The IPCC and most climate scientists are obsessed with Linear Trends. They are encouraged by the fact that the only regularly available statistical treatment of irregular information on “scientific” calculators and computer spreadsheets is a “linear regression” calculation, using the method of least squares. Most people do not appreciate that its results are unreliable unless the original data are from a representative sample, uniform in time and place, and they approximately fit the Gaussian “bell” curve in every way. including by being symmetrical
Yet anybody who has attended any lecture on mathematical statistics and even many popular introductions to the subject must know that there are several mathematical models that may be more successful in studying irregular data. For example, extreme observations may follow the binomial distribution, which was traditionally successful in explaining the frequency of deaths from horse kicks in Prussian army corps. There are some people who are prepared to support the exponential distribution, which predicts escalating change extending to infinity, an impossible outcome for any climate trend except for those projected by the most enthusiastic environmentalists.
There are many reasons why linear trends are not a useful means of studying climate events. Samples are usually not representative. Observations usually take place at different times, using different methods or instruments and often in different places. The distribution curve of observations is often not symmetrical. The calculated “linear trend” may be quite different, depending on the starting and finishing point of the sequence. Ignorance or deliberate disregard of these necessities means that many opinions on the climate which have disregarded them are unsound.
One important consideration which seems to be ignored by most climate scientists is the treatment of irregular changes to an otherwise fairly steady sequence from rare or unforeseeable events.
Some years ago I received an Email from John Christy on the subject of his MSU satellite temperature series to the effect that it has been dominated by two sets of essentially irregular events. These were the volcanic eruptions by El Chichon in 1982 and Pinatubo in 1991 which caused a cooling, and the various manifestations of the El Nino weather pattern which provided unusually high temperatures, particularly in 1998. He commented that if the two earthquakes had taken place later and the 1998 El Nino had been earlier, then the temperature trend would have been negative instead of positive.
Phil Jones at CRU actually took this idea up some years ago (1990s) when he published an extra set of temperature anomaly figures which had been “corrected” for the influence of El Nino. It was withdrawn so they could recruit the 1998 El Nino upwards blip to form part of a “linear trend” which demonstrated the effects of carbon dioxide emissions.
One strange fact is that the calculated projected results of increases in carbon dioxide follow a decreasing, logarithmic path, not a linear one.
Although the Mean Global Surface Temperature Anomaly Record is supposed to be “corrected” for such problems as site change, instrument change, time of observation bias, and for “gaps” in the record, nobody seems to consider that perhaps a more plausible “trend”:might be found by applying “corrections” for changes in the number of stations, and such irregular events as volcanic eruptions and the various ocean oscillations. It may be easily possible to correct for El Nino which disrupts the sequence for a fairly short time but more difficulty with the Pacific Decadal Oscillation which appears to have a periodicity of about 60 years. Then there is the sun, whose influence is currently not well understood, yet undoubtedly underestimated by the IPCC.
The “correction” of the surface record for urban and land change effects now seem to be admitted by CRU from Phil Jones but they continue to issue the biased figures.. It seems increasingly likely that if realistic corrections could be implemented for all the uncertainties, natural events, and human effects on the ground, there will not be much evidence left to support claims for effects from greenhouse gas emissions.
The most blatant example of the use of an unusual event to claim an otherwise non-existent “linear trend” is in the reports of the Pacific Island Sea Level and Climate Monitoring Project of the Australian Government.. The “Linear Trends” that they report for the 12 Pacific Islands, for instance at their latest Report here
depend on the recorded depression of the ocean in all of the islands that took place during the two Tropical Cyclones of 1991 and 1992. Without these two events there are no significant recorded changes in sea level at any of the 12 Pacific Islands since then. They must be praying hard that a similar cyclone does not turn up to ruin their precious “trends”
It is even possible that this recorded depression was an artifact caused by the disturbances in the sea in the vicinity of the instruments during the storms
All other sea level records have been affected by storms of one sort and another. by seismic actability, by building and removal of minerals and ground water, besides the usual geological correction for isostasy. As with the temperature, one wonders how much would be left for the effects of greenhouse gases if all of these were done realistically.