On the rural weather stations in Paraguay

It seems to me that both sides of the climate debate are becoming far too slapdash in their public claims, and also in their childish personalised dismissals of good contrary evidence. The recent little spat over the La Victoria weather station in central-north Paraguay seems a case in point. Both sides jumped to hasty conclusions and slipshod rebuttals without doing any historical research on the weather station in question. When a hour spent trawling the online archives and histories would have shown them they were both treading on very shaky ground, and for a number of different reasons. For instance, James Delingpole has recently wrote in the Spectator and Breitbart London

“Never has anyone offered any satisfactory explanation for these adjustments [to the Paraguay weather station data]. Perhaps it’s about time someone did.”

He was taking his lead from Christopher Booker’s column in the Telegraph, which itself rested on some perhaps rather over-hasty fuming by climate bloggers. The climate commentariat had a field day, filling the comment boxes with their hobby-horses, inanities and spite. None of them, the otherwise-excellent Delingpole included, bothered to send 60 minutes on what journalists and activists are supposed to do in such circumstances: in-depth historical research. I’ve done some of that research, and found it easy to knock down their claims that climate experts couldn’t know why or how or if the station had moved. Along the way I inadvertently uncovered evidence which also invalidated their opponents’ rebuttal.

First, here’s the core graph (BEST) for the weather station in question, La Victoria / Puerto Casada, a miserable little river port of 7,000 souls that sits in the jungle just below the northern part of Paraguay. Currently it’s lacking even running water, according to the news reports. But their weather station is important, because it’s said to be one of only three rural stations that have operated over the long-term in that nation.

The nay-sayers claimed that no-one could know when or how or why the station moved (marked by the red lozenges on the graph) in such a remote place. Some also claimed that nearby stations were mapped as being sited over water, so must be in error. But it’s not difficult to get much closer to the truth, with a bit of basic research.

La Victoria / Puerto Casada, Paraguay:

* 1970 move: The website of the Paraguay national meteorology directorate has an online history that neatly explains the break in this station’s data at 1970-71, when the Navy gave up the operation of the stations …

“[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)

* 1989-92: Turbulence at this point in the chart may be explained by the national transition to the new Paraguay national meteorology directorate which was formed in July 1990 and sits within the DINAC (the Civil Aviation National Directorate). Getting instrumentation correctly calibrated at a previously neglected station in such a remote spot may have involved long gaps and trial-and-error.

* 2005: A possible change of site, and/or a major overhaul of this station, would have been in response to the official 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.”

So the La Victoria / Puerto Casada station obviously had moved (probably out to the edge of the big pasture airstrip and helipad located east of the town, changed organisations several times, and likely had poor calibration and maintenance of instruments prior to 2005 if not later. And we now know that until 1970 some of the nation’s other stations would have been located on the Navy’s river ships — these were probably also moved to airports after 1970.

Facts such as these suggests the climate scientists were correct in feeling a need to adjust the historical data, from this and the other rural stations.

On the other hand, a video demonstration of an official adjustment method appears to show it being rather crudely done by reference to stations in nearby nations — while the official GCOS Regional Action Plan for South America implies these other stations were equally likely to have been “in such a state of disrepair”, before circa 2005, that they cannot be relied on for useful climate data. (Some of the video’s adjustment methodology has also been subject to a partial rebuttal).

It appears that neither side in the debate seems to want to undertake the sort of close historical research — which could go far deeper to include oral fieldwork, newspaper archive research on freak weather events, and interviews with the maintenance technicians and navy personnel — that would be needed to carefully adjust or discount the year-by-year historical readings from this contentious weather station.

To complicate matters the town and its riverine hinterland is effectively owned by the Moonies cult, although they at least appear to be keeping their vast stretches of jungle pristine. Elsewhere ranching, soy farming and deforestation have sparked a violent ‘eco warrior’ guerrilla movement that has operated for the last 10 years in the north of Paraguay and which is still active. In terms of the corrupting influences of big agriculture in the region, I note also that agricultural crop insurance is reported to have been introduced to Paraguay — a system which pays out based partly on the official weather record. Such crop insurance makes it imperative that weather stations and their operators are not ‘tampered with’ so as to favour insurance claims — yet Paraguay is one of the most corrupt nations on earth.

