Temperature Adjustments: More Madagascar Madness

This started out as a discussion point following E.M. Smith's blog post Mysterious Madagascar Muse. The jist of the original article centred around the availability of data after 1990 in the GHCN dataset and the NASA/GISS treatment of temperature on the island. Well Madagascar has a bit of a further story to tell. I had offered to plot a 'spaghetti' graph of the temperatures from the ten stations used on Madagascar, and this has proven interesting as an example of how data is adjusted and filled in by GISS.

To start, the annual mean temperatures plotted on a graph (Figure 1) show clearly the differences between the stations - Antananarivo is high altitude and relatively cool, with a cooling trend; of the other stations, some have cooling trends, most are warming. Also noticeable is the very sparse data after 1990. Note the darker blue data for Maintirano, of which more later.

Figure 1. Annual Mean Temperatures for Undadjusted Madagascar Stations.

With such temperature differences between sites, obviously you cannot just average the temperatures. This is what it looks like if you do (Figure 2), and it clearly does not work as an average temperature for the island.

Figure 2. Averaged Annual Mean Temperatures (Clearly Wrong!)

Normalizing each of the temperature series by calculating the mean temperature for that station for the baseline period of 1951-1980 allows plotting of an anomaly-based ‘spaghetti’ graph (Figure 3). This shows what looks like warming-cooling-warming climate cycles very clearly and it is possible to fit a third order polynomial trendline though the averaged data. I've seen this again and again for data I've plotted around the world (incidentally these were for WUWT regular TonyB).

Figure 3. Normalized Unadjusted Annual Mean Temperatures for Stations on Madagascar

Now for the interesting bit - how GIStemp adjusts the data. GIStemp takes rural datasets and uses them to correct for urban warming. In this set of ten unadjusted stations there were three rural ones: Maintirano and two overlapping but separate ones for Antalalava (why kept separate?). In the homogenized set, only Maintirano, which has a large warming trend of 1.16 deg. C/century, remains unadjusted and all the other stations (Figure 4) have the trend increased - it seems to match Maintirano.

E.M.Smith finds seven other rural stations within 1000km that may contribute to homogenization. They also show cooling to about 1965-1975, then a warming trend. This is lost from the homogenized data.

Figure 4. Annual Mean Temperatures for Adjusted Madagascar Stations.

So overall what effect does homogenization have? - well a big one. Having started into a better understanding of calculation of anomalies, I decided it was better to leave that for the present, but a straight average of the normalized unadjusted and homogenized overlaid with a 10 year moving average for each (Figure 5) shows just what homogenisation does for the 'anomaly' value for Madagascar calculated this way - it stabilises the base period and significantly warms the subsequent years.

Figure 5. Comparison of Unadjusted and Adjusted Normalised Annual Mean Temperatures for Madagascar.

Given that several of the stations show a cooling trend prior to homogenization, and that UHI correction should NEVER be in the wrong direction, this is nothing short of scandalous.

I originally looked at the temperature trends using a database that has been developed over the last two months, but when I checked for any up-dated data on the GISS site, I found the trends were different (Table 1). We've now found the reason for that and that is worth investigating in its own right. The answer is simple - bad data. The database QC system throws out any year with missing months of data, and after 1990 the data in most of the Madagascar stations is patchy at best, so the database ignored the data in plotting the temperature trends. It is amazing how much warmer Madagascar is with that patchy data included.

>Table 1. Temperature Trends for Data Madagascar Stations: Comparison of Sources with/without QC Control (see text).

One final thing. Even the patchy data stops in 2005, so after this date Madagascar too gets 'filled in' data from elsewhere - it seems from the rural stations up to 1000km away - again. And even the stations used to 'fill in' have patchy data - many have a gap then ONE DATA POINT in 2009. This is unbelievable. Rather than give an example, check the station hyperlinks below for yourself:

	Ile Juan De N		17.1 S	42.7 E	111619700000	rural area	1973 - 2009
	Dzaoudzi/Pama		12.8 S	45.3 E	163670050000	rural area	1951 - 2009
	Iles Glorieus		11.6 S	47.3 E	111619680000	rural area	1956 - 2009
	Ouani (Anjoua		12.1 S	44.4 E	111670040000	rural area	1963 - 1984
	Serge-Frolow		15.9 S	54.5 E	168619760000	rural area	1954 - 2009
	Ile Europa		22.3 S	40.3 E	111619720000	rural area	1951 - 2009
	Porto Amelia		13.0 S	40.5 E	131672150004	rural area	1987 - 200

Read more Entry>>

On climate analysis, blog hits and science communication....

