France – 2009 Temperature Stations

Marseille Airport

What Do We Actually Measure in France Today?

In a prior posting we found that there are all of 17 stations in France in 2009 that are used by the GHCN. The overall dT/dt graph was produced (using all historical data in GHCN) and that made a great “hockey stick”. With gently dropping temperatures turned into a rapid rise at the end.

A brief inspection of the station list show that it was almost all Airports. Only one station was not flagged as an airport (and a commenter stated that it was, in fact, an airport too).

So we surmised that putting all the thermometers at the Airports would cause much of the observed temperature rise (that ‘just happens’ to start right after the non-airports are dropped from the series…)

But it nagged at me that I’d not looked at these stations in depth. Was it possible there were some turly great stations here, asking to be contrasted with others? Would there be rampant indications of Tarmac Attack in the record? Just what did those temperatures look like?

Well, I’ve finally gotten the set done. Now we will see.

About The Graphs

These are simple plots of the 2010 vintage GHCN Version 2 temperature data for those stations in France that were still reporting in 2009. These stations, and this data, are all the real temperature data for France in the basic data set that feeds into CRUtemp, GIStemp, NOAA/NCDC and others. This is, literally, the only data that can say if France is showing warming or not in the present. But does it have something else to say too?

I have plotted each station monthly temperatures by year, and added a trend line for each month. There is also an annual average (labeled ‘Yr”) with an overall trendline. I’ve tried to keep the colors and patterns mostly the same, graph to graph, but there will be some variation. Also of note is that the particular years on each graph vary greatly. For most, I’ve added a ‘break” when a large number of years are missing, for a couple I’d already made the graph when I thought of doing this. But it is worth emphasis that many stations have on a small amount of recent data, even if the first date is in the 1800’s…

The graphs will be presented in ‘station ID order’. Perhaps after seeing what’s in them I may rearrange the order, but for now, we’re just going exploring…

Each graph is much larger than in the posting. You can click on it to get the larger version.

Rouen

This is one of the graphs with a “big gap” that I did not segment. The bulk of the data is from a very early time up to 1882, then it jumps to 2001. So there is no “Baseline” coverage. It can only be compared to some OTHER stations to decide if it is “warming” in codes such as GIStemp and CRUTemp. Given the ‘moth eaten look’ of some of the early 2000 data I’m not sure it’s really good for much. Mostly you can just see how there is a ‘step up’ in the cold months minimum excursions and no real change in other months.

Rouen Temperature Graph

Notice that some Summer months are cooling, some winter months are warming, and some are just flat.

For now, I’m just going to load the graphs ( I’ll add my analysis later, it’s 4 AM as it is…)
But see if you can spot any ‘sunshine’ effects in the southern stations and perhaps a bit of engine warmth and heaters in the northern ones…

Brest

A nice little bit of ‘post WWII recovery and growth’ but lacking in longer term context. It catches the cool 1950 – 1980 baseline period reasonably well. You can also readily see the ‘clipping’ of the cold excursions that begins in 1990 or so and mostly impacts winter months.

Brest France Temperature Graph

Strasbourg

This is another chart where I did not insert a “Gap” between the older data (that runs out in 1887) and the ‘modern’ data that starts in 1949. So “Mind The Gap!”. We see remarkable stability in the summers over the years, but the winter temps are less prone to cold excursions as we exit the Little Ice Age. The 1830’s look to have been particularly brutal though by the 1840’s the summer temperatures were about the same as now. Also, if you look at the “Hot Pink” annual average, you can see how the temperatures ‘sag’ below the trend line in the “baseline” period of 1950-1980 then the average “bounces up” on the 1990-1991 change of processing, but then fails to advance. Some months, like September, are remarkably stable through the decades and centuries and argue for a ‘heaters and A/C / Tarmac” driven change in the other months. But then there are months like February (the dashed red near the bottom) that are flat despite the clipping of ‘low going’ peaks recently. Compare it with January where the slope is clearly up (and with very strong change of low going excursions but with the max January not rising much at all, being about 2-4 C over many many decades.) With Novembers and Decembers also of remarkably flat trend, the bulk of the “warming” in winter comes only from the January trend.

