The Puerto Casado Story (it moved because there is change and there is a change because it moved)


A view from the river at Puerto Casado. Photo – Julio Balles

Puerto Casado is a small town along the border of Paraguay whose temperature history became a topic on Christopher Booker’s column in the Telegraph. Booker showed how a raw cooling trend had metamorphosed into its opposite after ‘adjustments’. This was contested quickly by climate scientist Ed Hawkins who pointed to the town’s record on BEST. It indicated the station had been moved.


BEST record: the red diamonds are marked ‘station moves’.

Perhaps the Puerto Casado station had been shifted to a cooler location… introducing a false cooling trend? The natural question to ask is – how does BEST know a rural station in the remote reaches of South America moved twice? Where does it get this information?

The answer, it turns out, is an entangled mess. The Puerto Casado record in BEST includes metadata from various sources. Metadata is what contains station location information. The sources collated by BEST show different latitude-longitude pairs for the station. The co-ordinates are slightly different, and they fall not far from one other.


The town labelled ‘La Victoria’ is Puerto Casado, for which it is another name. Some of the coordinates derived from BEST’s data sources for the station are shown.

BEST does not know the field reality of the station. Nor does it not know if the station truly moved or locations were wrongly recorded. Nor does BEST have information on the timing of any move. What it does is assume the station moved—given that different coordinates were recorded—and looks for breaks/shifts in the temperature. If breaks are present they are assumed to be due to moves. Plus, the breaks are assumed to have caused the station to look different from its neighbours.

In other words, what BEST records as ‘moves’ are not known documented moves.

Following this, BEST transforms the temperature series. It compensates for the ‘moves’ and tries to remove shifts. The result is a Puerto Casado record, which has its linear trend reversed by close to 2.7C per century.

Puerto Casado BEST

To answer our original questions about BEST: was the station moved? We don’t know for sure. When was it moved? We don’t know. What is the effect of the supposed moves? We don’t know but we think it changed the temperature. How do you know this? Because there is change. When do you think the moves happened? When the changes occurred. And what will do to ‘correct’ this? Make Puerto Casado look like every other station around it.

We can ask BEST further questions: Are you not data-peeking? How did you settle on such non-independent analysis? We can expect silence.

Remarkably enough, supporters of climate orthodoxy manage to top such circuitous circular reasoning. Kevin Cowtan, another global temperature adjustment practitioner, declares the station instruments at Puerto Casado must have suffered calibration errors—at the same points in time when BEST says the stations must have moved.

For all the hype, BEST’s methods produce local records that are no better than the NCDC, conjuring ad-hoc rationalizations for ‘adjustments’ from the temperatures themselves. The reasoning is circular and BEST and others do not even attempt to hide it. Journalists like Booker are right to question such methods and data.



  1. omanuel

    Thanks, Shub, for your efforts to understand why governments lie. Despite all their struggles, Big Brother is going down now!

    Our challenge today is to:

    1. Retain the benefits* of, and
    2. Eliminate the deception in

    The Great Social Experiment of 1945-2015

    We must not get side-tracked trying to punish those who used grant funds to deceive the public for the past seventy years.

    *Benefits: Reduced nationalism, racism, sexism, xenophobia, etc.

  2. Pingback: Shub Niggurath On The Paraguayan Adjustments | NOT A LOT OF PEOPLE KNOW THAT
  3. Paul Matthews

    I read the BEST methods paper. It is incredibly vague on station moves:

    “we incorporate a procedure that detects large discontinuities in time in a single station record. These could be caused by undocumented station moves,…”

    “The first is an examination of station metadata, such as documented station moves or instrumentation changes. For the current paper, the only metadata-based cut we use is based on gaps in the record; if a station failed to report temperature data for a year or more, then we consider that gap as evidence of a change in station conditions and break the time series into separate records at either side of the gap. In the future, we will extend the use of the scalpel to processes such as station moves and instrumentation changes; however, the analysis presented below is based on the GHCN dataset which does not provide the necessary metadata to make those cuts.”