I have 3D scatter plots of temperature trend data slowly turning before my mind's eye. The frustrating thing is that that is the only place I can see them; I can't reproduce them on the computer.  I have to find another way to show what I am able to see. I also need some serious thinking time about the best number crunching.  I need a break, to stand back and think about it all...


Digging in the Clay hit 8,000 page views on blog stats the other day.  I was going to wait until 10,000 (sometime mid-February at the present rate) to report it, but it fits with my reflective mood today I guess. That is 8,000 page views from 4,000 visits by 2,000 unique visitors in two and a half months - more than I dreamed possible.  Actually I'm not sure if I care about blog stats that much, although it is gratifying.  I just needed a place to 'think out loud', one where the pressure of possible public scrutiny keeps you on your toes in terms of both accuracy and clear communication.  And if others picked up on some of it and found it contributed to the bigger picture so much the better.

Friends have commented on the melange of styles of the blog posts (well mine - vjones). Yes - different styles for different purposes. At the moment the analytical stuff is getting the 'report what you did and what it shows with minimal opinion' treatment. This is deliberate - straight, boring and science-like; the opinion comes later when the bigger picture is complete. Part of my reason for this is that I am often put off by the framing of anything to do with climate science.  I am more likely to read something relatively neutral. I like to be informed, but to make up my own mind.  On the other hand I've let my opinions spill over into several posts on climate science methods and communication e.g. here, and experimented with different styles (Climate Fast Food) and means of communication (GIStemp Reloaded).

That brings me to a link I was sent yesterday (knowing my interest in science communication) to a book review in this month's Science, which includes Don’t Be Such a Scientist: Talking Substance in an Age of Style by Randy Olson. Actually, I found the review and the mention of Olson's film Sizzle: A Global Warming Comedy, a little off-putting, but the synopsis and chapter headlines on the book website look interesting so I will at least pick it up for a closer look next time I'm in a bookstore. Chapter 1 is entitled Don't Be So Cerebral - basically, appeal to other senses - and presumably the one that journalists go for - gut reaction ;-).  Plenty of that in climate science communication; all those poor, poor polar bears (/sarc off).
Read more Entry>>

Adjustment Effects on Temperature Trends: Part 3 - Effects by Latitude

Part 3 looks at a geographical distribution of how adjustments affect temperature trends.  I have not dug down to the country level, although I wouldn't rule it out for some future investigation. I did look at distribution by WMO region, but very little to see there. However, when I looked at distribution by lattitude band, a clear warming bias stood out for 24N to 44N (Figure 1). 

 

Figure 1. Distribution by latitude band showing how adjustments affect the warming or cooling trends of temperature data sets.

As with Part 1 and Part 2 I immediately wanted to see the detail of the magnitude of adjustment.  Figure 2 shows the bias very well - again for 24N to 44N. There is indeed a strong warming bias in the adjustments, with huge adjustments of 1-2.5 degC/Century.

Figure 2. A breakdown of the magnitude of temperature trend adjustment by latitude band.

There are only very slight biases in other latitidue bands, if they are detectable at all. I was tempted not to even show them, however here are the graphs for 64N to 90N (Figure 3) where there are cooling adjustments, and the bands either side of the Equator - 24N to 24S (Figure 4).  There is also a cooling bias in Equator to 24N.  Again these are subtle.

                            

Figure 3. Adjustments for 64N to 90N       Figure 4. Adjustments for bands spanning the Equator.

