Lomborg-errors

Maintaining that a flawed data series on global forest area is `the best available´  
      General issues                                                                                   How Lomborg cheats                                                                  Lomborg´s methods

 

    Many environmentalists are very concerned about worldwide loss and degradation of forests, especially tropical rainforests. As correctly stated by the WWF, "the area and quality of the world´s forests have continued to decline at a rapid rate" (cited by Lomborg in The Skeptical Environmentalist p. 110). Lomborg wants to oppose such statements by claiming that actually the total forest area is constant or slightly increasing. Statistics on global forest cover is given by FAO (the UN Food and Agriculture Organization). There are many difficulties in making such global estimates. One such difficulty is ambiguities regarding what actually constitutes a forest. `Closed forest´ is defined by FAO as areas where tree canopies cover at least 20 % of the ground (in developed countries) or 10 % of the ground (in developing countries).
`Forests and woodland´ includes areas where tree canopies cover as little as 5 % of the ground, and also includes areas with no trees, but with scrub, for instance bush steppe, as long as the bushes cover at least 10 % of the ground.
   FAO has data from 1948 onwards. The oldest figures are not very reliable, as governments in various countries have not always stuck carefully to the above definitions, so many areas which were at first not considered to be forests or woodland have later been found to fulfil the above criteria. Therefore previous forest estimates have sometimes been adjusted upward. For instance in 1959 there was a large upward adjustment of the forest area of Canada. Fairly reliable and uniform data series, partially based on satellite data, have been produced only from 1980 onwards. In addition, older estimates as far back as 1961 have later been revised. If we calculate on the basis of these revised figures for `Forests and woodlands´, we find that the global coverage has declined by about 0.2 % per year since 1961. This agrees very well with the trend shown by the more reliable data from 1980 onwards.
   However, Lomborg wants to say that everything is fine and that the global forest coverage is not shrinking. How he manages to do so, is explained here on Lomborg-errors (
example 2). He chooses a FAO data series that goes from 1950 to 1994, using for each year the first published, unrevised figure. This data series shows a slightly rising trend for the total forest coverage. This is partially because forest areas are sometimes revised upwards like for Canada in 1959, and partially because not all nations were included from start. For instance, the forest area of Papua New Guinea was only included from 1961 onwards. This gives a sudden jump upwards in the global sum, a jump which is of course purely an artifact. Although Lomborg knows this, he pretends that the upward adjustments are not artifacts and reflect real increases in forest coverage. This, of course, is unequivocally wrong - it is deliberate misuse of the data. He does that even though every volume of the FAO yearbook explicitly states that "It should be borne in mind that defintions used by reporting countries vary considerably and items classified under the same category often relate to greatly differing kinds of land. . . Thus the area specified is not intended to refer to or delineate "forest coverage."
   Lomborg first presented his provocative claim that the total forest area is slightly increasing in an article in a Danish newspaper in January 1998. He was at once heavily criticised for this misuse of the data in discussions on the internet, and that criticism has continued ever since. The criticism was repeated in the book written against Lomborg in Denmark in 1999 (`Fremtidens Pris´), it was repeated in the criticism in Scientific American in January 2002, and it was repeated in the complaint lodged to the Danish Committees for Scientitific Dishonesty in February 2002.
   How has Lomborg tackled this criticism? How does he defend the use of a data series full of artifacts, a data series which is printed by FAO with the warning that it precisely cannot be used for the purpose tha Lomborg uses it for?
   First, there is the trick used by any magician: move attantion away from the point where the trick is made. So when people debate with Lomborg, they discuss whether it is relevant to use a data series for `forest and woodland´ which includes even bush steppe, maquis and savanna, in addition to plantations. Lomborg evades this debate by simply stating that he thinks that `forest and woodland´ is what should be understood by forest - it ends up in a discussion of what may properly be designated `forest´, and that discussion leads nowhere. Furthermore, people say that although it may be true that the total forest area is roughly constant, the crucial thing is the disappearance of tropical rainforest - the expansion of Siberian coniferous forests is of no help to the lemurs of Madagascar or the rhinoceroses of Vietnam. If the debate goes in that direction, then people discuss as if Lomborg´s postulate about the global forest area were correct, and Lomborg wins.
   Second, Lomborg does all he can to postulate that the other FAO data series are unreliable. He mentions examples where global estimates for one year have been adjusted a few years later. He postulates that these adjustments are so large that in a short time series they are larger than the total real change. This problem can only be avoided, he claims, by using the longest possible time series where such adjustments become small relative to the total trend. This claim is obviously wrong - the long unrevised time series from 1950 to 1994 is objectively speaking completely unfit, when large obvious artifacts are not corrected for. When the forest area of Papua New Guinea is included in the total only from 1961 onwards, the resulting upwards jump can never be counted as if it were a real forest increase. But Lomborg does exactly that. He manages by drowning the debate in a lot of details which very few people can keep track of. He even cheats an extra time by making his own personal adjustment of the forest area of Russia when there is a drop that he does not like.
   Third, Lomborg does all he can to postulate that his preferred data set is the best because it has the longest time series. He writes  in `The Skeptical Environmentalist´ - and repeats everywhere - that `Data availability is poor but by far the best available´. It does not matter to him that this is an outright lie, which may be revealed by his critics. All he needs is to have an argument to justify his choice of data set. Whenever he is criticised, he stubbornly sticks to that argument in such a way that an outsider cannot know whether Lomborg or his opponents are right. The clue here is that revealing the lie requires that you go much into details with the data sources, and the general public has no patience to listen to such details. Therefore, even though Lomborg´s claims are outright lies, he relies on that they will never be effectively exploded in the eyes of the public. There will be some experts who know that his claims are lies, but the public does not know if they can trust these experts. Which leads to the fourth point:
   The fourth point is to make the public distrust his critics. When Lomborg and an opponent from the Worldwatch Institute were interviewed in 2001 by The New York Times (see
here on Lomborg-errors), the article says that the Worldwatch Institute in a report from 1998 said that the amount of forest is declining, after which the text continues: "But the longest data series of annual figures available from the United Nations´ Food and Agriculture Organization shows that global forest cover has in fact increased." So Lomborg has managed to persuade the journalist that his version is correct, because of the long data series. Next, the point of view of his opponent is presented as follows in the article: "Janet Abramovitz, Worldwatch's forest expert, said the world forest cover had shrunk significantly in the last 20 years. She based that contention on a different, shorter series of Food and Agriculture Organization statistics but declined to cite a percentage." So in the eyes of the public, Lomborg´s contention wins.
  
When Lomborg was attacked by a series of articles in Scientific American in January 2002, he had The Economist to write a leader to his defense, saying "Mr Lomborg defends these positions on the basis of official data and published science. Environmentalists typically use the same sources, but, as Mr Lomborg lays bare, are much less scrupulous about setting short runs of data in their long-term context . . . ". So Lomborg even manages to call his opponents unscrupulous.  To a scientist, this is a bogus argument  - there is no a priori reason why the longest data series should absolutely be the best, at least not when that data series is objectively known to be an artifact. But the lay reader assumes that since Lomborg speaks about such a principle about long data series, as if everybody is aware of that principle, it must be an important rule in science to use long data series. When Lomborg even says that the others who use short data series are less scrupulous, the lay reader will even assume that breaking this principle is something which honest scientists should not do.
   So Lomborg does the impossible - he successfully defends the use of a data series which cannot possibly be defended.