June (light blue at the top) is also interesting in that there is a big “jump up” in recent years (early 2000s) that looks like a ‘sunny June on the Tarmac’ to me. We’ll see how that behaves over time with increased cosmic rays and the Svensmark Theory. It would also be very interesting to see what the temperature is away from the Airport, tarmac, cars, jet engines, and June Travel to France…

Strasbourg France Temperature Graph

But at the end of the day, when I look at this data, I don’t see much of a generalized well distributed effect with onset over the life of the Industrial Revolution. I do see a lot of “artifacts” in the data the speak to me of local issues, particular months events, and “Land Use” via UHI and Airports in the sun.

Nantes

Warming winters but with a dead flat April.

Nantes France Temperature Graph

Summers pretty darned flat too. Artifacts of winter heaters and UHI?

Bourges

OK, about here I caught on to the need to flag data gaps so you will notice a “jump” of gap from about 1897 to 1949. I do have to wonder if there are any “equipment change” artifacts at this splice point, but we’ll present the data ‘as is’ here. In many ways this particular chart fascinates me. Summers cooling, winters warming, all in all a little bit of heaven!

So the Hot Pink line shows a couple of discontinuities. We’ve got a jump in 1990-1991 that’s fairly spectacular, but then is followed by the typical “Failure To Advance”. That just shouts to me “Equipment or Process Change!!!”. The other jump is at the “splice” of the 1800’s to the post W.W.II era.

Also interesting is that April Falling with October Rising wedge in the ‘transition months’. We’re not getting an earlier spring, but we are getting a later fall? Or maybe just a UHI effect is higher in October than in April?

Bourges France Temperature Graph

Then look at winter! Rising like Crazy! But form a LIA low into a modern optimum pleasant. Notice how much the ‘cold excursions’ are clipped. If that’s “Global Warming”, I want more of it. A Lot More. Notice that the Red February and the Blue January lines are NOT going higher at their peaks. They are tracking right along about 7 C and a trend line fitted to those peaks would not be rising. It’s the low excursions that are being suppressed.

Now I can’t say if that is a UHI artifact, a Quality Control artifact (as we saw in an earlier article where low going excursions will be much more likely to be replaced by an average in the NCDC published methods), and equipment change artifact (going to Stevenson Screens, then later to AWS / ASOS gear with it’s known problem with self warming). Or potentially even a tiny impact from CO2 (though not in keeping with the predictions of Doom In Our Time!). What I can say is that it is completely beneficial to life and make the place more pleasant, and is not a threat of any kind.

In my opinion, a close inspection of the Bourges really raw data (for the whole life of the station) along with a good equipment an process change history and a “land use” narrative would pretty much tell you what causes “Global Warming”. A great little “Dig Here” project! There are many such Poster Child stations that could be used, so you don’t have to be in France to do it, but this is a nice one.

Dijon

June and July dead flat, August rising. Do a lot of folks take August vacations to Dijon? Who knows… maybe CO2 likes to spend August there…

Dijon France Temperature Graph

Then look at winters! Dead flat! So we have another case where all the “warming” comes from just one or two ‘odd months’.

And if we average this with Bourges where winters are rising and summers not so much the result is exactly what again?

Even if done via “anomalies” all you are doing is taking two very different sets of local effects and averaging them to get no meaning at a global scale. If done with 2000 stations instead of 2, that issue is still present. That whole “put it all in one pot and stir” process is just making Mulligan Stew out of the data and hiding what it really has to say.

Limoges

A short record that captures only one cycle of the ocean currents, measuring from a cool phase to a warm phase. But even this site has some flat monthly trends. Look at Sept, April, and December, in particular.

Limoges France Temperature Graph

So once again all the “warming” comes predominantly from a subset of the months data.