I suppose until I put all this together I am not sure of the significance this bias in adjustment. At present I cannot say it is intentional that so many small warming biases have crept in with each adjustment, but I am more sure than ever that the adjustments now must be justified.

So where next with this?  Well, at the moment I have a hankering to examine the effect of the major fall off of stations after 1990, but I guess I'll have to play with the data for a while to see what it shows.

Read more Entry>>

Adjustment Effects on Temperature Trends: Part 2 - Magnitudes of Adjustment

Part 1 suggested there were more warming adjustments than cooling ones. Now does that not fly in the face of analysis done by others (e.g. Giorgio Gilestro and as discussed here)? Actually, no. It is just that I dug deeper! Very convenient little graph that one of GG's. It hides a lot. Here is my version (Figure 1), reproduced from the data and trends in the TEKtemp database:

Figure 1. Reproduction of distribution analysis of GHCN V2.mean temperature data adjustments.

Now, having spent weeks working on the data on and off, I know that many of the adjustments are quite small, so I decided to expand or rather sub-divide the frequancy bins in the areas of the smaller adjustments. I expanded the frequancy bins on either side to give better scaling on the y-axis and show up any discrepancy in symmetry. In Figure 2 the bias towards warming adjustments now shows up. Same data, but this time put under the microscope rather than viewed though the wrong end of a telescope. The bias is small, but it is there. The next question is - is it significant? Kevin will probably want to add something here as he had a lot to say on Giorgio Gilestro's thread at the time...

Figure 2. Distribution analysis of GHCN V2.mean temperature data adjustments with an increased number of (small magnitude) frequency bins.

Since Kevins world temperature trend maps use DegC/Century, I have got more used to looking at these units and the next thing I wanted to do was look at the distribution of trends that were more closely related to those used in the maps. Figure 3 shows there is a slightly larger number of data sets with warming trends of up to 2.49 degC/Century, in comparison to the number of ones with an equivalent rate of cooling. Overall there are 2319 sets with a warming trend and 1948 with a cooling trend (I have not included those with no trend in these counts, but I have included all those on either side of zero).

Figure 3. Distribution of trends across a reduced number of categories appropriate to TEKtemp trend maps.

In Part 1 it was obvious that adjustment makes a great deal of difference to the distribution of trends on a scatter plot, with seemingly a larger number of warming adjustments than cooling ones. Figure 3 confirms that, but in Part 1 I also showed there were six different types of adjustment, so that brought me to produce Figure 4.

Figure 4. Distribution of adjustment trends by both the type and magnitude of adjustment.

The first thing that struck me about Figure 4 - and Figure 3 - was how symmetrical they are, but not quite. I mean I was expecting to see a great deal more adjustment to cooling trends than we actually see. Perhaps that is just my abhorance of some of the warming trends associated with the growth of cities that seem to be an intrinsic part of this data: I was hoping the adjustments would evidence the correction of UHI, although in fairness I have not yet looked at many individual cities or generally yet at that level of detail. Figure 5 breaks down the data into paired graphs for each type of adjustment.

Figure 5. Distribution of adjustment trends by both the type and magnitude of adjustment:
[A] Warming to cooling; [B] Cooling to warming; [C] Warming to less warming; [D] Warming to more warming; [E] Cooling to increased cooling; [F] Cooling to less cooling.

So in each of the paired graphs, there appear to be more data sets with warming trends than for the equivalent magnitude of cooling trends. So for the trend category of 0.5 to 0.99 DegC/Century (Cooling to warming), there are 181 data sets, but for -0.99 to -0.5 DegC/Century there are only 103 (Warming to cooling). For increased warming [D] 1.0 to 2.49 DegC/Century there are 362 data sets, but for less warming [C] -1.0 to -2.49 DegC/Century there are only 215.

I am almost tempted to think that you couldn't plan this better, I mean where you adjust the raw data and add warming somewhere, you also adjust other raw data so that you add an equal and opposite amount of  cooling elsewhere so that overall you introduce only a small overall warming bias in the trends. I just can't see how so much warming can fit the 'need for correction' of the data. Yes, some stations will need to have a warming adjustment, but a slight majority? And these are often not small adjustments.