    So it’s completely unclear what they are doing. Yet they label the points on the graph as a “Station move”.

  4. omanuel

    Robert K. Wilcox, well-known author of a book on the intrigue at the end of WWII, recently described the fact that the UN was formed on 24 Oct 1945 to forbid public knowledge of the source of energy that destroyed Hiroshima – the source of energy in the core of the Sun – as “the greatest secret of the universe” !

  5. Shub Niggurath

    Paul, I agree about their Methods paper, overly focused on their methodologic cleverness, which may not be inappropriate given their proposing newew methods in temperature analysis, but a reduced focus on data integrity.

    Which is perhaps why we see them using such terms as ‘station move’ in a cavalier and non-specific manner.

    This is Mosher on what a BEST station move is:

    You have one record from GCOS that starts in year 19xx and reports
    position XYZ. it ends
    Then you have another source that picks up and reports a different location. it stops
    Then you have a third source that picks up and reports a third location.

    So, “station move” is one probable explanation. The other explanation
    is bad metadata or less accurate metadata.

    With different locations we make the decision to treat these records as different stations. If there was no move in actuality then this decision has zero effect on the data.

    “station move” is a simple but not necessarily lucid way of describing the process.

    In other words, a ‘move’ doesn’t convey any meaning in BEST. BEST’s labeling of their figures catches Ed Hawkins and such homogeneists as Victor Venema off-guard.

    The only sure way to find out is to learn the actual history of temperature recording in such stations.

  6. omanuel

    BEST drug the good name of UC Berkeley into the climate debate and tried to save the reputations of UN’s IPCC and the Norwegian Nobel Peace Prize Committee.

    Instead, the excellent reputation of the institution where I was in 1962-1964 for research on stable isotope mass spectrometry was badly damaged.

    Professor Judith Curry recognized the problem and very wisely put some distance between herself and BEST.

  7. Colorado Wellington

    For all the hype, BEST’s methods produce local records that are no better than the NCDC, conjuring ad-hoc rationalizations for ‘adjustments’ from the temperatures themselves.

    Shub, not only do they create these rationalizations ad-hoc but they back into them on the basis of a reverse-engineered post hoc fallacy:

    X must have happened because Y happened because if Y followed X then Y must have been caused by X.

    i.e. specifically

    The station must have moved because measured temperatures dropped because if measured temperatures dropped after a station move then the measured drop must have been caused by a station move.

    A straightforward “post hoc ergo propter hoc” would be bad enough to disqualify such nonsense. This convoluted mess is worse, if not the worst.

    Certainly not the BEST.

  8. Shub Niggurath

    In a way, yes. Why people have trouble calling circular reasoning in simple terms is beyond me.

    The reason why I hesitated on ‘post-hoc’: BEST are in the business of producing a global reconstruction. Their method essentially frees them from concerns of *local station data quality and integrity*. If a station record shows numerous shifts, as many do, BEST will label each one a separate station and incorporate the data into the gridded average calculation. That is all they are concerned about.

    What is therefore puzzling is, why BEST applies its breakpoint detection methods and reconstructs *local station temperature records*. The source of information that a move occurred is derived independently. The information is used as an excuse to data-peek, i.e., designate points in time as being when ‘moves’ occurred. You only know a move occurred if you know a move occurred. Otherwise you don’t. Everything else is cheating.

  9. David Haden

    The data file and series suggests a station move in 1971, and again in 2005 (appears to be 2006 in the graph). It would thus seem most likely that the station was first sited on the south of the hill located just a little to the east of the town. Then in 1971 it was moved across the river from the town. Circa 2005 it was moved again, out to what gazetteers call the “Puerto La Victoria Airport (PCJ)” – which appears on the satellite imagery to be a rather rough-looking grass airstrip cut into forest, east of the town. The most easterly co-ordinates you give in your blog post are close to the track that leads to the north end of this airstrip. Parallel to the road at this junction there is indeed a short well-defined track shown on the satellite photography, and a likely looking candidate for three weather station emplacements can seen at the end of this track. See these Google Maps links…

    I would hazard a guess that a 2005 station would have been in response to the “GCOS REGIONAL ACTION PLAN FOR SOUTH AMERICA” document (September 2004). This found that “climate observing systems in many South American nations are in such a state of disrepair that reliable assessment, quantification, and prediction of climatic conditions and their impacts has been compromised” and the plan “aim[d] to improve systematic climate observation programmes in South America.” Online at:

  10. Colorado Wellington

    I understand your “post hoc” hesitation as it applies to the primary method, Shub.