Clermont – Ferr

A little scrap of data recently, hardly worth having in the record. Why bother? But GIStemp and other codes will ‘fill in’ some nearby stations in the gap or splice this data onto the other stations via ‘homogenizing’ and ‘The Reference Station Method’ and make something out of it…

Clermont - Ferr Temperature Graph

Why the longer Station ID number in this graph? Because if you use the shorter form, you will pick up two stations, not just one. FWIW, I think that most of the “Global Warming” found in many locations is an artifact of ‘splicing’ anomalies together that are disjoint as are these two stations. Yes, techniques such as “The Reference Station Method” claim to fix that, but they are not perfect and some of that splice artifact will leak through. Here’s the two stations spliced:

Clermont-Ferr and Le Puy De Dome spliced

Puts a nice little dip right in the baseline period, doesn’t it?…

Bordeaux Meri

Another place with winters more or less flat and summers cooling. April is sure a challenge!

Bordeaux France Temperature Graph

But if you measure only from the cool cherry picked baseline, you could make some ‘warming’ out of that, I suppose.

Mont Aigoual

They take off a little over a decade then come back with what looks like a splice artifact to me.

Mont Aigoual France Temperature Graph

Even then, a lot of months are dead flat. Winters are not warmer. Though it does look like they might have managed to cherry pick a couple of warm summers in the last decade.

Toulouse Blag

Another of these wonderful long lived stations, but with a horrid data dropout between the late 1800s and the post W.W.II era. Just puzzling. Why leave out the 1920s and 1930? (The ’40s had that ‘war thing’ so that’s understandable I guess).

At any rate, we’re starting to see a familiar story. Many flat months. Some winter “warming” but not by actually getting warmer, just by having less very very cold excursions. The “Baseline” below the Hot Pink trend line, but a 1990-91 Great Leap to above the line, then Failure To Advance.

Toulouse Blag France Temperature Graph

To me it just shouts equipment and processing artifacts. CO2? I’m just not seeing it…

There is also an interesting spike in the summer temps at the very end. Perhaps a low cloud cover combined with lots of Tarmac at the Airport? A shift of the AMO or ??? But in any case, that spike is not global, it’s not annual, and it’s clear a local weather event, not a climate event. 2008 temps returned to trend and winters didn’t participate.

Montpelier

Another substantially useless scrap of recent data spliced onto an old record. Good only for inducing GIStemp and related codes to make up ‘missing data’ from whole cloth and find what is not there.

Montpelier France Temperature Graph

Does look to have a significant splice artifact at the join, though. The prior segments are more or less flat (with January dropping) and all the ‘trend’ comes from the splice. Splicing this to even more stations will not improve it, no matter how ‘elegant’ the splice.

Marseille Marignane

A nice picture. Reasonably long set, reasonably complete. A bit of a “dropout” in W.W.II with a bit of what looks like it may be a process or equipment change artifact at the “splice” after that.

But “the big lumps” are that post 1990-1991 “jump”. We prune the low going ranges in winter (notice that bottom red line no longer gets past the trend line for January (dotted blue). But the tops of the winter data do not rise above prior tops. We are losing winter cold excursions, but we are not getting warmer winters.

Marseille Marignane France Temperature Graph

Spring and fall are largely flat too, but look at that dotted green April. Peaks are still hanging around that 14 C to 15 C range, but the cool excursions are just gone. A nice consistent April for planting. I can live with that. As long as those are really nice days and not just a Q.A. artifact or the result of new tarmac at the Airport… I suppose it’s possible they’ve just had more spring and summer sun in the last decade. A comparison of the cloud cover records for that decade vs prior would be another nice “Dig Here!”.

Nice

With a Nice Splice ;-) but this time with the 1800’s ending a bit earlier.

A bit more even ‘warming’ but with September continuing to not participate. (Wonder what September is like in France. It often does not show much change…)

Nice France Temperature Graph

Winters continuing the pattern of ‘higher lows’ but not ‘higher highs’.