In Part 3 I will look at how these types of adjustments are distributed by latitude.

Read more Entry>>

Adjustment Effects on Temperature Trends: Part 1

This 'digging' into temperature data is getting kind of addictive.  The previous post asked the simple question " how many [temperature stations] have data adjusted?" and showed the proportions of adjusted data by WMO region. But, as soon as you ask the question - "Does adjustment affect temperature trends?" you immediately want to follow it up with "By how much?" and  "In what way?", then "Where on the globe?" So here is a taster.

A quick recap first.  Kevin's TEKtemp database (to which access is available for anyone wishing to look at the data in database form - just post a comment or email Kevin - more info here) and a bit of bespoke code, has allowed us to look at and compare the temperature trends at individual temperature stations across the globe and to produce maps that show regions with cooling and warming trends. The figures below show examples of  TEKtemp graphs for two of the stations in the NOAA GHCN dataset that have adjustments which affect the temperature trends. Note that Figure 1 (Orland, CA) has only minor adjustments that reduce the temperature trend slightly reducing the cooling trend;  Figure 2 (Rock Springs Airport, WY) has major adjustments that turn a warming trend into a cooling trend.

Figure 1. Temperature plots and trends for Orland, CA

 

 Figure 2. Temperature plots and trends for Rock Springs, WY

The piece of code Kevin has written to plot graphs and calculate trends has a minimum data QC level of 20 years; any year with one or more missing months is not included in the plot. Comparing the trends by station in a large table has been interesting to say the least.  Of the datasets that qualified for trend calculation, 1065 were not adjusted by NOAA, whereas 4508 were adjusted in some way. Of the 4508, 2319 showed an increased trend and 1955 showed a decreased trend after adjustment;  234 had minor adjustments which had no effect on the trend.

I have initially simply calculated the difference between the trends. This gives me a number for each station record that refects both the magnitude and direction of adjustment.  I've plotted that against the raw data trends (Figure 3) and separately against the adjusted data trends (Figure 4) so these are, in effect, 'before' and 'after' graphs.

 

Figure 3.  Distribution of adjustments vs raw data trends.  

Figure 4. Distribution of adjustments vs adjusted data trends.

These two figures made me go 'wow', but I'm not sure that I should be that surprised, after all we know that some data is warmed by adjustment and some is cooled by it. Figure 3 shows there is a lot of cooling data that will be warmed by adjustment and a lot of warm data that will be cooled by adjustment, overall there are more points to the right of the Y axis and therefore more warming than cooling station trends.  Figure 4 suggests that, after adjustment there is even more data with warmig trend, but perhaps this is just a 'by eye' bias.

Figure 5.  Distibution of data trends of raw data vs adjusted data.

Figure 5 on the other hand confirms this - definately more warming than cooling data and it SEEMS to be more warmed by adjustment.  Could this really be?

[Update] I had hesitated to add Figure 6 when I first wrote this post as it is complex and I was not confident about how to present it, however I think it shows the adjustments rather well.  To understand it, consider that there are six types of adjustments that affect the temperature trends:

• Warming to cooling
• Cooling to warming
• Warming to less warming
• Warming to increased warming
• Cooling to less cooling
• Cooling to increased cooling

 The six parts of Figure 6 below are the exploded out 'parts' of the Figures 3 and 4, showing each of these types of adjustment.  The comparison on each graph of the raw and adjusted slope gives an idea of just how much change there is to some of the individual temperature records.

Figure 6. Distribution of adjustments vs raw and adjusted data trends: exploded views examinng the six types of adjustments: reduced cooling [a]; cooling to warming [b]; increased warming [c]; increased cooling [d]; warming to cooling [e]; reduced warming [f].

I find this set of graphs quite persuasive.  I do believe adjustment is necessary, and as perhaps anticipated, the vast majority of adjustements are small and make only a small difference to the overall trend, however there are some very large adjustments and some of the most extreme trends are adjusted to produce an extreme trend of the opposite sign.  Kevin had previously explored reported many of these as physically unjustifiable. I concur.  In Part 2 I'll look more at the magnitude and distribution of these six types of adjustments.
Read more Entry>>

Temperature stations : how many have data adjusted?