    They created an attractive, seemingly independent franchise for themselves. They will protect their business as long as they can.

  11. Shub Niggurath

    David, excellent. It appears we have been along the same path. I’ve been working to re-find where I read that many South American stations had exposure problems – poorly sited, left open to sun etc. This was subsequently addressed. From Jo Nova’s site, where commenter Eliza has left details, it appears this could have been in the early 70s (he says his dad was involved in the project). From the WMO document above, it is plausible the second set of corrective actions (for whatever reason) were undertaken in the mid-2000s. Additionally, it is plausible the first site was behind what appears to be an old factory though the exact coordinates themselves lie over a pond with algal overgrowth – this is the ‘bald patch’ to the east and south of the large building. There is a second point where two coordinates lie – across the river at the town and to the east. The problem here is this is unlikely to have been a site for a station as the area gets flooded. The flooded areas can be seen on different satellite images. But it is hard to be categorical. The third site is likely the far east one, as you point out, which falls somewhere at the north end what appears to be a linear broad swath of cleared trees – the airport. Interestingly, from Panoramio, it is evident the there are two airstrips at the eastern end of the town itself (see: From the lower resolution satellite images it is evident the area has changed quite a bit and the airport station is probably the one reporting temperatures now.

    It’s a good story but it would be good to get confirmation.

  12. Pingback: Saturday silliness – the BEST adjustments of temperature | Watts Up With That?
  13. David Haden

    Thanks for that, Shub. Based on your new link and imagery it seems the town has: i) a basic dirt-field light aircraft landing strip on the eastern edge of the town, and ii) a larger military-grade pasture airstrip and helipad located out to the east of the town. I would imagine that even the small light aircraft strip has a basic weather station, as an aid to pilots. But would such data get collated and passed on… who knows?

    I’m should perhaps say that I’m just someone who’s stumbled in from a friend’s Facebook link. I noticed a certain lack of dogged online research here, and they encountered what seemed to be overly-flippant dismissals elsewhere. So I happened to get curious enough to contribute 30 minutes of research, to try to help straighten things out. As an independent observer — who’s also seen Kevin Cowtan’s YouTube rebuttal on the Paraguay data at — it seems to me like _both_ sides of this rather sloppy debate have got themselves in a tangle over the Paraguay stations and their data, due to a simple lack of investigation and knowledge of the facts ‘on the ground’.

    If the “GCOS Regional Action Plan for South America” document (September 2004) summarises the mid 2000s situation correctly as… “climate observing systems in many South American nations are in such a state of disrepair that reliable assessment, quantification, and prediction of climatic conditions and their impacts has been compromised” …then perhaps 25 years or more of South American weather station data is indeed shaky at best, and has been officially acknowledged as such. Yet Kevin Cowtan’s rebuttal video (link above) seems painfully unaware of this official history and its implications for the climate record.

    On the other hand, such widespread disrepair would mean that climate scientists are now warranted in trying to adjusting the data from South American nations. How they adjust seems to be the crux of your argument with them. Are they carefully adjusting on a station-by-station basis, based on an investigative database that logs local knowledge about station moves and oral history interviews about maintenance lapses, for all very-remote stations? It seems highly unlikely that is the case. Kevin Cowtan’s video (link above) appears to suggest the data adjustments are made on a ‘nearby proxy’ basis — pick a nearby ‘good’ station across the border in another nation and then use its data to adjust the ‘bad’ stations in a wide arc around that ‘good’ station. Is that good science? I have no idea, and I’m not qualified to answer. But it seems to me that a much tighter and more grounded case on South America needs to be made, and on both sides of the debate.

