New Paper: Trends in Arctic sea ice extent from CMIP5, CMIP3 and observations

[donsutherland1]

"Trends in Arctic Sea ice extent from CMIP5, CMIP3 and observations," (Julienne C. Stroeve, Vladimir Kattsov, Andrew Barrett, Mark Serreze, Tatiana Pavlova, Marika Holland, and Walter N. Meier, scientists affiliated with either NSIDC, the National Center for Atmospheric Research, or the Voeikov Main Geophysical Observatory in Russia) was published in the August 25, 2012 edition of Geophysical Research Letters.

http://www.agu.org/p...2GL052676.shtml

Some highlights:

- Arctic sea ice extent has had a declining linear trend in all months over the 1979-2011 timeframe

- The downward linear trend is largest in September

- Based on CMIP5, the estimated anthropogenic contribution ranges from 52% to 67% (+/- 1σ from the ensemble mean of around 60%), but the mean figure is "based on the assumption that the model simulations are members of the same distribution and that we have enough members to resolve the distribution" but that "assumption does not exactly hold since some models better represent the climate system than others."

- The impact of external anthropogenic forcing is "becoming increasingly clear in both the observations and model studies."

- CMIP5 better represents Arctic sea ice trends than CMIP3

- CMIP5's ensembles suggest that a seasonally ice-free Arctic (<1 million sq. km. ice extent) "within the next few decades is a distinct possibility." A single ensemble member is far more aggressive.

All in all, this is an interesting paper. It documents modeled vs. observed outcomes. It also supports the important role anthropogenic forcing is playing with respect to Arctic ice trends, even as uncertainties make it somewhat difficult to quantify the anthropogenic contribution with precision.

[ben4vols]

Some highlights:

- Arctic sea ice extent has had a declining linear trend in all months over the 1979-2011 timeframe

- The downward linear trend is largest in September

- Based on CMIP5, the estimated anthropogenic contribution ranges from 52% to 67% (+/- 1σ from the ensemble mean of around 60%), but the mean figure is "based on the assumption that the model simulations are members of the same distribution and that we have enough members to resolve the distribution" but that "assumption does not exactly hold since some models better represent the climate system than others."

- The impact of external anthropogenic forcing is "becoming increasingly clear in both the observations and model studies."

- CMIP5 better represents Arctic sea ice trends than CMIP3

- CMIP5's ensembles suggest that a seasonally ice-free Arctic (<1 million sq. km. ice extent) "within the next few decades is a distinct possibility." A single ensemble member is far more aggressive.

Models models models. So CMIP5 better represents current Arctic sea ice trends. Global climate models did pretty well with the last warming period (roughly 1970 to 2000) but failed to to come close to predicting any another time period outside of that 30 year span. I have a feeling that if we had reasonable Arctic ice data back to the late 1800's we would find that the CMIP5 and other models used to predict Arctic conditions fails to closely predict any period outside of the current period. So the 52% to 67% of anthropogenic contribution holds no real value. Much like the last study you posted.

[dabize]

Models models models. So CMIP5 better represents current Arctic sea ice trends. Global climate models did pretty well with the last warming period (roughly 1970 to 2000) but failed to to come close to predicting any another time period outside of that 30 year span. I have a feeling that if we had reasonable Arctic ice data back to the late 1800's we would find that the CMIP5 and other models used to predict Arctic conditions fails to closely predict any period outside of the current period. So the 52% to 67% of anthropogenic contribution holds no real value. Much like the last study you posted.

Post:

It is increasingly clear that AGW is driving Arctic Ice melt, and it may have changed the dynamic in a way that the models are having trouble with but are causing an accelerated change that could make the Arctic ice-free soon.

The Limbic View:

Models, models models.

Hey Doc, about that biopsy............

[donsutherland1]

Models models models. So CMIP5 better represents current Arctic sea ice trends. Global climate models did pretty well with the last warming period (roughly 1970 to 2000) but failed to to come close to predicting any another time period outside of that 30 year span. I have a feeling that if we had reasonable Arctic ice data back to the late 1800's we would find that the CMIP5 and other models used to predict Arctic conditions fails to closely predict any period outside of the current period. So the 52% to 67% of anthropogenic contribution holds no real value. Much like the last study you posted.