A previous post looked at the how the number of stations used in reporting climatic temperature data through the NOAA GHCN database has varied since 1880. Here, before starting to examing the effects of adjustment, I'm simply looking at how much of the data is adjusted. It's a lot actually. I'm not going to do too much discussion here; I just really want to let the graphs speak for themselves.

Figure 1. Number and breakdown by WMO Region of stations for which there is [A] raw data and [B] adjusted data in the NOAA/GHCN database.

Figure 1 shows how the total number of stations is divided by WMO region for both 'raw' and adjusted data over the period 1880 to 2009.  It is probably easier to look at Figure 2 to see how much of the data is adusted. On a percentage basis more than 70% of the data is adjusted in some way up to 1990, when there is not only a massive drop-off in the number of stations, but also a drop in adjustment.  After 2006, we are not only left with about 800 stations reporting temperature to the database from across the globe, but the percentage of those that are adjusted in some way falls to 20-30%. Why? Is it just 'high quality' stations that remain? 

Figure 2. Graph showing the total number of stations in the NOAA/GHCN database and the
(black line) percentage that are adjusted by year.

Looking at the adjustments by region is quite revealing too.  In Figure 3 there is quite a bit of variability in the percentage adjustments by region.  Antarctica starts with 100% adjustment becuase data from Base Orcadas (60S) is modified and pasted into the record prior to any stations below 64S reporting data in 1945.  Data from Europe, Asia and the South-West Pacific (Australia, New Zealand, Indonesia, Malaysia etc.) are adjusted most, whilst African data is adjusted least.

Figure 3. Regional percentages of adjusted data: numbers of adjusted stations for each region
are expressed as a percentage of the total number of stations.

The data for individual regions is perhaps easier to see in Figure 4. What intrigues me is why it is so variable by region (Africa vs Asia) and the reasons may come to light when we examine the adjustment of rural vs urban data.  And then what is up with 1990? Why that sudden drop off? (I must go and look at some station metadata).

A guest post at Die Klimazwiebel today from Reinhard Böhm of the Central Institute for Meteorology and Geodynamics (ZAMG) in Austria, discusses the need for adjustment (homogenization) and he is concerned that access to unadjusted data can result in its 'misuse' by those who do not understand the inherent biases that require adjustment. What he says is important and I agree with his reasoning for a lot of it. He then says:

"I can advise everyone to use original data only for controlling the quality or the respective homogenization attempts but not for analysis itself if the goal is a timeframe of 20 years or more – a length usually necessary to gain statistically significance at the given high frequent variability of climate."

Just one thing to point out here, there are adjustments and adjustments.  The NOAA GHCN 'Raw' data is already adjusted (for time of observation, station history etc.). There is then a further set of homogenisation done by either GHCN or GISS and these adjustment have been the focus of our analysis. [Update 23rd Jan. After checking NCDC documentation here I can see I was wrong - v2.mean is 'raw' data.]

Adjustments are an integral part of temperature station data and climate analysis, but they should be necessary and appropriate.  So far in our analysis we have found a lot of adjustment that seems to be neither, or at least it is not clear how some of the adjustements we see can be justified as either.  However, what our approach allows us to do is to isolate sub-sections of data very rapidly for comparison and analysis.  So far what we have found is interesting..... (but you'll still have to be patient, Andy).
Read more Entry>>

The 'Station drop out' problem

Now that I've produced a series of colour coded maps showing the warming/cooling trends in the NOAA/GISS GHCN data for three distinct time periods i.e. 1880 to 1939, 1940 to 1969 and 1970 to 2010 (as well as for the whole 1880 to 2010 period), I've noticed that a number people commenting on the 'Mapping global warming' thread here are unaware of the NOAA/GISS station 'drop out' issue and how it may affect the warming/cooling trends.

The primary purpose of this new thread is to show charts of the number (i.e. count) of stations by year in the NOAA (and so therefore more or less GISS also) GHCN raw and adjusted datasets.

