The second Sea Ice Outlook of this year has been published. The SIO is organized by the Sea Ice Prediction Network (as part of the Arctic research program 'Study of Environmental Arctic Change', or SEARCH), and is a compilation of projections for the September 2016 Arctic sea ice extent, based on NSIDC monthly extent values. These projections are submitted by professionals as well as amateurs (public outlooks).
Here's the summary for the July report:
This month the median pan-Arctic extent Outlook for September 2016 sea ice extent is 4.3 million square kilometers (essentially unchanged from June) with quartiles of 4.1 and 4.6 million square kilometers (See Figure 1 in the full report, below). If the median Outlook should agree with the observed estimate come September, this year would be the third lowest September in the satellite record. The spread in the Outlook contributions narrowed slightly from June to July, with an overall range this month of 3.6 to 5.2 million square kilometers.
The full range of Outlooks submitted this month lies within the range of the 10 lowest years of sea ice extent in the observational record. No Outlook is predicting a new record this year, despite the warm winter, record low extents for every month in 2016 except March, and evidence of thin ice in spring. Does this mean it won't be a record? Meteorological conditions suggest not, as surface temperature over the central Arctic has been near normal in the last month and forecasts of atmospheric circulation for the next few weeks suggest near normal surface temperature in the near future. At the same time, the subpolar seas are warmer than average and the land surrounding the Arctic Ocean has been warm and both are projected to continue to be warm in August and September. Finally, we note that no Outlook predicted a record low in 2012 either, and as a whole, the Outlooks tend to be miss the extreme low and high years compared to the long-term trend.
And here's the figure showing all the projections (click for a larger version):
The September minimums for the last 11 years (in million km2, found here):
Although the median for the SIO went up slightly, from 4.28 in June to 4.33 for July, the median for the poll on the Arctic Sea Ice Forum increased more, from 3.81 to 3.96 million km2:
If we look at the predictions from citizen scientists and ASIB commenters, Rob Dekker's prediction went up from 3.8 to 4.1 million km2, and Bosse went up from 4.1 to 4.2 million km2. Chris Reynolds doesn't seem to have submitted a prediction this month, but a few weeks ago he wrote: "My April prediction was for a September 2016 extent of 4.0 to 5.3 million kmsq, my expectation is that this September will be in the lower part of that range".
As for the professional predictions: I had already noted in the latest Melting momentum blog post that the CPOM team increased their prediction to 5.2 (from 4.5 last month) and it is now one of the highest predictions. Zhang and Schweiger (from the UW Polar Science Center running the PIOMAS sea ice volume model) even went up from 4.2 to 5.2 million km2. On the other hand the Ionita-Grosfeld prediction went down from 4.7 to 3.9 million km2. Interesting swings.
One of the scientists who did quite well in the Sea Ice Outlooks of the past few years, is Dr Andrew Slater from the NSIDC (check out his awesome website). This graph shows how his 50-day forecast has performed so far this melting season:
Dr Slater explains his method for the SIO:
Persistence can be computed in several ways. I have looked out to Sept. for the sake of comparison and as a very basic benchmark.
- 1) Daily anomaly persistence at 112 days lead time (so that I can go all the way to Sep 30th), then compute mean for Sept = 4.95
- 2) Persist the absolute anomaly from May to Sept (using NSIDC monthly value, not mean of daily). Sept = 5.06
- 3) Persist the standard normal deviate from May to Sept (using NSIDC monthly). Labeled as "Variance Persistence". Sept = 3.59
None of these methods have true skill at this long lead time. For April-June the daily persistence at 85-day lead time looks like a nice forecast (and is giving better results than a 50-day lead time), however, this is pure coincidence.
His 85-day forecast (SPIE) is slightly lower at 4.4 million km2.
All in all, most contributions point to this melting season ending up as a top 3 year. It would be a remarkable thing to happen, as atmospheric patterns haven't been conducive to sea ice reduction at all. Then again, a mild winter, early openings and plenty of heat in the system may have compensated for that. More on that in upcoming posts.
Read the entire 2016 July Outlook Report, a summary of current conditions and more detailed predictions here. And the August Call for Sea Ice Outlook contributions can be found here. There will be a new poll on the Arctic Sea Ice Forum at the start of next month.