I’ve moved to WordPress. This post can now be found at A Look At The Individual Proxies Used In Kaufman et al (2009)
##################On its publication in September 2009, Kaufman et al (2009) “Recent Warming Reverses Long-Term Arctic Cooling” made its rounds in blogs.
The abstract reads, “The temperature history of the first millennium C.E. is sparsely documented, especially in the Arctic. We present a synthesis of decadally resolved proxy temperature records from poleward of 60°N covering the past 2000 years, which indicates that a pervasive cooling in progress 2000 years ago continued through the Middle Ages and into the Little Ice Age. A 2000-year transient climate simulation with the Community Climate System Model shows the same temperature sensitivity to changes in insolation as does our proxy reconstruction, supporting the inference that this long-term trend was caused by the steady orbitally driven reduction in summer insolation. The cooling trend was reversed during the 20th century, with four of the five warmest decades of our 2000-year-long reconstruction occurring between 1950 and 2000.”
A BBC News article “Arctic 'warmest in 2,000 years'”…
http://news.bbc.co.uk/2/hi/science/nature/8236797.stm
…included the following graph:
http://i45.tinypic.com/2dcahd3.gifKaufman et al Graph In BBC Article
KAUFMAN et al PROXY DATA IS AVAILABLE FROM THE NOAA PALEOCLIMATOLOGY PROGRAM
NOAA Paleoclimatology Program includes a webpage for Kaufman et al (2009):
http://lwf.ncdc.noaa.gov/paleo/pubs/kaufman2009/kaufman2009.html
Toward the bottom of the page is a link to a spreadsheet that includes the standardized data of the 23 individual proxies included in Kaufman et al (2009):
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/reconstructions/arctic/kaufman2009arctic.xls
The last column (Column Y) in the spreadsheet is the average of the standardized proxies, which I’ve plotted below. Though the scales are obviously different, the curves of the BBC graph and the Average of the Standardized Proxies are similar in shape, a few minor differences, but they are similar.
http://i50.tinypic.com/2ptdpuv.pngAverage Of Kaufman et al Standardized Proxies
The differences may reflect the October 2009 corrections made to 7 of the 23 proxies by Kaufman et al (noted in the spreadsheet). They included:
"October 2009 Update
"Record 2 was revised using the correct time scale of McKay et al. (S16)
"Record 12 was revised to omit the high-pass filter used by Andersen et al. (S25)
"Record 20 was corrected to reflect the interpretation of Tijander et al. (S32) that X-ray density is related inversely to temperature
"Record 21 was corrected to reflect the interpretation of Haltia-Hovi et al. (S33) that varve thickness is related inversely to temperature
"Records 3 and 10 were revised to correct rounding errors
"Record 19 was truncated at 1799 to exclude the 1805 bin, which was based on only one year (1800)"
In some respects, a spaghetti graph of the 23 proxies shows the same basic curve, but at least one of the proxies had higher values 2000 years ago and others appear to be dropping, not rising, in recent years.
http://i49.tinypic.com/4h9u6x.pngSpaghetti Graph Of Kaufman et al Standardized Proxies
That prompted me to plot the individual proxies to see what they revealed.
GRAPHS OF INDIVIDUAL PROXIES
The following are graphs of the 23 proxies used in Kaufman et al. There is a classic hockey stick, Proxy 22, which is a graph of the Briffa et al Yamal tree ring data. There’s also a reverse hockey stick, Proxy 12, which is the Andersen et al DYE3 South Greenland Ice Isotope data, where the upward blade is in the first few decades; then the curve remains relatively flat for the remainder of the term. Some of the datasets illustrate warmer temperatures in recent decades, Proxies 9 and 23. There are others that are relatively flat but there are many others that show warmer temperatures in centuries past. The graphs of the individual proxies follow, without commentary. The titles include the location, proxy type, author(s), and years covered.
