dendroclimatogy

a briefing document

site map

On housing and making living systems ecological

Tectonics: tectonic plates - floating on the surface of a cauldron

Index
what is dendroclimatology?

anatomy of a tree cross-section

variables
hockey sticks and all that
estimating temperature via dendrochronology - recent review article
bibliography
end notes

what is dendroclimatology

This is the science that reconstructs historic aspects of climate using tree ring records, particularly for times before instrumental records.
Dendroclimatology is the study of tree rings to determine the relationship between tree rings and environmental variations, including climate.

Dendrochronology is different, not to be confused with dendroclimatology. Dendrochronology is the comparison of tree ring series back through history for dating purposes. It is a very reliable dating mechanism. A series is a number of tree rings on a trunk cross-section, perhaps fifty rings.

A cross-section from a maritime pine showing tree rings.
Each ring is made of a lighter part that is early growth, and a darker part of later growth.
Growth earlier in the growth season produces larger cells with thinner walls,
while later growth produces smaller cells with thicker walls.
[Note that maritime pines are fast-growing trees.]

Some available series go back two thousand years, and there are series that go back several thousand years, with some parts of the series floating. Floating means that there are gaps in the series. With floating series, radio-carbon dating is used to check the ages of the tree rings.

With continuous series, the dating is so accurate that it is used to calibrate the radio-carbon dating. An anchored bristle-cone pine series in south-west USA goes back eight thousand years. Fully anchored chronologies extending back ten thousand years exist for oak trees in Germany.

There is much work proceeding to try and line up dendrochronologic series with older series such as stalagmite, coral, sedimentary deposits, ice cores and so on.

Dendroclimatology is a far more difficult and unreliable science that attempts to estimate climate variables, such as temperature and rainfall, from differences in the width and density of year by year tree-ring variations.

anatomy of a tree cross-section

Cork oak trunk cross-section, with simplified labelling

When you cut through a tree trunk, as you see, most tree cross-sections are not symmetrical. At present, I can find little authoritative data on the causes of these asymmetries. A large number of growing conditions are suspected, and they may well be compound. Cores are taken from trees by drilling with an auger [5 millimetres diameter] but, of course, the people who are analysing the tree rings are not usually the people doing the sampling. A height of 1.3 metres above the ground is recommended. Without cutting down the tree there is no way to know with certainty the degree and directions of asymmetries.

Note the degree of asymmetry

Tree ring analysis master lists are also developed from archeological samples. These samples are often in poor condition, dragged out of swamps, or difficult to access, as in the roof of a cathedral. In such circumstances, the wood may be waterlogged and the sapwood will be friable. This makes the use of an auger often impracticable. More on this subject can be read in Timber: dendrochronology of roof timbers in Lincoln Cathedral.

Also you can see in the pictures above, the thickness of the rings varies considerably according to the symmetries. The ring thickness also rambles randomly round the circumference of the trunk. These various factors should introduce some degree of caution in trusting these measurements. The best that can be hoped for is that these problems can be somewhat mitigated by averaging out processes. It is common to strive for a precision of one one-hundredth of a millimetre; to my judgment, this suggests ‘delusions of accuracy’ [a belief that the measurements are more precise than reality allows].

To give you some idea of how difficult is the measuring and matching in the real world, here is a quote from Timber: Dating of the Roof Timbers at Lincoln Cathedral, p.7.

“...sequences with less than 50 rings are, normally, not even considered for dating. The dating results from the roofs of Lincoln Cathedral are a good illustration. About 75% of the timbers with 50 to 65 rings failed to date, 45% with 70 to 80 rings failed to date, 25% with 85 to 170 rings failed to date while of those with more than 170 rings none failed to date. Fortunately, most of the measured samples had more than 80 rings.”

Illustration of heartwood-sapwood transition and knotting.
Note also the dark blue staining from fungus.

For some further background, the photograph just above makes clear some other details of a cross-section through a tree trunk. This trunk has been left in the open for a while after felling, and you can see very clearly the distinction between the heartwood (‘dead wood’) and the sapwood. In the sapwood, the capillaries are still open and, thus, water can ingress. You can see the signs of damp and mould (the bluish darkening). Naturally, the sapwood is more vulnerable to deterioration from rotting and attack, especially over long time periods.