FYI, in all knowledge/information/quantitative-intensive fields models are used. They help explain the things that they were designed to represent (simplified representations of complex phenomena). They add insight into addressing scenarios for the future e.g., "what if" questions that might not be possible to answer if one relies strictly on the historical record in circumstances where no prior cases or precedents exist. They also make it possible to distribute expertise beyond what would be possible if one relied strictly on the handful of people who might hold expertise in a given area.

Of course, models, because they are simplified representations and knowledge is imperfect, have limitations. They are not perfect. However, they are continually improved as the knowledge base and computing power expand. Most importantly, they add value. Otherwise fields ranging from economics to meteorology to engineering to medicine would not utilize them.

Finally, with respect to the study in question, the analysis focused on the best-performing models. Stroeve et al explained:

A subset of the models was selected for calculating the multi-model ensemble mean and seasonal cycle. Models were selected by comparing the distribution of the simulated September extent with the observed range as computed over the period 1953-1995. Models with more than 75% of their distribution falling outside the observed range of 6.13 to 8.43 X 10⁶ km² were rejected.

...For completeness, we...show September trends for all CMIP5 models, including the rejected ones, and list multi-model ensemble means for both teh subset and the full model suite in Table S1 in the auxialiary material.

In other words, the analysis was performed with the models that best represented the observed data. At the same time, for purposes of completeness, a supplementary analysis was also performed that included all the models, including those that did not perform as well against the observed data.

[ben4vols]

In other words, the analysis was performed with the models that best represented the observed data. At the same time, for purposes of completeness, a supplementary analysis was also performed that included all the models, including those that did not perform as well against the observed data.

That still doesn't help the fact that there is nothing to compare the model results with past the mid to late 1900's to present in regards to Arctic conditions. Just because the models represent one time period well doesn't mean they will model all periods well.

[donsutherland1]

That still doesn't help the fact that there is nothing to compare the model results with past the mid to late 1900's to present in regards to Arctic conditions. Just because the models represent one time period well doesn't mean they will model all periods well.

There are earlier datasets that document an even longer-term decline (with fluctuations, which were largely due to internal variability when anthropogenic forcing was weaker):

http://www.ipcc.ch/p...igure-4-10.html

That a growing body of independent scientific research is all concluding that there is an anthropogenic contribution strengthens the case for such an explanation.

Aside from the Stroeve, et al., paper, one finds at least two other recent papers concluding that there is an anthropogenic role:

Notz, et al.: http://www.agu.org/p...2GL051094.shtml

Day , et al.: http://iopscience.io.../034011/article

If the observations increasingly cannot be explained by natural forcings and internal variability, some other factor must be responsible. If it is a natural factor, one must both identify it, which has yet to be done by those dismissing an AGW role, and then also explain why that factor didn't seem to play a large role until recently when it came to Arctic sea ice trends. Why now has it increasingly displaced the natural forcings (solar, orbital fluctuations, volcanoes) and internal variability (AMO, etc.) to the extent that those variables have decreasing explanatory power?

Scientists are concluding that anthropogenic forcing is the driver in question. Moreover, when anthropogenic forcing is combined with the natural variables, the newest modeling reasonably explains the Arctic sea ice trends (though still winding up on the conservative side of things in most cases). If anthropogenic forcing were irrelevant, then those seeking to argue against it need to explain why the anthropogenic element resolves the growing disparity between the outcomes expected solely from the natural forcings and the observed outcomes. The correlation is too strong and too persistent for the explanatory power of anthropogenic forcing to be merely a matter of temporary and random coincidence.

Scientists are publishing more and more works demonstrating a growing role for anthropogenic forcing, even amidst the natural variables and continuing uncertainties. Those dismissing the anthropogenic role have yet to offer a coherent and reasonably comprehensive alternative that explains the trends at least as well as the anthropogenic forcing.