Arguments
Tree-ring proxies and the divergence problem
What the science says...
The divergence problem is a physical phenomenon - tree growth has slowed or declined in the last few decades, mostly in high northern latitudes. The divergence problem is unprecedented, unique to the last few decades, indicating its cause may be anthropogenic. The cause is likely to be a combination of local and global factors such as warming-induced drought and global dimming. Tree-ring proxy reconstructions are reliable before 1960, tracking closely with the instrumental record and other independent proxies.
Climate Myth...
Tree-rings diverge from temperature after 1960
Actual reconstructions "diverge" from the instrumental series in the last part of 20th century. For instance, in the original hockey stick (ending 1980) the last 30-40 years of data points slightly downwards. In order to smooth those time series one needs to "pad" the series beyond the end time, and no matter what method one uses, this leads to a smoothed graph pointing downwards in the end whereas the smoothed instrumental series is pointing upwards — a divergence (Climate Audit).
Tree growth is sensitive to temperature. Consequently, tree-ring width and tree-ring density, both indicators of tree growth, serve as useful proxies for temperature. By measuring tree growth in ancient trees, scientists can reconstruct temperature records going back over 1000 years. Comparisons with direct temperature measurements back to 1880 show a high correlation with tree growth. However, in high latitude sites, the correlation breaks down after 1960. At this point, while temperatures rise, tree-ring width shows a falling trend. This divergence between temperature and tree growth is called, imaginatively, the divergence problem.
The divergence problem has been discussed in the peer reviewed literature since the mid 1990s when it was noticed that Alaskan trees were showing a weakened temperature signal in recent decades (Jacoby 1995). This work was broadened in 1998 using a network of over 300 tree-ring records across high northern latitudes (Briffa et al. 1998). From 1880 to 1960, there is a high correlation between the instrumental record and tree growth. Over this period, tree-rings are an accurate proxy for climate. However, the correlation drops sharply after 1960. At high latitudes, there has been a major, wide-scale change in tree-growth over the past few decades.
Figure 1: Twenty-year smoothed plots of tree-ring width (dashed line) and tree-ring density (thick solid line), averaged across a network of mid-northern latitude boreal forest sites and compared with equivalent-area averages of mean April to September temperature anomalies (thin solid line). (Briffa et al. 1998)
Has this phenomenon happened before? In other words, can we rely on tree-ring growth as a proxy for temperature? Briffa et al. 1998 shows that tree-ring width and density show close agreement with temperature back to 1880. To examine earlier periods, one study split a network of tree sites into northern and southern groups (Cook 2004). While the northern group showed significant divergence after the 1960s, the southern group was consistent with recent warming trends. This has been a general trend with the divergence problem - trees from high northern latitudes show divergence while low latitude trees show little to no divergence. The important result from Cook 2004 was that before the 1960s, the groups tracked each other reasonably well back to the Medieval Warm Period. Thus, the study suggests that the current divergence problem is unique over the past thousand years and is restricted to recent decades.
This suggests the decline in tree growth may have an anthropogenic cause. A thorough review of the many peer reviewed studies investigating possible contributing factors can be found in On the ’divergence problem’ in northern forests: A review of the tree-ring evidence and possible causes (D’Arrigo 2008). Some of the findings:
-
Various studies have noted the drop in Alaskan tree-growth coincides with warming-induced drought. By combining temperature and rainfall records, growth declines were found to be more common in the warmer, drier locations.
-
Studies in Japan and Bavaria suggest increasing sulfur dioxide emissions were responsible.
- As the divergence is widespread across high northern latitudes, Briffa et al. (1998) suggests there may be a large scale explanation, possibly related to air pollution effects. A later study by Briffa proposed that falling stratospheric ozone concentration is a possible cause of the divergence, since this observed ozone decline has been linked to an increased incidence of ultraviolet (UV-B) radiation at the ground (Briffa et al. 2004).
-
Connected to this is global dimming (a drop in solar radiation reaching the ground). The average amount of sunlight reaching the ground has declined by around 4 to 6% from 1961 to 1990.
-
One study suggests that microsite factors are an influence on whether individual trees are vulnerable to drought stress. Eg - the slope where the tree is located, the depth to permafrost and other localised factors (Wilmking & Singh 2008). This paper amusingly refers to the divergence problem as the "divergence effect" so as "to not convey any judgement by the wording" (you wouldn't want to offend those overly sensitive Alaskan trees).
There is evidence for both local and regional causes (e.g. drought stress) as well as global scale causes (e.g. global dimming). It's unlikely there's a single smoking gun to explain the divergence problem. More likely, it's a complex combination of various contributing factors, often unique to different regions and even individual trees.