You will also see the remains of old branches, commonly known as knots - the pointed ovals or boat shapes. In the management of forestry, knots are mostly considered as a flaw in the wood (artists and other woodworkers may sometimes take a different view), and so knotted wood fetches lower prices in the market. Hence, lower branches are usually sheared off as the tree grows, in order to improve wood quality and price.

Thus, in the photo above, you may note that some of the knots finish within the heartwood and do not extend to the surface of the trunk. The most important branches, from the point of view of the tree, are those that access the light. Because of this, many trees shed lower branches as they fight for height in the canopy in dense forest and plantations.

When trees have limited light availability because they are planted close together, the trees tend to grow straight towards towards the strongest light source - the sky. Industrial foresters take advantage of this behaviour to ensure straight tree trunks, and so optimal plank production.

advertising disclaimer

variables

I was once amazed when I asked a very capable botanist what a particular flower was. It looked to me like a typical garden variety. I was told it was a common meadow flower (weed!?). I said it was nothing like the flowers I’d seen under that name - the one in the garden was at least twice as high and the flowers at least two or three times the size

I was informed that the difference was solely down to better soil and fertilisers and attention.

Soil varies over a few feet, the flower had been taken from a nearby flood plain field. Trees a few metres apart can be in very different conditions, let alone on different continents.

Two main variables are currently under examination as temperature proxies/surrogates:

• ring width, and
• the density of late growth (examined by x-ray).
  [see first illustration above.]

However, a great number of variables effect the growth of trees: microclimates, soil conditions, shelter, root depth, crowding, but above all rainfall and temperature. Light is also a factor - photosynthesis is not possible without it. When trying to obtain useful samples as temperature proxies, trees at the edge of their range are sought out, because that is the sort of area where the trees are most likely to be stressed. Even then, the proxies are not very reliable for any particular year, even conditions in previous years can have effects of later years of growth.

Sometimes there are false rings, and sometimes the year’s rings may be close to invisible, especially in individual tree samples. Thus, it is usual to take many samples to establish averages for a particular growing area.

In order to try to discern whether the growing conditions have changed over long periods, say hundreds of years, sample averages and variations from groups of years are compared one with another to check consistencies. One even wonders whether the increasing CO₂ in the atmosphere in modern times would confound any changes that you may wish to attribute to temperature!

During the life of a tree, in the first year or seven, the rings grow more quickly and are, therefore, wider. So ring sequences are standardised, according to the age of the ring, to a notional width for say first-year or fifth-year growth. As different tree species have different growing patterns and other characteristics, for dendroclimatology single species are used. However, it is not unusual to see series from different species used in an attempt to establish regional conditions.

Notice that there is a lot of averaging and adjusting going on to set up a ring or density series, at which point you are no longer dealing with raw data or individual trees, but with synthesised data.

This becomes more dubious. We only have thermometer temperatures going back one or two hundred years. Many of those temperature series are far from reliable, and are very often nowhere near the most useful trees that may reflect the variable such as temperature that you are attempting to model. And remember each sample represents only a small, local area, so attempts are made to combine averages from many areas. To make things still more difficult, the critical temperature differences usually lie in only one, or a small number of summer months. Thus the proxy temperatures are also similarly limited. The degree of difference between seasonal temperatures in different areas is, therefore, a factor.

Even when all this done, correlation between the measurement, such as temperature, and the tree measurement may correlate reasonably well, but are still well away from 100% correlation.

Note the black lines towards the end of the tree ring surveys, which represent real temperatures.
This represents attempts to examine the viability of tree rings as surrogates/proxies for measured temperatures.

Thus you will hope for too much, if you expect to use tree ring data for more than having an idea of temperature trends over a period of time. Do not expect to be able to pin down accurate temperatures in any particular year. Remember that even with the trends, there are error factors that should be shown, as can be seen in examples given on understanding graphs and charts. [2]

hockey sticks and all that

Ring-width chronologies from Briffa (2000) Image: Jansen et al. (2007) Paleoclimate

Note the wild variations of the plots.
Note the 20-year moving average applied to the various series. (Often 30- and 40-year trends are used).
Note that each series tends to include only a small number of trees.
Remember that these are measurements of tree-ring characteristics. They are not measures of temperature. The assumption is that these characteristics are related to temperature changes which are presumed to change growth rates.