Further, I note a U.S. Army monograph Area Handbook for Paraguay (Foreign Area Studies of the American University, 1972), which identified the intense variability of the nation’s weather from year to year. At that time (the book was researched 1971, admittedly before the current deforestation) the authors would have been partly basing their information on the weather monitoring of the Paraguay Navy, which — as I’ve established — until 1970 operated all the nation’s weather stations. Presumably with a certain level of military precision. The U.S. Army Handbook usefully informs readers that…

“Because of the lack of pronounced topographic barriers within Paraguay, these opposite prevailing winds frequently meet with little modification to bring about abrupt and irregular changes in the usually moderate subtropical weather conditions that may also vary considerably from year to year. […] The number of days with temperatures falling below freezing ranges from as few as three to as many as sixteen yearly, and variations deep in the interior are even more pronounced. Some winters are very mild, with winds blowing constantly from the north and frosts being unusual. At other extremes, tongues of Antarctic air during the winter months bring subfreezing temperatures to all weather stations.” — Area Handbook for Paraguay, 1972.

So there’s very high natural variability. This may be relevant in terms of what we know about the adjustment of the early Paraguay weather station data. It appears to have been heavily adjusted down using a computer algorithm, forcing more recent readings up, then marked as “Quality Controlled Adjusted (QCA) TAVG”. This is what caused the initial complaints and allegations. The apparently hazily documented method of homogenising adjustment used, so far as I can discover, first compares candidate data points across… “three general categories: basic integrity, outlier, and spatial consistency” in order to automatically flag data as being suspect. Yet we’ve just read (above) that the U.S. Army stated Paraguay’s topography makes it unusually subject to a notable number of unexpected freak weather events and dramatic changes, year-on-year. Could these freak outliers in the data cause the climate database computers to flag some of Paraguay’s pre-1970 station data as ‘suspect’, even when in fact the data was from an accurate reading made by the Navy? It certainly appears that the Paraguay station data is not being adjusted per-station in response to a detailed history of each particular station, despite the importance of rural stations such as Puerto Casada as proxies for a vast area in the centre of South America. Nor does it seem to be being adjusted according to a close and detailed account of any extreme weather circumstances in which the station collected the data, as recorded in the local press archives. Some kind of blanket mathematical modelling is evidently being employed across stations, rather than asking pertinent historical questions such as: how efficiently run were Paraguay’s Navy weather stations from the 1920s to 1970? Did they get better or worse over that time? What do their Navy manuals suggest their station instrumentation and regimen was like? Did the mariners compensate the readings due to running a station on a metal ship, above water? Did they check and calibrate their on-deck and mast measurements against weather balloon readings? Or were the weather stations sited near the shore?

Finally, there is another worrying aspect of the graph for Puerto Casado, beyond the basic question of station moves. The graph’s run of temperature sequences appear to understand the station as being de facto three different stations, presumably due to the station moves. Yet the computer algorithm adjustment appears to have ignored this splitting, treating the station as if it had always had a single fixed location. To make matters worse, at the same time the algorithm was presumably trying to adjust toward a regional expected value, but that value can only have been based on nearby regional weather stations — stations that the 2004 GCOS Regional Action Plan for South America clearly stated were… “in such a state of disrepair that reliable assessment, quantification, and prediction of climatic conditions and their impacts has been compromised”.

All of the above findings do rather muddy the waters for those who want to assume that either “the weather station record is 100% useful, with a bit of slight adjustment” or “the weather station record is 100% falsified, because of gross adjustment”. As with most things, the only honest answers are “It depends…” and “it’s complex…”.


2 thoughts on “On the rural weather stations in Paraguay

  1. David Haden February 10, 2015 / 12:23 am

    Perhaps also relevant…

    South America Climate Data: Average Monthly Air Temperature Time Series — “Average monthly air temperature series from 1960 to 1990”. “The South America Air Climate (SAC) data set (372 snapshots) collects monthly-measures of air temperature at 6477 weather stations in South America.”

    Compare “6477 weather stations in South America” with…

    “… surface weather stations in South America add up to a few hundred stations, of which ca. 260 are included in the Global Climate Observing System (GCOS) under the World Meteorological Organization (WMO).” (Report on the PISAC Initiative, WMO)


Comments are closed.