Falling snow flakes are humbling. Not just because of their beauty, but also because of their gentle ability to completely paralyse a developed and overpopulated country with traffic jams, clogged railroad switches and closed runways.
But although that (together with a repeatedly forecasted but ever-postponed ‘horror winter’) may make for a great annual media hype, it does not imply winters are getting colder. To the contrary – winter warming in the Netherlands* is increasing – but it’s not as simple as a ‘local seasonal case’ of global climate warming.
[*) Here we use the Netherlands as case study, but the general climate analysis and the shown seasonal trends go for Northwest Europe (in which the Netherlands lie central) as a whole, including winter development for Great Britain, Ireland, Belgium, Germany, France, Denmark and (southern) Norway & Sweden.]
‘Medieval Cool Period’ and Little Ice Age Dutch winters
Winters in Northwest Europe have been warming for a very long time. The much-romanticised ‘winters of yesteryear’ when marine ports would remain ice-locked for months and big rivers like (famously) the Thames and the Rhine could stay frozen until March are in fact in no one’s living memory. Such winters define the Little Ice Age, which ended [officially, you have to draw the line somewhere] in 1850.
For the Netherlands probably the coldest winters (and years) occurred in the 16th and 17th century – a period during which the wetland-rich country established its ice-skating tradition – a tradition which has survived quite a bit of winter warming since.
Average winter temperature (December previous year + January & February) graph plotted from the ‘Buisman temperature reconstruction’ carried out by A.F.V. van Engelen, J. Buisman and F. IJnsen, of the Royal Netherlands Meteorological Institute (KNMI). Line shows the 20-year average winter temperature. The Buisman record continues back in time to around 700AD, but suffers from data scarcity before the year 1400. It clearly shows a couple of dips in the average winter temperature for the Netherlands (notably second two thirds of the 16th century, the entire 17th century and a shorter period around 1800, but also several such dips (according to the Buisman temperature reconstruction) from the early to the late Middle Ages – stressing the so-called Medieval Warm Period would perhaps indeed be more appropriately simply named ‘Medieval Period’ – as the graph for example shows grouped cold winters occurring before 1450. [Reason why for instance the KNMI uses a relatively early starting date for the Little Ice Age (LIA), from 1430-1850.] Whether that one really cold 20th century winter (the exact winter season is that of 1962-1963) would qualify as a proper ‘Little Ice Age winter’ remains questionable, because the coldest winters of the LIA may have been colder than shown in the graph, as (from 1706 the data is comprised of temperature measurements) people may have had trouble to record the actual minima during the LIA winters – so the really cold winters of that time may have still been considerably colder (overall) than that of 1963 [we are speculating here]. The cold winter of 1783-1784 has been connected to large volcanic eruptions in Iceland in 1783. Also boreal cold winters of 1452, 1809 and 1815 can be related to volcanic eruptions (judging by a tree ring record) – although these also quite perfectly correlate with the Spörer (1450-1550) and Dalton (1790-1820) solar minima respectively. [Also the Pleistocene temperature record disputes volcanoes are strong climate coolers.] Overall Little ‘Ice Age’ is still something of a climatic misnomer as it easily leads to dramatisation of imagined winter conditions (and expected climate cooling under possible future grand solar minima). What is also clear from the graph is that winter variability has been high throughout the centuries, which it still is – although the upward trend throughout the 19th and especially 20th century is unmistakable, as shown in more detail below. Data: KNMI | Graph: Rolf Schuttenhelm, www.bitsofscience.org
From the year 1706 the much more detailed collection by Labrijn et al. (Royal Netherlands Meteorological Institute (KNMI)) of historic temperature measurements allows for monthly differentiation. These daily and systematic weather observations (the Dutch have a relatively well-established observatory record) were carried out in Delft/Rijnsburg (1706-1734), Zwanenburg (1735-1800 & 1811-1848), Haarlem (1801-1810) and Utrecht (1849-1897) – and from 1898 at the current KNMI headquarters in De Bilt. [Possibly relevant for winter measurements is that especially the 1706-1848 range was carried out in western coastal areas that tend to experience slightly milder winters than the inland provinces.] A couple of things are especially noteworthy from this graph. Firstly that January, which is clearly the coldest winter month with the highest chance of substantial frost periods, has been warming faster than December and February – especially over the course of the 19th century. The other remarkable difference between the winter months is that December and January temperatures seem much more susceptible to multidecadal trends, which (to some extent) seem to correlate with the Maunder and Dalton solar minima. Data: KNMI | Graph: Dorien Bartels, www.bitsofscience.org
As expressed by the above graphs Dutch average winter temperatures (and from 1706 also average temperatures for the three winter months) show considerable natural climatic variation, different multidecadal climatic changes, and a long-term warming trend. Chronologically the winter temperatures can be divided in the Medieval (Warm) Period, the Little Ice Age, followed by the modern period, with gradually increasing temperatures from 1850AD.
60-year record does not show an increase of cold starts to winter season
Zooming in on the last 60 years shows that for the Netherlands the warming-winters trend continues, and is in fact accelerating compared to the 1850-present timeframe.