Because I've further broken down the station counts by 'WMO Region' and still further by Country for each of the separate charts for each WMO Region, I've chosen to display the charts as both 'stacked' area charts and 'unstacked' area charts.Because there are many countries in any given WMO Region, I've also chosen to show separate series for the 'Top 10' countries in a region and have lumped the remainder into an 'All Others' category. To see a much larger readable version of a chart just click on the appropriate chart.

I'll now present each set of charts in turn, starting with the overall raw/adjusted charts first, and followed by some observations that are evident within the charts. Note that in each case the 'No. of stations' is the count of the no. of stations that have recorded temperature data available in the GHCN dataset in that given year.

Raw data - No. of stations by year for All WMO Regions

Figure 1 - Raw stations by year (stacked) Figure 2 - Raw stations by year (unstacked)

Figures 1 and 2 above show for how many stations there is raw temperature data available in the NOAA GHCN raw dataset (the v2.mean file) for each year from 1880 to 2009. The station count data has been broken down further to show the station counts for each WMO Region as a separate series. As can be seen from the charts, there are relatively few (less than 500) stations that have raw temperature data prior to 1880. From 1880 onwards there is a more or less linear increase in the no. of reporting stations (i.e. stations that have raw data) from 1880 to about 1950 when the number reaches a little over 3300. After this point, within the space of four years, there is a sudden expansion in the number to over 4500, which then reaches a peak of 5348 stations in 1966. Its worth mentioning that there are 7250 records in the GHCN station inventory file (v2.temperature.inv) some of which are for 'Ships' but it is clear from this peak count that there isn't raw temperature data available in the GHCN v2.mean file for all the stations listed in the NOAA GHCN station inventory file.

After peaking in 1966 the total raw data station count then declines in a more or less linear fashion to about 3750 in 1989. Over the next couple of years there is a sudden 'drop out' of stations from the total station count to about 1900 in 1992. Figure 2 shows that this 'drop out' is for non-North American stations i..e Asian, European and South West Pacific (Australian) stations. Why the sudden preciptious drop in the station count post 1989/1990? Further fine details for this 'drop out' can be seen in the later charts for the individual WMO regions (see Figures 8,10 and 12). After 1992 there is a more or less linear decline in the raw data station count to about 1630 stations in 2005. There is then a further sudden inexplicale 'drop out' to about 960 stations in 2006 with in 2009 the total station count reduced to a mere 840 stations. What on earth is going on here? What caused the sudden increase in the number of reporting stations around 1950 and what caused the equally precipitous 'drop out' of many of these stations around 1989/1990?

Adjusted data - No. of stations by year for All WMO Regions



Figure 3 - Adjusted stations by year(stacked) Figure 4 - Adjusted stations by year (unstacked)

As can be seen from Figures 3 and 4 above, the general shape of the total adjusted data station count is very similar to that for the raw data station counts in Figures 1 and 2. There are however some interesting further observations to be noted. If you look at Figure 4 relative to Figure 2, you can see that, despite the addition of a significant number of raw data stations in North America (primarily the US) in about 1950, there is no corresponding step increase in the station count for the adjusted stations. In particular note the station 'drop out' post 1989/1990. After 1990 almost all the North American (US and Canadian) adjusted data stations have 'dopped out'. Also if you look at Figure 3 , in 2006, the number of adjusted stations plummets to less than 190 and of those only 27 are North American (US) stations; in 2008 the adjusted station count reaches as low as 55 (10 North American, 8 Asian, 2 European and 35 South West Pacific (Australia).

For anyone interested, the peak value in the number of adjusted stations is 4018 in 1966. How on earth is anyone supposed to know what happened to 'global warming' in the first decade of the 21st century when the station counts have been reduced to such a ridiculously lower level in the NOAA GHCN dataset? Particularly when you bare in mind that it is this dataset that forms the input to GISTemp that produces all those scary 'red almost everywhere' colour contoured anomaly maps?

