need to make a case to NASA for more support for polar research in a competitive funding environment.
Prasad Gogineni gave us a financial breakdown of how polar research money is now spent (roughly equal support for ice sheet mass balance and sea ice research). A goal of sea ice research is improving the simulation of high latitudes in GCM's; another is understanding the role of ice-covered oceans in global deepwater formation. Roughly 70% of research money is directed towards process studies, with the balance devoted to algorithm development. Gogineni said that his goals for the workshop were to obtain answers to: (1) Where are we? and (2) Where do we go from here? He requested recommendations for observations in the next 5-10 years. He ended by noting NASA's strong investment in RADARSAT Geophysical Processor System (RGPS), and sought feedback on the value of this investment to the research community.
Kenneth Bergman said that his goal for the workshop was to understand how to parameterize sea ice into global coupled GCM's.
Kim Partington said that his goal for the workshop was to create a prioritized list of issues (algorithms, modeling, etc) to be addressed. He also noted the new Navy PIP3.0 model development as an example of potential interagency synergy.
Michael Steele opened the scientific talks by summarizing recent observations of decadal climate change in the Arctic region. He focused on oceanographic observations and models that show subsurface warming and near-surface salinity shifts. The former probably originates in the open waters of the Norwegian Sea, where air temperatures are recently warmer owing to shifts in the Arctic Oscillation. The latter (salinity shifts) are probably a more local effect, created by wind patterns in the Arctic Ocean that control where fresh river waters flow into the deep basins. Evidence was also presented that showed how the usual Beaufort high pressure cell has weakened in recent years. Concurrently, a variety of sea ice changes have occurred: extent decreases (measured by passive microwave satellites), volume decreases (model output) and Fram Strait flux increases (model output). Much of this change can probably be attributed to shifts in the Arctic Oscillation; however a full understanding of the problem is still elusive.
Douglas Martinson gave a comprehensive summary of the differences between Arctic and Antarctic climate. Of particular note are the air temperature regimes over sea ice: it is generally warmer over Antarctic sea ice, although temperature gradients near the coast are probably greater there. Also, the Antarctic pack is divergent, which necessarily leads to a seasonal ice pack in view of the above-freezing summer temperatures. Martinson discussed both the Arctic and Antarctic Oscillations recently discovered by Thompson and Wallace, which show remarkably similar patterns.
David Rind discussed the role of sea ice in global climate simulations. He asked three questions: (1) What is the influence of sea ice on global climate? (2) How sensitive is sea ice to climate change? (3) What factors affect sea ice influence and sensitivity? To answer the first question, Rind first showed results from a simulation in which sea ice concentration was artificially varied by a given percentage; surface air temperatures were significantly affected only in polar regions, as the sea surface temperatures were not allowed to change in other areas. When they are allowed to change, other simulations show a significant effect in lower latitudes on the climatological time-scale. To achieve this response requires the interaction between sea ice and polar clouds: a GCM must allow both to vary in order to obtain a realistic climate warming simulation; without sea ice changes, the global sensitivity is reduced by 35-40%. Rind stressed that it was cloud physics that communicates sea ice changes to the rest of the globe, at which point the water vapor response becomes strongly involved. Addressing his second question, Rind noted that sea ice changes in the Southern Hemisphere differ greatly between q-flux models and coupled models, and even different versions of coupled models produce different results. Hence results in this area are very sensitive to model formulation, both for the ocean and sea ice codes. Addressing the third question, Rind stressed the importance of a realistic control simulation when examining sea ice sensitivity to climate warming. Those simulations with the highest sensitivity have relatively thin Arctic ice and relatively large Antarctic ice extents; under warming scenarios these simulations create catastrophic loss of sea ice mass and thus large climate changes.
James Morison extended Steele's and Martinson's discussions of Arctic climate change, emphasizing that signals have appeared in numerous physical parameters such as permafrost, the upper ocean, the sea ice, the troposphere and the stratosphere. He presented a schematic picture of how these changes might be linked, as well as a plan for how they might be studied in the future. Morison is spearheading a new multi-agency initiative known as SEARCH (Study of the Environmental ARctic CHange) which will focus on decadal scale changes of these physical parameters. He noted that this type of climate study might be modeled after similar efforts to understand ENSO, wherein efforts were focused on observations, models, and process studies. Morison closed by proposing that models might provide a crucial component in this initiative via sea ice thickness assimilation studies, in which sparse observations would be merged with model output to provide optimal estimates of the sea ice mass balance.
Hansen mentioned that the perceived discrepancy in model estimates of the magnitude of the sea ice feedback in global climate is more a matter of bookkeeping rather than a real difference. Some climate model studies testing the effect of changed sea ice "credit" the entire temperature change to sea ice while others divide it among the sea ice, water vapor, and cloud changes that occurred in the experiment. If the impacts of water vapor and cloud changes induced by sea ice change are included, the global climate effect of sea ice change is large. He noted that there should be no disagreement that sea ice is a major player in global climate sensitivity. He suggested that people look at the poster by Nick Tausnev, et al., which shows that the GISST Sea Ice changes for 1950-1998 (even if they are exaggerated by a factor of two) indicate that sea ice has been a major contributor to global temperature change in the past 50 years.
Parkinson asked whether the Arctic Oscillation might reverse soon and whether that could lead to cooling and thickening of the ice cover. Morison responded by noting the presence of both decadal oscillations and a longer term trend in recent decades. Serreze noted that the Arctic Oscillation has been reversing during the past 2 years.
The rest of the discussion was largely devoted to descriptions and explanations of the Arctic Oscillation. Schmidt noted that stratospheric dynamics appear to be necessary for the model trend in sea-level pressure (SLP) to resemble the observed Arctic Oscillation (AO). Flato stated that he was able to obtain a realistic model Arctic Oscillation with no stratosphere. Serreze suggested that ozone was a key ingredient. Miller noted that anthropogenic greenhouse forcing caused the SLP and AO trends in the GISS model, and that ozone seemed to have no effect upon either trend. Overland noted that the troposphere and the stratosphere are intimately linked, and that a signal which originates in one will inevitably propagate to the other. Shindell said that observations indicate that the SLP trend is "top down," i.e. of stratospheric origin.