http://i46.tinypic.com/120m4pe.pngProxy 1
#########################
http://i49.tinypic.com/o7vs51.pngProxy 2
#########################
http://i47.tinypic.com/1h6n9c.pngProxy 3
#########################
http://i50.tinypic.com/jtthtc.pngProxy 4
#########################
http://i47.tinypic.com/jpeura.pngProxy 5
#########################
http://i46.tinypic.com/35nbr85.pngProxy 6
#########################
http://i48.tinypic.com/aff40k.pngProxy 7
#########################
http://i48.tinypic.com/1pxb4p.pngProxy 8
#########################
http://i48.tinypic.com/2nki4g7.pngProxy 9
#########################
http://i48.tinypic.com/21kz8ye.pngProxy 10
#########################
http://i45.tinypic.com/2usfjp4.pngProxy 11
#########################
http://i48.tinypic.com/mx259k.pngProxy 12
#########################
http://i45.tinypic.com/2d77eih.pngProxy 13
#########################
http://i50.tinypic.com/34gn888.pngProxy 14
#########################
http://i50.tinypic.com/2qlgylx.pngProxy 15
#########################
http://i48.tinypic.com/eia1hj.pngProxy 16
#########################
http://i50.tinypic.com/xc3ipg.pngProxy 17
#########################
http://i48.tinypic.com/2mhv2xg.pngProxy 18
#########################
http://i47.tinypic.com/21kw4r8.pngProxy 19
#########################
http://i49.tinypic.com/24yn7nt.pngProxy 20
#########################
http://i45.tinypic.com/2eqevb8.pngProxy 21
#########################
http://i50.tinypic.com/2daej3b.pngProxy 22
#########################
http://i45.tinypic.com/2cnvbcy.pngProxy 23
CLOSING
Viewed individually, the proxies do not necessarily illustrate “that a pervasive cooling in progress 2000 years ago continued through the Middle Ages and into the Little Ice Age,” or that the “cooling trend was reversed during the 20th century, with four of the five warmest decades of our 2000-year-long reconstruction occurring between 1950 and 2000.” Many of the proxies illustrate past temperatures rivaling or exceeding current temperatures and indicate long-term variability, not a long-term decline or pervasive cooling.
UPDATE 12-17-09: I modified the closing sentence with the addition of ...and indicate long-term variability, not a long-term decline or pervasive cooling.
9 comments:
Bob,
Very informative post, thanks for the link at Climate Audit. The standardized graphical views of the Kaufman figure, the summed proxies, and the individual records is helpful.
Re: the Tiljander (Lake Korttajarvi, Finland) varve proxy (and certain other lakebed sediment proxies), Kaufman's correction truncates the series at 1800, prior to the start of the instrumented record. This was the right thing to do, as Tiljander noted that the climate signal is accompanied and then surpassed by (spurious) human-activity signals, starting in ~1720. (This action was in addition to flipping the X-Ray Density interpretation from the upside-down Mann orientation to that suggested by Tiljander.)
So my question is how you weight these no-post-1800-data proxies in calculating the average, equally-weighted signal. Is it simply a matter of normalizing each proxy such that the Y-axis is SDs, and then summing for each year or decade for which the proxy is available?
IIRC, Steve McIntyre once noted in passing that Tiljander's data record contains an anomolous set of readings for one varve, for 1326 I believe. At NOAA's FTP site for Tiljander et al (2003), XRD is as follows:
1324 - 94.3
1325 - 84.3
1326 - 172.6
1327 - 120.0
1328 - 100.7
Values for Lightsum, Darksum, and Thickness (used by Mann et al 2008 but not by Kaufman et al 2009) for 1326 are much farther from the neighboring years. Mann et al might have adjusted 1326 values (I'm traveling and can't check the zipped files here). I'll follow up later.
A work-in-progress compilation of Tiljander-related resources is here.
AMac: You asked, "So my question is how you weight these no-post-1800-data proxies in calculating the average, equally-weighted signal. Is it simply a matter of normalizing each proxy such that the Y-axis is SDs, and then summing for each year or decade for which the proxy is available?"
If a cell has no data, EXCEL excludes it in its averages.
Thanks for your (as usual) beautiful graphs:
The average give a general idea for the periferic Arctic Region(periferic because none of this data comes from the Arctic Ocean itself, but from the nearby land), but to have a complete undestanding of then trends one must descend to the regional-to-local level.
A few questions and suggestions for a second part on this subject:
1)Do you have any idea why the scale is so different in your avarages (several°C) than in the press release(less than 1°C)?
2)This one of the best paleoclimate studies publicated so far for the Arctic, as it is a synthesis of 23 multi-proxy studies.
It will be great to average the proxies from the same region(Alaska, Canadian Archipielago, Greenland, Scandinavia, Siberia).