Frankly, this all looks to me like what you would get from putting a thermometer near a pile of concrete and tarmac. Summer sun warms the tarmac more, thus the sunny months ‘warm’ over the Jet Age Airport Growth. Winter lows are suppressed by all the thermal mass and fuel burn along with UHI in general.

Perpignan

Another dataset with a splice / gap. More of the same. Winter lows clipped, but highs not rising. The “1990-91” artifacts. Perhaps a sunny few years in the 2000s.

Perpignan France Temperature Graph

Ajaccio

A very short data set that captures the rise out of a cold baseline to the present (but omits the prior warm periods). Oddly, January warming, but February and December not so much… Strange how that happens… An interesting ‘step function’ higher in the 1990-91 area with significant impact on summers. (September not so much… does it get cloudy in September?)

Adjaccio France Temperature Graph

To me it looks like a lot of specific seasonal effects, likely from changes of how the measuring is done or land use. Other than some kind of cloud modulation, I can not see how CO2 could do this (and as I understand it CO2 is not the major driver of clouds.) Frankly, even the ‘clouds did it’ hypothesis would lead me to suspect the Solar / Cosmic Ray process far more than the CO2 thesis.

Conclusions

We have the suppression of cold going spikes in recent years, especially in winter. We have sunny places with tarmac in the sun and cold places with what looks like added winter heat. We have different months rising or flat in different places, and even the sporadic cooling months…

And we have a bunch of thermometers that only have very old or very recent data but with the middle gone. Basically useless. GIStemp tosses any segment with less than 20 years of data… but will ‘make up bits’ when possible. I’d rather use real data and not made up bits.

I see lots of opportunities for ‘splice artifacts’ given that many of these will be ‘filled in’ with data from other places, and almost all of them will be compared to some OTHER thermometer in the ‘grid / box anomaly creation step’. The truth in the data, that it’s by month and context dependent change, will be lost in that homogenizing and infilling process, then the ‘grid box’ will make up fantasy “anomalies” comparing thermometers that exist now to different thermometers in the past. Pretty much a broken idea.

But one thing I don’t see is the broadly and evenly rising temperatures you would expect from CO2 generally keeping heat in. And given that different places have warmer or colder summers, and others have warmer or colder winters, it looks more like a ‘selection bias’ artifact in the summary than any broadly acting “climate change”.

The Meta Data

The GHCN Meta Data can be a bit fanciful some times, but here it is. I’ll present it two ways. As a “ragged right” where you can see all the fields and as a “fixed format” where the right side gets truncated, but the data fields line up better for easier comparison.

First up, fixed format:

61507015000 LILLE                           50.57    3.10   52   33U  171FLxxno-9x-9WARM CROPS      C   24
61507037000 ROUEN                           49.38    1.18  157  131U  114HIxxno-9A 3WARM CROPS      C   13
61507110000 BREST                           48.45   -4.42  103   78U  164FLxxCO 7A 3WARM CROPS      C   15
61507190000 STRASBOURG                      48.55    7.63  154  170U  252FLxxno-9A 3WARM DECIDUOUS  C   22
61507222000 NANTES                          47.17   -1.60   27   51U  253FLxxno-9A 3WARM CROPS      C   30
61507255000 BOURGES                         47.07    2.37  166  152U   75HIxxno-9A 1WARM CROPS      C   32
61507280000 DIJON                           47.27    5.08  227  241U  150HIxxno-9A 4WARM FOR./FIELD C   17
61507434000 LIMOGES                         45.87    1.18  402  335U  136HIxxno-9A 5WARM CROPS      C   13
61507460000 CLERMONT-FERR                   45.78    3.17  330  473U  153MVxxno-9x-9WARM CROPS      C   41
61507510000 BORDEAUX/MERI                   44.83   -0.70   61   44U  220FLxxCO30A 3WARM DECIDUOUS  C   25
61507560000 MONT AIGOUAL                    44.12    3.58 1565 1019R   -9MTxxno-9x-9MED. GRAZING    A    0
61507630000 TOULOUSE/BLAG                   43.63    1.37  153  160U  371FLxxno-9A 3WARM GRASS/SHRUBC   61
61507643000 MONTPELLIER                     43.58    3.97    6   38U  178HIxxCO 8x-9WARM CROPS      C   34
61507650000 MARSEILLE/MARIGNANE FRANCE      43.30    5.40    8   95U  901HIxxCO10A10WATER           C   88
61507690000 NICE                            43.65    7.20   10  142U  331MVxxCO 1A 5WARM CROPS      C   17
61507747000 PERPIGNAN                       42.73    2.87   48   45U  101FLxxCO12x-9WARM CROPS      C   32
61507761000 AJACCIO                         41.92    8.80    9   80S   47MVxxCO 1A 3MED. GRAZING    C   17

