Friday :: May 4, 2007

faster than forecast

by Christina Hulbe

A paper published this week in the journal Geophysical Research Letters (GRL press release) has received a fair bit of coverage in the popular media, though not exactly for the main points made by its authors. Julienne Stroeve and her co-authors compared model simulations with observations of Arctic sea ice extent (most reliably since 1979, when good satellite observation begins) and found that the models underestimate the rate of sea ice decline at two diagonostic times, September (the end of the melt season) and March (the end of winter).

The rapid decline in Arctic sea ice is not new (we've discussed it here in the past). What's new is the opportunity to gain insight into the processes underlying the change and into model fidelity to the "real world." Here are the key points made by Stroeve and colleagues regarding the Arctic sea ice:

1. if the IPCC AR4 multi-model mean time series properly reflect the response to [greenhouse gas] loading, then both natural variability and forced change have been strong players in the observed September and march trends, with the [forced change] becoming more dominant during 1979-2006;

2. given evidence that the IPCC models as a group are too conservative regarding their [greenhouse gas] response, the [greenhouse gas] imprint may be larger [than the above conclusion suggests];

The National Snow and Ice Data Center press release is here. More from me after the break.

Arctic variability
Interannual variability in Arctic climate is dominated by a strong atmospheric variability called the northern annular mode (NAM). The NAM is an oscillation of atmospheric mass between high and middle latitudes that results in distinct patterns of change in sea level pressure, temperature, storm tracks, and other weather phenomena.

From 1989 to 1995, the winter NAM was in an exceptionally strong positive phase. This means higher highs and lower lows, driving strong atmospheric circulation. The strong winds flushed thick multi-year sea ice out of the Arctic basin and into the north Atlantic via Fram Strait, leaving the Arctic ocean with relatively thin ice, vulnerable to summer melting. Atmospheric circulation and ocean heat transport have continued to favor sea ice loss since the mid 90's.

It is clear that natural variability is an important (perhaps leading) component of the recent overall downward trend in sea ice extent. It is not, however, all of the story.

anthropogenic forcing
While they underestimate its magnitude, state-of-the-art climate models all reproduce the overall negative trend in September sea ice extent when the anthropogenic greenhouse gas forcing is included. The models fail to reproduce the trend when greenhouse gas concentrations are held constant at pre-industrial values. Thus, we must conclude that both anthropogenic forcing and natural variability are required to explain the observed changes in the Arctic sea ice.

model performance
So what's missing from the models? Why do they underestimate the magnitude of the downward trend? One possibility is that the models might be underestimating NAM-type variability (pdf). Another possibility is that the resolution of the climate models is insufficient to capture all the relevant processes.

Some important sea ice processes take place over length scales that are smaller than climate model resolution (pdf of a technical paper). Stroeve and colleagues note that the two models in the IPCC ensemble that best reproduce the the observed decline since 1979 have relatively advanced sea ice components. Additionally, vertical structure of the ocean, which is important to heat transfer, might not be well enough resolved to capture important Arctic processes.

In the end, the story here is not that the Arctic sea ice is retreating "faster than expected," as some news outlets have reported, but that it is retreating faster than climate models can simulate. The distinction between these two is important. The current generation of climate system models are good but they need to be better. Analyses like the one conducted by Stroeve and her colleagues show us where effort should be spent to attain that goal.

Christina Hulbe :: 8:20 AM :: Comments (28) :: Digg It!