NW Europe and De Bilt

RUTI: De Bilt and the North West European low lands

Central Europe north of Paris has only limited temperature data available from GHCN. Central Northern Europe appears to have had almost same treatment as Turkey and the Levant: One/few long running temperature series per country (often for the capital) to show previous warm peak 1925-50.


Fig 1

Here are some of the sad temperature GHCN and Hadcrut temperature stations showing mostly nothing:
Fig 2
In addition, cities like Amsterdam, Rotterdam, Utrecht, Den Haag, Groningen, Antwerp, Liege, Brügge, Cologne, Düsseldorph etcetc. are not “represented” at all. Belgium only has the Bruxelles station: Uccle.

In general not for Germany, Czech rep. and Poland there are a huge number of small town or rural stations not displayed above. These only show mostly useless fractions of temperature series often even before year 1900.

Is the missing North Europe temperature data not existing?

To briefly consider this let’s check out what data CHGN uses for NW Europe, precipitation:

Below for example Belgium where GHCN use one station for temperature (Bruxelles/Uccle) 

Fig 4
It appears fair to conclude that the missing temperature data of NW Europe is the result of GHCN´s decision not to show most temperature data of NW Europe?

In the middle of this area of scarce temperature data pops up a near rural station: De Bilt.

 “De Bilt” in the heart of central Holland - no wonder this station has attention! De Bilt is world known since it goes back to 1706, and (for some reason) De Bilt is available still.


Fig 5
However, the adjusted version - here from Hadcrut - shows no warm period before 1950 unlike other rural stations over most of the world.
The De Bilt adjustment is commented here: http://climateaudit.org/2007/06/06/de-bilt-adjustments/

The De Bilt station is supposed to have been moved 100-200 meters in 1950 and therefore, the old warm period is gone from the adjusted data.

Just 140 km South of De Bilt, in a similar geographical location (not too far of close to the coast) we have the Bruxelles/Uccle temperature station. Obviously, this is an Urban heat contaminated station, but to evaluate a significant adjustment over the one year 1950, any urban status should hardly be a problem. (Urban effects are normally seen over decades).

Here are De Bilt Vs. Bruxelles/Uccle, both from “unadjusted” GHCN:

We could not have demanded a better confirmation of the 1950 De Bilt temperature drop, a perfect match in trend around 1950. This match obviously goes far beyond random similarity. And then it is also interesting to see the strong gradual trend difference between the rural De Bilt and the Urban Uccle, around 1 K from 1890 to 1960, 70 years. Somewhat like the UHI effect generally seen in RUTI France.
So case closed - the adjusted Hadcrut is just wrong?
Well, Adjustments seems to have hit Uccle data too:

Fig 7: The same year the De Bilt station is supposed to have been moved a little caused a 1 K change in data, in exact this year 1950 something similar is supposed to have happened to the Uccle temperature station!
And if possible even more bizarre: Now the Uccle Urban warming trend is simply gone for the years 1900 to 1950!!

What a coincidence! Same year the De Bilt and the Uccle had a significant change resulting in the same significant change in 1950! And as a miracle on top, the urban trend in Uccle data has vanished in the period 1900-1950 !?
Result: Now the 2 adjusted graphs happens to look similar.

Lets move on: In the dried out data dessert of Northern Europe, suddenly is an oasis from heaven:
For some reason the Wiesbaden series – that shows only 17 years – happens to show the very drop in temperatures around 1950 (includes years 1949 and 1950!).


Below: Wiesbaden and Frankfurt is around 30 km apart geographically, however, Frankfurt is Urban and Wiesbaden is nearly rural.

Frankfurt and Wiesbaden data in absolute values: Despite the 30 km in distance, I can’t wish for a better match for stitching data. Frankfurt-Wiesbaden stiched (averaged) and put on top of De Bilt and other temperature series in the area:

Fig 10: On the graph above I have added Luxembourg Airport from Hadcrut 1880-1975. I cut Luxembourg airport Hadcrut data in 1975 because the trend of Luxembourg data after 1950 has been adjusted in a manner I cannot see supported, adding no less than 0,8K warming trend:


And finally I also show Netherland Hadcrut Maastricht data in the graph above. For some reason (an error??) the whole Maastricht dataset is available from Hadcrut online, and I cannot spot any sign of adjustments here.