The “Air Station Flag” is that “A” (or an x if not an airport) just after the “47MVxxCO 1” for Ajaccio or the x in “101FLxxCO12x” just before -9WARM for Perpignan. So the format is 12 digits of StationID ( 3 country code, 5 station, 3 sub-station), Station Name, LAT, LONG, Altitude in Meters from reports, Altitude in Meters from maps, RSU Rural Suburban Urban flag, Population, then a set of coded bits about distance from water, on an island, etc, the already discussed Air Station Flag, a couple of more flags, the Terrain Type NEARBY from a map, and a couple of more flags.

So that last entry for Ajaccio says it is in country 615 (France), is station 07761 with no substation number, the name is AJACCIO and it is at LAT 41.92 LONG 8.80 with a reported elevation of 9 meters and a terrain map elevation of 80 meters. It is a Suburban station with 47,000 population and is an Airport near Mediterranean Grazing land. (I’ll spare you the interpretation of the other flags that mostly have to do with distance to water, if on an island, and similar details).

And ragged right:

61507015000 LILLE 50.57 3.10 52 33U 171FLxxno-9x-9WARM CROPS C 24
61507037000 ROUEN 49.38 1.18 157 131U 114HIxxno-9A 3WARM CROPS C 13
61507110000 BREST 48.45 -4.42 103 78U 164FLxxCO 7A 3WARM CROPS C 15
61507190000 STRASBOURG 48.55 7.63 154 170U 252FLxxno-9A 3WARM DECIDUOUS C 22
61507222000 NANTES 47.17 -1.60 27 51U 253FLxxno-9A 3WARM CROPS C 30
61507255000 BOURGES 47.07 2.37 166 152U 75HIxxno-9A 1WARM CROPS C 32
61507280000 DIJON 47.27 5.08 227 241U 150HIxxno-9A 4WARM FOR./FIELD C 17
61507434000 LIMOGES 45.87 1.18 402 335U 136HIxxno-9A 5WARM CROPS C 13
61507460000 CLERMONT-FERR 45.78 3.17 330 473U 153MVxxno-9x-9WARM CROPS C 41
61507510000 BORDEAUX/MERI 44.83 -0.70 61 44U 220FLxxCO30A 3WARM DECIDUOUS C 25
61507560000 MONT AIGOUAL 44.12 3.58 1565 1019R -9MTxxno-9x-9MED. GRAZING A 0
61507630000 TOULOUSE/BLAG 43.63 1.37 153 160U 371FLxxno-9A 3WARM GRASS/SHRUBC 61
61507643000 MONTPELLIER 43.58 3.97 6 38U 178HIxxCO 8x-9WARM CROPS C 34
61507650000 MARSEILLE/MARIGNANE FRANCE 43.30 5.40 8 95U 901HIxxCO10A10WATER C 88
61507690000 NICE 43.65 7.20 10 142U 331MVxxCO 1A 5WARM CROPS C 17
61507747000 PERPIGNAN 42.73 2.87 48 45U 101FLxxCO12x-9WARM CROPS C 32
61507761000 AJACCIO 41.92 8.80 9 80S 47MVxxCO 1A 3MED. GRAZING C 17