It is from data of this type that the recent temperature advance is calculated, and this trend is calculated against a long-term downward trend.

Note the dark black line which shows you the combined average of all the original series. It is from summary plots of this sort of data that squeals about ‘the hockey stick’ come from the denialist industry ☺. As you will see, the data does suggest a recent (industrial age) advance, as exhibited by the recent turn-up in the first graph.

The supposed hockey-stick graph [first one in this section] comes from a paper by Briffa and one other in response to criticism by McIntyre. Here is an earlier summary of similar work.

As for ‘the hockey stick’, that depends on just how you select the scales for your graphs. I have seen suggestions that only a 95% significance level is ‘scientifically acceptable’. This is nonsense. The interpretation of statistical tests of ‘significance’ are always a matter of judgment.

Consider your best beloved is apparently dying of some dread disease. The best medicine can do is to guess that the disease is one of two possibilities. A treatment is known for both diseases, but you know that disease A is four times as common/likely as disease B. You also know that if you give the wrong treatment, the person will die, and if you do not treat, they will die within the year. In this situation, you have only an 80% chance of being correct if you apply treatment for A.

80% becomes highly significant in making your choices. In fact, well-presented work should not use the word ‘significant’ loosely, but should use phrases such as ‘statistically significant at the 95% level, or the 80% level’, or whatever. And very likely, the presenter should give other details such as the statistical test being used, and the sample size.

It is only in context that you can decide whether a trend is ‘significant’ or not. It is only in context that you can decide whether to do anything about it. What you decide to do depends entirely on your judgement of costs and benefits. There isn’t a magical formula. Each person must make judgments according to their own values.

Each person must decide whether a statistical pattern reflects real world phenomenon, or whether it is just a statistical/numerical artefact.

The general consensus is that there is global warming, and that it is mainly human caused. The simple reason for this consensus is because the data convinces most people who bother to dig deeply enough, that anthropogenic global warming is far the most likely explanation of the data.

Whether it is a correct interpretation does not depend on whether you want it to be so, or on whether you have interests in filthy fossil fuels, or on whether your tenure at university depends on studying climate science, or whether you just want to keep on running that 4x4 and don’t want your fun curtailed. Old men often prefer to ignore the mess they are leaving for the next generation. Witness the huge build-up of debt by irresponsible politicians. The young are likely to be less sanguine.

estimating temperature via dendrochronology - recent review article

The tree ring data is ‘wrung out’ of the original raw data by typical/various statistical techniques in which I do not place great trust.

“...To verify results, tree-ring chronologies were compared to the instrumental climatic data averaged over a large sector of the sub-arctic. Finally, the longer-term tree-ring temperature reconstructions were compared to other proxy data including long-term variations in solar radiation[19], long-term variations in volcanic activity derived from ice-core measurements[20], and variations in CO2 concentrations in air trapped in the GISP2 ice core (central Greenland)[21].”
—
“ ...The range in estimates obtained from different sources of the degree to which reconstructed temperature at its post-glacial maximum exceeded the maximum warmth of the 20th or early 21st centuries is significant: from 0.6°C (glacier and stalagmite layers and bottom deposits) to 3 to 3.5°C...”
—
“Millennium length tree-ring chronologies were constructed from samples at two sites close to the northern tree line: east Taymir (71° 00’ N, 102° 00’ E) and northeast Yakutia (69° 24’ N, 148° 25’ E). These stands are made up of Gmelin larch (Larix gmelinii) and Cajander larch (L. cajanderi) in which the oldest living trees are up to 1,100 years old[26]. Well-preserved dead tree trunks allowed extension of the record farther back than the maximum age of the living samples, allowing the construction of absolutely dated tree-ring chronologies for the periods from 431 BC to AD 1999 (Taymir) and from 359 BC to AD 1998 (Yakutia). The RCS approach was applied to distinguish variation caused by climatic change. These representative tree samples were highly responsive to recorded temperature, allowing the reconstruction of temperatures over the last two millennia with annual resolution. With this long-term record, the instrumental record of 20th-century temperatures can be compared with temperatures during the MWP. Earlier results[27] showed that tree-ring chronologies were highly correlated across distances up to 200 km (up to 500 km in northern regions), so these millennial chronologies represent temperature variations over a large sector of the Siberian sub-arctic.”
—
“The relationship of climate to tree-ring variability in Scots pine in Scandinavia has been studied in a wide range of growth environments. A comprehensive study was made of pine growing on peat lands along a north–south profile through Sweden to see if the trees contained high-resolution climate information[66]. Peat land pines were also compared to pines growing on dry sites. Pines growing on peat lands are dependent on growing-season temperature and precipitation, as well as on local water-table variations, which are influenced by longer-term trends in both temperature and precipitation. There is a lag of up to several decades in the response of the pines to water levels, such that trees are integrating the immediate effects of growing-season climate as well as a delayed effect from the water table, making them unsuitable for high-frequency climate reconstruction. The sensitivity of pines growing on peat land also changes depending on climate. When the growing season is wet and cold, temperature is more important and trees respond positively to temperature, in particular to July temperature. Precipitation response increases to the south but is never as important as for pines growing on dry soils. Precipitation is important mainly in controlling water-table levels[67].”