Now we have a good match between central Germany and central Benelux not only supporting the De Bilt data, but also telling a story of NW European temperatures.


The “most rural” site I have found for this central German area – with data available (!) combining 1950 and 2010 - is Saarbruecken:

Fig 13

However, Nurnberg and Trier-P are near rural too it seems:


From 1950 to 2010, the available 2006, 7, 9, 10 data shows that Saarbruecken has 0,7 K less warming than Frankfurt. In addition, Nurnberg and Trier has 0,35 K less warming than Frankfurt. This supports the impression from the graphs above, that there is hardly any warming in the area from the 1930-40 peak till today.

Further East, for Prague, then we have one long running near rural station:

Fig 15.

So this station should better reflect the lower flat rural Bohemia than Urban Prague. The idea, that modern temperatures are not significantly higher than the warming peak 1925-50 appears supported, but let’s compare with all other rural Czech temperatures stations from the lower Bohemia:
Fig 16.

Result: No data available from GHCN of lower rural sites, only the Prague station… Seems like the De Bilt situation, except here Hadcrut did not use the long series at all, only GHCN has the long Prague station.

Back to central Germany, going a little north to Erfurt, a smaller town with a temperature station near rural conditions, nice:


Fig18: Inserting Erfurt Data resembles Frankfurt-Wiesbaden data well. GHCN did not give us the warm peak 1930-50 for Erfurt, But why assume that something totally different happened in Erfurt just in the 1930-50 period?

Going north approaching the latitude again for De Bilt, in my search for mostly rural data I came across Dresden, I would call it Suburban,
Well outside Dresden city, but near airport and road.

Fig19: The Dresden series fits well, especially, the Dresden data comply with Early de Bilt 1910-20 period.

There are 2 temperature stations for Stuttgart, again the peak 1930-50 not to be seen for GHCN (-and Hadcrut worse, shows no data prior to 1953). For the post 1950-series we can compare the 2 GHCN graphs for Stuttgart:

Fig20: We see a significant trend difference, after 1960.
The cold-trend graph – we could call “Stuttgart 1” is around 10 km from Stuttgart city centre, and the Deutsche wetter station appears placed on natural vegetation, on open field. We are in a little town called “Berneim”.
Stuttgart 1, colder trend:

The warm trend graph from Deutsche Wetterdienst is located 5 km from Stuttgart city centre, and the thermometer on some red substance (??) and not natural vegetation. Who knows when they changed the foundation and what effect it had? And why doesn’t Deutsche Wetterdienst know the temperature for this station 2000-2009??
Stuttgart 2, warmer trend: 


Fig23: There are huge differences between the two Stuttgart series just few kilometres apart. Remember that each point represent a whole year with more measurements per day. Both Stuttgart series are located in around 3-400 meters elevation.

Both series has some suburban character; however, the warm series is half the distance to the city centre. In the 1950´ies, the 2 dataseries are 100% identical each year. This means that the most rural series can be stitched with the old part of the more urban series, and so I will. This decision all readers are welcome to criticize and comment.


Fig24: The large dive in temperatures for Stuttgart 1930 down to 1950 level, is not dependent on my choice of stitching, since this dive was already present in the warm trend Stuttgart 2 data. The Stuttgart series resembles the Prague series and De Bilt.

Hannover and Emden are both rather well placed near rural stations, but they differ a little from what we have seen in this writing so far. In stead, we have 4 rural temperature series for southern Denmark and Sweden (3 from Nordklim, 1 from GHCN) that appears to Match Emden and Hannover better:


Fig 26: Hannover and Emden has more heat around 1910-25 than the south Scandinavian series, but otherwise a useful match.


Fig27, Two Berlin GHCN series:At first glance, this looks like a rural and an urban station with different trends. However, things are strange indeed. Many years 1880-1905 and1930-1950 has no trend difference. For the 1905-1930 period, the Tempel series is very near to identical to the Dahlem except the period is lowered around 0,8 K permanently. Then the 1950-1990 period the series are again near identical, but now Tempel data is pushed 0,5 K up compared to Dahlem.