3)Then compare the proxy data with the GISSTEMP station data.
A FINAL REMARK: as one improve the resolution and "zoom" to regional-to-local level, weather and climate variability greatly increases.
So isn't surprising that there are notable variations between the different places. Some warm and other cool at the same time, so while the improved resolution permits a better analysis, it may be misleading if one forgets the Big Picture(all-Arctic): some places may have been warmer in the past than now, but this is expected as the inter-regional variation increases together with the spatial resolution (for example between Alaska and Siberia).
One of the characteristics of contemporary Global Warming, specially since the 1980s, is its WIDESPREAD nature, with big warm anomalies covering entire Continents or Ocean Basins(in this case, the whole Arctic) for entire seasons of the year (in the Arctic, specially in Fall and Winter).
This greatly contrast with previous climate anomalies, when some areas warmed and other cooled.
Anonymous 12:36PM: You wrote, “A few questions and suggestions for a second part on this subject.”
I have no plans for a second part. I really have a very limited interest in paleoclimatological studies.
You asked, “Do you have any idea why the scale is so different in your avarages (several°C) than in the press release(less than 1°C)?”
The data from the NOAA Paleoclimatological Program website is standardized. The graph from the BBC is not.
You wrote, “This one of the best paleoclimate studies publicated so far for the Arctic, as it is a synthesis of 23 multi-proxy studies.”
Personally, to the skeptical side of me, it appears to be a grouping of proxies that were hand selected to provide a desired outcome, which was the graph of the long-term decline in temperatures, followed by a steep rise in recent decades. My opinion in based on the fact that there are many other long-term proxies available but they were not included in this study.
You wrote, “It will be great to average the proxies from the same region(Alaska, Canadian Archipielago, Greenland, Scandinavia, Siberia).”
The link to the spreadsheet is here (or above):
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/reconstructions/arctic/kaufman2009arctic.xls
Regards
Bob
Are you fricking kidding me!!
Surely, Kaufman did not just average the proxies together to get the BBC graph?
I just removed the Yamal Series and averaged. The early part of the chart almost equals the latter (late 20th century) part. ie what's the concern.
It can't be that simple.
Anonymous: You asked, “Are you fricking kidding me!!
Surely, Kaufman did not just average the proxies together to get the BBC graph?”
Again, my opinion in based on the fact that there are many other long-term proxies available but they were not included in this study. Lindqvist also published a collection of two-millennia proxies in 2009.
http://www3.interscience.wiley.com/cgi-bin/fulltext/122225084/PDFSTART
Looking at Lindqvist’s Table 1, there are 37 Proxies North of 60N that cover the vast majority of the two-millennia time span. Why didn’t Kaufman include the others? And if Kaufman et al was a study of two millennia, why didn’t they limit the proxies to that period? In other words, many of the Kaufman proxies do not begin until mid-to-late in the first millennium. Why are they included?
Again, Kaufman appears to include a grouping of proxies that were hand selected to provide a desired outcome, which was the graph of the long-term decline in temperatures, followed by a steep rise in recent decades.
The Linqvist data is here:
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/contributions_by_author/ljungqvist2009/ljungqvist2009recons.txt
Are these Briffa series the ones which have actual temperatures spliced on at 1960? If so, you can lop off the post-1960 Briffa data -- indeed, you would need to.
NZ Willy: Correct me if I'm wrong, but wasn't the Briffa reconstruction you're referring to a composite of multiple Northern Hemisphere proxies? The three Briffa datasets shown above are individuals. And two of the three Briffa datasets included do show declines after 1945.
Anonymous: "I just removed the Yamal Series and averaged. The early part of the chart almost equals the latter (late 20th century) part. ie what's the concern."
I suppose if you just say all we have done is come back to where we were 2000 years ago, you could say that. But the fact that we essentially recouped the temperature drop that occurred over the past 1900 years in a short 100 year period should be raise some sort of flag shouldn't it?
It not the fact that we may have raised temperatures back to where they were in the distant past, but the rapid rate in which the temperatures have increased is of concern.
I also suggest plotting out the same 23 proxy data (not just the averages in column Y) by using a running 100-year average period for all the proxy data and see what the graph looks like, starting with the year 95.
The noted rapid rate of increase in proxy temperatures during the past 100 years should be fairly obvious.
Dennis H.
Post a Comment