Every winter the Netherlands hopes there will be another Elfstedentocht (Eleven Cities Tour). Scientists from KNMI and the Netherlands Environmental Assesment Agency (PBL) computed that the probability to have a period cold enough for an Elfstedentocht is now about 1 in 12 years, against roughly 1 in 5 years a century ago.

If we manage to keep global warming within 2ºC of the late 19th century temperature, as agreed in Paris, the probability remains 5% per year. However, if global warming continues unchecked the probability is about 1% in 2050 and even smaller afterwards, so that only one or two more of these skating tours will ever be possible.

This Elfstedentocht is a skating race and tour along the traditional eleven cities of Frysia in the north of the country. It requires thick enough ice along almost the whole 199 km route over canals and lakes. This in turn requires two weeks of very cold weather. These periods have always been scarce, and have become even rarer due to global warming.

The exact criteria to organise an Elfstedentocht have changed over the years. The first few tours could be skated on thinner ice as not many people participated. Forty years ago it was only organised when the ice was already 15 cm thick, and three days were needed after that to organise it. Nowadays this has been reduced to two days and the weather forecast combined with an ice growth model are used to estimate how thick the ice will be on the day. We therefore compute the probability of a "potential Elfstedentocht", a period that is cold enough to organise one. It turns out that a period of 15 days with an average temperature of −4.2 ºC or lower at De Bilt is usually cold enough to organise an Elfstedentocht. It can still go wrong: the winter of 2012 was cold enough, but snow on the ice kept it too thin.

It is hard to compute trends in winter extremes as the differences between harsh and mild winters are much larger than the trend up to now. Still, a trend to higher temperature is visible by eye in the graph of the coldest 15-day period of the winter (Fig. 1). Cold extremes, denoted by blue downward excursions, are getting less cold. We used two statistical methods to compute how rare they are now. Both agree that the probability of a period cold enough for an Elfstedentocht is about 8% per year now, with an uncertainty range of 5% to 19%. This can also be denoted by a 1 in 12 year probability. (This does not mean that we get one every 12 years, but that the probability is 1 in 12 every year.) A century ago this probability was still 1 in 5 years. As mentioned before, the rules were different so that not all of these could be used to skate the tour.

We have also computed the probabilities in each of the four KNMI'14 climate scenarios. In all scenarios the probability decreases as the globe and hence the Netherlands warms. If the world reaches the goal of the Paris accord of limiting global warming to 2 ºC above the temperature at the end of the 19th century the probability stays about 5%, depending on how the weather reacts to this warming. However, if we let the world warm by 4 ºC the probability for an Elfstedentocht becomes very small after 2050 and we expect no more than one or two tours will be possible until the end of the century and none afterwards for a long time.