What is then supposed to have happened to cause all these jumps in temperature? Did they move the station around on their lot of land? Perhaps they have a mobile temperature station for off-road use?
I believe I have the right to choose the Dalhem series for Berlin because it fits nicely with the bulk of German-Benelux data series so far. Especially the nearby Dresden temperature series confirms my use of Berlin Dalhem:


Fig 29: Also we see that the Berlin Dalhem series shows some Scandinavian characteristic like a bigger dive around 1940-42.

On Google Maps I found that the Potsdam station was located near the “Potsdam Institut Für Klimafolgenforschung”, that is “Potsdam research institute for consequences of climate change”.  
Potsdam is located very near Berlin (30km?) so it makes sense to compare with Berlin Dalhem:


Here are the 2 adjusted temperature series Potsdam and Berlin Tempel against nearby Dresden representing the bulk of central German data series and then the Berlin Dalhem:

Fig 33: Dresden and many other German temperature series are wrong too? They should have an adjustment 0,9K down 1880-1930 too? No, at least for now, Berlin Tempel and Potsdam will not be used in RUTI.

From one (roof) top to another:

Fichtelberg temperature mountain-station.

- The only fully rural German station where data 1890-2010 is available without interruptions. How can his be?


Fig34: Average yearly temperatures on the Fichtelberg top (1214 m)  is 2 – 3 degrees Celsius.

Snow cover is present a  larger part of the year and changes in precipitation and snow cover periods are a dominant factor for temperatures on the very top. The mountain top area is limited to a thin line of tops between Czech Republic and Germany so it will not be considered for now. It represents a small area compared to the lower land stations. To read more about the differences low and highland, see RUTI Alps where Fichtelberg will be compared to the high Alpine area temperatures.


Fig 36.  RUTI Graph Central Benelux, Germany + Czech republic . (The Central Benelux and Germany avg is similar with Southern Swedish trends.)

Fig 37: RUTI graph, Northern Holland, Germany:


Had Crut´s De Kooy and Vlissingen temperature stations:
First a cut-and-paste from RUTI Scandinavia (Denmark):



..The weakness for the west Jutland coastal stations might be that in cold periods they equals average Danish temperatures well while in warm periods they don’t show the heat well...

Hadcrut´s De Kooy and Vlissingen stations is located on the western coast of Holland and is thus likely to face the same problems as mentioned for the Jutland west coast. This may not be the first choice of station if you want to reflect warm periods of mainland temperature trends, the bulk of the country:

Fig 39: GHCN does not use De Kooy and Vlissingen temperature stations, so while hardly any data has been made available, Hadcrut is actively introducing these 2 stations them selves for some reason.

In the entire central Europe, Hadcrut has only introduced 4 stations not available in GHCN, the 2 other are also stations placed in a location not well for representing the bulk of Europe. See RUTI: Alps.


Fig 40: De Kooy shows a remarkable “Scandinavian dive” 1940-42, and good similarity in general. However, the De Kooy data has a significant colder trend than Nordby Fanoe. Nordby is located close to the city Esbjerg which is the fastest growing city in Denmark. From 1900 to 2010 the urban Esbjerg area went from near zero (!) to around 150.000 today. I do not know if this has an influence, but I could imagine that things has developed on Fanoe as well.
Sadly, Hadcrut has chosen not to show us 1990-2010 for De Kooy and Vlissingen… This is a problem, because if I used the rather low De Kooy temperatures in my Northern Holland – Germany avg. we would get a faulty uptrend in the average graph after 1990.

Finally, here a compare with the Vlissingen and De Kooy stations compared to the average England temperature trend:

Vlissingen and especially the De Kooy station appears to belong best to the English climate zone. The English average is based on numerous rural long series and is likely of high quality.


The De Bilt and NW Europe scene with some strong adjustments reducing the heat of the past while omitting numerous stations near by is not unique at all. The same scene we find in RUTI: The levante, where Beirout temperatures before 1950 is severely cold-adjusted and many tempearture stations near by lacks vital data or is not available at all.


For issues with coastal stations see: RUTI coastal temperature stations

PS! In addition, it appears that the  Non-Coastal area of Morocco had a significant similarity with the West European non-coastal areas. Here a compare of European Non-coastal trends with the non-coastal Morocco temperature data from Marakesh:

From: Morocco and West Sahara.