Sea Ice Forecast Glitch

A huge unseasonal Arctic storm has contributed to record low ice levels which themselves have triggered a significant numerical glitch in our computer forecasts for Arctic swell.

It's been widely reported that 2012 will see a new record low for Arctic sea ice concentration since satellite observations began. There's strong evidence that this is part of a long term trend that has accelerated rapidly in recent years.

Dr Edmond Hansen of the Norwegian Polar Institute explains:

"This is not some short-lived phenomenon - this is an ongoing trend. You lose more and more ice and it is accelerating - you can just look at the graphs, the observations, and you can see what's happening."

With a strong positive feedback effect, in which dark sea water reflects far less warming sunlight than white ice, this acceleration means the Arctic is now the fastest warming area of the planet.

© 2017 - NASA image by Jeff Schmaltz

Over the month of August, the Arctic lost 35,400 square miles of ice a day. Accelerated by temperature 1-3 degrees above seasonal norms, but also by a large, unseasonable Arctic storm occurring in early August.

© 2017 - MSW Period Chart

A powerful 965mb low like this can affect the ice in a number of detrimental ways. Breaking ice from the main pack and pushing it into warmer water, churning the ice and making it slushier and stirring warmer water from deeper in the ocean increasing surface temperature. All three effects will be potentially increased in the case of already warm local conditions and fragile ice cover.

While storms like this are common in the winter their occurrence in summer, when the ice is at its most fragile, are less frequent. Paul A. Newman, chief scientist for Atmospheric Sciences at NASA's Goddard Space Flight Center estimates just eight such storms in 34 years of satellite records.

For surfers the long term implications of these changes are unclear, but scientist Dr. Jennifer summarises: "The question is not whether sea ice loss is affecting the large-scale atmospheric circulation... it's how can it not?". There is clear evidence linking warming seas and an absence of ice to changes in mid-latitude weather patterns and a suggestion that there's a link between the melting ice and the unseasonably low jet stream that brings storm systems further south in the Atlantic summer.

Model Issues and Your Forecast

Beyond these hypotheticals there's one clear consequence of this completely unforeseen ice pattern. The main swell model we use for our MSW forecasts, the NOAA NWW3, calculates swell generation on the northern boundary of its domain, based on the latest sea ice data obtained from satellite scans. The sudden and rapid change in ice concentration over the last few days revealed an area of ocean for which there exists a significant numerical issue in the mathematics responsible for calculating south bound swell. The results were a seemingly endless series of unfeasibly long period swells propagating south into Northern Europe and beyond. To clarify, it wasn't that the model was forecasting these swells as likely, BECAUSE of the lack of ice, but that the lack of the ice exposed a bug or error in the system that hadn't appeared in testing with a broad range of historic conditions.

The area affected can be seen on the Northern Atlantic swell charts as a long period (25-28 second swell). Impacting a very small area of the the Northern UK and Ireland with propagation down through the Atlantic in the dramatic red colour.

Our issue has been how to deal with this problem. We can roll over to the low resolution NOAA swell model that many other surf forecasting sites use - but this means you'll lose all swell data other than a combined swell height and period - something we feel is significant for reliable surf forecasting and the reason we don't use this product normally. Given the NOAA are working on a permanent fix to their model and that the sea ice itself is likely to rapidly reform as we head into Arctic winter, we've opted to modify the local forecast to remove swell partitions that appear to be as a result of this issue, while leaving the charts and other product unchanged. We'll continue to update our blog with the latest information regarding this issue and apologies for the problem. In over a decade of using the operational NWW3 model for surf forecasting this is the first issue and only issue of this kind we've seen.