One erroneous characterization is that scientists have been hiding the divergence problem. In fact, tree-ring divergence has been openly discussed in the peer-reviewed literarure since 1995. A perusal of the many peer reviewed papers (conveniently summarised in D’Arrigo et al. 2008) reveal the following:
-
The divergence problem is a physical phenomenon - tree growth has slowed or declined in the last few decades, mostly in high northern latitudes.
-
The divergence problem is unprecedented, unique to the last few decades, indicating its cause may be anthropogenic.
-
The cause is likely to be a combination of local and global factors such as warming-induced drought and global dimming.
-
Tree-ring proxy reconstructions are reliable before 1960, tracking closely with the instrumental record and other independent proxies.
Intermediate rebuttal written by John Cook
Update July 2015:
Here is a related lecture-video from Denial101x - Making Sense of Climate Science Denial
Additional videos from the MOOC
Interviews with various experts
Expert interview with Tim Osborne
Last updated on 29 October 2016 by pattimer. View Archives
To summarize, supporters of Michael Mann are annoyed at the dust up about him splicing direct NOAA data on the end of 960 years of proxy data when we know now that many climate scientists knew about this, called it "the divergence problem," and were OK with Mann not clearly labeling what he was doing for the world policy makers. The big story about 2009 is need for better email security and need to catch science hackers. Because the "data trick" to "hide the decline" got miscontrued by many due to bad non-scientists like Sarah Palin. All real scientists know that the Briffa data set was perfectly valid until 1960 when modern interferences caused it to be invalid from there on. And, we all know that warming must be occuring becuase atmosheric CO2 in known to cause rapid warming (not gradual or minimal) because we see in the ice core record that warming is always associated with higher CO2, even though the "lag problem." The CO2 increased always after the warming but this is an even more alarming observation because it means that warming can be triggered by a small amount of CO2 that then causes a run-away spiral of more and more CO2. And we don't know how the warming stopped, the CO2 declined and the oceans and land recovered. Conclusion: we need to take action to freeze industrialization or better, reverse it along with population but currently lack necessary political structure to do so.
[PS] Welcome to Skeptical Science. Please take time to review the comments policy. Comments that do not conform will be deleted. In particular, please note the "No sloganeering" and "No Accusations" section. You are very welcome to discuss the science and article content, supporting your arguments with appropriate data and references. However, it would pay to first read the articles.
That was thoughtful of you to protect your blog from my post. Surely most of the reader already know that CO2 has a shielding limit once the critical concentration is reached that completely blocks the IR the narrow band areas not overlapped by water vapor. But this post only makes sense if you allow my prior post back.
[PS] Offtopic. If you want to comment here, you must abide the comments policy. If you do not, then posting rights will be forfeited. The right place to comment (after you have read the science and can back your comment) is "CO2 effect is saturated". Use the search function on top left to find appropriate place. Repetition of long-debunked myths without new data is simply sloganneering.
A skeptic site claims that "almost all" proxy data diverges from the temperature record, and that therefore (1) the temperature record is wrong - too hot - and (2) the proxy record is misinterpreted, i.e. it was hotter in the past. They show this graph:
This particular graph doesn't seem to show even a single proxy matching the instrumental record.
SkS says that in fact some tree ring proxies still track the temperature record accurately, and I assume there are other proxies tracking accurately as well. I have some questions:
1. Can I get a graph of all the proxies "trusted by scientists" that zooms in on the period where an instrumental record exists (150 years) rather than the usual 1000+ years (assuming that's not too much to show on one chart)? (Btw - I assume some proxies apply only to certain latitudes, so maybe what is needed is a series of proxy-vs-instrumental-record charts, each for a different latitude range...)
2. Other than southern tree rings, which proxies are tracking accurately?
3. Are there proxies other than northern tree rings that are no longer tracking the instrumental record? If so, which ones? If they are no longer tracking now, then why would scientists have confidence in past readings?
4. The SkS argument list doesn't mention proxies at all, so you guys should add a general page about proxy myths. For instance, global warming diverges by latitude - and some proxies are only available at certain latitudes, which implies that in fact those proxies that "diverge" from the global temperature record are not necessarily diverging at all, but are doing exactly what scientists expect.
[PS] See this page for list of proxy datasets used for paleoclimatology. If deniers are pushing some myth around proxies, then we would address it but there are other sites eg here dealing with summarizing the complex problems of paleoclimatology.
[RH] Fixed image width that was breaking page format. Try to keep your images 500px wide or less.