Raw data stations - North and Central America



Figure 5 - North and Central America (stacked) Figure 6 - North and Central America (unstacked)

The key obsevations in Figures 5 and 6 are, firstly, that the US stations are dominant. Secondly, note the -sudden 'drop out' in the Canadian stations post-1989. Down from 407 in 1989 to only 35 in 1991! Why? Have NOAA fallen out with Environment Canada? What happened to the Mexican stations after 1985? Why the precipitous drop in the number of reporting US stations from 842 in 2005 to only 124 in 2006? I think this has something to do with USHCN version 2 but I'm not sure. If so why haven't NOAA just copied all the raw data they have for the US stations in the USHCNv2 raw dataset into the NOAA GHCN raw dataset? GISS takes the NOAA GHCN raw dataset as its input and replaces all the US stations with data it takes from the USHCNv2 dataset (and other Antarctic station from SCAR). Why therefore doesn't NOAA save GISS the job of having to read in and merge together data from their two separate datasets?


Raw data stations - Asia



Figure 7 - Asia (stacked) Figure 8 - Asia (unstacked)

The key obsevation from Figures 7 and 8 is that Chinese stations are the dominant contribution in Asia. It is also clear that are the primary cause of the sudden increase in the number of reporting stations after 1950. Note that the increase in the numbers for Japan and the Russain Federations is much more gradual. All three countries show the 'precipitous drop out' of reporting stations around 1989/1990. Why? The Chinese stations in particular drop from a high number of 361 in 1990 to only 14 in 1991. Very odd? In addition to upsetting Environment Canada, has NOAA also broken off diplomatic relations with the Chinese? It looks like diplomatic relations with Mongolia were broken off a little earlier than they were with China as the Mongolian station 'drop out' occurs after 1982/83. Meanwhile, having enjoyed good relations with South Korea from 1973 when the number of reporting stations increased to over 60, sadly after 1993, the South Koreans appear to have also fallen out with NOAA with the numbers dropping to only 10 in the subsequent years.

Raw data stations - Europe


Figure 9 - Europe (stacked) Figure 10 - Europe (unstacked)

Now who said "Turkeys never vote for Christmas"? Well according to Figures 10 and 11 the population of Turkish thermometers was thriving between 1961 and 1990. It looks like there was a dramatic change in Turkish voting patterns post 1990 - the Turkey thermometers clearly opted to vote for Christmas and as a consequence their population was culled dramatically with about only 10% of their population survivng post 1990. The post-1950 step increase in the number of reporting stations is evident for most of the European countries, particularly Poland, Germany, Italy and France. It seems the French (according to NOAA) didn't care very much about measuring temperature prior to 1950.. Clearly everyone in Europe decided that measuring temperature was less important after 1989/1990. Did the EU bureaucrats decide to introduce a tax on measuring temperature at about that time?


Raw data stations - South West Pacific



Figure 11 - South West Pacific (stacked) Figure 12 - South West Pacific (unstacked)

Well the WMO obviously doesn't like to upset people who live in the South West Pacific by just referring to it as Australia! But for a small and brief contribution from the Philippine and Indonesian stations from 1950 to 1975, the whole region is clearly dominated by the Australian stations with just a smidgeon of help from New Zealand and Malaysia. Quite why there is the extreme preciptitous (but now all too familar) 'drop out' after 1991/2 I don't know. There were ZERO reporting Australian stations in 1994 according to NOAA by the way! Why?



Raw data stations - Africa



Figure 13 - Africa (stacked)Figure 14 - Africa (unstacked)

There's not that much to see in Africa really. With the possible exception of South Africa, no one country dominates. What is evident, as with most of the other WMO regions, is the step increase in the number of reporting stations in what seems like most if not all of the African countries in around 1950. There appear to be two stages to the station 'drop outs' . The familarly 1989/1990 'drop out' appears to be preceded by a further 'drop out' period at around 1978/79. There's also an observable doubling in the number of reporting South African stations from 1961 onwards until the 1989/1990 'drop out'. From about 2002/3 almost 50% of the contribution towards teh African station count comes from only one country - Algeria. What happened to the South African stations?
Read more Entry>>