Tomorrow, April 1st, I'll be doing a short presentation on the Sea Ice Prediction Workshop that will be webcast by UCAR. I'll be talking 10-15 minutes about the ASIB, ASIG and ASIF, and about increasing public interest in Arctic sea ice. You can view the webcast here.
It was always clear how difficult it is to forecast the Arctic sea ice melting season, but this realization reached an even deeper level in the last two years (more about that a couple of paragraphs below). A lot of scientists are working hard to improve the science of predicting the yearly sea ice minimum, and some of them have now evaluated these forecasting efforts, using data from the interagency "system-scale, cross-disciplinary, long-term arctic research program" SEARCH (Study of Environmental Arctic Change) Sea Ice Outlook, which was set up after the drastic sea ice decline of 2007. These monthly SIO predictions have been extensively covered on the ASIB the past 4 melting seasons.
The study by Stroeve et al. (including ASIB guest blogger Larry Hamilton) is called Predicting September Sea Ice Ensemble Skill of the Search Sea Ice Outlook 2008–2013, has just been published in Geophysical Research Letters, and its abstract reads as follows:
Since 2008, the SEARCH Sea Ice Outlook has solicited predictions of September sea ice extent from the Arctic research community. Individuals and teams employ a variety of modeling, statistical and heuristic approaches to make these predictions. Viewed as monthly ensembles each with one or two dozen individual predictions, they display a bimodal pattern of success. In years when observed ice extent is near its trend, the median predictions tend to be accurate. In years when the observed extent is anomalous, the median and most individual predictions are less accurate. Statistical analysis suggests that year-to-year variability, rather than methods, dominate the variation in ensemble prediction success. Furthermore, ensemble predictions do not improve as the season evolves. We consider the role of initial ice, atmosphere and ocean conditions, and summer storms and weather in contributing to the challenge of sea ice prediction.
There's a very informative press release on the NSIDC website, from which I'd like to repeat a quote there by Dr. Stroeve:
“Short term predictions are achievable, but challenges remain in predicting anomalous years, and there is a need for better data for initialization of forecast models,” Stroeve said. “Of course there is always the issue that we cannot predict the weather, and summer weather patterns remain important.”
This summarizes in one sentence one of last year's big lessons for me. I made the mistake of thinking that the ice was now so thin that it wouldn't matter much what kind of weather it was exposed to, it would still decrease to or even past the September 2007 level. This turned out to be wrong, weather still precedes ice thickness when it comes to determining the ice pack's total extent in September.
Paradoxically it's the decrease in ice thickness that has led to this situation, in the sense that it was probably easier to forecast the minimum extent in the past, where overall thicker ice would reduce the possibility for large negative outliers. Now large swings can go either way, as we saw in 2012 and 2013. Does this make it impossible to forecast the melting season accurately? Not necessarily.
The other big lesson I learned from last year is how important the start of the melting season is. 2012 had such a good start (for melting) that spells of weather that was less conducive to sea ice decrease later on didn't matter all that much, the trend lines kept going down. Last year it was the exact opposite. Despite an average winter and a record amount of first-year ice, the start to the melting season was so cold and cloudy that the 2-3 weeks of weather conducive to sea ice decrease (high temps, lots of sunshine) later on, just weren't enough to make the melting season get into the top 3 of lowest minimums.
I think that knowledge of initial ice conditions (thickness, snow cover, distribution of multi-year ice) and of the impact of what happens in April and May (open water, melt ponds, terrestrial snow cover) could already give a good idea of what to expect in the 3 months that follow. Weather patterns might even become easier to forecast if their so-called 'stickiness' - where the weather in a particular region gets stuck - becomes a feature, rather than a bug, and scientists start to have a clue about how that works.
But forecasting the minimum sea ice extent is going to remain a very difficult task, until volume has decreased even more and reduces the possibility of large positive outliers. I think that by that time we'll be busier trying to forecast the consequences.