  14. Timo Soren

    Temp adjustments for a move are just silly. They are an areas temp. If I gave you 3 different stations simultaneously with exactly those measurements, they should use all 3 unadjusted. What needs to assumed by them is that it is 3 stations, each with a beginning and an end and that they are representative of the area. The insane idea that temps are equal in a limited area is just stupid.

  15. omanuel

    FEAR is probably the most powerful emotion and easily overwhelms logic.

    FEAR of nuclear annihilation in 1945 is the root of today’s social insanity induced by common knowledge politicians, world leaders, their puppet scientists and perhaps even members if the medical and banking community are lying.

  16. David Haden

    The precise dates and location of the La Victoria / Puerto Casado station’s circa-2005 move can probably be had from the Paraguay national meteorology directorate. The directorate was formed in July 1990 and sits within the DINAC (the Civil Aviation National Directorate). They have a website and even a short history online at: This online history neatly explains a break in this weather station’s data at 1970-71, thus…

    “[In] 1938, the Bureau of Meteorology was created by reorganizing meteorological services under a single Directorate under the Ministry of War and Navy, now the Ministry of Defence. … There were systematic records made of meteorological observations in the Department of Hydrography and Beacon of the Navy, from September 1928 under the Ministry of War and Navy and operated until 1970.” (text via Google Translate)

    So there’s the probable reason for a 1970-71 data break in the Puerto Casado graph — the Navy gave up responsibility for the weather stations in 1970. The takeover of the stations by the new national meteorology directorate in July 1990 may also explain some or the graph jerkiness at Puerto Casado in 1989-92, but that’s just my guess.

    I note the Dept. also has a variety of PDF Bulletin publications, freely available online, which appear to include climate data:

  17. Pingback: AndThenTheresPhysics on Paraguayan Temperature Data | ManicBeancounter
  18. manicbeancounter

    Thanks for the tweet on my ATTP post. It does not matter to me about censorship. More important is that proper understanding of actual temperature data is achieved. I have therefore posted a long comment to ATTP’s blog, cross-posted to my own post, challenging ATTP to counter Paul Homewood’s arguments; admit that he is wrong on this one; or take other courses of action. If I am wrong, ATTP and others will need to up their game quite a bit to properly answer Homewood. In so doing, collective understand of temperature data sets will be enhanced.

  19. Pingback: Time To Start Jailing Global Warming Proponents |
  20. Shub Niggurath

    That’s right. When I refer to censorship, I draw attention to its corrupting nature – how people use it to protect their wrong positions and push them further – when they know they are wrong. They know they may be wrong because they’ve read what they censored, even though they prevented everyone else from reading it, perpetrating the impression their position is unchallenged.

    People take censorship to be personally insulting and ego-dystonic and the aggrieved party to be complaining as a result, but that is not my concern. The effects of censorship are more insidious and run deeper – they affect understanding and the claims being made.

    I did not mention AndPhysics and Victor Venema in my post because though he has a thread with >500 comments on a post named ‘Puerto Casado’, there is no original contribution. It would have taken just one comment to point this out but the way this person operates is by getting rid of every single person who would do it.

  21. manicbeancounter

    Thanks Shub for your clarification. I have slightly different views, which I will develop further. However, I entirely agree the effects of censorship run deep. Long before I heard about climate change I realized that a major part of the development of all knowledge – science and non-science subjects – is through comparing and contrasting different ideas and the way those ideas fit to our emerging knowledge of the world. That is far from original as the Ancient Greeks developed the concept. Censorship stops or retards this process.

  22. daveburton

    David Haden, Dr. Cowtan’s video shows that adjustments increased global land surface temperature warming by 35%, when compared to the unadjusted data. Here’s the graph from 3:44 in Dr. Cowtan’s video:

    The unadjusted trend is in red, adjusted in pink. At the left end, for year 1900, the graph shows adjusted temperature is 0.26°C cooler than unadjusted temperature.