Average temperatures for December, January and February for the Netherlands, including trendline for each month, between 1950 and 2012. Conclusion: all winter months seem to be warming equally fast. Data: Royal Netherlands Meteorological Institute (KNMI) | Graph: Dorien Bartels, www.bitsofscience.org
Despite a couple of recent relatively cold Decembers (2008, 2009 and 2010) there seems to be little difference in the 60-year trend for the Dutch winter months of December, January and February. This does not support the Arctic-warming-cold-winter hypothesis – which suggest Arctic sea ice melting (perhaps through increased October Siberian snow cover) would favour negative phases in both the Arctic Oscillation (AO) and the North Atlantic Oscillation (NAO) at the onset of the boreal winter, and therefore increase the chance of (Northwest European) cold spells, especially in December.
Two centuries of warming – comparing Dutch winter temperatures of 1780-1810, 1880-1910 and 1990-2010
As the below winter temperature table again shows Dutch winters have been on a warming trend for at least two centuries. Shown is a comparison of official climate averages (which is calculated over three full decades) at the start of the 19th, 20th and 21st century – for three-month winter season, and as average temperature for each of the individual winter months December, January and February.
Not only warmer, also less snow and ice
It is not just average seasonal and winter monthly temperatures that are increasing, it is also the cold that is declining. Shown below is the 60-year trend for the Hellmann number for winters in the Netherlands, a measurement for cumulative frost during the entire winter season (from November to March). This means of monitoring winter development is of separate importance, because it would be theoretically possible that the occurrence and magnitude of cold spells would develop differently from seasonal and winter monthly climatic averages. Seeing a declining Hellmann trend provides further evidence against any dominant climate-warming-winter-cooling hypotheses, like the one mentioned above.
Not only are average winter temperatures on the rise, cumulative frost (from November-March) is in decline. [See also our special article about Hellmann development of Dutch winters, which presents the KNMI dataset from 1900 to present – and shows that also when focusing not on average temperature but on cumulative frost recent perceived Dutch ‘cold winters’ were in fact ‘normal to mild’ – and near-doubling of the timescale does not change the downward direction of the frost occurrence trend.] Data: Royal Netherlands Meteorological Institute (KNMI) | Graph: Dorien Bartels, www.bitsofscience.org.
But why? At least partially because climate change causes westerlies to increase – NAO record shows
Winters in Northwest Europe are strongly influenced by N-S or NW-SEA-running air pressure differences over the North Atlantic, captured in what is called the North Atlantic Oscillation (NAO). In its positive phase (low pressure systems around Iceland, high pressure around the Azores) it increases the Atlantic influence on the European climate, with dominant mild (and wet) westerlies. In the negative phase (high pressure over Northern Europe) the normal west circulation is blocked, and cold air moves in from Siberia or the Arctic.
Shown above is the NAO development over the same 60-year timeframe, for all three winter months, per winter, between 1950-2012. Yes, there seems to be an increasing trend, favouring a positive NAO during the winter. Notable is that not only the median range seems to be increasing, also very deep monthly values are becoming rarer, and extreme high values have newly developed. Data: National Oceanic and Atmospheric Administration (NOAA) | Graph: Dorien Bartels, www.bitsofscience.org
What to expect – continuation of temperature trends, return of historic extreme winters very unlikely
Now what to expect? There is no reason to assume NAO’s upward trend will continue, although above zero values could keep prevailing. This is because longer NAO datasets show cyclical behaviour (which as shown below also explains cold winters (with deeply negative NAO) of the 1950s and for instance the extremely cold winter of 1962-1963.
Winter NAO index record derived from air pressure measurements at land stations (Iceland and Portugal) since 1864 shows a multidecadal sine wave. This cycle exceeds the 11-year normal solar cycle timescale – and much sooner suggests a connection with the Atlantic Meridional Overturning Circulation, which creates relatively long-lasting water temperature anomalies in the North Atlantic Ocean, which affects convection strength – and therefore the favoured location of depressions and high pressure weather systems (and thus NAO). Data: UCAR/NCAR | Graph: Wikimedia Commons.
All in all there is no reason to assume winter warming in coming years will either speed up or dramatically decrease. Under continued warming of European winter land temperatures also winters in the Netherlands will keep getting milder, with ever-decreasing chances of substantial frost periods (although climate variability will likely remain high).
Higher chance for normal-to-cold winters if a prolonged solar minimum would develop (next minimum no sooner than 2015), but no breach of multidecadal upward trend
If the Sun will move (as some predict) into another grand solar minimum, this is very unlikely to force a return of ‘Little Ice Age winters’ – because the solar minima largely work through favouring a negative NAO (which as shown above is also affected by other factors) – and because the difference in average winter temperatures between historic silent sun episodes and historic periods of higher solar activity is smaller than the difference between historic winter temperatures and modern winter temperatures as a whole – in other words global climate warming is a stronger forcer than the solar cycle, even locally for European winters (for which the solar cycle is still a relatively strong forcer – as its climatic influence is weaker elsewhere and for other seasons).
© Rolf Schuttenhelm | www.bitsofscience.org