Qwertie @78, the first figure you show appears to be a close up of this graph from the IPCC AR4:
You will note that the coloured lines represent reconstructions, not proxies. The difference is that a proxy is the data from a single core or tree ring record that is expected to covary with temperature. Tree ring records will typically be taken from multiple trees within a small region, including dead or fossil trees which are used to extend the record back in time beyond the lifespan of an individual tree. The reconstructions differ from each other because of different methodologies and/or data sets. In all cases they have a small number of proxies - typically in the tens to low hundreds. This contrasts with the thousands of thermometer records used in determining the Global Mean Surface Temperature. Because each thermometer/proxy only records. To get some idea of the impact of using a small number of proxies,
is a comparison of the land surface record (CRUTem3) with just 61 long record rural stations:
In addition to the limited number of proxies, reconstructions also face difficulties because the proxies do not follow temperature perfectly. High latitude or altitude tree rings are significantly impacted by temperature, but they are also effected by precipitation, cloud cover and no doubt other effects. Using multiple proxies will average out these effects to get a better temperature signal than from any single proxy, but again reconstructions will not be perfect as a result. There are further difficulties because not all proxies have records over the full period. In particular, records only extend to the period in which they where collected, often in the 1980s or earlier. Consequently reconstructions face a drop of accuracy in the final few decades of the reconstruction.
Finally, here is a reconstruction of GMST from 1880-2010 using 173 temperature sensitive proxies, compared to the NOAA NCDC Merged Land Ocean Surface Temperature record (MLOST):
As you can see the warming trend in the paleo record continues after 1980.
From the essay at the top, "Tree growth is sensitive to temperature." I've been a professional forester since Nixon was in the White House and I know that there are many variables that influence tree growth. I wouldn't place too much faith in the relationship between tree growth and temperature. Other factors include soil moisture, age of the tree, competition with other trees (is the forest dense or has it been thinned either naturally or by cutting?), pathologies which may be temporary like gypsy moth, air pollution and the level of atmospheric CO2. Some research shows trees growing faster with higher levels of CO2.
[PS] Funnily enough, the importance of other factors is well known and using tree rings as proxies requires careful selection. See here for example and more detail.
The relationship between temperature and tree rings seems quite complex. It doesn’t sound like anyone can properly explain the divergence problem. So how come climate scientists are so confident that past tree ring data is reliable?
I think the article and the comments in the discussion thread here sum up things pretty well:
Much ado about not much at all.
If tree rings are only reliable for part of the instrumental record wht are they considered very reliable prior to the instrumental record?
Tree rings are a useful tool, both for the early 20th century (to help ground-truth the overlap period with the instrumental temperature record and to therefore give them weight for before that overlap period) and earlier.
However, scientists have far more proxy records than just tree ring cores. So the arsenal that scientists rely upon has many tools beyond just tree rings.
Temperature measurements began in 1659. Stations were added throughout the centuries since then, becoming a truly global network beginning in 1850. Proxy records extend that record literally many hundreds of millions of years into the past.
Have you read the OP and worked your way through the linked articles and everything discussed in the comments section here? If not, why not?
I have read the article above and had a look at some of the papers. I’m just not sure why the pre 1880 tree ring data should be considered so reliable if nearly half the tree ring data after 1880 isn’t reliable and we don’t really know why.
as for the tree ring data being just one small piece, I thought it was quite important in producing the hockey stick? Without the trees is much more fuzzy isn’t it?
TomJanson @85,
The early reconstructions of pre-1880 temperatures were indeed dependent on tree ring data. Groveman & Landsberg (1979) was entirely tree ring data with a reconstruction back to AD1579 while Bradley & Jones (1993) did also employ ice cores in reconstructions back to AD1400, as did the Hockey Stick iteself (Mann et al 1998). But things have moved on a lot since then with many other proxy types giving confirmation that the tree ring reconstructions are providing useful data. The graphic below is from PAGES2k Consortium (2017).
The 'tree ring divergence problem' continues to be investigated but without resolution in sight. And those creating tree ring reconstructions are well aware of the issue. The tree ring reconstruction 'fit' to the instrument record is now a lot more impressive than that shown in the OP above. See for instance Fig 1b of St George & Esper (2019). (A rather tiny version of Fig 1 is below, 1a being the top graph.)
How come the trees chart above doesn't show the divergence?
Tom Janson @87,
If you look (& the paper itself affords the best view) you will see that there is divergence, just not as much and as early as in the OP's figure 1. To quote from the paper:-
The caption for Fig 1b runs:-
And be aware that divergence is not a universal phenomenon but although it is wide-spread enough to appear in any reconstructions that span large areas.