    At the right end (2013? 2014?) it shows adjusted temperature is 0.045°C warmer than unadjusted temperature.

    That’s a 0.30°C difference, which is just over 35% of the 0.86°C total warming seen in the unadjusted version.

    If you’re not convinced that’s accurate then blow up the graph and digitize it with something like WebPlotDigitizer (an excellent tool, which I heartily recommend, btw).

    Dr. Cowtan’s analysis doesn’t debunk Booker’s point, it proves it.

  23. Steven Mosher

    Still wrong.

    we do not compensate for the move.

    The metadata indicates there may have been a move.

    All that does is PREVENT us from asserting that the station IS ONE CONTINUOUS RECORD.

    In short, the meta data says we have no evidence to COMBINE segments.

    So the segments are treated as different stations.

    IF there is no difference between these segments, then the algorithm is a NOP

    This of course was tested.


    well you take stations and arbitrarily slice them… or segment them

    False segmenting doesnt do anything.

    false combining is the real problem.

    In any case you can remove all adjustments if you like and the global answer doesnt change.

    read that again.

    Adjustments only effect the local detail. Globally they have no scientifically interesting effect.

  24. Shub Niggurath

    ,…the global answer doesnt change…

    Your first problem is you don’t read.

    Where, in the post above, do I write about the global answer?


    On ATTP’s blog, you made further pretty statements.

    You said:

    Faced with this answer skeptics will of course look at the extreme cases of those histograms.
    With 40000 stations trust me we do have some extreme cases. And statistically I know that we probably have some of those cases wrong. That means that any skeptic out there can happen upon or look for a case that looks weird and then impugn the whole dataset or approach.

    When the whole thing started, Paul Homewood did not do what you say. You would have known this if you had read Booker’s story and followed the trail back to Paul’s blog. It was two clicks away.

    Paul’s question was: ‘there’s this huge blob of temperature sitting across, Peru, Argentina and eastern Brazil. Let’s leave the urban stations. Let’s pick the rural UHI unaffected stations and compare raw to adjusted.

    In other words, Paul did not go fishing for some ‘extreme’ case and find Puerto Casado. Puerto Casado is bang in the middle of a large area of warmth contributing to the global record.

    Now, what contribution it may have to the gridded average is a different question and issue. But, the station was not picked randomly.

    You didn’t know this.

    The question on people’s minds is: what is the impact of local stations and large data-sparse regions on the global average in a situation where government agencies are cutting it close and making such calls as ‘hottest year ever’?

    If the adjustments can be removed and the global answer doesn’t change – why do it?

    Your concerns as you participate in the temperature debate are different from mine. Yours are to uphold BEST’s methodology and issues surrounding calculation of global averages. Mine, at the present moment, is to precisely determine the station history of Puerto Casado. For you stations are numbers on a spreadsheet or flat-file. For me stations are real places on the earth with their own unique history and life.

    If I show, for example, that your local station grinder converted a negative trend into a positive one with no basis in local history, the fact that such changes had or have no effect on the global average, won’t matter. You would be in the wrong. Your method would be wrong.

    Any scientific method has to reconcile reality at all scales. No point saying the operation was a success but the patient died.

  25. Pingback: The Propaganda methods of ….and Then There’s Physics on Temperature Homogenisation | ManicBeancounter
  26. Pragmatist

    Interesting discussion. Larger point: many laypeople don’t trust the publicly-funded scientists to make reasonable assumptions when adjusting raw data. Furthermore, the raw data themselves from far-flung stations maintained by people of questionable expertise do not exactly inspire confidence. Finally, feeding tweaked data into fairly inaccurate climate models to project future warming requires yet another leap of faith on the part of laypeople. When we’re talking about a supposed warming trend of only .6º C over the last 60 years or so, it’s hard to fathom the hysterical tone coming out of the scientific community.

  27. Pingback: Is there a Homogenisation Bias in Paraguay’s Temperature Data? | ManicBeancounter
  28. Pingback: Temperature Homogenization at Puerto Casado | ManicBeancounter