related material
tree growth and agw
the present stage of global temperature measurement
on tree-ring data and peer-reviewed studies

For more problems with statistical analysis, see ‘intelligence’: misuse and abuse of statistics.

bibliography

	English Heritage Research Transactions Vol 7: Timber: Dating of the Roof Timbers at Lincoln Cathedral by Jeanne Marie Teutonico, R.R. Laxton, C.D. Litton, and R.E. Howard £33.25 [amazon.co.uk] {advert} amazon.com [amazon.com] {advert} James & James (Science Publishers) Ltd; Volume 7, 2001, pbk ISBN-10: 1902916034 ISBN-13: 978-1902916033 £33.25 [amazon.co.uk] {advert} amazon.com [amazon.com] {advert} James & James (Science Publishers) Ltd; Volume 7, 2001, pbk ISBN-10: 1902916034 ISBN-13: 978-1902916033	A short, 80-page introduction for enthusiasts.
	Paleoclimatology: reconstructing Climates of the Quaternary by Raymond S. Bradley £60.60 [amazon.co.uk] {advert} $64.60 [amazon.com] {advert} Academic Press,1999, hbk ISBN-10: 012124010X ISBN-13: 978-0121240103	A useful first-year student’s summary and notes on the subject. There is a longer review at Anthropogenic global warming.

end notes

1. These proxies are also used for guessing at precipitation. Yet another dubiety arises: a major driver of plant growth is light, not temperature. Thus, you must assume correlations between heat and available light.
   
   
2. As you will see from this recent review article, it is rare for the variance to rise above 50%, some of the samples are not much larger than 30, the minimum level recommended for any sort of parametric correlation measurement. Note: if r is the correlation, then r² is deemed as a measure of the proportion of variance explained.
   
   
3. There are very few people in the life sciences who have an adequate grasp of using statistics. People doing this sort of work would be well advised to consult with a statistician before publishing. It is poor statistical work that has led to recent [this note: late 2009] hysteria, such as Steve McIntyre’s criticism of sloppy work by Mann on dendroclimatogy [See Hockey sticks, principal components and spurious significance by Stephen McIntyre and Ross McKitrick.] and the fuss over “hide the decline” referring to the perfectly reasonable absence of Briffa’s most recent dendroclimatogy estimates in favour of the obviously far more useful empiric thermometer readings!
   
   
4. In layman’s terms, this amounts to listing the components of the data that can be accounted for by different phenomena, whether seasonal rainfall, soil fertility or pollution. Then what remains unaccounted can be surmised to be caused by, say, increased climate temperature.

email email_abelard [at] abelard.org

© abelard, 2009,20 december

all rights reserved

the address for this document is https://www.abelard.org/briefings/dendroclimatogy.php

720 words
prints as 3 A4 pages (on my printer and set-up)